1 2018-10-29 Martin Liska <mliska@suse.cz>
3 * Makefile.in: Make dependency to json.o.
4 * doc/gcov.texi: Document new JSON format, remove
5 old intermediate format documentation.
6 * gcov.c (struct function_info): Come up with m_name and
8 (function_info::function_info): Initialize it.
9 (function_info::~function_info): Release it.
10 (main): Rename flag_intermediate_format to flag_json_format.
11 (print_usage): Describe --json-format.
12 (process_args): Set flag_json_format.
13 (output_intermediate_line): Remove.
14 (output_intermediate_json_line): Likewise.
15 (get_gcov_intermediate_filename): Return new extension
17 (output_intermediate_file): Implement JSON emission.
18 (output_json_intermediate_file): Implement JSON emission.
19 (generate_results): Use ::get_name for function name.
20 Handle JSON output file.
21 (read_graph_file): Use ::get_name instead of cplus_demangle.
22 (read_count_file): Likewise.
23 (solve_flow_graph): Likewise.
24 (add_line_counts): Likewise.
25 (accumulate_line_counts): Use new flag_json_format.
26 (output_function_details): Use ::get_name instead of cplus_demangle.
27 (output_lines): Likewise.
28 * json.cc (test_writing_literals): Add new tests.
29 * json.h (class literal): Add new boolean constructor.
31 2018-10-29 Segher Boessenkool <segher@kernel.crashing.org>
33 PR rtl-optimization/87701
34 PR rtl-optimization/87780
35 * combine.c (make_more_copies): Rewrite.
37 2018-10-28 Kugan Vivekanandarajah <kuganv@linaro.org>
39 * doc/generic.texi (ABSU_EXPR): Document.
40 * match.pd (absu(x)*absu(x) -> x*x): Handle.
41 (absu(absu(X)) -> absu(X)): Likewise.
42 (absu(-X) -> absu(X)): Likewise.
43 (absu(X) where X is nonnegative -> X): Likewise.
45 2018-10-28 Iain Buclaw <ibuclaw@gdcproject.org>
47 * Makefile.in (tm_d_file_list, tm_d_include_list): New variables.
48 (TM_D_H, D_TARGET_DEF, D_TARGET_H, D_TARGET_OBJS): New variables.
49 (tm_d.h, cs-tm_d.h, default-d.o): New rules.
50 (d/d-target-hooks-def.h, s-d-target-hooks-def-h): New rules.
51 (s-tm-texi): Also check timestamp on d-target.def.
52 (generated_files): Add TM_D_H and d-target-hooks-def.h.
53 (build/genhooks.o): Also depend on D_TARGET_DEF.
54 * config.gcc (tm_d_file, d_target_objs, target_has_targetdm): New
56 * config/aarch64/aarch64-d.c: New file.
57 * config/aarch64/aarch64-linux.h (GNU_USER_TARGET_D_CRITSEC_SIZE):
59 * config/aarch64/aarch64-protos.h (aarch64_d_target_versions): New
61 * config/aarch64/aarch64.h (TARGET_D_CPU_VERSIONS): Define.
62 * config/aarch64/t-aarch64 (aarch64-d.o): New rule.
63 * config/arm/arm-d.c: New file.
64 * config/arm/arm-protos.h (arm_d_target_versions): New prototype.
65 * config/arm/arm.h (TARGET_D_CPU_VERSIONS): Define.
66 * config/arm/linux-eabi.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
67 * config/arm/t-arm (arm-d.o): New rule.
68 * config/default-d.c: New file.
69 * config/glibc-d.c: New file.
70 * config/gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
71 * config/i386/i386-d.c: New file.
72 * config/i386/i386-protos.h (ix86_d_target_versions): New prototype.
73 * config/i386/i386.h (TARGET_D_CPU_VERSIONS): Define.
74 * config/i386/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
75 (GNU_USER_TARGET_D_CRITSEC_SIZE): Define.
76 * config/i386/t-i386 (i386-d.o): New rule.
77 * config/kfreebsd-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
78 * config/kopensolaris-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
79 * config/linux-android.h (ANDROID_TARGET_D_OS_VERSIONS): Define.
80 * config/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
81 * config/mips/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
82 * config/mips/mips-d.c: New file.
83 * config/mips/mips-protos.h (mips_d_target_versions): New prototype.
84 * config/mips/mips.h (TARGET_D_CPU_VERSIONS): Define.
85 * config/mips/t-mips (mips-d.o): New rule.
86 * config/powerpcspe/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
87 * config/powerpcspe/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
88 * config/powerpcspe/powerpcspe-d.c: New file.
89 * config/powerpcspe/powerpcspe-protos.h (rs6000_d_target_versions):
91 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
92 Support GNU D by using 0 as the language type.
93 * config/powerpcspe/powerpcspe.h (TARGET_D_CPU_VERSIONS): Define.
94 * config/powerpcspe/t-powerpcspe (powerpcspe-d.o): New rule.
95 * config/riscv/riscv-d.c: New file.
96 * config/riscv/riscv-protos.h (riscv_d_target_versions): New
98 * config/riscv/riscv.h (TARGET_D_CPU_VERSIONS): Define.
99 * config/riscv/t-riscv (riscv-d.o): New rule.
100 * config/rs6000/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
101 * config/rs6000/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
102 * config/rs6000/rs6000-d.c: New file.
103 * config/rs6000/rs6000-protos.h (rs6000_d_target_versions): New
105 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
106 Support GNU D by using 0 as the language type.
107 * config/rs6000/rs6000.h (TARGET_D_CPU_VERSIONS): Define.
108 * config/rs6000/t-rs6000 (rs6000-d.o): New rule.
109 * config/s390/s390-d.c: New file.
110 * config/s390/s390-protos.h (s390_d_target_versions): New prototype.
111 * config/s390/s390.h (TARGET_D_CPU_VERSIONS): Define.
112 * config/s390/t-s390 (s390-d.o): New rule.
113 * config/sparc/sparc-d.c: New file.
114 * config/sparc/sparc-protos.h (sparc_d_target_versions): New
116 * config/sparc/sparc.h (TARGET_D_CPU_VERSIONS): Define.
117 * config/sparc/t-sparc (sparc-d.o): New rule.
118 * config/t-glibc (glibc-d.o): New rule.
119 * configure: Regenerated.
120 * configure.ac (tm_d_file): New variable.
121 (tm_d_file_list, tm_d_include_list, d_target_objs): Add substitutes.
122 * doc/contrib.texi (Contributors): Add self for the D frontend.
123 * doc/frontends.texi (G++ and GCC): Mention D as a supported language.
124 * doc/install.texi (Configuration): Mention libphobos as an option for
125 --enable-shared. Mention d as an option for --enable-languages.
126 (Testing): Mention check-d as a target.
127 * doc/invoke.texi (Overall Options): Mention .d, .dd, and .di as file
128 name suffixes. Mention d as a -x option.
129 * doc/sourcebuild.texi (Top Level): Mention libphobos.
130 * doc/standards.texi (Standards): Add section on D language.
131 * doc/tm.texi: Regenerated.
132 * doc/tm.texi.in: Add @node for D language and ABI, and @hook for
133 TARGET_CPU_VERSIONS, TARGET_D_OS_VERSIONS, and TARGET_D_CRITSEC_SIZE.
134 * dwarf2out.c (is_dlang): New function.
135 (gen_compile_unit_die): Use DW_LANG_D for D.
136 (declare_in_namespace): Return module die for D, instead of adding
137 extra declarations into the namespace.
138 (gen_namespace_die): Generate DW_TAG_module for D.
139 (gen_decl_die): Handle CONST_DECLSs for D.
140 (dwarf2out_decl): Likewise.
141 (prune_unused_types_walk_local_classes): Handle DW_tag_interface_type.
142 (prune_unused_types_walk): Handle DW_tag_interface_type same as other
144 * gcc.c (default_compilers): Add entries for .d, .dd and .di.
145 * genhooks.c: Include d/d-target.def.
147 2018-10-28 Iain Sandoe <iain@sandoe.co.uk>
150 * config/rs6000/darwin.h (STACK_BOUNDARY): New.
151 (RS6000_STARTING_FRAME_OFFSET): Adjust to preserve 16byte alignment.
152 (STACK_DYNAMIC_OFFSET): Likewise.
154 2018-10-27 Sandra Loosemore <sandra@codesourcery.com>
157 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Fix
160 2018-10-26 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
162 * doc/extend.texi (PowerPC builtins): Fix __builtin_unpack_ibm128
163 return type and other typos.
165 2018-10-26 Aaron Sawdey <acsawdey@linux.ibm.com>
167 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Change to
168 a shorter sequence with fewer branches.
169 (emit_final_str_compare_gpr): Ditto.
171 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
173 * config/rs6000/tmmintrin.h: New file.
174 * config.gcc (powerpc*-*-*): Add tmmintrin.h to extra_headers.
176 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
178 * config/rs6000/mmintrin.h: Enable 32bit compilation.
179 * config/rs6000/xmmintrin.h: Likewise.
181 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
183 * config/rs6000/xmmintrin.h (_mm_extract_pi16): Fix for big-endian.
185 2018-10-26 Richard Biener <rguenther@suse.de>
187 * tree-vect-slp.c (vect_mark_slp_stmts): Add visited hash_set
189 (vect_mark_slp_stmts_relevant): Likewise.
190 (vect_detect_hybrid_slp_stmts): Likewise.
191 (vect_bb_slp_scalar_cost): Likewise.
192 (vect_remove_slp_scalar_calls): Likewise.
194 2018-10-26 Jan Hubicka <jh@suse.cz>
196 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix recursion.
197 (warn_types_mismatch): Fix walk of DECL_NAME.
198 (odr_types_equivalent_p): Fix overactive assert.
200 2018-10-26 Richard Biener <rguenther@suse.de>
202 PR tree-optimization/87105
203 * tree-vectorizer.h (_slp_tree::refcnt): New member.
204 * tree-vect-slp.c (vect_free_slp_tree): Decrement and honor
206 (vect_create_new_slp_node): Initialize refcnt to one.
208 (scalar_stmts_set_t, bst_fail): Remove.
209 (vect_build_slp_tree_2): Add bst_map argument and adjust calls.
210 (vect_build_slp_tree): Add bst_map argument and lookup
211 already created SLP nodes.
212 (vect_print_slp_tree): Handle a SLP graph, print SLP node
214 (vect_slp_rearrange_stmts): Handle a SLP graph.
215 (vect_analyze_slp_instance): Adjust and free SLP nodes from
216 the CSE map. Fix indenting.
217 (vect_schedule_slp_instance): Add short-cut.
219 2018-10-26 Martin Liska <mliska@suse.cz>
222 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Hash with
223 addr_expr and not with pointers.
225 2018-10-26 Jan Hubicka <jh@suse.cz>
227 * tree.c (free_lang_data_in_type): Only check main variants.
228 * ipa-devirt.c (warn_odr): Make static.
229 (types_same_for_odr): Drop strict variant.
230 (types_odr_comparable): Likewise.
231 (odr_or_derived_type_p): Look for main variants.
232 (odr_name_hasher::equal): Cleanup comment.
233 (odr_subtypes_equivalent): Add warn and warned arguments; check main
235 (type_variants_equivalent_p): break out from ...
236 (odr_types_equivalent): ... here; go for main variants where needed.
237 (warn_odr): ... here; turn static.
238 (warn_types_mismatch): Compare mangled names of main variants.
239 * ipa-utils.h (types_odr_comparable): Drop strict parameter.
240 (type_with_linkage_p): Sanity check that we look at main variant.
241 * lto.c (lto_read_decls): Only consider main variant to be ODR type.
242 * tree.h (types_same_for_odr): Drop strict argument.
244 2018-10-26 Richard Biener <rguenther@suse.de>
246 PR tree-optimization/87746
247 * tree-vect-data-refs.c (vect_update_misalignment_for_peel):
248 Simplify and fix WRT strided store groups with size not
249 equal to step in element count.
250 (vect_analyze_group_access_1): Dump the whole group.
252 2018-10-25 Carl Love <cel@us.ibm.com>
254 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_VSCEDPGT,
255 P9V_BUILTIN_VEC_VSCEDPLT, P9V_BUILTIN_VEC_VSCEDPEQ,
256 P9V_BUILTIN_VEC_VSCEDPUO): Rename base overloaded name. Add quad
257 precicion entry for each overloaded builtin.
258 * config/rs6000/rs6000-builtin.def (VSCEDPGT, VSCEDPLT, VSCEDPEQ,
259 VSCEDPUO): Rename overloaded name.
260 (VSCEDPGT, VSCEQPGT, VSCEDPLT, VSCEQPLT, VSCEDPEQ, VSCEQPEQ,
261 VSCEDPUO, VSCEQPUO): Add defitions for overloaded builtins.
262 * config/rs6000/vsx.md (xscmpexpqp_<code>_<mode>): Add
263 define_expand for xscmpexqp instruction.
264 (*xscmpexpqp): Add define_insn for the xscmpexqp instruction.
266 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
267 Jinsong Ji <jji@us.ibm.com>
269 * config/rs6000/emmintrin.h (_mm_slli_epi16): Replace deprecated
270 function with vec_sl.
271 (_mm_slli_epi32): Likewise.
272 (_mm_slli_epi64): Likewise.
273 (_mm_srai_epi16): Replace deprecated function with vec_sra.
274 (_mm_srai_epi32): Likewise.
275 (_mm_srli_epi16): Replace deprecated function with vec_sr.
276 (_mm_srli_epi32): Likewise.
277 (_mm_srli_epi64): Likewise.
278 (_mm_sll_epi16): Replace deprecated function with vec_sl.
279 (_mm_sll_epi32): Likewise.
280 (_mm_sll_epi64): Likewise.
281 (_mm_sra_epi16): Replace deprecated function with vec_sra.
282 (_mm_sra_epi32): Likewise.
283 (_mm_srl_epi16): Replace deprecated function with vec_sr.
284 (_mm_srl_epi32): Likewise.
285 (_mm_srl_epi64): Likewise.
287 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
288 Jinsong Ji <jji@us.ibm.com>
290 * gcc/config/rs6000/emmintrin.h (_mm_sll_epi16): Replace
291 comparison operators with vec_cmp* for compatibility due to
292 unfortunate history; clean up formatting and use types more
294 (_mm_sll_epi32): Likewise.
295 (_mm_sll_epi64): Likewise.
296 (_mm_srl_epi16): Likewise.
297 (_mm_srl_epi32): Likewise.
298 (_mm_srl_epi64): Likewise.
300 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
301 Jinsong Ji <jji@us.ibm.com>
303 * config/rs6000/emmintrin.h (_mm_sll_epi64): Remove wrong cast.
304 * config/rs6000/xmmintrin.h (_mm_min_ps): Change m's type to
305 __vector __bool int. Use vec_cmpgt in preference to deprecated
306 function vec_vcmpgtfp.
307 (_mm_max_ps): Likewise.
309 2018-10-25 Jeff Law <law@redhat.com>
311 * config/rl78/rl78.c (insn_ok_now): Always re-recognize the insn
314 2018-10-25 Martin Sebor <msebor@redhat.com>
316 * doc/extend.texi (aligned): Expand attribute description.
317 (Alignment): Rename section. Discuss function arguments.
319 2018-10-25 Jan Hubicka <jh@suse.cz>
321 * ipa-devirt.c (main_odr_variant): Remove.
322 (hash_odr_name, types_same_for_odr, types_odr_comparable,
323 odr_name_hasher::equal, odr_subtypes_equivalent_p):
324 Drop use of main_odr_variant.
325 (add_type_duplicate): Silence confused warnings on integer types.
326 (get_odr_type): Always look for main variant.
327 (register_odr_type): Simplify.
329 2018-10-25 Richard Biener <rguenther@suse.de>
331 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses):
332 Initialize ng to silence error with release checking bootstrap.
334 2018-10-25 Richard Biener <rguenther@suse.de>
336 * tree-if-conv.c: Include tree-ssa-sccvn.h.
337 (tree_if_conversion): Run CSE on the if-converted loop body.
339 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
341 * config/s390/constraints.md (ZL): New constraint.
342 * config/s390/s390.c (legitimate_pic_operand_p): Accept LARL
344 * config/s390/s390.md (movdi_larl): Remove.
345 (movdi_64): Add the LARL alternative.
347 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
350 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
351 (RTX_CODE_HWINT_P): New macro.
352 (rtx_code_size): Use RTX_CODE_HWINT_P ().
354 2018-10-25 Jan Hubicka <jh@suse.cz>
356 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
359 2018-10-25 Richard Biener <rguenther@suse.de>
361 PR tree-optimization/87665
362 PR tree-optimization/87745
363 * tree-vectorizer.h (get_earlier_stmt): Remove.
364 (get_later_stmt): Pick up UID from the original non-pattern stmt.
366 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
368 * options.texi (Deprecated): Move list to Var section.
370 2018-10-24 Bill Schmidt <wschmidt@linux.ibm.com>
371 Jinsong Ji <jji@us.ibm.com>
373 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Change deprecated
374 __vector long to __vector long long.
375 (_mm_cvtpd_ps): Likewise.
376 (_mm_cvttpd_epi32): Likewise.
377 (_mm_cvtpi32_pd): Likewise.
378 (_mm_unpackhi_epi64): Likewise.
379 (_mm_unpacklo_epi64): Likewise.
381 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
383 PR rtl-optimization/87720
384 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
386 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
388 * gimple-ssa-isolate-paths.c
389 (find_implicit_erroneous_behavior): Do not change code if the
390 pass is running for warnings only.
391 (find_explicit_erroneous_behavior): Likewise.
393 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
395 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
396 Define as rs6000_mangle_decl_assembler_name.
397 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
398 long double to IEEE long double, switch the names of the long
399 double built-in functions to be <func>f128 instead of <func>l.
401 2018-10-24 Martin Sebor <msebor@redhat.com>
403 * doc/extend.texi (nonnull): List no-argument form. Reference
404 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
406 2018-10-24 Richard Biener <rguenther@suse.de>
408 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
410 2018-10-24 Martin Liska <mliska@suse.cz>
412 PR tree-optimization/84436
413 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
415 (switch_conversion::contains_linear_function_p): New.
416 (switch_conversion::build_one_array): Support linear
417 transformation on input.
418 * tree-switch-conversion.h (struct switch_conversion): Add
419 contains_linear_function_p declaration.
421 2018-10-24 Richard Biener <rguenther@suse.de>
423 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
424 if its argument is CONSTANT_CLASS_P.
426 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
428 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
429 it is wrong for forward declarations.
431 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
433 * config/s390/s390.c (s390_check_qrst_address): Add the missing
434 SYMBOL_REF_P () check.
436 2018-10-24 Richard Biener <rguenther@suse.de>
438 PR tree-optimization/87105
439 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
441 (vect_analyze_data_ref_accesses): Handle duplicate loads and
442 stores by splitting the affected group after the fact.
443 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
444 fail the SLP build because of size constraints.
446 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
448 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
449 * configure: Regenerate.
450 * config.in: Regenerate.
451 * varasm.c (mergeable_string_section): Use readonly_data_section
452 if linker doesn't support SHF_MERGE with alignment > 8.
453 (mergeable_constant_section): Likewise.
455 2018-10-24 Richard Biener <rguenther@suse.de>
457 PR tree-optimization/84013
458 * tree-ssa-structalias.c (struct msdi_data): New struct for
459 marshalling data to walk_stmt_load_store_ops.
460 (maybe_set_dependence_info): Refactor as callback for
461 walk_stmt_load_store_ops.
462 (compute_dependence_clique): Set restrict info on all stmt kinds.
464 2018-10-24 Martin Liska <mliska@suse.cz>
466 * cgraph.c (cgraph_node::dump):
467 Remove reduntant dumps and make tp_first_run dump more compact.
469 2018-10-24 Richard Biener <rguenther@suse.de>
471 PR tree-optimization/87665
472 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
475 2018-10-12 Jeff Law <law@redhat.com>
477 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
480 2018-10-23 Richard Biener <rguenther@suse.de>
482 * tree-vrp.c (add_assert_info): Guard dump_printf with
484 * gimple-ssa-evrp-analyze.c
485 (evrp_range_analyzer::record_ranges_from_incoming_edge):
486 Use value_range::ignore_equivs_equal_p.
488 2018-10-23 Richard Biener <rguenther@suse.de>
490 PR tree-optimization/87105
491 PR tree-optimization/87608
492 * passes.def (pass_all_early_optimizations): Add early phi-opt
494 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
495 addition to debug stmts.
496 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
497 and abs replacement early.
498 * tree-cfg.c (gimple_empty_block_p): Likewise.
500 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
503 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
504 specified to configure.
505 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
507 2018-10-23 Richard Biener <rguenther@suse.de>
509 PR tree-optimization/87700
510 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
512 2018-10-23 Jakub Jelinek <jakub@redhat.com>
515 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
516 second argument from __mmask16 to __mmask8.
517 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
518 _mm_mask_packs_epi32): Likewise.
519 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
521 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
523 2018-10-23 Richard Biener <rguenther@suse.de>
525 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
527 2018-10-23 Richard Biener <rguenther@suse.de>
529 PR tree-optimization/86144
530 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
533 2018-10-23 Richard Biener <rguenther@suse.de>
535 PR tree-optimization/87693
536 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
537 the case we do not find the taken edge.
539 2018-10-22 Bill Schmidt <wschmidt@linux.ibm.com>
540 Jinsong Ji <jji@us.ibm.com>
542 * config/rs6000/emmintrin.h (_MM_SHUFFLE2): Comment cleanup.
543 (_mm_store_pd): Use unaligned vector type for pointer cast.
544 (_mm_maskmoveu_si128): Likewise.
545 * config/rs6000/xmmintrin.h (__m128_u): New typedef.
546 (_mm_store_ps): Use unaligned vector type for pointer cast.
548 2018-10-22 Paul Koning <ni1d@arrl.net>
550 * symtab.c (symtab_node::increase_alignment): Correct max
553 2018-10-22 Yury Gribov <tetra2005@gmail.com>
555 PR tree-optimization/87633
556 * match.pd: Do not generate unordered integer comparisons.
558 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
560 PR rtl-optimization/87600
561 * combine.c: Add include of expr.h.
562 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
563 register to a pseudo.
564 (make_more_copies): New function, add a copy to a new pseudo after
565 the moves from hard registers into pseudos.
566 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
567 later. Call make_more_copies.
569 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
571 * lra-constraints.c (process_alt_operands): New local array,
572 matching_early_clobber. Check matching_early_clobber before
573 decrementing reject, and set matching_early_clobber after.
575 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
578 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
579 call output_addr_const and hope for the best.
581 2018-10-22 Richard Biener <rguenther@suse.de>
583 * gimple-ssa-evrp-analyze.c
584 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
585 smarter about what ranges to use.
586 * tree-vrp.c (add_assert_info): Dump here.
587 (register_edge_assert_for_2): Instead of here at multiple but
590 * gcc.dg/tree-ssa/evrp12.c: New testcase.
591 * gcc.dg/predict-6.c: Adjust.
592 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
593 * gcc.dg/tree-ssa/vrp02.c: Likewise.
594 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
596 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
597 Richard Biener <rguenther@suse.de>
599 * bitmap.h: Update data structure documentation, including a
600 description of bitmap views as either linked-lists or splay trees.
601 (struct bitmap_element_def): Update comments for splay tree bitmaps.
602 (struct bitmap_head_def): Likewise.
603 (bitmap_list_view, bitmap_tree_view): New prototypes.
604 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
606 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
607 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
608 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
609 released bitmap element here.
610 (bitmap_element_free): Remove.
611 (bitmap_elt_clear_from): Work on splay tree bitmaps.
612 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
613 this function similar ones such that linked-list bitmap implementation
614 functions are grouped.
615 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
616 and moved for grouping.
617 (bitmap_list_insert_element_after): Renamed from
618 bitmap_elt_insert_after, and moved for grouping.
619 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
620 (bitmap_tree_link_left, bitmap_tree_link_right,
621 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
622 bitmap_tree_link_element, bitmap_tree_unlink_element,
623 bitmap_tree_find_element): New functions for splay-tree bitmap
625 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
626 Renamed and moved, see above entries.
627 (bitmap_tree_listify_from): New function to convert part of a splay
628 tree bitmap to a linked-list bitmap.
629 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
630 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
631 (bitmap_find_bit): Remove.
632 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
633 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
634 Handle splay tree bitmaps.
635 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
636 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
637 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
638 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
639 bitmap_intersect_compl_p, bitmap_ior_and_compl,
640 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
641 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
642 corresponding changes to use linked-list specific bitmap_element
643 manipulation functions as applicable for efficiency.
644 (bitmap_tree_to_vec): New function.
645 (debug_bitmap_elt_file): New function split out from ...
646 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
647 (bitmap_print): Likewise.
649 PR tree-optimization/63155
650 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
652 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
655 2018-10-22 Martin Liska <mliska@suse.cz>
657 PR tree-optimization/87686
659 2018-08-29 Martin Liska <mliska@suse.cz>
661 * tree-switch-conversion.c (switch_conversion::expand):
662 Strenghten assumption about gswitch statements.
664 2018-10-22 Martin Liska <mliska@suse.cz>
666 * ipa-icf.c (sem_item::compare_attributes): Remove.
667 (sem_item::compare_referenced_symbol_properties): Use
668 attribute_list_equal instead.
669 (sem_function::equals_wpa): Likewise.
670 * ipa-icf.h: Remove compare_attributes.
672 2018-10-22 Richard Biener <rguenther@suse.de>
675 * mem-stats.h (mem_usage::operator==): Fix pasto.
677 2018-10-22 Richard Biener <rguenther@suse.de>
679 PR tree-optimization/87640
680 * tree-vrp.c (set_value_range_with_overflow): Decompose
682 (extract_range_from_binary_expr_1): Adjust.
684 2018-10-22 Martin Jambor <mjambor@suse.cz>
686 * tree-eh.h (stmt_could_throw_p): Add function parameter.
687 (stmt_can_throw_external): Likewise.
688 (stmt_can_throw_internal): Likewise.
689 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
690 (lower_eh_constructs_2): Likewise.
691 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
692 (stmt_can_throw_external): Likewise.
693 (stmt_can_throw_internal): Likewise.
694 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
695 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
696 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
697 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
698 (pass_lower_eh_dispatch::execute): Pass cfun to
699 stmt_can_throw_external.
700 (cleanup_empty_eh): Likewise.
701 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
702 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
703 stmt_can_throw_external instead of pushing it to cfun.
704 (symbol_table::create_edge): Likewise.
705 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
706 stmt_can_throw_internal.
707 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
708 to stmt_could_throw_p.
709 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
710 stmt_can_throw_internal.
711 (pass_store_merging::execute): Likewise.
712 * gimple-ssa-strength-reduction.c
713 (find_candidates_dom_walker::before_dom_children): Pass cfun to
715 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
716 stmt_can_throw_internal.
717 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
718 to stmt_can_throw_external.
719 (check_stmt): Pass cfun to stmt_could_throw_p.
720 (check_stmt): Pass cfun to stmt_can_throw_external.
721 (pass_nothrow::execute): Likewise.
722 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
723 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
724 stmt_can_throw_internal.
725 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
726 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
727 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
728 * tree-complex.c (expand_complex_libcall): Pass cfun to
729 stmt_could_throw_p and to stmt_can_throw_internal.
730 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
731 * tree-inline.c (copy_edges_for_bb): Likewise.
732 (maybe_move_debug_stmts_to_successors): Likewise.
733 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
735 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
736 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
737 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
738 stmt_can_throw_internal.
739 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
740 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
742 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
743 stmt_can_throw_internal.
744 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
745 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
747 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
748 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
749 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
750 (convert_mult_to_fma_1): Likewise.
751 (convert_to_divmod): Likewise.
752 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
753 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
754 * tree-ssa-propagate.c
755 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
756 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
757 (maybe_optimize_range_tests): Likewise.
758 (linearize_expr_tree): Likewise.
759 (reassociate_bb): Likewise.
760 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
761 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
762 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
763 (handle_char_store): Likewise.
764 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
765 stmt_can_throw_internal.
766 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
768 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
769 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
770 (vectorizable_simd_clone_call): Likewise.
771 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
772 (gimple_stringop_fixed_value): Likewise.
774 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
776 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
777 literal pool references.
778 (s390_check_qrst_address): Adapt to the new behavior of
779 s390_loadrelative_operand_p ().
781 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
784 * config/i386/sse.md (*andnot<mode>3_bcst): New.
786 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
789 * config/i386/sse.md (*<code><mode>3_bcst): New.
791 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
794 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
795 V4DI, V16SI and V8DI.
796 (*sub<mode>3<mask_name>_bcst): New.
797 (*add<mode>3<mask_name>_bcst): Likewise.
799 2018-10-21 Bill Schmidt <wschmidt@linux.ibm.com>
800 Jinsong Ji <jji@us.ibm.com>
802 * config/rs6000/emmintrin.h (_mm_movemask_pd): Replace __vector
803 __m64 with __vector unsigned long long for compatibility.
804 (_mm_movemask_epi8): Likewise.
805 * config/rs6000/xmmintrin.h (_mm_cvtps_pi32): Likewise.
806 (_mm_cvttps_pi32): Likewise.
807 (_mm_cvtpi32_ps): Likewise.
808 (_mm_cvtps_pi16): Likewise.
809 (_mm_loadh_pi): Likewise.
810 (_mm_storeh_pi): Likewise.
811 (_mm_movehl_ps): Likewise.
812 (_mm_movelh_ps): Likewise.
813 (_mm_loadl_pi): Likewise.
814 (_mm_storel_pi): Likewise.
815 (_mm_movemask_ps): Likewise.
816 (_mm_shuffle_pi16): Likewise.
818 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
821 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
822 __builtin_ia32_vfnmsubpd512_mask.
823 (_mm512_mask_fnmsub_round_pd): Likewise.
824 (_mm512_fnmsub_pd): Likewise.
825 (_mm512_mask_fnmsub_pd): Likewise.
826 (_mm512_maskz_fnmsub_round_pd): Use
827 __builtin_ia32_vfnmsubpd512_maskz.
828 (_mm512_maskz_fnmsub_pd): Likewise.
829 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
830 (_mm512_mask_fnmsub_round_ps): Likewise.
831 (_mm512_fnmsub_ps): Likewise.
832 (_mm512_mask_fnmsub_ps): Likewise.
833 (_mm512_maskz_fnmsub_round_ps): Use
834 __builtin_ia32_vfnmsubps512_maskz.
835 (_mm512_maskz_fnmsub_ps): Likewise.
836 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
837 __builtin_ia32_vfnmsubpd256_mask.
838 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
839 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
840 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
841 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
842 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
843 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
844 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
845 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
846 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
847 __builtin_ia32_vfnmsubpd.
848 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
849 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
850 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
851 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
852 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
853 * config/i386/i386-builtin.def: Add
854 __builtin_ia32_vfnmsubpd256_mask,
855 __builtin_ia32_vfnmsubpd256_maskz,
856 __builtin_ia32_vfnmsubpd128_mask,
857 __builtin_ia32_vfnmsubpd128_maskz,
858 __builtin_ia32_vfnmsubps256_mask,
859 __builtin_ia32_vfnmsubps256_maskz,
860 __builtin_ia32_vfnmsubps128_mask,
861 __builtin_ia32_vfnmsubps128_maskz,
862 __builtin_ia32_vfnmsubpd512_mask,
863 __builtin_ia32_vfnmsubpd512_maskz,
864 __builtin_ia32_vfnmsubps512_mask,
865 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
866 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
867 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
868 __builtin_ia32_vfnmsubpd256.
869 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
870 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
871 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
873 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
875 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
877 (fmai_vmfnmsub_<mode><round_name>): Likewise.
879 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
882 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
883 __builtin_ia32_vfnmaddpd512_mask.
884 (_mm512_mask_fnmadd_round_pd): Likewise.
885 (_mm512_fnmadd_pd): Likewise.
886 (_mm512_mask_fnmadd_pd): Likewise.
887 (_mm512_maskz_fnmadd_round_pd): Use
888 __builtin_ia32_vfnmaddpd512_maskz.
889 (_mm512_maskz_fnmadd_pd): Likewise.
890 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
891 (_mm512_mask_fnmadd_round_ps): Likewise.
892 (_mm512_fnmadd_ps): Likewise.
893 (_mm512_mask_fnmadd_ps): Likewise.
894 (_mm512_maskz_fnmadd_round_ps): Use
895 __builtin_ia32_vfnmaddps512_maskz.
896 (_mm512_maskz_fnmadd_ps): Likewise.
897 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
898 __builtin_ia32_vfnmaddpd256_mask.
899 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
900 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
901 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
902 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
903 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
904 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
905 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
906 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
907 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
908 __builtin_ia32_vfnmaddpd.
909 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
910 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
911 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
912 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
913 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
914 * config/i386/i386-builtin.def: Add
915 __builtin_ia32_vfnmaddpd256_mask,
916 __builtin_ia32_vfnmaddpd256_maskz,
917 __builtin_ia32_vfnmaddpd128_mask,
918 __builtin_ia32_vfnmaddpd128_maskz,
919 __builtin_ia32_vfnmaddps256_mask,
920 __builtin_ia32_vfnmaddps256_maskz,
921 __builtin_ia32_vfnmaddps128_mask,
922 __builtin_ia32_vfnmaddps128_maskz,
923 __builtin_ia32_vfnmaddpd512_mask,
924 __builtin_ia32_vfnmaddpd512_maskz,
925 __builtin_ia32_vfnmaddps512_mask,
926 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
927 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
928 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
929 __builtin_ia32_vfnmaddpd256.
930 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
931 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
932 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
934 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
936 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
938 (fmai_vmfnmadd_<mode><round_name>): Likewise.
940 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
943 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
944 __builtin_ia32_vfmsubpd512_mask.
945 (_mm512_mask_fmsub_round_pd): Likewise.
946 (_mm512_fmsub_pd): Likewise.
947 (_mm512_mask_fmsub_pd): Likewise.
948 (_mm512_maskz_fmsub_round_pd): Use
949 __builtin_ia32_vfmsubpd512_maskz.
950 (_mm512_maskz_fmsub_pd): Likewise.
951 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
952 (_mm512_mask_fmsub_round_ps): Likewise.
953 (_mm512_fmsub_ps): Likewise.
954 (_mm512_mask_fmsub_ps): Likewise.
955 (_mm512_maskz_fmsub_round_ps): Use
956 __builtin_ia32_vfmsubps512_maskz.
957 (_mm512_maskz_fmsub_ps): Likewise.
958 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
959 __builtin_ia32_vfmsubpd256_mask.
960 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
961 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
962 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
963 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
964 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
965 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
966 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
967 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
968 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
969 __builtin_ia32_vfmsubpd.
970 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
971 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
972 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
973 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
974 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
975 * config/i386/i386-builtin.def: Add
976 __builtin_ia32_vfmsubpd256_mask,
977 __builtin_ia32_vfmsubpd256_maskz,
978 __builtin_ia32_vfmsubpd128_mask,
979 __builtin_ia32_vfmsubpd128_maskz,
980 __builtin_ia32_vfmsubps256_mask,
981 __builtin_ia32_vfmsubps256_maskz,
982 __builtin_ia32_vfmsubps128_mask,
983 __builtin_ia32_vfmsubps128_maskz,
984 __builtin_ia32_vfmsubpd512_mask,
985 __builtin_ia32_vfmsubpd512_maskz,
986 __builtin_ia32_vfmsubps512_mask,
987 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
988 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
989 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
990 __builtin_ia32_vfmsubpd256.
991 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
992 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
993 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
995 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
997 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
999 (fmai_vmfmsub_<mode><round_name>): Likewise.
1001 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1003 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
1004 Remove plus. Renamed to ...
1005 (*sub<mode>3<mask_name>_bcst): This.
1006 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
1007 (*add<mode>3<mask_name>_bcst): This.
1009 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1012 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
1014 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1017 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
1018 (_mm_or_epi32): Likewise.
1019 (_mm256_xor_epi32): Likewise.
1020 (_mm_xor_epi32): Likewise.
1021 (_mm256_or_epi64): Likewise.
1022 (_mm_or_epi64): Likewise.
1023 (_mm256_xor_epi64): Likewise.
1024 (_mm_xor_epi64): Likewise.
1026 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
1029 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
1031 2018-10-20 Jakub Jelinek <jakub@redhat.com>
1034 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
1036 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
1038 * doc/ux.texi: Move @section directly after @node.
1040 2018-10-19 Jakub Jelinek <jakub@redhat.com>
1044 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
1045 depend closely nested inside of loop with ordered clause with
1048 2018-10-19 David Malcolm <dmalcolm@redhat.com>
1050 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
1051 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
1052 * doc/ux.texi: New file.
1054 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
1056 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
1057 be the first CR field allocated.
1059 2018-10-19 Richard Biener <rguenther@suse.de>
1062 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
1063 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
1065 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
1068 * config/i386/sse.md
1069 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
1070 (*add<mode>3<mask_name>_bcst_2): Likewise.
1072 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
1074 * config/i386/sse.md
1075 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
1076 Replace nonimmediate_operand with register_operand.
1077 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
1079 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
1082 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
1084 PR rtl-optimization/87596
1085 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
1086 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
1087 for instructions in FROM..TO range.
1089 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
1091 * cfgexpand.c (expand_one_var): Use specific wording in error message
1092 for non-local frame variables.
1093 * stor-layout.c (layout_decl): Do not issue a warning for them.
1095 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
1097 * haifa-sched.c (priority): Add force_recompute parameter.
1098 (apply_replacement): Call priority () with force_recompute = true.
1099 (restore_pattern): Likewise.
1101 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1103 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
1104 HOST_BITS_PER_WIDE_INT.
1105 (test_vector_ops_duplicate): Likewise.
1107 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1110 * config/i386/sse.md (VF_AVX512): New.
1111 (avx512bcst): Likewise.
1112 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
1114 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
1116 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
1119 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
1121 * doc/invoke.texi (-dumpversion): Improve grammar.
1122 (-dumpfullversion): Make more consistent with -dumpversion.
1124 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
1126 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
1127 Set JUMP_LABEL to the jump insn.
1128 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
1129 Predict emitted jump and add label to jump insn.
1131 2018-10-18 David Malcolm <dmalcolm@redhat.com>
1133 PR tree-optimization/87562
1134 * input.c (get_substring_ranges_for_loc): Use
1135 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
1136 getting the linemap for the endpoint. Verify that it's either
1137 in the same linemap as the start point's spelling location, or
1138 at least in the same file.
1140 2018-10-18 Richard Biener <rguenther@suse.de>
1142 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
1143 feed width-specific load/store costs through ix86_vec_cost.
1144 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
1145 (k8_cost): Likewise.
1146 (bdver_cost): Likewise.
1147 (znver1_cost): Likewise.
1148 (btver1_cost): Likewise.
1149 (btver2_cost): Likewise.
1151 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1153 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
1154 to simplify subreg of vec_merge.
1156 2018-10-18 Richard Biener <rguenther@suse.de>
1158 * config/i386/i386.c: Fix costing of vector FMA.
1160 2018-10-18 Richard Biener <rguenther@suse.de>
1162 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
1164 (ix86_builtin_vectorization_cost): For vec_construct properly
1165 cost insertion into SSE regs.
1166 (...): Adjust calls to ix86_vec_cost.
1168 2018-10-18 Richard Biener <rguenther@suse.de>
1172 2018-02-07 Richard Biener <rguenther@suse.de>
1174 PR tree-optimization/84204
1175 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
1178 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1181 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
1183 (test_vector_ops_duplicate): Add test for a scalar subreg of a
1184 VEC_MERGE of a VEC_DUPLICATE.
1186 2018-10-17 Joseph Myers <joseph@codesourcery.com>
1188 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
1189 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
1190 * doc/standards.texi (C Language): Document C2X.
1191 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
1192 (rl78_option_override): Handle "GNU C2X" language name.
1194 2018-10-17 Joseph Myers <joseph@codesourcery.com>
1196 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
1197 Document C17 as published in 2018.
1199 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
1202 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
1203 bail out if both sides do not have the same storage order.
1205 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
1207 * bitmap.c (bitmap_head::dump): New.
1208 * bitmap.h (bitmap_head): Add dump().
1209 * gimple-ssa-evrp-analyze.c
1210 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
1211 (evrp_range_analyzer::set_ssa_range_info): Same.
1212 (evrp_range_analyzer::record_ranges_from_phis): Same.
1213 (evrp_range_analyzer::record_ranges_from_stmt): Same.
1214 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
1215 * gimple-ssa-sprintf.c (get_int_range): Same.
1216 (format_integer): Same.
1217 (sprintf_dom_walker::handle_gimple_call): Same.
1218 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
1219 (ipcp_vr_lattice::top_p): Same.
1220 (ipcp_vr_lattice::bottom_p): Same.
1221 (ipcp_vr_lattice::set_to_bottom): Same.
1222 (ipa_vr_operation_and_type_effects): Same.
1223 (propagate_vr_across_jump_function): Same.
1224 (ipcp_store_vr_results): Same.
1225 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
1226 (ipa_print_node_jump_functions_for_edge): Same.
1227 (ipa_get_value_range): Same.
1228 (ipa_compute_jump_functions_for_edge): Same.
1229 (ipa_write_jump_function): Same.
1230 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
1231 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
1233 * vr-values.c (set_value_range_to_nonnegative): Same.
1234 (set_value_range_to_truthvalue): Same.
1235 (vr_values::get_value_range): Same.
1236 (vr_values::set_defs_to_varying): Same.
1237 (vr_values::update_value_range): Same.
1238 (symbolic_range_based_on_p): Same.
1239 (vr_values::op_with_boolean_value_range_p): Same.
1240 (vr_values::extract_range_for_var_from_comparison_expr): Same.
1241 (vr_values::extract_range_from_ssa_name): Same.
1242 (vr_values::extract_range_from_binary_expr): Same.
1243 (vr_values::extract_range_from_unary_expr): Same.
1244 (vr_values::extract_range_from_cond_expr): Same.
1245 (vr_values::extract_range_from_comparison): Same.
1246 (vr_values::check_for_binary_op_overflow): Same.
1247 (vr_values::extract_range_basic): Same.
1248 (vr_values::extract_range_from_assignment): Same.
1249 (compare_ranges): Same.
1250 (compare_range_with_value): Same.
1251 (vr_values::adjust_range_with_scev): Same.
1252 (vrp_valueize): Same.
1253 (vrp_valueize_1): Same.
1254 (vr_values::get_vr_for_comparison): Same.
1255 (vr_values::compare_name_with_value): Same.
1256 (vr_values::compare_names): Same.
1257 (vr_values::vrp_evaluate_conditional): Same.
1258 (find_case_label_ranges): Same.
1259 (vr_values::vrp_visit_switch_stmt): Same.
1260 (vr_values::extract_range_from_phi_node): Same.
1261 (vr_values::simplify_div_or_mod_using_ranges): Same.
1262 (vr_values::simplify_bit_ops_using_ranges): Same.
1263 (test_for_singularity): Same.
1264 (range_fits_type_p): Same.
1265 (vr_values::simplify_cond_using_ranges_1): Same.
1266 (vr_values::simplify_switch_using_ranges): Same.
1267 (vr_values::simplify_float_conversion_using_ranges): Same.
1268 (vr_values::two_valued_val_range_p): Same.
1269 (vr_values::add_equivalence): Move to value_range::equiv_add.
1270 * vr-values.h (vr_values::add_equivalence): Remove.
1271 (VR_INITIALIZER): Remove.
1272 * tree-vrp.c (value_range::set): New.
1273 (value_range::equiv_add): New.
1274 (value_range::value_range): New.
1275 (value_range::deep_copy): New.
1276 (value_range::check): New.
1277 (value_range::equal_p): New.
1278 (value_range::ignore_equivs_equal_p): New.
1279 (value_range::operator==): New.
1280 (value_range::operator!=): New.
1281 (value_range::symbolic_p): New.
1282 (value_range::numeric_p): New.
1283 (value_range::set_undefined): New.
1284 (value_range::set_varying): New.
1285 (value_range::may_contain_p): New.
1286 (value_range::equiv_clear): New.
1287 (value_range::singleton_p): New.
1288 (value_range::intersect): New.
1289 (value_range::dump): New.
1290 (value_range::set_and_canonicalize): New.
1291 (set_value_range): Adjust for value_range API.
1292 (set_value_range_to_undefined): Same.
1293 (set_value_range_to_varying): Same.
1294 (set_and_canonicalize_value_range): Same.
1295 (set_value_range_to_nonnull): Same.
1296 (set_value_range_to_null): Same.
1297 (range_is_null): Same.
1298 (range_is_nonnull): Same.
1299 (range_int_cst_p): Same.
1300 (range_int_cst_singleton_p): Same.
1301 (symbolic_range_p): Same.
1302 (range_includes_zero_p): Same.
1303 (value_range_constant_singleton): Same.
1304 (vrp_set_zero_nonzero_bits): Same.
1305 (ranges_from_anti_range): Same.
1306 (extract_range_into_wide_ints): Same.
1307 (extract_range_from_multiplicative_op): Same.
1308 (set_value_range_with_overflow): Same.
1309 (extract_range_from_binary_expr_1): Same.
1310 (extract_range_from_unary_expr): Same.
1311 (dump_value_range): Same.
1312 (debug_value_range): Same.
1313 (vrp_prop::check_array_ref): Same.
1314 (vrp_prop::check_mem_ref): Same.
1315 (vrp_prop::vrp_initialize): Same.
1316 (vrp_prop::visit_stmt): Same.
1317 (intersect_ranges): Same.
1318 (vrp_prop::visit_phi): Same.
1319 (vrp_prop::vrp_finalize): Same.
1320 (determine_value_range_1): Same.
1321 (determine_value_range): Same.
1322 (vrp_intersect_ranges_1): Rename to...
1323 (vrp_intersect_1): this.
1324 (vrp_intersect_ranges): Rename to...
1325 (value_range::intersect_helper): ...this.
1326 (vrp_meet_1): Rename to...
1327 (value_range::union_helper): ...this.
1328 (vrp_meet): Rename to...
1329 (value_range::union_): ...this.
1330 (copy_value_range): Remove.
1331 * tree-vrp.h (struct value_range): Rewrite into a proper class.
1332 (value_range::vrtype): New.
1333 (value_range::type): New.
1334 (value_range::equiv): New.
1335 (value_range::min): New.
1336 (value_range::max): New.
1337 (value_range::varying_p): New.
1338 (value_range::undefined_p): New.
1339 (value_range::null_p): New.
1340 (value_range::equiv_add): New.
1341 (copy_value_range): Remove.
1343 2018-10-17 David Malcolm <dmalcolm@redhat.com>
1345 * Makefile.in (SELFTEST_TARGETS): New.
1346 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
1347 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
1348 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
1350 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
1351 (selftest-c++-gdb, selftest-c++-valgrind): Move to
1353 * configure: Regenerate.
1354 * configure.ac (selftest_languages): New.
1356 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
1358 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
1359 overflow wraps argument.
1360 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
1361 wide_int_range_multiplicative_op.
1362 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
1363 overflow wraps argument.
1364 (wide_int_range_multiplicative_op): Same.
1365 (wide_int_range_lshift): Same.
1366 (wide_int_range_div): Same.
1367 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
1368 (wide_int_range_lshift): Same.
1369 (wide_int_range_div): Same.
1371 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
1373 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
1374 use sign as argument.
1375 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
1376 wide_int_range_shift_undefined_p.
1378 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1380 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
1382 (@despeculate_copy<ALLI_TI:mode>): ... This.
1383 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
1386 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1388 * config.gcc: Obsolete *-*-solaris2.10*.
1389 * doc/install.texi (Specific, *-*-solaris2*): Document it.
1391 2018-10-12 Jeff Law <law@redhat.com>
1393 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
1394 reg + sym +- const_int addressing modes.
1396 2018-10-15 David Malcolm <dmalcolm@redhat.com>
1398 * common.opt (fdiagnostics-minimum-margin-width=): New option.
1399 * diagnostic-show-locus.c (layout::layout): Apply the minimum
1401 (layout::start_annotation_line): Only print up to 3 of the
1402 margin character, to avoid touching the left-hand side.
1403 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
1404 minimum margin width, as set by test_diagnostic_context's ctor.
1405 (selftest::test_fixit_insert_containing_newline): Likewise.
1406 (selftest::test_fixit_insert_containing_newline_2): Likewise.
1407 (selftest::test_line_numbers_multiline_range): Clear
1408 dc.min_margin_width.
1409 * diagnostic.c (diagnostic_initialize): Initialize
1411 * diagnostic.h (struct diagnostic_context): Add field
1413 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
1414 * opts.c (common_handle_option): Handle
1415 OPT_fdiagnostics_minimum_margin_width_.
1416 * selftest-diagnostic.c
1417 (selftest::test_diagnostic_context::test_diagnostic_context):
1418 Initialize min_margin_width to 6.
1419 * toplev.c (general_init): Initialize global_dc->min_margin_width.
1421 2018-10-15 David Malcolm <dmalcolm@redhat.com>
1423 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1424 Fix usage of "error_at_rich_loc" in the comment.
1426 2018-10-15 Renlin Li <renlin.li@arm.com>
1429 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
1430 if-conversioned loop when it contains ifn with types not
1431 supported by backend.
1432 * internal-fn.c (expand_direct_optab_fn): Add an assert.
1433 (direct_internal_fn_supported_p): New helper function.
1434 * internal-fn.h (direct_internal_fn_supported_p): Declare.
1436 2018-10-15 Jakub Jelinek <jakub@redhat.com>
1439 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
1440 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
1441 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
1443 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
1445 PR tree-optimization/87022
1446 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
1447 bits in dist vector rather than the first one.
1449 2018-10-15 Richard Biener <rguenther@suse.de>
1452 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
1453 (visit_loadstore): When a used restrict tag escaped verify that
1454 the points-to solution of "other" pointers do not include
1456 (compute_dependence_clique): If a used restrict tag escaped
1457 communicated that down to visit_loadstore.
1459 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
1461 * config/s390/s390.c (s390_expand_vec_init): Force vector element
1462 into reg if it isn't a general operand.
1464 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
1467 * config/i386/sse.md (*vec_dupv2di): Add register source to
1470 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
1473 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
1474 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
1475 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
1476 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
1478 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
1480 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
1481 (notice_args_size): Set it in the current trace if no insn that can
1482 throw internally has been seen yet.
1483 (connect_traces): When connecting args_size between traces, allow the
1484 incoming values not to match if there is an insn setting it before the
1485 first insn that can throw internally; in that case, force the creation
1486 of a CFI note on this latter insn.
1488 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
1490 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
1491 the base class, because using the injected-class-name was not clearly
1492 specified until DR 176.
1494 2018-10-12 Paul Koning <ni1d@arrl.net>
1496 * config/pdp11/pdp11.md (doloop_end): New expander.
1497 (doloop_end_insn): renamed from "doloop_end".
1498 (addqi3): New pattern.
1499 (subqi3): New pattern.
1500 * config/pdp11/predicates.md (incdec_operand): New predicate.
1502 2018-10-12 Yury Gribov <tetra2005@gmail.com>
1505 * real.c (format_helper::can_represent_integral_type_p): New function
1506 * real.h (format_helper::can_represent_integral_type_p): Ditto.
1507 * match.pd: New pattern.
1509 2018-10-12 Alexandre Oliva <oliva@adacore.com>
1511 * configure.ac: Introduce --enable-large-address-aware
1512 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
1513 * doc/install.texi: Document it.
1514 * configure, config.in: Rebuilt.
1515 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
1516 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
1517 (LINK_SPEC): Insert it.
1518 * config/i386/mingw-264.h: Likewise.
1520 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
1522 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
1524 PR rtl-optimization/87600
1525 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
1527 2018-10-12 Paul Koning <ni1d@arrl.net>
1529 * doc/md.texi (doloop_end): Document that the pattern code may
1530 need to check operand mode.
1532 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
1534 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
1535 to zero-extend between int and floating-point registers.
1536 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
1537 ldp into floating-point registers. Add type and arch attributes.
1538 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
1539 Use f_loads for type attribute.
1541 2018-10-11 Martin Sebor <msebor@redhat.com>
1543 * doc/extend.texi (attribute packed): Correct typos.
1545 2018-10-11 Martin Sebor <msebor@redhat.com>
1547 * doc/extend.texi (attribute flatten): Mention interaction with
1550 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
1553 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
1554 Set new_decl virtual flag to zero.
1556 2018-10-11 Martin Sebor <msebor@redhat.com>
1559 * doc/extend.texi (attribute format_arg): Discuss using multiple
1560 attributes on a single function.
1562 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
1564 PR tree-optimization/86829
1565 * match.pd (sin (atan (x))): New simplification rules.
1566 (cos (atan (x))): Likewise.
1567 * real.c (build_sinatan_real): New function.
1568 * real.h (build_sinatan_real): Prototype.
1570 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
1572 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
1574 (fold_mergeeo_helper): New helper function.
1575 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
1576 intrinsics. Correct some whitespace indentation issues.
1578 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
1581 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
1582 Use HOST_WIDE_INT_1U for shift.
1584 2018-10-11 Doug Rupp <rupp@adacore.com>
1585 Olivier Hainque <hainque@adacore.com>
1587 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
1588 Pass --relax to the linker for RTPs.
1589 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
1591 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
1592 Jan Hubicka <jh@suse.cz>
1593 Martin Jambor <mjambor@suse.cz>
1595 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
1596 the same elements are repeated rather than printing all of them.
1597 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
1598 "repeated" elements.
1599 * read-rtl-function.c (test_loading_repeat): New function.
1600 (read_rtl_function_c_tests): Call test_loading_repeat.
1601 * rtl-tests.c (test_dumping_repeat): New function.
1602 (rtl_tests_c_tests): Call test_dumping_repeat.
1604 2018-10-11 Richard Biener <rguenther@suse.de>
1606 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
1607 bdver?_cost): Unify to ...
1608 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
1609 * config/i386/i386.c (processor_cost_table): Adjust.
1611 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
1614 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
1615 the thunk when expanding to GIMPLE.
1617 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1619 * varasm.c (mergeable_string_section): Don't try to move zero-length
1620 strings to the merge section.
1622 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
1625 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
1627 2018-10-10 Jakub Jelinek <jakub@redhat.com>
1630 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
1631 to special_args set.
1633 2018-10-10 Richard Biener <rguenther@suse.de>
1635 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
1636 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
1637 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
1638 and recursing and pattern terminating the recursion on SSE
1639 vector width using ix86_expand_reduc.
1640 (reduc_sminmax_scal_<mode>): Split into part reducing to half
1641 width and recursing and SSE2 vector variant doing the final
1642 reduction with ix86_expand_reduc.
1643 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
1644 with terminating the recursion at AVX level, splitting that
1647 2018-10-09 David Malcolm <dmalcolm@redhat.com>
1649 * genmatch.c (error_cb): Rename to...
1650 (diagnostic_cb): ...this, converting int params to enums.
1651 (fatal_at): Update for renaming.
1652 (warning_at): Likewise.
1654 * input.c (selftest::ebcdic_execution_charset::apply):
1655 Update for renaming of...
1656 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
1658 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
1659 converting level and reason to enums.
1660 (class selftest::lexer_error_sink): Rename to...
1661 (class selftest::lexer_test_options): ...this, renaming field
1662 "m_errors" to "m_diagnostics".
1663 (selftest::lexer_test_options::apply): Update for renaming of...
1664 (selftest::lexer_test_options::on_error): ...this, renaming to...
1665 (selftest::lexer_test_options::on_diagnostic): ...this
1666 converting level and reason to enums.
1667 (selftest::test_lexer_string_locations_raw_string_unterminated):
1668 Update for renamings.
1669 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
1672 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
1674 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
1675 * config/rs6000/pmmintrin.h: New file.
1677 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
1679 PR tree-optimization/86659
1680 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
1682 2018-10-09 Richard Biener <rguenther@suse.de>
1684 PR tree-optimization/63155
1685 * tree-ssa-structalias.c: Include tree-ssa.h.
1686 (get_constraint_for_ssa_var): For undefs return nothing_id.
1687 (find_func_aliases): Cleanup PHI handling.
1689 2018-10-09 Richard Biener <rguenther@suse.de>
1691 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
1694 2018-10-09 Martin Liska <mliska@suse.cz>
1696 * asan.c (asan_emit_stack_protection): If a stack variable
1697 is located in a same file as current function, then emit
1698 line info into variable definition string.
1700 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
1702 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
1705 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
1707 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
1710 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
1712 PR tree-optimization/86659
1713 * gimple-match.h (struct gimple_match_op): Add reverse field.
1714 (gimple_match_op::set_op): New overloaded method.
1715 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
1716 the REF_REVERSE_STORAGE_ORDER flag on the value.
1717 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
1718 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
1720 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
1723 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
1724 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
1726 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
1729 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
1730 Defined with __builtin_ia32_vfmaddsubpd512_mask.
1732 2018-10-08 Richard Biener <rguenther@suse.de>
1734 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
1735 cost the same as AVX128 ones.
1737 2018-10-08 Paul Koning <ni1d@arrl.net>
1739 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
1740 (expand_block_move): New function.
1741 * config/pdp11/pdp11.c (output_block_move): Remove.
1742 (expand_block_move): New function.
1743 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
1744 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
1745 (*movmemhi1): Remove.
1747 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
1749 * config/s390/2827.md: Increase latencies for some FP instructions.
1751 2018-10-08 Richard Biener <rguenther@suse.de>
1753 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
1755 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
1756 * tree-vectorizer.h (dump_stmt_cost): Adjust.
1757 (add_stmt_cost): Dump return value of the hook.
1759 2018-10-08 Richard Biener <rguenther@suse.de>
1761 PR tree-optimization/63155
1762 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
1763 (ssa_propagation_engine::ssa_propagate): Remove redundant
1764 bitmap bit clearing.
1766 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
1768 PR rtl-optimization/86939
1769 PR rtl-optimization/87479
1770 * ira.h (non_conflicting_reg_copy_p): New prototype.
1771 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
1772 (make_hard_regno_dead): Don't add conflicts for register
1773 ignore_reg_for_conflicts.
1774 (make_object_dead): Likewise.
1775 (non_conflicting_reg_copy_p): New function.
1776 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
1777 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
1778 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
1779 (make_hard_regno_dead): Don't add conflicts for register
1780 ignore_reg_for_conflicts. Remove special conflict handling of
1781 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
1782 check_pic_pseudo_p and update callers.
1783 (mark_pseudo_dead): Don't add conflicts for register
1784 ignore_reg_for_conflicts.
1785 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
1787 2018-10-05 Andrew Waterman <andrew@sifive.com>
1788 Jim Wilson <jimw@sifive.com>
1790 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
1791 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
1792 new pattern using HONOR_SNANS that emits one extra instruction.
1794 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
1796 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
1797 patterns): Merge SI and DI patterns to a GPR pattern.
1798 (unnamed define_insn and define_split for record form of that): Merge
1799 to a single define_insn_and_split pattern.
1801 2018-10-05 David Malcolm <dmalcolm@redhat.com>
1804 * input.c (expand_location_to_spelling_point): Add param "aspect"
1805 and use rather than hardcoding LOCATION_ASPECT_CARET.
1806 (get_substring_ranges_for_loc): Handle the case of a single token
1807 within a macro expansion.
1808 * input.h (expand_location_to_spelling_point): Add "aspect" param,
1809 defaulting to LOCATION_ASPECT_CARET.
1811 2018-10-05 Paul Koning <ni1d@arrl.net>
1813 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
1814 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
1815 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
1816 (pdp11_guard_type): New function.
1818 2018-10-05 Paul Koning <ni1d@arrl.net>
1820 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
1821 * config/pdp11/pdp11.opt (mfloat32): Remove.
1823 * doc/invoke.texi (pdp11 -mfloat32): Remove:
1824 (pdp11 -mfloat64): Remove.
1826 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
1828 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
1829 (*cmp<mode>_cc_i387): Ditto.
1830 (*cmpu<mode>_cc_i387): Ditto.
1831 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
1832 * config/i386/i386.c (ix86_expand_fp_compare): Remove
1834 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
1835 Emit x86_sahf_1 pattern.
1836 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
1837 (ix86_expand_carry_flag_compare): Ditto.
1839 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
1841 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
1842 to reg_or_0_operand. Add "C" constraint.
1843 (*cmpxf_cc_i387): Ditto.
1844 (*cmp<mode>_i387): Change operand 2 predicate
1845 to nonimm_or_0_operand. Add "C" constraint.
1846 (*cmp<mode>_cc_i387): Ditto.
1847 (*cmp<mode>_0_i387): Remove insn pattern.
1848 (*cmp<mode>_0_cc_i387): Ditto.
1850 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
1852 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
1853 * config/i386/predicates.md (nonimm_or_0_operand): Rename
1854 from vector_move_operand. Update all uses.
1856 2018-10-05 Martin Sebor <msebor@redhat.com>
1858 PR tree-optimization/87490
1859 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
1862 2018-10-05 Richard Biener <rguenther@suse.de>
1864 PR tree-optimization/63155
1865 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
1866 vertical space in dumpfiles.
1867 * tree-ssa-propagate.h
1868 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
1869 * tree-ssa-propagate.c (cfg_blocks_back): New global.
1870 (ssa_edge_worklist_back): Likewise.
1871 (curr_order): Likewise.
1872 (cfg_blocks_get): Remove abstraction.
1873 (cfg_blocks_add): Likewise.
1874 (cfg_blocks_empty_p): Likewise.
1875 (add_ssa_edge): Add to current or next worklist based on
1877 (add_control_edge): Likewise.
1878 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
1880 (ssa_propagation_engine::ssa_propagate): ... here. Unify
1881 iteration from CFG and SSA edge worklist so we process
1882 everything in RPO order, prioritizing forward progress
1884 (ssa_prop_init): Allocate new worklists, do not dump
1886 (ssa_prop_fini): Free new worklists.
1888 2018-10-05 Richard Biener <rguenther@suse.de>
1890 * tree-core.h (tree_block::abstract_flag): Remove.
1891 (tree_block::block_num): Make full 32bits.
1892 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
1893 * tree.h (BLOCK_ABSTRACT): Remove.
1894 * dwarf2out.c (gen_lexical_block_die): Remove dead code
1895 resulting from BLOCK_ABSTRACT being always false.
1896 (gen_inlined_subroutine_die): Likewise.
1897 (gen_block_die): Likewise.
1898 * tree.c (block_ultimate_origin): Likewise.
1899 * tree-pretty-print.c (dump_block_node): Remove code dealing
1900 with BLOCK_ABSTRACT.
1901 * tree-ssa-live.c (dump_scope_block): Likewise.
1902 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
1903 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
1905 2018-10-05 Richard Biener <rguenther@suse.de>
1907 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
1908 is asked for initialize mode to the component mode of the
1911 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
1914 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
1915 assembler for -mavx.
1916 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
1918 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
1921 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
1923 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
1924 to be DImode. When using mffscrn, force the operand to a register.
1926 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
1928 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
1929 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
1930 X87MODEF mode iterator.
1931 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
1932 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
1933 X87MODEF mode iterator.
1935 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
1937 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
1938 -Wno-prio-ctor-dtor.
1940 2018-10-04 David Malcolm <dmalcolm@redhat.com>
1942 * Makefile.in (OBJS): Add opt-problem.o.
1943 * dump-context.h: Include "selftest.h.
1944 (selftest::temp_dump_context): New forward decl.
1945 (class dump_context): Make friend of class
1946 selftest::temp_dump_context.
1947 (dump_context::dump_loc_immediate): New decl.
1948 (class dump_pretty_printer): Move here from dumpfile.c.
1949 (class temp_dump_context): Move to namespace selftest.
1950 (temp_dump_context::temp_dump_context): Add param
1951 "forcibly_enable_dumping".
1952 (selftest::verify_dumped_text):
1953 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
1954 (selftest::verify_item):
1955 (ASSERT_IS_TEXT): Move here from dumpfile.c.
1956 (ASSERT_IS_TREE): Likewise.
1957 (ASSERT_IS_GIMPLE): Likewise.
1958 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
1960 (dump_context::dump_loc_immediate): ...this new function.
1961 (class dump_pretty_printer): Move to dump-context.h.
1962 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
1963 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
1964 (temp_dump_context::temp_dump_context): Move to "selftest"
1965 namespace. Add param "forcibly_enable_dumping", and use it to
1966 conditionalize the use of m_pp;
1967 (selftest::verify_dumped_text): Make non-static.
1968 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
1969 (selftest::verify_item): Make non-static.
1970 (ASSERT_IS_TEXT): Move to dump-context.h.
1971 (ASSERT_IS_TREE): Likewise.
1972 (ASSERT_IS_GIMPLE): Likewise.
1973 (selftest::test_capture_of_dump_calls): Pass "true" for new
1974 param of temp_dump_context.
1975 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
1976 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
1978 * opt-problem.cc: New file.
1979 * opt-problem.h: New file.
1980 * optinfo-emit-json.cc
1981 (selftest::test_building_json_from_dump_calls): Pass "true" for
1982 new param of temp_dump_context.
1983 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
1984 (optinfo::emit_for_opt_problem): New function.
1985 (optinfo::emit): Clarity which emit_item is used.
1986 * optinfo.h (optinfo::get_dump_location): New accessor.
1987 (optinfo::emit_for_opt_problem): New decl.
1988 (optinfo::emit): Make const.
1989 * selftest-run-tests.c (selftest::run_tests): Call
1990 selftest::opt_problem_cc_tests.
1991 * selftest.h (selftest::opt_problem_cc_tests): New decl.
1992 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
1993 bool to opt_result, converting fprintf messages to
1994 opt_result::failure_at calls. Add "stmt" param for use by the
1996 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
1997 (runtime_alias_check_p): Convert return type from bool to
1998 opt_result, converting dump_printf calls to
1999 opt_result::failure_at, using the statement DDR_A for their
2001 (find_data_references_in_stmt): Convert return type from bool to
2002 opt_result, converting "return false" to opt_result::failure_at
2004 * tree-data-ref.h: Include "opt-problem.h".
2005 (dr_analyze_innermost): Convert return type from bool to opt_result,
2006 and add a const gimple * param.
2007 (find_data_references_in_stmt): Convert return type from bool to
2009 (runtime_alias_check_p): Likewise.
2010 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
2011 dr_analyze_innermost.
2012 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
2013 Convert return type from bool to opt_result, adding a message for
2014 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
2015 (vect_analyze_data_ref_dependence): Convert return type from bool
2016 to opt_result. Change sense of return type from "false"
2017 effectively meaning "no problems" to "false" meaning a problem,
2018 so that "return false" becomes "return opt_result::success".
2019 Convert "return true" calls to opt_result::failure_at, using
2020 the location of statement A rather than vect_location.
2021 (vect_analyze_data_ref_dependences): Convert return type from bool
2023 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
2024 calls to opt_result::failure_at, using the stmt location rather
2026 (vect_verify_datarefs_alignment): Convert return type from bool
2028 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
2029 into multiple more-tightly-scoped copies.
2030 (vect_analyze_data_refs_alignment): Convert return type from bool
2032 (vect_analyze_data_ref_accesses): Likewise, converting a
2033 "return false" to a "return opt_result::failure_at", adding a
2035 (vect_prune_runtime_alias_test_list): Convert return type from
2036 bool to opt_result, converting dump_printf_loc to
2037 opt_result::failure_at. Add a %G to show the pertinent statement,
2038 and use the stmt's location rather than vect_location.
2039 (vect_find_stmt_data_reference): Convert return type from
2040 bool to opt_result, converting dump_printf_loc to
2041 opt_result::failure_at, using stmt's location.
2042 (vect_analyze_data_refs): Convert return type from bool to
2043 opt_result. Convert "return false" to "return
2044 opt_result::failure_at", adding messages as needed.
2045 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
2046 type from bool to opt_result.
2047 (vect_determine_vf_for_stmt): Likewise.
2048 (vect_determine_vectorization_factor): Likewise, converting
2049 dump_printf_loc to opt_result::failure_at, using location of phi
2050 rather than vect_location.
2051 (vect_analyze_loop_form_1): Convert return type from bool to
2052 opt_result, converting dump_printf_loc calls, retaining the use of
2054 (vect_analyze_loop_form): Convert return type from loop_vec_info
2055 to opt_loop_vec_info.
2056 (vect_analyze_loop_operations): Convert return type from bool to
2057 opt_result, converting dump_printf_loc calls, using the location
2058 of phi/stmt rather than vect_location where available. Convert
2059 various "return false" to "return opt_result::failure_at" with
2060 "unsupported phi" messages.
2061 (vect_get_datarefs_in_loop): Convert return type from bool to
2062 opt_result. Add a message for the
2063 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
2064 (vect_analyze_loop_2): Convert return type from bool to
2065 opt_result. Ensure "ok" is set to a opt_result::failure_at before
2066 each "goto again;", adding new messages where needed.
2067 Add "unsupported grouped {store|load}" messages.
2068 (vect_analyze_loop): Convert return type from loop_vec_info to
2070 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
2072 * tree-vect-stmts.c (process_use): Likewise, converting
2073 dump_printf_loc call and using stmt location, rather than
2075 (vect_mark_stmts_to_be_vectorized): Likeise.
2076 (vect_analyze_stmt): Likewise, adding a %G.
2077 (vect_get_vector_types_for_stmt): Convert return type from bool to
2078 opt_result, converting dump_printf_loc calls and using stmt
2079 location, rather than vect_location.
2080 (vect_get_mask_type_for_stmt): Convert return type from tree to
2081 opt_tree, converting dump_printf_loc calls and using stmt location.
2082 * tree-vectorizer.c: Include "opt-problem.h.
2083 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
2084 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
2085 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
2086 enabled, use it to report at the top level "couldn't vectorize
2087 loop" followed by the problem.
2088 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
2089 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
2091 (vect_analyze_stmt): Likewise.
2092 (vect_get_vector_types_for_stmt): Likewise.
2093 (tree vect_get_mask_type_for_stmt): Likewise.
2094 (vect_analyze_data_ref_dependences): Likewise.
2095 (vect_enhance_data_refs_alignment): Likewise.
2096 (vect_analyze_data_refs_alignment): Likewise.
2097 (vect_verify_datarefs_alignment): Likewise.
2098 (vect_analyze_data_ref_accesses): Likewise.
2099 (vect_prune_runtime_alias_test_list): Likewise.
2100 (vect_find_stmt_data_reference): Likewise.
2101 (vect_analyze_data_refs): Likewise.
2102 (vect_analyze_loop): Convert return type from loop_vec_info to
2104 (vect_analyze_loop_form): Likewise.
2105 (vect_analyze_slp): Convert return type from bool to opt_result.
2107 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2109 * doc/invoke.texi (-fopt-info): Document new "internals"
2111 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
2112 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
2114 (optinfo_verbosity_options): Add "internals".
2115 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
2116 (dump_context::apply_dump_filter_p): New member function.
2117 (dump_context::dump_loc): Use apply_dump_filter_p rather than
2118 explicitly masking the dump_kind.
2119 (dump_context::begin_scope): Increment the scope depth first. Use
2120 apply_dump_filter_p rather than explicitly masking the dump_kind.
2121 (dump_context::emit_item): Use apply_dump_filter_p rather than
2122 explicitly masking the dump_kind.
2123 (dump_dec): Likewise.
2124 (dump_hex): Likewise.
2125 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
2126 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
2127 (opt_info_switch_p): Update handling of default
2128 MSG_OPTIMIZED_LOCATIONS to cope with default of
2129 MSG_PRIORITY_USER_FACING.
2130 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
2131 masking the dump_kind.
2132 (selftest::test_capture_of_dump_calls): Update test_dump_context
2133 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
2134 than MSG_ALL. Generalize scope test to be run at all four
2135 combinations of with/without MSG_PRIORITY_USER_FACING and
2136 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
2137 for each of the two values.
2138 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
2139 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
2140 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
2141 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
2142 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
2143 with MSG_PRIORITY_*.
2144 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
2145 dump messages as MSG_PRIORITY_USER_FACING.
2146 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
2147 about the interaction with MSG_PRIORITY_*.
2149 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2151 * varasm.c (output_constant): Add new parameter merge_strings.
2152 Make strings properly zero terminated in merge string sections.
2153 (mergeable_string_section): Don't fail if the last char is non-zero.
2154 (assemble_variable_contents): Handle merge string sections.
2155 (assemble_variable): Likewise.
2156 (assemble_constant_contents): Likewise.
2157 (output_constant_def_contents): Likewise.
2158 (output_constructor_array_range,
2159 output_constructor_regular_field): Adjust call to output_constant.
2160 (output_object_block): Adjust call to assemble_constant_contents
2161 and assemble_variable_contents.
2163 2018-10-04 Martin Liska <mliska@suse.cz>
2166 * cgraphunit.c (process_function_and_variable_attributes):
2167 Warn about a function with alias attribute and a body.
2169 2018-10-04 Martin Liska <mliska@suse.cz>
2172 * multiple_target.c (redirect_to_specific_clone): New function.
2173 (ipa_target_clone): Use it.
2174 * tree-inline.c: Fix comment.
2176 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2178 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
2180 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
2181 (gcc::dump_manager::register_pass): New member function, adapted
2182 from loop body in gcc::pass_manager::register_pass, adding a
2183 call to update_dfi_for_opt_info.
2184 (gcc::dump_manager::opt_info_enable_passes): Store the
2185 -fopt-info options into the new fields. Move the loop
2187 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
2189 * dumpfile.h (struct opt_pass): New forward decl.
2190 (gcc::dump_manager::register_pass): New decl.
2191 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
2192 (class gcc::dump_manager): Add fields "m_optgroup_flags",
2193 "m_optinfo_flags", and "m_optinfo_filename".
2194 * passes.c (gcc::pass_manager::register_pass): Move all of the
2195 dump-handling code to gcc::dump_manager::register_pass.
2197 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
2199 PR rtl-optimization/87466
2200 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
2201 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
2202 * doc/tm.texi: Regenerate.
2203 * ira-lives.c (process_bb_node_lives): Use the new target hook.
2204 * lra-lives.c (process_bb_lives): Likewise.
2205 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
2208 2018-10-04 Tamar Christina <tamar.christina@arm.com>
2210 * params.c (add_params): Fix initialization.
2212 2018-10-04 Martin Liska <mliska@suse.cz>
2214 PR gcov-profile/84107
2215 * tree-profile.c (init_ic_make_global_vars):
2216 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
2217 Come up with new ic_tuple* variables. Emit
2218 __gcov_indirect_call{,_topn} variables.
2219 (gimple_gen_ic_profiler): Access the variable
2221 (gimple_gen_ic_func_profiler): Access
2222 __gcov_indirect_call.callee field.
2223 (gimple_init_gcov_profiler): Use ptr_type_node.
2224 * value-prof.c (gimple_ic): Use ptr_type_node.
2226 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2228 PR tree-optimization/85787
2229 * ipa-pure-const.c (malloc_candidate_p_1): Move most of malloc_candidate_p
2230 into this function and add support for detecting multiple phis.
2231 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
2233 2018-10-04 Martin Liska <mliska@suse.cz>
2236 * ipa-inline.c (inline_to_all_callers_1):
2237 Call ultimate_alias_target for node being inlined.
2239 2018-10-03 Jeff Law <law@redhat.com>
2241 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
2243 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
2244 * tree.h (get_typenode_from_name): Prototype.
2246 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
2248 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
2249 Change operand 2 predicate to nonimmediate_operand.
2250 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2252 2018-10-03 Martin Sebor <msebor@redhat.com>
2253 Jeff Law <law@redhat.com>
2255 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
2257 (get_string_length): Detect unterminated arrays.
2258 (format_string): Same.
2259 (format_directive): Warn about unterminated arrays.
2260 (handle_gimple_call): Mark statements with no_warning as needed.
2262 2018-10-03 Jim Wilson <jimw@sifive.com>
2264 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
2265 also define __riscv_abi_rve. Delete trailing white space.
2267 2018-10-03 Paul Koning <ni1d@arrl.net>
2269 Enable LRA register allocator for PDP11.
2270 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
2273 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
2274 * config/pdp11/pdp11.opt (-mlra): New option.
2275 * doc/invoke.texi (PDP-11 Options): Document -mlra.
2277 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
2279 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
2280 (*<absneg:code>extend<mode>xf2): Ditto.
2282 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
2284 PR tree-optimization/87415
2285 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
2288 2018-10-02 Jeff Law <law@redhat.com>
2290 * gimple-fold.c (get_range_strlen): Only set *nonstr when
2291 an unterminated string is discovered. Bubble up range
2292 even for unterminated strings.
2293 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
2294 indicates the string was not terminated via NONSTR.
2296 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
2298 * tree-vrp.c (extract_range_from_unary_expr): Special case all
2299 pointer conversions.
2300 Do not do anything special for anti-ranges.
2302 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
2304 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
2305 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
2307 2018-10-03 Martin Liska <mliska@suse.cz>
2309 PR gcov-profile/86109
2310 * coverage.c (coverage_begin_function): Do not
2311 mark lambdas as artificial.
2312 * tree-core.h (struct GTY): Remove tm_clone_flag
2313 and introduce new lambda_function.
2314 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
2316 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
2319 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
2320 P8_VECTOR and VSX are enabled.
2322 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
2324 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
2325 0x3907 as CPU model number.
2327 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
2329 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
2330 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
2331 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
2332 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
2333 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
2334 * config/s390/s390.md: Likewise. Rename also the cpu attribute
2335 value from arch12 to z14.
2337 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
2339 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
2341 (isinf<mode>2): Ditto.
2343 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
2345 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
2346 before emitting fxam. Perform calculations in XFmode.
2348 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
2350 * match.pd (((X /[ex] A) +- B) * A): New transformation.
2352 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
2355 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
2356 * tree.c (signed_or_unsigned_type_for): Handle complex.
2358 2018-10-02 Jeff Law <law@redhat.com>
2360 * gimple-fold.c (get_range_strlen): Remove dead code.
2362 2018-10-02 Martin Sebor <msebor@redhat.com>
2363 Jeff Law <law@redhat.com>
2365 * builtins.c (unterminated_array): Add new arguments.
2366 If argument is not terminated, bubble up size and exact
2368 (expand_builtin_strnlen): Detect, avoid expanding
2369 and diagnose unterminated arrays.
2370 (c_strlen): Fill in offset of start of unterminated strings.
2371 * builtins.h (unterminated_array): Update prototype.
2373 2018-10-02 Richard Biener <rguenther@suse.de>
2375 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
2376 of haddv4df, first reduce to SSE width and exploit the fact
2377 that we only need element zero with the reduction result.
2378 (reduc_plus_scal_v2df): Likewise.
2380 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
2382 * dojump.h (do_jump): Delete.
2383 (do_jump_1): Likewise.
2384 (split_comparison): Move around.
2385 * dojump.c (do_jump): Make static.
2386 (do_jump_1): Likewise.
2387 (jumpifnot): Move around.
2388 (jumpifnot_1): Likewise.
2390 (jumpif_1): Likewise.
2391 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
2393 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
2395 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
2396 insns in the delay slot and add_insn_after for the jump insn.
2398 2018-10-02 Richard Biener <rguenther@suse.de>
2400 * tree-inline.c (expand_call_inline): Use the location of
2401 the callee declaration for the inline-entry marker.
2402 * final.c (notice_source_line): Remove special-casing of
2403 NOTE_INSN_INLINE_ENTRY.
2405 2018-10-01 Carl Love <cel@us.ibm.com>
2408 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
2409 (__builtin_mtfsb0): New.
2410 (__builtin_mtfsb1): New.
2411 ( __builtin_set_fpscr_rn): New.
2412 (__builtin_set_fpscr_drn): New.
2413 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
2414 (rs6000_expand_set_fpscr_rn_builtin): Add.
2415 (rs6000_expand_set_fpscr_drn_builtin): Add.
2416 (rs6000_expand_builtin): Add case statement entries for
2417 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
2418 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
2419 RS6000_BUILTIN_MFFSL.
2420 (rs6000_init_builtins): Add ftype initialization and def_builtin
2421 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
2422 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
2423 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
2424 rs6000_mffscdrn): Add define_insn.
2425 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
2426 * doc/extend.texi: Add documentation for the builtins.
2428 2018-10-01 Richard Biener <rguenther@suse.de>
2430 PR tree-optimization/87465
2431 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
2432 causing branch miscounts.
2434 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2436 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
2437 aarch64_option_default_param): New.
2438 (params.h): Include.
2439 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
2440 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
2441 stack-clash protection validation code.
2443 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2445 * params.c (validate_param): New.
2446 (add_params): Use it.
2447 (set_param_value): Refactor param validation into validate_param.
2448 (diagnostic.h): Include.
2449 * diagnostic.h (diagnostic_ready_p): New.
2451 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2453 * params.c (set_param_value):
2454 Add index of parameter being validated.
2455 * common/common-target.def (option_validate_param): New.
2456 * common/common-targhooks.h (default_option_validate_param): New.
2457 * common/common-targhooks.c (default_option_validate_param): New.
2458 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
2459 * doc/tm.texi: Regenerate.
2461 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2464 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2465 Add validation for stack-clash parameters and set defaults.
2467 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2470 * configure.ac: Add stack-clash-protection-guard-size.
2471 * doc/install.texi: Document it.
2472 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
2473 * params.def: Update comment for guard-size.
2474 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
2475 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
2476 * configure: Regenerate.
2478 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2481 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
2482 STACK_DYNAMIC_OFFSET): New.
2483 * config/aarch64/aarch64.c (aarch64_layout_frame):
2484 Update outgoing args size.
2485 (aarch64_stack_clash_protection_alloca_probe_range,
2486 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
2488 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2491 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
2493 * target.def (stack_clash_protection_alloca_probe_range): New.
2494 (stack_clash_protection_final_dynamic_probe): Remove.
2495 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
2496 (default_stack_clash_protection_final_dynamic_probe): Remove.
2497 * targhooks.c: Likewise.
2498 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
2499 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
2500 * doc/tm.texi: Regenerate.
2502 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2505 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
2506 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
2507 aarch64_clamp_to_uimm12_shift): New.
2508 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
2509 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
2511 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2514 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
2516 2018-10-01 Jeff Law <law@redhat.com>
2517 Richard Sandiford <richard.sandiford@linaro.org>
2518 Tamar Christina <tamar.christina@arm.com>
2521 * config/aarch64/aarch64.md
2522 (probe_stack_range): Add k (SP) constraint.
2523 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
2524 STACK_CLASH_MAX_UNROLL_PAGES): New.
2525 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
2526 stack probes for stack clash.
2527 (aarch64_allocate_and_probe_stack_space): New.
2528 (aarch64_expand_prologue): Use it.
2529 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
2530 (aarch64_sub_sp): Add emit_move_imm optional param.
2532 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
2534 PR tree-optimization/87261
2535 * match.pd: Remove trailing whitespace.
2536 Add (x & y) | ~(x | y) -> ~(x ^ y),
2537 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
2539 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
2541 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
2543 (ashlsi3_insn): Update instruction constraints.
2544 (ashrsi3_insn): Likewise.
2545 (rotrsi3): Likewise.
2546 (add_shift): Likewise.
2547 * config/arc/constraints.md (Csz): New 32 bit constraint. It
2548 avoids placing in the limm field small constants which, otherwise,
2549 could end into a small instruction.
2551 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
2553 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
2554 destination register is not odd-even.
2555 (umaddsidi4_split): Likewise.
2557 2018-10-01 Richard Biener <rguenther@suse.de>
2559 * tree-inline.c (expand_call_inline): Store origin of fn
2560 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
2561 * tree.c (block_ultimate_origin): Simplify and do some
2564 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
2566 * config/i386/mmx.md (EMMS): New int iterator.
2567 (emms): New int attribute.
2568 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
2569 EMMS int iterator. Explicitly declare clobbers.
2570 (mmx_emms): Remove expander.
2572 * config/i386/predicates.md (emms_operation): Remove predicate.
2573 (vzeroall_pattern): New predicate.
2574 (vzeroupper_pattern): Rename from vzeroupper_operation.
2575 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
2576 vzeroupper_pattern and vzeroall_pattern predicates.
2578 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
2580 PR rtl-optimization/86939
2581 * ira-lives.c (make_hard_regno_born): Rename from this...
2582 (make_hard_regno_live): ... to this. Remove update to conflict
2583 information. Update function comment.
2584 (make_hard_regno_dead): Add conflict information update. Update
2586 (make_object_born): Rename from this...
2587 (make_object_live): ... to this. Remove update to conflict information.
2588 Update function comment.
2589 (make_object_dead): Add conflict information update. Update function
2591 (mark_pseudo_regno_live): Call make_object_live.
2592 (mark_pseudo_regno_subword_live): Likewise.
2593 (mark_hard_reg_dead): Update function comment.
2594 (mark_hard_reg_live): Call make_hard_regno_live.
2595 (process_bb_node_lives): Likewise.
2596 * lra-lives.c (make_hard_regno_born): Rename from this...
2597 (make_hard_regno_live): ... to this. Remove update to conflict
2598 information. Remove now uneeded check_pic_pseudo_p argument.
2599 Update function comment.
2600 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
2601 to conflict information. Update function comment.
2602 (mark_pseudo_live): Remove update to conflict information. Update
2604 (mark_pseudo_dead): Add conflict information update.
2605 (mark_regno_live): Call make_hard_regno_live.
2606 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
2607 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
2609 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
2612 * config/i386/i386.c (construct_container): Use TImode for
2613 BLKmode values in 2 integer registers.
2615 2018-09-29 Jeff Law <law@redhat.com>
2617 * builtins.c (unterminated_array): Pass in c_strlen_data * to
2618 c_strlen rather than just a tree *.
2619 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
2620 Update recursive calls appropriately. If caller did not provide a
2621 suitable data pointer, create a local one. When a non-terminated
2622 string is discovered, bubble up information about the string via the
2623 c_strlen_data object.
2624 * builtins.h (c_strlen): Update prototype.
2625 (c_strlen_data): New structure.
2626 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
2627 For a type 2 call, if c_strlen indicates a non-terminated string
2628 use the length of the non-terminated string.
2629 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
2631 2018-09-29 Jakub Jelinek <jakub@redhat.com>
2634 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
2635 __m512d type for __A argument rather than __m512.
2637 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
2639 * match.pd (simple_comparison): Don't optimize if either operand is
2640 a function pointer when target needs function pointer canonicalization.
2642 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
2644 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
2645 power5 .. power9 to remove indirection.
2646 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
2647 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
2648 ASM_CPU_476_SPEC): Delete.
2649 (ASM_CPU_SPEC): Adjust.
2650 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
2651 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
2653 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
2655 * config.in: Delete HAVE_AS_DCI.
2656 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
2657 * config/rs6000/rs6000.h: Ditto.
2658 * configure.ac: Delete HAVE_AS_DCI.
2659 * configure: Regenerate.
2661 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
2663 * config.in (HAVE_AS_LWSYNC): Delete.
2664 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
2665 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
2667 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
2668 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
2670 * configure.ac: Delete HAVE_AS_LWSYNC.
2671 * configure: Regenerate.
2673 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
2674 Pierre-Marie de Rodat <derodat@adacore.com>
2676 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
2677 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
2678 (cgraph_node::create_thunk): Add indirect_offset parameter.
2679 (thunk_adjust): Likewise.
2680 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
2681 and initialize the corresponding field with it.
2682 (cgraph_node::dump): Dump indirect_offset field.
2683 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
2684 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
2685 (thunk_adjust): Add indirect_offset parameter and deal with it.
2686 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
2687 pass it to thunk_adjust. Do not call the target hook if it's non-zero
2688 or if the thunk is external or local. Fix formatting. Do not chain
2689 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
2690 if any, in the GIMPLE representation.
2691 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
2692 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
2693 (input_node): Read indirect_offset field.
2694 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
2695 call to thunk_adjust.
2696 * tree-nested.c (struct nesting_info): Add thunk_p field.
2697 (create_nesting_tree): Set it.
2698 (convert_all_function_calls): Copy static chain from targets to thunks.
2699 (finalize_nesting_tree_1): Return early for thunks.
2700 (unnest_nesting_tree_1): Do not finalize thunks.
2701 (gimplify_all_functions): Do not gimplify thunks.
2703 2018-09-28 David Malcolm <dmalcolm@redhat.com>
2705 * opt-suggestions.c (option_proposer::build_option_suggestions):
2706 Release "option_values".
2708 2018-09-28 David Malcolm <dmalcolm@redhat.com>
2710 * coverage.c (get_coverage_counts): Convert problem-reporting dump
2711 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
2712 * dumpfile.c (kind_as_string): New function.
2713 (dump_loc): Rather than a hardcoded prefix of "note: ", use
2714 kind_as_string to vary the prefix based on dump_kind.
2715 (selftest::test_capture_of_dump_calls): Update for above.
2717 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
2719 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
2720 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
2722 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
2724 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
2725 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
2726 INVALID_REGNUM instead of FPSR_REG.
2727 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
2728 * config/i386/i386.md: Update comment of FP compares.
2729 (fldenv): Do not clobber FPSR_REG.
2731 2018-09-28 Richard Biener <rguenther@suse.de>
2733 * tree.h (BLOCK_ORIGIN): New.
2734 * omp-expand.c (grid_expand_target_grid_body): Assign
2735 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
2736 * tree-inline.c (remap_block): Likewise.
2737 * auto-profile.c (get_function_decl_from_block): Simplify
2738 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
2739 * langhooks.c (lhd_print_error_function): Likewise.
2740 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
2742 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
2743 * tree.c (block_nonartificial_location): Likewise.
2744 (block_ultimate_origin): Likewise.
2745 * tree-pretty-print.c (percent_K_format): Likewise. Remove
2746 no longer needed LTO case.
2748 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
2749 Jan Hubicka <jh@suse.cz>
2750 Martin Jambor <mjambor@suse.cz>
2752 * simplify-rtx.c (simplify_merge_mask): New function.
2753 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
2754 same masks are used in op1 or op2.
2755 (test_vec_merge): New function.
2756 (test_vector_ops): Call test_vec_merge.
2758 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
2760 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
2761 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
2762 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
2763 (cypress_costs): Set it.
2764 (supersparc_costs): Likewise.
2765 (hypersparc_costs): Likewise.
2766 (leon_cost): Likewise.
2767 (leon3_costs): Likewise.
2768 (sparclet_costs): Likewise.
2769 (ultrasparc_costs): Likewise.
2770 (ultrasparc_costs): Likewise.
2771 (niagara_costs): Likewise.
2772 (niagara2_costs): Likewise.
2773 (niagara3_costs): Likewise.
2774 (niagara4_costs): Likewise.
2775 (niagara7_costs): Likewise.
2776 (m8_costs): Likewise.
2777 (TARGET_CAN_FOLLOW_JUMP): Define.
2778 (pass_work_around_errata::gate): Minor tweak.
2779 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
2780 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
2781 Automaitcally clear MASK_FSMULD mask for V7 processors.
2782 (sparc_can_follow_jump): New static function.
2783 (output_ubranch): Deal with CROSSING_JUMP_P.
2784 (sparc_use_sched_lookahead): Rewrite using switch statement.
2785 (sparc_issue_rate): Reorder.
2786 (sparc_branch_cost): New function.
2788 2018-09-27 Martin Sebor <msebor@redhat.com>
2790 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
2791 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
2792 (int_fits_type_p): Same.
2794 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
2796 * config/i386/i386.md (FPCR_REG): Remove.
2797 (UNSPEC_FLDCW): Remove.
2798 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
2799 (x86_fldcw_1): Remove insn pattern.
2800 (fnstenv): Do not clobber FPCR_REG.
2802 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
2803 (CALL_USED_REGISTERS): Ditto.
2804 (REG_ALLOC_ORDER): Ditto.
2805 (REG_CLASS_CONTENTS): Ditto.
2806 (HI_REGISTER_NAMES): Ditto.
2807 (ADDITIONAL_REGISTER_NAMES): Use defines instead
2808 of numerical constants.
2809 * config/i386/i386.c (regclass_map): Remove fpsr register.
2810 (dbx_register_map): Ditto.
2811 (dbx64_register_map): Ditto.
2812 (svr4_dbx_register_map): Ditto.
2813 (print_reg): Do not handle FPCR_REG.
2815 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
2818 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
2819 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
2820 Delete, always treat as true.
2821 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
2822 Ditto. Simplify remaining code.
2823 * config/powerpcspe/powerpcspe.h: Ditto.
2824 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
2825 Simplify remaining code.
2826 (rs6000_expand_builtin): Ditto.
2827 * config/rs6000/rs6000.h: Ditto.
2828 * configure.ac: Ditto.
2829 * configure: Regenerate.
2831 2018-09-27 Martin Liska <mliska@suse.cz>
2833 * coverage.c (get_coverage_counts): Revert the formatting
2834 of missing profile opt info.
2836 2018-09-27 Richard Biener <rguenther@suse.de>
2840 * dwarf2out.c (set_block_origin_self): Do not mark outermost
2841 block as we do not output that.
2842 (gen_inlined_subroutine_die): Elide the originally outermost
2843 block, matching what we do for concrete instances.
2844 (decls_for_scope): Add parameter specifying whether to recurse
2847 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
2848 Tom de Vries <tom@codesourcery.com>
2852 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
2853 STORE_FLAG_VALUE == 1.
2855 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
2857 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
2858 constant definitions.
2859 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
2860 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
2861 ("speculation_barrier"): New expander definition.
2863 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
2865 PR gcov-profile/86957
2866 * common.opt: New warning option -Wmissing-profile.
2867 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
2868 * doc/invoke.texi: Document -Wmissing-profile.
2870 2018-09-26 Jim Wilson <jimw@sifive.com>
2872 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
2873 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
2875 2018-09-26 Martin Sebor <msebor@redhat.com>
2877 * tree.c (zerop): Change return type to bool.
2878 (integer_zerop, integer_onep, integer_each_onep): Same.
2879 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
2880 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
2881 (real_onep, real_minus_onep, chain_index): Same.
2882 (print_type_hash_statistics, type_list_equal): Same.
2883 * tree.h (zerop): Same.
2884 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
2885 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
2886 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
2887 (real_onep, real_minus_onep, chain_index): Same.
2888 (print_type_hash_statistics, type_list_equal): Same.
2890 2018-09-26 Jim Wilson <jimw@sifive.com>
2892 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
2894 2018-09-26 Jakub Jelinek <jakub@redhat.com>
2897 * config/i386/i386.c: Include debug.h and dwarf2out.h.
2898 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
2901 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
2903 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
2905 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
2907 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
2908 and FP_SECOND_SSE_REGS.
2909 (REG_CLASS_NAMES): Ditto.
2910 (REG_CLASS_CONTENTS): Ditto.
2911 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
2912 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
2913 (ix86_preferred_output_reload_class): Ditto.
2914 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
2915 clobber constraint to "=&f".
2916 (fix_truncdi_i387): Ditto.
2917 (lrintxfdi2): Ditto.
2918 (fistdi2_<rounding>): Ditto.
2919 (fpremxf4_i387): Change "=u" constraint to "=f".
2920 (fprem1xf4_i387): Ditto.
2922 (fptanxf4_i387): Ditto.
2923 (fxtractxf3_i387): Ditto.
2924 (fscalexf4_i387): Ditto.
2925 (atan2xf3): Change "u" constraint to "f".
2926 (fyl2xxf3_i387): Ditto.
2927 (fyl2xp1xf3_i387): Ditto.
2929 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
2932 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
2933 for removed I387_MASK_PM entity.
2936 2018-09-26 Jeff Law <law@redhat.com>
2938 2018-09-26 Alexey Neyman <stilor@att.net>
2940 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
2941 headers are no longer pulled in by <isl/val.h>.
2943 2018-09-26 Richard Biener <rguenther@suse.de>
2946 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
2947 or concrete instance DIE to the tree. Create abstract origin
2948 attributes also for concrete instances.
2950 2018-09-26 Alexey Neyman <stilor@att.net>
2952 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
2953 headers are no longer pulled in by <isl/val.h>.
2955 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
2957 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
2958 Use new helper functions.
2959 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
2960 Use new helper functions.
2961 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
2962 aarch_mm_needs_release): New declarations.
2963 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
2964 aarch_mm_needs_release): New.
2966 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
2968 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
2969 (arm32_output_mi_thunk): Deal with long calls.
2971 2018-09-26 Richard Biener <rguenther@suse.de>
2975 * tree-inline.c (expand_call_inline): When the location
2976 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
2977 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
2978 the inserted BLOCK to make inlined_function_outer_scope_p
2980 * dwarf2out.c (add_call_src_coords_attributes): Do not add
2981 coords for reserved locations.
2983 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
2985 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
2986 (*call_indirect_nonlocal_sysv<mode>): Ditto.
2987 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
2988 (*sibcall_nonlocal_sysv<mode>): Ditto.
2989 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
2990 (<bd>_<mode>): Ditto.
2991 (<bd>tf_<mode>): Ditto.
2993 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
2995 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
2996 control string as a list of templates instead of as C code.
2997 (*altivec_movti): Ditto.
2998 * config/rs6000/darwin.md (movdf_low_di): Ditto.
3000 2018-09-25 Jim Wilson <jimw@sifive.com>
3002 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
3003 when target symbol is weak.
3005 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
3008 * builtins.c (unterminated_array): Simplify.
3009 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
3010 where pointer arithmetic is safe.
3012 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3015 * config/rs6000/altivec.md (altivec_vspltb): Use
3016 const_0_to_15_operand instead of u5bit_cint_operand.
3017 (*altivec_vspltb_internal): Ditto.
3018 (altivec_vspltb_direct): Ditto.
3019 (altivec_vsplth): Use const_0_to_7_operand instead of
3021 (*altivec_vsplth_internal): Ditto.
3022 (altivec_vsplth_direct): Ditto.
3023 (altivec_vspltw): Use const_0_to_3_operand instead of
3025 (*altivec_vspltw_internal): Ditto.
3026 (altivec_vspltw_direct): Ditto.
3027 (altivec_vspltsf): Ditto.
3028 (*altivec_vspltsf_internal): Ditto.
3029 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
3030 various splats with the proper size immediate. Reorder the various
3031 cases by ascending size of immediate, and put all such together.
3033 2018-09-25 Richard Biener <rguenther@suse.de>
3036 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
3038 (maybe_create_die_with_external_ref): Do not create
3039 DW_TAG_imported_unit here.
3040 (add_abstract_origin_attribute): Handle external BLOCK refs.
3041 (dwarf2out_abstract_function): Simplify LTO case.
3042 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
3043 rather than using maybe_create_die_with_external_ref.
3045 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
3048 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
3049 (frndintxf2_mask_pm_i387): Ditto.
3050 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
3051 Enable for !flag_trapping_math.
3052 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
3053 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
3054 Change operand 1 predicate to nonimmediate_operand.
3055 (attr "i387_cw"): Remove mask_pm.
3056 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
3057 (enum ix86_entity): Remove I387_MASK_PM.
3058 * config/i386/i386.c (ix86_i387_mode_needed): Do not
3059 handle I387_MASK_PM.
3060 (ix86_mode_needed): Ditto.
3061 (ix86_mode_after): Ditto.
3062 (ix86_mode_entry): Ditto.
3063 (ix86_mode_exit): Ditto.
3064 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
3066 2018-09-25 Jakub Jelinek <jakub@redhat.com>
3068 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
3069 to_update_switch_stmts to vNULL instead of calling create on them
3072 2018-09-25 Richard Biener <rguenther@suse.de>
3074 PR tree-optimization/87402
3075 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
3076 (visit_phi): Re-instantiate handling of supposed to be VARYING
3077 but non-VARYING backedge value.
3079 2018-09-25 Richard Biener <rguenther@suse.de>
3082 * dwarf2out.c (struct sym_off_pair): New.
3083 (external_die_map): New global.
3084 (lookup_decl_die): When in LTO create DIEs lazily from the
3086 (lookup_block_die): New function, create DIEs lazily in LTO.
3087 (equate_block_to_die): New function.
3088 (dwarf2out_die_ref_for_decl): During WPA get the association
3089 from the external DIE map.
3090 (dwarf2out_register_external_die): Record mapping into the
3092 (maybe_create_die_with_external_ref): New function split out from
3093 DIE generation part of old dwarf2out_register_external_die.
3094 (add_abstract_origin_attribute): Do not return the DIE. When
3095 in LTO reference externals directly.
3096 (dwarf2out_abstract_function): When in LTO ignore calls for
3097 decls with external DIEs (already present abstract instances).
3098 (gen_call_site_die): Adjust.
3099 (add_high_low_attributes): Likewise.
3100 (gen_lexical_block_die): Likewise.
3101 (gen_inlined_subroutine_die): Likewie.
3102 (gen_block_die): Likewise.
3103 (dwarf2out_inline_entry): Likewise.
3104 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
3107 2018-09-25 Martin Liska <mliska@suse.cz>
3109 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
3110 integers and not by a float value.
3112 2018-09-25 Martin Liska <mliska@suse.cz>
3115 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
3117 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
3118 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
3119 * godump.c (go_format_type): Remove extra parenthesis.
3121 2018-09-25 Martin Liska <mliska@suse.cz>
3123 * alias.c (set_dest_equal_p): Remove unused function.
3124 * config/i386/i386.c (def_builtin_pure2): Likewise.
3125 * diagnostic-show-locus.c (class layout): Remove
3127 (layout::layout): Likewise here.
3128 * dump-context.h (class temp_dump_context): Likewise.
3129 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
3130 (add_AT_loclistsptr): Likewise.
3131 (add_AT_offset): Likewise.
3132 (get_AT_hi_pc): Likewise.
3133 (is_comdat_die): Likewise.
3134 (type_is_enum): Likewise.
3135 (ceiling): Likewise.
3136 (add_AT_vms_delta): Likewise.
3137 (is_class_die): Likewise.
3138 * edit-context.c (class line_event): Remove unused field.
3139 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
3141 * ipa-cp.c (ipa_get_vr_lat): Likewise.
3142 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
3143 (ok_for_base_p_nonstrict): Likewise.
3144 * tree-chrec.c (is_not_constant_evolution): Likewise.
3145 (chrec_fold_poly_cst): Likewise.
3146 * tree-if-conv.c (has_pred_critical_p): Likewise.
3147 * tree-ssa-coalesce.c (print_exprs): Likewise.
3148 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
3149 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
3150 * tree-vrp.c (value_ranges_intersect_p): Likewise.
3151 (value_range_nonnegative_p): Likewise.
3153 2018-09-25 Martin Liska <mliska@suse.cz>
3155 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
3156 Do not handle "GNU Pascal".
3157 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
3159 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
3160 from documentation. Likewise.
3161 * dbxout.c (dbxout_range_type): Likewise.
3162 * doc/cpp.texi: Likewise.
3163 * doc/extend.texi: Likewise.
3164 * doc/frontends.texi: Likewise.
3165 * doc/invoke.texi: Remove Pascal entry.
3166 * tree.def (CLEANUP_POINT_EXPR): Likewise.
3167 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
3169 2018-09-25 Martin Liska <mliska@suse.cz>
3172 * doc/invoke.texi: Document all parameters and remove default
3175 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
3178 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
3179 contains HOST_WIDE_INTs when computing its size.
3181 2018-09-24 Jim Wilson <jimw@sifive.com>
3184 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
3186 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
3188 2018-09-24 Andrew Pinski <apinski@marvell.com>
3190 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
3191 access prev before checking it for NULLness in the
3192 AARCH64_FUSE_CMP_BRANCH case.
3194 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
3197 * config/i386/i386.c (rest_of_insert_endbranch): Set
3198 endbr_queued_at_entrance to true and don't insert ENDBR if
3199 x86_function_profiler will be called.
3200 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
3202 * config/i386/i386.h (machine_function): Add
3203 endbr_queued_at_entrance.
3205 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
3207 * genattrtab.c (mk_attr_alt): Use alternative_mask.
3208 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
3210 (check_attr_test): Use alternative_mask.
3211 (get_attr_value): Likewise.
3212 (compute_alternative_mask): Use alternative_mask and XWINT.
3213 (make_alternative_compare): Use alternative_mask.
3214 (attr_alt_subset_p): Use XWINT.
3215 (attr_alt_subset_of_compl_p): Likewise.
3216 (attr_alt_intersection): Use alternative_mask and XWINT.
3217 (attr_alt_union): Likewise.
3218 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
3219 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
3220 (simplify_test_exp): Use alternative_mask and XWINT.
3221 (write_test_expr): Use alternative_mask and XWINT, adjust bit
3222 number calculation to support 64 bits. Generate code that
3223 checks 64-bit masks.
3224 (main): Use alternative_mask.
3225 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
3227 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
3230 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
3231 RETURN+USE when returning via %r14.
3233 2018-09-24 Martin Liska <mliska@suse.cz>
3235 * gcov.c (output_lines): Print colorization legend
3236 for both flag_use_colors and flag_use_hotness_colors.
3239 2018-09-24 Martin Liska <mliska@suse.cz>
3241 * coverage.c (get_coverage_counts): Use warning_at
3242 with current_function_decl location. Use %qD in warning
3245 2018-09-24 Martin Liska <mliska@suse.cz>
3247 * memory-block.h (memory_block_pool::release): Annotate with
3248 valgrind that the memory is not accessible.
3250 2018-09-24 Martin Liska <mliska@suse.cz>
3253 * asan.c: Make asan_handled_variables extern.
3255 * cfgexpand.c (expand_stack_vars): Make sure
3256 a representative is unpoison if another
3257 variable in the partition is handled by
3258 use-after-scope sanitization.
3260 2018-09-24 Richard Biener <rguenther@suse.de>
3262 PR tree-optimization/63155
3263 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
3264 the worklist when the edge of the respective argument isn't
3267 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3269 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
3270 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
3271 (MASK_CLASS_P): Update for rename.
3272 (MAYBE_MASK_CLASS_P): Ditto.
3273 (REG_CLASS_NAMES): Update.
3274 (REG_CLASS_CONTENT): Update.
3275 * config/i386/i386.c (regclass_map): Update for MASK_REG
3276 and ALL_MASK_REGS rename.
3277 * config/i386/constraints.md (Yk): Update for rename.
3280 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3282 * config/i386/i386.h (enum reg_class): Remove
3283 EVEX_SSE_REGS and MOD4_SSE_REGS.
3284 (REG_CLASS_NAMES): Update.
3285 (REG_CLASS_CONTENT): Update.
3286 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
3287 registers as ALL_SSE_REGS.
3288 (ix86_additional_allocno_class_p): Remove.
3289 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
3290 (ix86_register_priority): Lower priority of EVEX SSE registers.
3291 Use IN_RANGE macro where appropriate.
3292 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
3293 AVX-5124VNNIW checks.
3294 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
3295 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
3296 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
3297 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
3298 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
3299 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
3300 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
3301 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
3302 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
3303 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
3304 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
3305 * config/i386/constraints.md (Yh): Remove.
3307 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3309 * config/i386/i386.c (regclass_map): Declare integer REX registers
3312 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
3314 * doc/service.texi (Service): Switch the fsf.org link to https.
3316 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
3319 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3320 Define to speculation_safe_value_not_needed.
3322 2018-09-21 Florian Weimer <fweimer@redhat.com>
3325 * doc/extend.texi (Common Function Attributes): Mention that
3326 noreturn suppresses tail call optimization.
3328 2018-09-21 Jeff Law <law@redhat.com>
3330 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
3331 vr_values::cleanup_edges_and_switches.
3332 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
3334 (identify_jump_threads): Remove EDGE_IGNORE handling.
3335 (execute_vrp): Move handling of to_remove_edges and
3336 to_update_switch_stmts into vr_values class member functions.
3337 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
3338 (to_update_switch_stmts): Likewise.
3339 * vr-values.c: Include cfghooks.h.
3340 (vr_values::vr_values): Initialize to_remove_edges and
3341 to_update_switch_stmts.
3342 (vr_values::~vr_values): Verify to_remove_edges and
3343 to_update_switch_stmts are empty.
3344 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
3345 (vr_values::cleanup_edges_and_switches): New member function.
3346 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
3347 function. Add new data members.
3349 2018-09-21 David Malcolm <dmalcolm@redhat.com>
3351 PR tree-optimization/87309
3352 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
3353 calls with pflags and alt_flags.
3354 (selftest::test_capture_of_dump_calls): Add test of interaction of
3355 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
3357 2018-09-21 Olivier Hainque <hainque@adacore.com>
3359 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
3361 2018-09-21 Olivier Hainque <hainque@adacore.com>
3363 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
3365 2018-09-21 Olivier Hainque <hainque@adacore.com>
3367 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
3368 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
3369 (VXWORKS_LIBS_RTP): Minor reordering.
3371 2018-09-21 Olivier Hainque <hainque@adacore.com>
3373 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
3374 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
3376 2018-09-21 Olivier Hainque <hainque@adacore.com>
3378 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
3379 (PTRDIFF_TYPE): Likewise.
3381 2018-09-21 Olivier Hainque <hainque@adacore.com>
3383 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
3384 triplet, similar to support for VxWorks7.
3385 * config/vxworks-dummy.h: Provide a default definition
3386 of TARGET_VXWORKS64 to 0.
3388 2018-09-21 Olivier Hainque <hainque@adacore.com>
3390 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
3391 * config/vxworks-dummy.h: here.
3393 2018-09-21 Olivier Hainque <hainque@adacore.com>
3395 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
3397 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
3398 Bo Zhou <zbo.zhou@hisilicon.com>
3400 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
3401 * config/aarch64/aarch64-tune.md: Regenerated.
3402 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
3403 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
3404 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
3406 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
3407 Julian Brown <julian@codesourcery.com>
3409 * builtins.c (get_builtin_sync_mem): Handle address spaces.
3411 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
3413 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
3414 if the call takes a static chain.
3416 2018-09-21 Martin Liska <mliska@suse.cz>
3418 * auto-profile.c (autofdo_source_profile::read): Do not
3420 (read_profile): Do not add working sets.
3421 (read_autofdo_file): Remove sum_all.
3422 (afdo_callsite_hot_enough_for_early_inline): Remove const
3424 * coverage.c (struct counts_entry): Remove gcov_summary.
3425 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
3426 do not support GCOV_TAG_PROGRAM_SUMMARY.
3427 (get_coverage_counts): Remove summary and expected
3429 * coverage.h (get_coverage_counts): Likewise.
3430 * doc/gcov-dump.texi: Remove -w option.
3431 * gcov-dump.c (dump_working_sets): Remove.
3432 (main): Do not support '-w' option.
3433 (print_usage): Likewise.
3434 (tag_summary): Likewise.
3435 * gcov-io.c (gcov_write_summary): Do not dump
3437 (gcov_read_summary): Likewise.
3438 (gcov_histo_index): Remove.
3439 (gcov_histogram_merge): Likewise.
3440 (compute_working_sets): Likewise.
3441 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
3443 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
3444 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
3445 (GCOV_HISTOGRAM_SIZE): Remove.
3446 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
3447 (struct gcov_summary): Simplify rapidly just
3448 to runs and sum_max fields.
3449 (gcov_histo_index): Remove.
3450 (NUM_GCOV_WORKING_SETS): Likewise.
3451 (compute_working_sets): Likewise.
3452 * gcov-tool.c (print_overlap_usage_message): Remove
3453 trailing empty line.
3454 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
3455 (output_lines): Remove program related line.
3456 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
3457 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
3459 (input_profile_summary): Do not read it.
3460 (merge_profile_summaries): And do not merge it.
3461 (input_symtab): Do not call removed function.
3462 * modulo-sched.c (sms_schedule): Do not print sum_max.
3463 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
3464 removed when histogram method was invented.
3465 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
3467 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
3469 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
3470 and dump selected value.
3471 * profile.c (add_working_set): Remove.
3472 (get_working_sets): Likewise.
3473 (find_working_set): Likewise.
3474 (get_exec_counts): Do not work with working sets.
3475 (read_profile_edge_counts): Do not inform as sum_max is removed.
3476 (compute_branch_probabilities): Likewise.
3477 (compute_value_histograms): Remove argument for call of
3478 get_coverage_counts.
3479 * profile.h: Do not make gcov_summary const.
3481 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
3483 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
3485 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
3487 PR tree-optimization/86990
3488 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
3489 Check that the entire merged store group is made of constants only for
3492 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
3494 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
3495 (VTABLE_VERIFICATION_SPEC): Likewise.
3496 (SANITIZER_EARLY_SPEC): Likewise.
3497 (SANITIZER_SPEC): Likewise.
3498 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
3499 * doc/invoke.texi (Link Options): Document -r.
3501 2018-09-20 Richard Biener <rguenther@suse.de>
3504 * gimplify.c (gimplify_expr): Retain alignment of
3505 addressable lvalue in dereference.
3507 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
3510 * configure.ac: Check for .loc is_stmt support.
3511 * configure, config.in: Rebuilt.
3512 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
3515 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3517 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
3518 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
3521 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3523 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
3525 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
3528 2018-09-20 Marek Polacek <polacek@redhat.com>
3530 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
3532 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
3534 PR tree-optimization/87288
3535 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
3536 into account when determining PEELING_FOR_NITERS.
3538 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
3540 PR tree-optimization/86877
3541 * tree-vect-loop.c (vect_analyze_loop_2): Call
3542 vect_verify_datarefs_alignment.
3544 2018-09-19 Marek Polacek <polacek@redhat.com>
3546 * doc/invoke.texi: Document -Wclass-conversion.
3548 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
3550 * config/pa/pa.c (pa_adjust_priority): Delete.
3551 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
3553 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
3554 (atomic_storehi): Likewise.
3555 (atomic_storesi): Likewise.
3556 (atomic_loaddi): Restore compare and swap exchange loop code.
3558 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
3560 PR rtl-optimization/86902
3561 * combine.c (try_combine): When changing the CC mode used, don't change
3562 an unrelated mode in other_insn to that new CC mode.
3564 2018-09-19 David Malcolm <dmalcolm@redhat.com>
3566 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
3567 with %T in place of calls to dump_generic_expr.
3568 (prune_runtime_alias_test_list): Likewise.
3569 (create_runtime_alias_checks): Likewise.
3570 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
3571 (vect_analyze_data_ref_dependence): Likewise.
3572 (vect_slp_analyze_data_ref_dependence): Likewise.
3573 (vect_record_base_alignment): Likewise. Use %G in place of call
3574 to dump_gimple_stmt.
3575 (vect_compute_data_ref_alignment): Likewise.
3576 (verify_data_ref_alignment): Likewise.
3577 (vect_find_same_alignment_drs): Likewise.
3578 (vect_analyze_group_access_1): Likewise.
3579 (vect_analyze_data_ref_accesses): Likewise.
3580 (dependence_distance_ge_vf): Likewise.
3581 (dump_lower_bound): Likewise.
3582 (vect_prune_runtime_alias_test_list): Likewise.
3583 (vect_find_stmt_data_reference): Likewise.
3584 (vect_analyze_data_refs): Likewise.
3585 (vect_create_addr_base_for_vector_ref): Likewise.
3586 (vect_create_data_ref_ptr): Likewise.
3587 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
3588 (vect_can_advance_ivs_p): Likewise.
3589 (vect_update_ivs_after_vectorizer): Likewise.
3590 (vect_gen_prolog_loop_niters): Likewise.
3591 (vect_prepare_for_masked_peels): Likewise.
3592 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
3593 (vect_determine_vectorization_factor): Likewise.
3594 (vect_is_simple_iv_evolution): Likewise.
3595 (vect_analyze_scalar_cycles_1): Likewise.
3596 (vect_analyze_loop_operations): Likewise.
3597 (report_vect_op): Likewise.
3598 (vect_is_slp_reduction): Likewise.
3599 (check_reduction_path): Likewise.
3600 (vect_is_simple_reduction): Likewise.
3601 (vect_create_epilog_for_reduction): Likewise.
3602 (vect_finalize_reduction:): Likewise.
3603 (vectorizable_induction): Likewise.
3604 (vect_transform_loop_stmt): Likewise.
3605 (vect_transform_loop): Likewise.
3606 (optimize_mask_stores): Likewise.
3607 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
3608 (vect_split_statement): Likewise.
3609 (vect_recog_over_widening_pattern): Likewise.
3610 (vect_recog_average_pattern): Likewise.
3611 (vect_determine_min_output_precision_1): Likewise.
3612 (vect_determine_precisions_from_range): Likewise.
3613 (vect_determine_precisions_from_users): Likewise.
3614 (vect_mark_pattern_stmts): Likewise.
3615 (vect_pattern_recog_1): Likewise.
3616 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
3617 (vect_record_max_nunits): Likewise.
3618 (vect_build_slp_tree_1): Likewise.
3619 (vect_build_slp_tree_2): Likewise.
3620 (vect_print_slp_tree): Likewise.
3621 (vect_analyze_slp_instance): Likewise.
3622 (vect_detect_hybrid_slp_stmts): Likewise.
3623 (vect_detect_hybrid_slp_1): Likewise.
3624 (vect_slp_analyze_operations): Likewise.
3625 (vect_slp_analyze_bb_1): Likewise.
3626 (vect_transform_slp_perm_load): Likewise.
3627 (vect_schedule_slp_instance): Likewise.
3628 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
3629 (vect_mark_stmts_to_be_vectorized): Likewise.
3630 (vect_init_vector_1): Likewise.
3631 (vect_get_vec_def_for_operand): Likewise.
3632 (vect_finish_stmt_generation_1): Likewise.
3633 (vect_check_load_store_mask): Likewise.
3634 (vectorizable_call): Likewise.
3635 (vectorizable_conversion): Likewise.
3636 (vectorizable_operation): Likewise.
3637 (vectorizable_load): Likewise.
3638 (vect_analyze_stmt): Likewise.
3639 (vect_is_simple_use): Likewise.
3640 (vect_get_vector_types_for_stmt): Likewise.
3641 (vect_get_mask_type_for_stmt): Likewise.
3642 * tree-vectorizer.c (increase_alignment): Likewise.
3644 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
3646 * doc/rtl.texi: Adjust vec_select description.
3647 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
3648 non-constant selectors.
3650 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
3652 * config/aarch64/aarch64-protos.h
3653 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
3654 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
3655 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
3656 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
3657 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
3658 (AARCH64_FL_PROFILE): Move index so flags are ordered.
3659 (AARCH64_ISA_RCPC8_4): New flag.
3660 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
3661 to aarch64_offset_9bit_signed_unscaled_p.
3662 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
3664 * config/aarch64/constraints.md (Ust): New constraint.
3665 * config/aarch64/predicates.md.
3666 (aarch64_9bit_offset_memory_operand): New predicate.
3667 (aarch64_rcpc_memory_operand): New predicate.
3669 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
3671 PR rtl-optimization/87361
3672 * rtlanal.c (nonzero_bits1): Revert accidental change.
3674 2018-09-19 Richard Biener <rguenther@suse.de>
3676 PR tree-optimization/87349
3677 PR tree-optimization/87342
3678 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
3680 2018-09-18 Marek Polacek <polacek@redhat.com>
3682 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
3683 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
3685 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
3687 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
3689 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
3691 PR rtl-optimization/86882
3692 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
3694 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
3696 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
3697 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
3699 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
3702 * doc/invoke.texi (Link Options): Fix formatting and grammar.
3704 2018-09-18 Richard Biener <rguenther@suse.de>
3707 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
3708 (compute_samebase_partition_bases): Likewise.
3709 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
3710 (gimple_can_coalesce_p): Simplify.
3712 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
3714 Handle a library implementation of ffs calling __builtin_ffs.
3715 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
3716 (mmix_init_libfuncs): New function: make __builtin_ffs expand
3719 2018-09-17 David Malcolm <dmalcolm@redhat.com>
3721 * diagnostic-show-locus.c (class layout_range): Add field
3723 (layout_range::layout_range): Add "original_idx" param and use it
3724 to initialize new field.
3725 (make_range): Use 0 for original_idx.
3726 (layout::layout): Pass in index to calls to
3727 maybe_add_location_range.
3728 (layout::maybe_add_location_range): Add param "original_idx" and
3729 pass it on to layout_range.
3730 (layout::print_any_labels): Pass on range->m_original_idx to
3732 (gcc_rich_location::add_location_if_nearby): Use 0 for
3734 * gcc-rich-location.h (text_range_label::get_text): Update for new
3736 (range_label_for_type_mismatch::get_text): Likewise.
3738 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
3740 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
3742 2018-09-17 David Malcolm <dmalcolm@redhat.com>
3744 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
3745 format_string_diagnostic_t.
3746 (fmtwarn_n): Likewise.
3747 * substring-locations.c
3748 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
3749 (format_warning_n_va): Convert to...
3750 (format_string_diagnostic_t::emit_warning_n_va): ...this.
3751 (format_warning_va): Convert to...
3752 (format_string_diagnostic_t::emit_warning_va): ...this.
3753 (format_warning_at_substring): Convert to...
3754 (format_string_diagnostic_t::emit_warning): ...this.
3755 (format_warning_at_substring_n): Convert to...
3756 (format_string_diagnostic_t::emit_warning_n): ...this.
3757 * substring-locations.h (class format_string_diagnostic_t): New
3759 (format_warning_va): Convert to
3760 format_string_diagnostic_t::emit_warning_va.
3761 (format_warning_n_va): Convert to
3762 format_string_diagnostic_t::emit_warning_n_va.
3763 (format_warning_at_substring): Convert to
3764 format_string_diagnostic_t::emit_warning.
3765 (format_warning_at_substring_n): Convert to
3766 format_string_diagnostic_t::emit_warning_n.
3768 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
3769 Bernd Schmidt <bernds_cb1@t-online.de>
3771 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
3774 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
3776 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
3777 operand 0 predicate to nonimmediate operand.
3778 (rint<mode>2_frndint): Remove insn pattern.
3779 (rint<mode>2): Change operand 1 predicate to general_operand.
3780 Extend operand 1 to XFmode and generate rintxf2 insn.
3781 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
3782 Do not use X87MODEF mode macro.
3783 (frndintxf2_<rounding>_i387): Rename from
3784 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
3785 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
3786 to XFmode and generate significandxf3 insn.
3788 2018-09-17 Richard Biener <rguenther@suse.de>
3790 PR tree-optimization/87328
3791 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
3792 visiting unexecutable backedges when not iterating.
3793 (do_rpo_vn): Mark all edges not executable even when not
3796 2018-09-17 Martin Jambor <mjambor@suse.cz>
3799 * doc/invoke.texi (Warning Options): Likewise.
3801 2018-09-17 Richard Biener <rguenther@suse.de>
3803 PR tree-optimization/87301
3804 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
3805 clean EH info from leftover copy assignments.
3807 2018-09-17 Martin Liska <mliska@suse.cz>
3809 PR gcov-profile/85871
3810 * gcov.c (output_intermediate_file): Fix out of bounds
3813 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
3815 * config/arc/arc.c: Object attributes for core4 not reflected
3817 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
3820 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
3822 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
3824 2018-09-17 Martin Liska <mliska@suse.cz>
3826 * doc/gcov.texi: Document new option --use-hotness-colors.
3827 * gcov.c (struct source_info): Declare new field.
3828 (source_info::source_info): Set default for maximum_count.
3829 (print_usage): Add new -q option.
3830 (process_args): Process it.
3831 (accumulate_line_info): Save src->maximum_count.
3832 (output_line_beginning): Make color line number if
3833 flag_use_hotness_colors is set.
3834 (output_line_details): Pass default argument value.
3835 (output_lines): Pass src->maximum_count.
3837 2018-09-17 Martin Liska <mliska@suse.cz>
3839 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
3840 Use processor_names table.
3841 * config/i386/i386.c (ix86_default_align): Use
3842 processor_cost_table for alignment values.
3843 (ix86_option_override_internal): Use processor_names.
3844 (ix86_function_specific_print): Likewise.
3845 * config/i386/i386.h (struct processor_costs):
3846 Add alignment values.
3847 (struct ptt): Remove and replace with const char *.
3848 * config/i386/x86-tune-costs.h (struct processor_costs):
3849 Declare default alignments for all costs.
3851 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
3853 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
3854 symbolics or VR_VARYING ranges for ABS_EXPR.
3855 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
3856 when range will wrap.
3858 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
3861 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
3862 before and after a JUMP_TABLE_DATA.
3864 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
3867 * dojump.c (do_compare_and_jump): Canonicalize function pointers
3868 when one operand is a function pointer. Use POINTER_TYPE_P and
3869 FUNC_OR_METHOD_TYPE_P.
3870 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
3871 * fold-const.c (build_range_check): Likewise.
3872 * match.pd (simple_comparison): Likewise.
3874 2018-09-14 David Malcolm <dmalcolm@redhat.com>
3877 * spellcheck.c (get_edit_distance_cutoff): New function.
3878 (selftest::test_edit_distance_unit_test_oneway): Rename to...
3879 (selftest::test_get_edit_distance_one_way): ...this.
3880 (selftest::test_get_edit_distance_unit): Rename to...
3881 (selftest::test_get_edit_distance_both_ways): ...this.
3882 (selftest::test_edit_distances): Move tests to this new function,
3883 and test some more pairs of strings. Update for above renaming.
3884 (selftest::get_old_cutoff): New function.
3885 (selftest::test_get_edit_distance_cutoff): New function.
3886 (selftest::assert_suggested_for): New function.
3887 (ASSERT_SUGGESTED_FOR): New macro.
3888 (selftest::assert_not_suggested_for): New function.
3889 (ASSERT_NOT_SUGGESTED_FOR): New macro.
3890 (selftest::test_suggestions): New function.
3891 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
3892 tests to selftest::test_edit_distances and call it. Add calls to
3893 selftest::test_get_edit_distance_cutoff and
3894 selftest::test_suggestions.
3895 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
3896 (best_match::consider): Replace hard-coded cutoff calculation with
3898 (best_match::get_cutoff): New declaration.
3899 (best_match::get_best_meaningful_candidate): Likewise.
3901 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3903 * builtins.c (fold_builtin_strlen): Remove TODO comment.
3905 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3908 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
3910 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
3911 terminated string literal.
3913 2018-09-14 Martin Sebor <msebor@redhat.com>
3915 * builtins.c (unterminated_array): Handle ARRAY_REF.
3916 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
3917 * builtins.h (unterminated_array): Declare extern.
3918 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
3920 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
3923 2018-09-14 Martin Sebor <msebor@redhat.com>
3924 Jeff Law <law@redhat.com>
3926 * builtins.c (unterminated_array): New.
3927 (expand_builtin_strcpy): Adjust.
3928 (expand_builtin_strcpy_args): Detect unterminated arrays.
3929 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
3930 unterminated arrays.
3931 * gimple-fold.h (get_maxval_strlen): Add argument.
3932 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
3934 * gimple-fold.c (get_range_strlen): Add argument.
3935 (get_maxval_strlen): Adjust.
3936 * gimple-fold.h (get_range_strlen): Add argument.
3938 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
3940 * config/i386/movdirintrin.h: Fix copyright year.
3942 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
3944 * reg-stack.c: Include regs.h.
3945 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
3946 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
3947 FIRST_STACK_REG, not DFmode.
3948 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
3949 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
3950 (change stack): Default register mode to the reg_raw_mode of
3951 FIRST_STACK_REG, not DFmode.
3952 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
3953 (*swapxf): Rename from swapxf.
3955 2018-09-14 Carl Love <cel@us.ibm.com>
3957 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
3958 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
3960 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
3963 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
3966 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
3969 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
3971 2018-09-14 Jason Merrill <jason@redhat.com>
3973 Fix --enable-gather-detailed-mem-stats.
3974 * hash-table.c (hash_table_usage): Change from variable to function.
3975 * hash-table.h: Adjust.
3976 * Makefile.in: Add missing dependencies on hash-table.h.
3978 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3980 PR tree-optimization/87259
3982 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
3983 execute_cse_reciprocals_1 has tried transforming.
3985 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
3987 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
3988 VR_VARYING for PLUS/MINUS_EXPR.
3990 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
3992 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
3995 2018-09-14 Richard Biener <rguenther@suse.de>
3998 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
3999 bits for the merged partition.
4001 2018-09-13 Martin Sebor <msebor@redhat.com>
4002 Bernd Edlinger <bernd.edlinger@hotmail.de>
4004 * builtins.h (c_srlen): Add argument.
4005 * builtins.c (warn_string_no_nul): New function.
4006 (c_strlen): Add argument and use it. Update recursive calls.
4007 Pass DECL argument to string_constant to get info on non
4008 terminated strings. Update *NONSTR as needed.
4009 (fold_builtin_strlen): Add argument to calls to c_strlen.
4010 Warn for unterminated arrays.
4011 (warn_string_no_null): Add prototype.
4012 * expr.c (string_constant): Update arguments. Update recursive
4013 calls appropriately. Detect missing NUL terminator and outermost
4014 declaration its missing in.
4015 Improve checks for arrays with nonzero lower bound or elements
4016 that are not a single byte. Simplify offset computation.
4017 Simplify checks for non-NUL terminated strings.
4018 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
4019 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
4021 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
4023 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
4025 * fold-const.c (c_getstr): Fix function comment. Remove unused third
4026 argument. Fix range checks.
4027 * fold-const.h (c_getstr): Adjust protoype.
4028 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
4029 string is constant but contains no NUL byte.
4031 * expr.c (string_constant): Adjust function comment.
4032 Remove bogus check for zero termination.
4034 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
4036 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
4037 (get_constant_size): Don't make STRING_CSTs larger than they are.
4038 (check_string_literal): New check function for STRING_CSTs.
4039 (output_constant): Use it.
4041 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
4044 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
4046 2018-09-13 Richard Biener <rguenther@suse.de>
4048 PR tree-optimization/87263
4049 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
4050 (struct unwind_state): Add max_rpo field.
4051 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
4052 Compute max_rpo, the max RPO number a block can be backwards reached
4053 from. Re-write non-iterating mode to a RPO ordered worklist approach,
4054 separating it from the iterating mode.
4056 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
4058 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
4059 (rfs_decision): New scheduling decision.
4061 2018-09-13 Richard Biener <rguenther@suse.de>
4064 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
4065 (vn_nary_op_insert_pieces_predicated): Do not write useless
4066 valid_dominated_by_p entry outside of the allocated storage.
4068 2018-09-13 Omar Sandoval <osandov@osandov.com>
4069 Tom de Vries <tdevries@suse.de>
4072 * dwarf2out.c (is_c): New function.
4073 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
4075 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
4078 * config/aarch64/aarch64.md (*aarch64_bfxil):
4080 * config/aarch64/constraints.md (Ulc): Define.
4081 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
4083 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
4086 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
4088 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
4089 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
4090 aarch64_layout_frame call.
4091 (aarch64_expand_epilogue): Likewise.
4092 (aarch64_initial_elimination_offset): Likewise.
4093 (aarch64_get_separate_components): Likewise.
4094 (aarch64_use_return_insn_p): Likewise.
4095 (aarch64_layout_frame): Remove unneeded check.
4097 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4099 * configure.ac: Only append
4100 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
4101 gcc_config_arguments if it was never reconfigured or last reconfigure
4102 was with different arguments.
4103 * configure: Regenerated.
4105 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4106 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4109 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
4110 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
4112 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4114 PR tree-optimization/87287
4115 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
4116 X % C == 0 to X % (unsigned) C == 0 optimization to ...
4117 * match.pd (X % C == 0): ... here. New optimization.
4119 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4122 * expr.h (maybe_optimize_mod_cmp): Declare.
4123 * expr.c (mod_inv): New function.
4124 (maybe_optimize_mod_cmp): New function.
4125 (do_store_flag): Use it.
4126 * cfgexpand.c (expand_gimple_cond): Likewise.
4128 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
4129 Julian Brown <julian@codesourcery.com>
4132 * gimplify.c (gimplify_scan_omp_clauses): Set
4133 target_firstprivatize_array_bases in OpenACC parallel and kernels
4134 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
4135 OpenACC data regions.
4137 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
4139 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
4140 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
4142 2018-09-12 Richard Biener <rguenther@suse.de>
4144 PR tree-optimization/87280
4145 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
4146 edge but unreachable target.
4147 (do_rpo_vn): For conservatively handling a PHI only mark
4148 the backedge executable but not the block reachable.
4150 2018-09-12 Richard Biener <rguenther@suse.de>
4152 PR tree-optimization/87266
4153 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
4156 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
4158 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
4160 ("trunc<BFP:mode><DFP_ALL:mode>2")
4161 ("trunc<DFP_ALL:mode><BFP:mode>2")
4162 ("extend<BFP:mode><DFP_ALL:mode>2")
4163 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
4164 according to the target operand type.
4166 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4167 Andreas Krebbel <krebbel@linux.ibm.com>
4169 PR tree-optimization/86844
4170 * gimple-ssa-store-merging.c
4171 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
4172 there are any overlapping stores in between them, make sure they are
4173 also coalesced or we give up completely.
4175 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4178 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
4179 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
4181 2018-09-12 Tom de Vries <tdevries@suse.de>
4183 * common.opt (gdescribe-dies): Add option.
4184 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
4185 attribute for artifical and nameless decls.
4186 (dwarf2out_register_external_die): Add description attribute to
4187 external reference die.
4188 (add_desc_attribute): New functions.
4189 (gen_subprogram_die): Add description attribute to
4190 DW_TAG_call_site_parameter.
4191 * tree-pretty-print.c (print_generic_expr_to_str): New function.
4192 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
4193 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
4195 (@item -gdescribe-dies): Add.
4197 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
4199 * tree-vrp.c (vrp_shift_undefined_p): Remove.
4200 (extract_range_from_binary_expr_1: Call
4201 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
4202 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
4205 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
4207 * gimple-ssa-warn-alloca.c
4208 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
4209 field for ALLOCA_BOUND_*_LARGE.
4211 2018-09-11 Nathan Sidwell <nathan@acm.org>
4213 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
4215 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4217 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
4218 for clobbers. Remove obsolete comment.
4220 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4222 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
4223 mpxchk, mpxld and mpxst types.
4224 (define_attr length_immediate): Remove all processing of mpx types.
4225 (define_attr prefix_0f): Ditto.
4226 (define_attr memory): Ditto.
4228 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4230 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
4231 (log<mode>2): Change operand 1 predicate to general_operand.
4232 Extend operand 1 to XFmode and generate logxf3 insn.
4233 (log10<mode>2): Change operand 1 predicate to general_operand.
4234 Extend operand 1 to XFmode and generate log10xf3 insn.
4235 (log2<mode>2): Change operand 1 predicate to general_operand.
4236 Extend operand 1 to XFmode and generate log2xf3 insn.
4237 (fyl2xp1_extend<mode>xf3_i387): Remove.
4238 (log1p<mode>2): Change operand 1 predicate to general_operand.
4239 Extend operand 1 to XFmode and generate log1pxf3 insn.
4240 (fxtract_extend<mode>xf3_i387): Remove.
4241 (logb<mode>2): Change operand 1 predicate to general_operand.
4242 Extend operand 1 to XFmode and generate logbxf3 insn.
4243 (ilogb<mode>2): Change operand 1 predicate to general_operand.
4244 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
4245 (significand<mode>2): Change operand 1 predicate to general_operand.
4246 Extend operand 1 to XFmode and generate significandxf3 insn.
4248 2018-09-11 Nathan Sidwell <nathan@acm.org>
4250 * gcc.c (perror_with_name, pfatal_with_name): Delete.
4251 (load_specs): Use fatal_error.
4252 (DELETE_IF_ORDINARY, process_command): Use error.
4253 (execute, run_attempt): Use fatal_error.
4255 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
4257 * diagnostic-core.h (sorry_at): New prototype.
4258 * diagnostic.c (sorry_at): New function.
4260 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
4262 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
4263 by zero as VR_UNDEFINED.
4265 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
4267 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
4268 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
4269 (<sincos>mode2): New expander.
4270 (sincos_extend<mode>xf3_i387): Remove insn pattern.
4271 (sincos -> sin, cos splitters): Remove splitter patterns.
4272 (sincos<mode>3): Change operand 2 predicate to general_operand.
4273 Extend operand 2 to XFmode and generate sincosxf3 insn.
4274 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
4275 Change operand 3 predicate to const1_operand.
4276 (fptan_extend<mode>xf4_i387): Remove insn pattern.
4277 (tanxf2): Update operands in the call to fptanxf4_i387.
4278 (tan<mode>2): Change operand 1 predicate to general_operand.
4279 Extend operand 1 to XFmode and generate tanxf3 insn.
4280 (atan2xf3): Rename from *fpatanxf3_i387.
4281 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
4282 (atan2xf3): Remove expander.
4283 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
4284 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
4285 (atan<mode>2): Change operand 1 predicate to general_operand.
4286 Extend operand 1 to XFmode and generate atanxf3 insn.
4288 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
4290 * config/i386/i386.md (x87/SSE constant load splitter): Use
4291 memory_operand instead of nonimmediate_operand for input operand
4294 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
4296 * config/i386/i386.md (float partial SSE register stall splitter): Move
4297 splitter near its instruction pattern.
4298 (float_extend partial SSE register stall splitter): Ditto.
4299 (float_truncate partial SSE register stall splitter): Ditto.
4301 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
4304 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
4305 to speculation_safe_value_not_needed.
4308 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
4309 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
4310 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
4311 leaf_function_p, instead use has_hard_reg_initial_val.
4313 2018-09-09 Nathan Sidwell <nathan@acm.org>
4315 * gcc.h (pfatal_with_name): Don't declare here.
4316 * gcc.c (pfatal_with_name): Make static.
4318 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
4320 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
4323 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
4325 PR rtl-optimization/85458
4326 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
4327 priority hook to reduce the priority of EXPR.
4329 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
4331 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
4332 DImode for x87 on 32bit targets. Conditionally disable x87 modes
4333 with X87_ENABLE_FLOAT. Remove preparation code.
4334 (*float<SWI48:mode><MODEF:mode>2): Rename from
4335 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
4336 math using "enabled" attribute.
4337 (*floatdi<MODEF:mode>2_i387): Rename from
4338 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
4339 enable for 32bit targets only.
4340 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
4342 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
4343 as operand 1 predicate. Rewrite as define_insn_and_split.
4344 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
4346 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
4348 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
4349 to fallthru to FLOAT case.
4351 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
4354 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
4355 around folding of vec_sl to handle out of range shift values.
4357 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
4359 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
4360 Update callers to gen_fix_trunc<mode>_i387_fisttp
4361 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
4362 nonimmediate_operand.
4363 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
4364 and corresponding splitters.
4365 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
4366 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
4367 (fix_truncdi_i387_with_temp): Remove insn pattern
4368 and corresponding splitters.
4369 (fix_trunc<mode>_i387): Change operand 0 predicate to
4370 nonimmediate_operand.
4371 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
4372 and corresponding splitters.
4373 (*fistdi2_1): Remove.
4375 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
4376 (lrintxfdi2): Remove expander. Reimplement as define_insn.
4377 (*fist<mode>2_1): Remove.
4378 (fist<mode>2): Ditto.
4379 (fist<mode>2_with_temp): Remove insn pattern and corresponding
4381 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
4382 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
4383 (fistdi2_<rounding>): Change operand 0 predicate to
4384 nonimmediate_operand.
4385 (fistdi2_<rounding>_with_temp): Remove insn pattern
4386 and corresponding splitters.
4387 (fist<mode>2_<rounding>): Change operand 0 predicate to
4388 nonimmediate_operand.
4389 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
4390 and corresponding splitters.
4392 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
4394 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
4396 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
4399 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
4401 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4402 early gimple folding of vec_splat().
4403 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
4404 * gimple-fold.h: Add an extern define for tree_vec_extract().
4406 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
4408 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
4409 wrappers around TREE_TYPE comparisons.
4411 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
4414 * config/s390/predicates.md: Add nonsym_memory_operand.
4415 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
4416 contains a SYMBOL_REF, load it into an intermediate pseudo.
4417 (s390_emit_compare_and_swap): Legitimize operand.
4418 * config/s390/s390.md: Use the new nonsym_memory_operand
4419 with UNSPECV_CAS patterns.
4421 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
4424 * config/s390/s390-passes.def: New file.
4425 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
4427 (make_pass_s390_early_mach): Add declaration.
4428 * config/s390/s390.c (make_pass_s390_early_mach):
4429 (s390_option_override): Remove dynamic registration.
4430 * config/s390/t-s390: Add s390-passes.def.
4432 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
4434 * config/s390/s390.c (s390_decompose_constant_pool_ref):
4435 Remove UNSPEC_LTREL_BASE check.
4436 (annotate_constant_pool_refs): Likewise.
4437 (find_constant_pool_ref): Likewise.
4438 (find_ltrel_base): Removed.
4439 (replace_ltrel_base): Removed.
4440 (s390_mainpool_finish): Remove replace_ltrel_base call.
4441 (s390_chunkify_start): Remove pending LTREL_BASE logic.
4442 (s390_chunkify_finish): Remove replace_ltrel_base call.
4443 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
4445 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
4448 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
4449 to speculation_safe_value_not_needed.
4451 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
4452 Bernd Schmidt <bernds_cb1@t-online.de>
4454 * config/nvptx/nvptx-opts.h: New file.
4455 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
4456 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
4458 (TARGET_SM35): New macro.
4459 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
4461 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
4463 (misa=): New option.
4464 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
4466 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
4468 * config/i386/i386.md (truncdfsf2): Remove expander.
4469 (truncdfsf2_with_temp): Ditto.
4470 (truncxf<mode>2): Ditto.
4471 (*truncdfsf_fast_mixed): Remove insn pattern.
4472 (*truncdfsf_fast_i387): Ditto.
4473 (*truncdfsf_mixed): Ditto.
4474 (*truncdfsf_i387): Ditto.
4475 (*truncdfsf2_i387_1): Ditto.
4476 (*truncxfsf2_mixed): Ditto.
4477 (*truncxfdf2_mixed): Ditto.
4478 (*truncxf<mode>2_i387_noop): Ditto. Update callers
4479 to call gen_truncxf<mode>2 instead.
4480 (*truncxf<mode>2_i387): Remove.
4481 (reg->reg splitters): Remove splitter pattern.
4482 (reg->mem splitters): Ditto.
4484 (truncdfsf2): New insn pattern.
4485 (truncxf<mode>2): Ditto.
4487 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4489 * tree-ssa-math-opts.c (is_mult_by): New function.
4490 (is_square_of): Use the above.
4491 (optimize_recip_sqrt): New function.
4492 (pass_cse_reciprocals::execute): Use the above.
4494 2018-09-05 Richard Biener <rguenther@suse.de>
4497 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
4498 to zero-init the emplaced vec.
4500 2018-09-05 Martin Liska <mliska@suse.cz>
4502 PR tree-optimization/87205
4503 * tree-switch-conversion.c (pass_lower_switch::execute):
4504 Group cases for switch statements.
4506 2018-09-05 Richard Biener <rguenther@suse.de>
4508 PR tree-optimization/87217
4509 * tree-ssa-sccvn.c (vuse_valueize): New.
4510 (vn_reference_lookup_pieces): Use it.
4511 (vn_reference_lookup): Likewise.
4513 2018-09-05 Nathan Sidwell <nathan@acm.org>
4516 * stor-layout.c (place_field): Scan forwards to check last
4517 bitfield when ms_bitfield_placement is in effect.
4519 2018-09-05 Richard Biener <rguenther@suse.de>
4522 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
4525 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
4526 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
4528 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
4529 * config.gcc (extra_objs): Build it.
4530 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
4532 * config/aarch64/aarch64-passes.def
4533 (pass_tag_collision_avoidance): New pass.
4534 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
4535 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
4536 (aarch64_classify_address): Remove static qualifier.
4537 (aarch64_address_info, aarch64_address_type): Move to...
4538 * config/aarch64/aarch64-protos.h: ... here.
4539 (make_pass_tag_collision_avoidance): New function.
4540 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
4543 2018-09-05 Martin Liska <mliska@suse.cz>
4545 * doc/gcov.texi: Update documentation of humar
4547 * gcov.c (format_count): Print one decimal place, it provides
4548 more fine number of situations like '1G' vs. '1.4G'.
4550 2018-09-05 Martin Liska <mliska@suse.cz>
4553 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
4554 * optc-gen.awk: Allow 'Var' for Deprecated options in order
4555 to generate a MASK value.
4557 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
4560 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
4561 if frame pointer isn't used.
4562 (compute_frame_pointer_to_fb_displacement): Likewise.
4564 2018-09-04 Jakub Jelinek <jakub@redhat.com>
4567 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
4568 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
4569 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
4570 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
4571 and OPTION_MASK_ISA_XSAVEC_UNSET.
4573 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
4575 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
4576 XOR operations in NAND case.
4578 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
4580 * wide-int-range.cc (wide_int_range_convert): New.
4581 * wide-int-range.h (wide_int_range_convert): New.
4582 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
4583 code into wide_int_range_convert.
4584 (extract_range_into_wide_ints): Do not munge anti range constants
4585 into the entire domain. Just return the range back.
4587 2018-09-04 Martin Liska <mliska@suse.cz>
4589 * genmatch.c (output_line_directive): Add new argument
4591 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
4593 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
4595 * doc/invoke.texi (Option Summary): Add whitespace.
4597 * doc/invoke.texi (Option Summary): Add -Waligned-new.
4599 2018-09-04 Richard Biener <rguenther@suse.de>
4601 PR tree-optimization/87211
4602 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
4603 backedge value we're supposed to treat as VARYING also number
4604 the PHI to VARYING in case it got a different value-number already.
4606 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
4608 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
4609 (extract_range_from_binary_expr_1): Do not call
4610 vrp_can_optimize_bit_op.
4611 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
4613 (wide_int_range_get_mask_and_bounds): New.
4614 (wide_int_range_optimize_bit_op): New.
4615 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
4616 (wide_int_range_bit_and): Same.
4617 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
4618 (wide_int_range_optimize_bit_op): New.
4619 (wide_int_range_get_mask_and_bounds): New.
4621 2018-09-04 Richard Biener <rguenther@suse.de>
4623 PR tree-optimization/87176
4624 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
4625 variable. When value-numbering a virtual PHI node make sure
4626 to not value-number to the backedge value.
4628 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
4630 * doc/extend.texi (Long Long, Hex Floats): Document support for
4631 long long and hex floats in more recent versions of ISO C++.
4633 2018-09-03 Richard Biener <rguenther@suse.de>
4635 PR tree-optimization/87177
4636 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
4639 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4641 * bb-reorder.c (edge_order): Convert to C-qsort-style
4642 tri-state comparator.
4643 (reorder_basic_blocks_simple): Change std::stable_sort to
4646 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4648 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
4649 tri-state comparator.
4650 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
4652 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4654 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
4655 (mergesort): ... here as maximum count for using netsort.
4656 (gcc_qsort): Set nlim to 3 if stable sort is requested.
4657 (gcc_stablesort): New.
4658 * system.h (gcc_stablesort): Declare.
4660 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4662 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
4663 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
4664 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
4666 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
4668 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
4669 lxsdx and stxsdx alternatives.
4670 (*mov<mode>_hardfloat64): Ditto.
4671 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
4673 2018-09-03 Richard Biener <rguenther@suse.de>
4675 PR tree-optimization/87200
4676 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
4679 2018-09-03 Martin Liska <mliska@suse.cz>
4681 PR tree-optimization/87201
4682 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
4683 Fix parenthesis in an expression.
4685 2018-09-03 Richard Biener <rguenther@suse.de>
4687 PR tree-optimization/87197
4688 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
4689 visited. CSE the VN_INFO hashtable lookup.
4691 PR tree-optimization/87169
4692 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
4693 iterating make sure there's no extra backedges from irreducible
4694 regions feeding the header. Mark the destination block
4697 2018-09-03 Martin Liska <mliska@suse.cz>
4700 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
4701 * common/common-targhooks.c (default_get_valid_option_values):
4703 * common/common-targhooks.h (default_get_valid_option_values):
4705 * common/config/i386/i386-common.c: Move processor_target_table
4707 (ix86_get_valid_option_values): New function.
4708 (TARGET_GET_VALID_OPTION_VALUES): New macro.
4709 * config/i386/i386.c (struct ptt): Move to i386-common.c.
4710 (PTA_*): Move all defined masks into i386-common.c.
4711 (ix86_function_specific_restore): Use new processor_cost_table.
4712 * config/i386/i386.h (struct ptt): Moved from i386.c.
4713 (struct pta): Likewise.
4714 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
4715 * doc/tm.texi.in: Likewise.
4716 * opt-suggestions.c (option_proposer::suggest_option):
4717 Pass prefix to build_option_suggestions.
4718 (option_proposer::get_completions): Likewise.
4719 (option_proposer::build_option_suggestions): Use the new target
4721 * opts.c (struct option_help_tuple): New struct.
4722 (print_filtered_help): Use the new target hook.
4724 2018-09-03 Martin Liska <mliska@suse.cz>
4727 * predict.c (set_even_probabilities): Add likely_edges
4728 argument and handle cases where we have precisely one
4730 (combine_predictions_for_bb): Catch also likely_edges.
4731 (tree_predict_by_opcode): Handle gswitch statements.
4732 * tree-cfg.h (find_case_label_for_value): New declaration.
4733 (find_taken_edge_switch_expr): Likewise.
4734 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
4735 Find pivot in decision tree based on probabily, not by number of
4738 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
4740 * doc/standards.texi (Standards): Update Objective-C reference.
4742 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
4744 * doc/install.texi (Prerequisites): Update link for MPC.
4746 2018-09-01 Michael Matz <matz@suse.de>
4748 PR tree-optimization/87074
4749 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
4750 PHIs for outer-loop uses.
4752 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
4754 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
4755 * doc/invoke.texi (C Dialect Options): Ditto.
4757 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
4759 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
4761 2018-08-31 Richard Biener <rguenther@suse.de>
4763 PR tree-optimization/87168
4764 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
4765 (rpo_elim::eliminate_avail): When OP was not visited it must
4768 2018-08-31 David Malcolm <dmalcolm@redhat.com>
4770 * tree-vrp.c (copy_value_range): Convert param "from" from
4771 "value_range *" to "const value_range *".
4772 (range_is_null): Likewise for param "vr".
4773 (range_int_cst_p): Likewise.
4774 (range_int_cst_singleton_p): Likewise.
4775 (symbolic_range_p): Likewise.
4776 (value_ranges_intersect_p): Likewise for both params.
4777 (value_range_nonnegative_p): Likewise for param "vr".
4778 (value_range_constant_singleton): Likewise.
4779 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
4780 (extract_range_into_wide_ints): Likewise for param "vr".
4781 (extract_range_from_multiplicative_op): Likewise for params "vr0"
4783 (vrp_can_optimize_bit_op): Likewise.
4784 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
4786 (extract_range_from_unary_expr): Likewise.
4787 (debug_value_range): Likewise for param "vr".
4788 (value_range::dump): Add "const" qualifier.
4789 (vrp_prop::check_array_ref): Convert local "vr" from
4790 "value_range *" to "const value_range *".
4791 (vrp_prop::check_mem_ref): Likewise.
4792 (vrp_prop::visit_stmt): Likewise for local "old_vr".
4793 (vrp_intersect_ranges_1): Likewise for param "vr_1".
4794 (vrp_intersect_ranges): Likewise.
4795 (simplify_stmt_for_jump_threading): Likewise for local "vr".
4796 (vrp_prop::vrp_finalize): Likewise.
4797 * tree-vrp.h (value_range::dump): Add "const" qualifier.
4798 (vrp_intersect_ranges): Add "const" qualifier to params as above.
4799 (extract_range_from_unary_expr): Likewise.
4800 (value_range_constant_singleton): Likewise.
4801 (symbolic_range_p): Likewise.
4802 (copy_value_range): Likewise.
4803 (extract_range_from_binary_expr_1): Likewise.
4804 (range_int_cst_p): Likewise.
4805 (vrp_set_zero_nonzero_bits): Likewise.
4806 (range_int_cst_singleton_p): Likewise.
4808 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
4810 * config/aarch64/arm_neon.h (vabsd_s64): New.
4811 (vnegd_s64): Likewise.
4813 2018-08-31 Martin Jambor <mjambor@suse.cz>
4815 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
4817 2018-08-31 Martin Liska <mliska@suse.cz>
4819 * ipa-icf.c (sem_item::add_type): Use
4820 sem_item::m_type_hash_cache.
4821 * ipa-icf.h: Move the cache from sem_item_optimizer
4824 2018-08-31 Nathan Sidwell <nathan@acm.org>
4826 * doc/extend.texi (Backwards Compatibility): Remove implicit
4827 extern C leeway of () being (...).
4829 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4831 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
4833 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
4837 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
4839 2018-08-31 Jakub Jelinek <jakub@redhat.com>
4842 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
4843 gen_int_mode. Formatting fixes.
4845 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
4847 * target.def (custom_function_descriptors): Improve documentation.
4848 * doc/tm.texi.in (Trampolines): Expand discussion of function
4849 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
4850 beginning of the section.
4851 * doc/tm.texi: Regenerated.
4853 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
4855 * cfg.h (class auto_edge_flag): Spell out the template-id of the
4856 base class in the initializer list. This is a workaround for
4857 building with older GCC.
4858 (class auto_bb_flag): Likewise.
4860 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
4862 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
4863 (altivec_vcmpequ<VI_char>_p): Remove star.
4864 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
4866 (expand_strncmp_vec_sequence): New function.
4867 (emit_final_str_compare_vec): New function.
4868 (expand_strn_compare): Add support for vector strncmp.
4869 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
4870 length specification to bytes.
4871 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
4872 (vcmpnezb_p): New pattern.
4873 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
4874 for option -mstring-compare-inline-limit.
4876 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
4878 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
4879 (PTA_SKYLAKE): Add PTA_AES.
4880 (PTA_GOLDMONT): Likewise.
4882 2018-08-29 Jan Hubicka <jh@suse.cz>
4885 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
4886 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
4888 2018-08-29 Jan Hubicka <jh@suse.cz>
4890 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
4892 (hash_tree): Do not visit TYPE_STUB_DECL.
4893 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
4894 stream TYPE_STUB_DECL.
4895 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
4896 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
4897 after free_lang_data.
4898 (type_in_anonymous_namespace_p): Likewise.
4900 2018-08-29 Jan Hubicka <jh@suse.cz>
4902 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
4903 comment that it has to be even number.
4904 (class sreal): Change m_sig type to int32_t.
4905 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
4906 int64_t for temporary calculations.
4907 (sreal_verify_basics): Drop one bit from minimum and maximum.
4909 2018-08-30 Richard Biener <rguenther@suse.de>
4911 PR tree-optimization/87147
4912 * tree-ssa-sccvn.c (SSA_VISITED): New function.
4913 (visit_phi): When the degenerate result is from the backedge and
4914 we didn't visit its definition yet drop to VARYING.
4915 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
4917 2018-08-29 Jan Hubicka <jh@suse.cz>
4919 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
4921 (hash_tree): Likewise.
4923 2018-08-29 Jan Hubicka <jh@suse.cz>
4925 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
4926 and TYPE_NEXT_REF_TO.
4928 2018-08-29 Jan Hubicka <jh@suse.cz>
4930 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
4931 comment that it has to be even number.
4932 (class sreal): Change m_sig type to int32_t.
4933 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
4934 int64_t for temporary calculations.
4935 (sreal_verify_basics): Drop one bit from minimum and maximum.
4937 2018-08-30 Tamar Christina <tamar.christina@arm.com>
4939 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
4941 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
4944 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
4945 if to_add is negative.
4947 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
4950 * builtins.c (c_strlen): Improve range checks.
4952 2018-08-29 Martin Sebor <msebor@redhat.com>
4953 Jeff Law <law@redhat.com>
4955 PR tree-optimization/86714
4956 PR tree-optimization/86711
4957 * builtins.c (c_strlen): Add arguments to call to string_constant.
4958 * expr.c (string_constant): Add argument. Detect missing nul
4959 terminator and outermost declaration it's missing in.
4960 * expr.h (string_constant): Add argument.
4961 * fold-const.c (read_from_constant_string): Add arguments to call to
4963 (c_getstr): Likewise.
4964 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
4966 * tree-ssa-strlen.c (get_stridx): Likewise.
4968 2018-08-29 Jan Hubicka <jh@suse.cz>
4970 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
4971 Do not stream DECL_VINDEX.
4972 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
4973 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
4974 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
4976 2018-08-29 Richard Biener <rguenther@suse.de>
4978 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
4979 virtual operands that are not default defs to honor region
4981 (rpo_vn_valueize): Remove ineffective code here.
4983 2018-08-29 Richard Biener <rguenther@suse.de>
4985 PR tree-optimization/87132
4986 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
4987 when skipping defs reachable over backedges.
4989 2018-08-29 Richard Biener <rguenther@suse.de>
4991 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
4992 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
4993 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
4994 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
4995 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
4996 (vn_reference_lookup_3): Remove use of const_parms.
4997 (free_rpo_vn): Do not free const_parms.
4998 (do_rpo_vn): Do not call init_const_parms.
4999 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
5000 SSA_NAME_POINTS_TO_READONLY_MEMORY.
5001 (call_may_clobber_ref_p_1): Likewise.
5003 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
5006 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
5008 (-ftree-scev-cprop): Document.
5010 2018-08-29 Jan Hubicka <jh@suse.cz>
5012 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
5014 (sreal constructor): Update.
5015 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
5016 sreal:operator/): Update.
5018 2018-08-29 Martin Liska <mliska@suse.cz>
5020 * tree-switch-conversion.c (switch_conversion::expand):
5021 Strenghten assumption about gswitch statements.
5023 2018-08-29 Richard Biener <rguenther@suse.de>
5025 PR tree-optimization/87117
5026 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
5027 re-value-number released SSA VDEFs.
5029 2018-08-29 Richard Biener <rguenther@suse.de>
5031 PR tree-optimization/87126
5032 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
5034 2018-08-28 Jim Wilson <jimw@sifive.com>
5036 * config/riscv/pic.md: Rewrite.
5037 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
5039 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
5040 (SOFTF, default_load, softload, softstore): New.
5042 2018-08-28 Jeff Law <law@redhat.com>
5044 * fold-const.c (fold_binary_loc): Remove recently added assert.
5046 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
5048 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
5049 to OP parmeter of generated function.
5051 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
5053 PR tree-optimization/87009
5054 * match.pd: Add boolean optimizations.
5056 2018-08-28 Martin Sebor <msebor@redhat.com>
5059 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
5060 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
5061 (pass_walloca::gate): Use it.
5062 (alloca_call_type): Same.
5063 (pass_walloca::execute): Same.
5064 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
5066 2018-08-28 David Malcolm <dmalcolm@redhat.com>
5068 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
5069 GCC_VERSION for usage of "__gcc_dump_printf__" format from
5072 2018-08-28 Richard Biener <rguenther@suse.de>
5074 PR tree-optimization/87124
5075 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
5076 constants before looking up avail.
5078 2018-08-28 Jakub Jelinek <jakub@redhat.com>
5081 * calls.c (maybe_warn_nonstring_arg): Punt early if
5082 warn_stringop_overflow is zero. Don't call get_range_strlen
5083 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
5084 Swap comparison operands to have constants on rhs. Only use
5085 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
5086 increment lenrng[0].
5088 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
5090 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
5091 use of tree_to_shwi. Remove duplicated test for the size being
5092 a whole number of bytes.
5094 2018-08-28 Richard Biener <rguenther@suse.de>
5096 PR tree-optimization/87117
5097 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
5098 Handle removed stmt without LHS (GIMPLE_NOP).
5100 2018-08-28 Richard Biener <rguenther@suse.de>
5102 PR tree-optimization/87117
5103 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
5104 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
5106 2018-08-28 Richard Biener <rguenther@suse.de>
5108 PR tree-optimization/87117
5109 * tree-ssa-pre.c (compute_avail): Do not make expressions
5110 with predicated values available.
5111 (get_expr_value_id): Assert we do not run into predicated value
5114 2018-08-28 Richard Biener <rguenther@suse.de>
5116 PR tree-optimization/87117
5117 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
5118 get virtual operands.
5119 (get_expr_operands): Handle STRING_CST like other decls.
5121 2018-08-28 Martin Liska <mliska@suse.cz>
5123 * tree.h: Update documentation of fndecl_built_in_p
5127 2018-08-27 Jeff Law <law@redhat.com>
5128 PR tree-optimization/87110
5129 * tree-ssa-dse.c (compute_trims): Handle non-constant
5132 2018-08-27 Martin Sebor <msebor@redhat.com>
5134 PR tree-optimization/86914
5135 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
5137 2018-08-27 Martin Sebor <msebor@redhat.com>
5139 PR tree-optimization/87112
5140 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
5141 the type of the bound argument.
5143 2018-08-27 Jeff Law <law@redhat.com>
5145 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
5146 type does not have a TYPE_SIZE_UNIT.
5148 2018-08-27 Steve Ellcey <sellcey@cavium.com>
5150 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
5151 with include of backend.h.
5153 2018-08-27 Richard Biener <rguenther@suse.de>
5155 PR tree-optimization/86927
5156 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
5157 use const cond reduction code.
5159 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
5161 PR tree-optimization/85758
5162 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
5164 2018-08-27 David Malcolm <dmalcolm@redhat.com>
5167 * diagnostic-show-locus.c (class layout_range): Update for
5168 conversion of show_caret_p to a tri-state.
5169 (layout_range::layout_range): Likewise.
5170 (make_range): Likewise.
5171 (layout::maybe_add_location_range): Likewise.
5172 (layout::should_print_annotation_line_p): Don't show annotation
5173 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
5174 (layout::get_state_at_point): Update for conversion of
5175 show_caret_p to a tri-state. Bail out early for
5176 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
5177 underlining or source colorization.
5178 (gcc_rich_location::add_location_if_nearby): Update for conversion
5179 of show_caret_p to a tri-state.
5180 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
5181 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
5183 (selftest::test_one_liner_labels): Likewise.
5184 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
5185 conversion of show_caret_p to a tri-state.
5186 * pretty-print.c (text_info::set_location): Likewise.
5187 * pretty-print.h (text_info::set_location): Likewise.
5188 * substring-locations.c (format_warning_n_va): Likewise.
5189 * tree-diagnostic.c (default_tree_printer): Likewise.
5190 * tree-pretty-print.c (newline_and_indent): Likewise.
5192 2018-08-27 David Malcolm <dmalcolm@redhat.com>
5195 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
5196 line above for line-insertion fix-it hints.
5197 (selftest::test_fixit_insert_containing_newline): Update the
5198 expected results, and add a test with line-numbering enabled.
5200 2018-08-27 Martin Liska <mliska@suse.cz>
5203 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
5204 params with DECL_HAS_VALUE_EXPR_P.
5206 2018-08-27 Martin Liska <mliska@suse.cz>
5208 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
5209 selected expansion strategy.
5211 2018-08-27 Martin Liska <mliska@suse.cz>
5213 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
5214 * builtins.c (is_builtin_fn): Likewise.
5215 * attribs.c (diag_attr_exclusions): Use new function
5216 fndecl_built_in_p and remove check for FUNCTION_DECL if
5218 (builtin_mathfn_code): Likewise.
5219 (fold_builtin_expect): Likewise.
5220 (fold_call_expr): Likewise.
5221 (fold_builtin_call_array): Likewise.
5222 (fold_call_stmt): Likewise.
5223 (set_builtin_user_assembler_name): Likewise.
5224 (is_simple_builtin): Likewise.
5225 * calls.c (gimple_alloca_call_p): Likewise.
5226 (maybe_warn_nonstring_arg): Likewise.
5227 * cfgexpand.c (expand_call_stmt): Likewise.
5228 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
5229 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
5230 (cgraph_node::verify_node): Likewise.
5231 * cgraphclones.c (build_function_decl_skip_args): Likewise.
5232 (cgraph_node::create_clone): Likewise.
5233 * config/arm/arm.c (arm_insert_attributes): Likewise.
5234 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
5235 * dse.c (scan_insn): Likewise.
5236 * expr.c (expand_expr_real_1): Likewise.
5237 * fold-const.c (operand_equal_p): Likewise.
5238 (fold_binary_loc): Likewise.
5239 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
5240 * gimple-low.c (lower_stmt): Likewise.
5241 * gimple-pretty-print.c (dump_gimple_call): Likewise.
5242 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
5244 * gimple.c (gimple_build_call_from_tree): Likewise.
5245 (gimple_call_builtin_p): Likewise.
5246 (gimple_call_combined_fn): Likewise.
5247 * gimplify.c (gimplify_call_expr): Likewise.
5248 (gimple_boolify): Likewise.
5249 (gimplify_modify_expr): Likewise.
5250 (gimplify_addr_expr): Likewise.
5251 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
5252 * ipa-cp.c (determine_versionability): Likewise.
5253 * ipa-fnsummary.c (compute_fn_summary): Likewise.
5254 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
5255 * ipa-split.c (visit_bb): Likewise.
5256 (split_function): Likewise.
5257 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
5258 * lto-cgraph.c (input_node): Likewise.
5259 * lto-streamer-out.c (write_symbol): Likewise.
5260 * omp-low.c (setjmp_or_longjmp_p): Likewise.
5261 (lower_omp_1): Likewise.
5262 * predict.c (strip_predict_hints): Likewise.
5263 * print-tree.c (print_node): Likewise.
5264 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
5265 * trans-mem.c (is_tm_irrevocable): Likewise.
5266 (is_tm_load): Likewise.
5267 (is_tm_simple_load): Likewise.
5268 (is_tm_store): Likewise.
5269 (is_tm_simple_store): Likewise.
5270 (is_tm_abort): Likewise.
5271 (tm_region_init_1): Likewise.
5272 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
5273 * tree-cfg.c (verify_gimple_call): Likewise.
5274 (move_stmt_r): Likewise.
5275 (stmt_can_terminate_bb_p): Likewise.
5276 * tree-eh.c (lower_eh_constructs_2): Likewise.
5277 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
5278 * tree-inline.c (remap_gimple_stmt): Likewise.
5279 (copy_bb): Likewise.
5280 (estimate_num_insns): Likewise.
5281 (fold_marked_statements): Likewise.
5282 * tree-sra.c (scan_function): Likewise.
5283 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
5284 (optimize_stack_restore): Likewise.
5285 (pass_fold_builtins::execute): Likewise.
5286 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5287 (mark_all_reaching_defs_necessary_1): Likewise.
5288 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
5289 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
5290 (pass_forwprop::execute): Likewise.
5291 * tree-ssa-loop-im.c (stmt_cost): Likewise.
5292 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5293 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
5294 * tree-ssa-strlen.c (get_string_length): Likewise.
5295 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
5296 (find_func_aliases_for_call): Likewise.
5297 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
5298 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
5299 * tree-tailcall.c (find_tail_calls): Likewise.
5300 * tree.c (need_assembler_name_p): Likewise.
5301 (free_lang_data_in_decl): Likewise.
5302 (get_call_combined_fn): Likewise.
5303 * ubsan.c (is_ubsan_builtin_p): Likewise.
5304 * varasm.c (incorporeal_function_p): Likewise.
5305 * tree.h (DECL_BUILT_IN): Remove and replace with
5307 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
5308 (fndecl_built_in_p): New.
5310 2018-08-27 Martin Liska <mliska@suse.cz>
5312 PR tree-optimization/86847
5313 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
5314 Dump also subtree probability.
5315 (switch_decision_tree::do_jump_if_equal): New function.
5316 (switch_decision_tree::emit_case_nodes): Handle special
5317 situations in balanced tree that can be emitted much simpler.
5318 Fix calculation of probabilities that happen in tree expansion.
5319 * tree-switch-conversion.h (struct cluster): Add
5321 (struct simple_cluster): Likewise.
5322 (struct case_tree_node): Add new function has_child.
5323 (do_jump_if_equal): New.
5325 2018-08-27 Martin Liska <mliska@suse.cz>
5327 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
5328 Add new argument to bit_test_cluster constructor.
5329 (bit_test_cluster::emit): Set bits really number of values
5331 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
5332 probability argument.
5333 * tree-switch-conversion.h (struct bit_test_cluster):
5334 Add m_handles_entire_switch.
5336 2018-08-27 Martin Liska <mliska@suse.cz>
5338 PR tree-optimization/86702
5339 * tree-switch-conversion.c (jump_table_cluster::emit):
5340 Make probabilities even for values in jump table
5341 according to number of cases handled.
5342 (switch_decision_tree::compute_cases_per_edge): Pass
5343 argument to reset_out_edges_aux function.
5344 (switch_decision_tree::analyze_switch_statement): Likewise.
5345 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
5348 2018-08-27 Martin Liska <mliska@suse.cz>
5350 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
5351 cfun argument explicitly.
5352 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
5353 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
5354 function gimple_switch_default_bb.
5355 (convert_switch_statements):
5357 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
5358 * stmt.c (label_to_block_fn): Use label_to_block and pass
5359 cfun argument explicitly and use gimple_switch_label_bb.
5360 (expand_case): Likewise.
5361 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
5362 cfun argument explicitly. Likewise.
5363 (make_edges_bb): Likewise.
5364 (make_cond_expr_edges): Likewise.
5365 (get_cases_for_edge): Likewise.
5366 (make_gimple_switch_edges): Likewise.
5367 (label_to_block_fn): Likewise.
5368 (label_to_block): Likewise.
5369 (make_goto_expr_edges): Likewise.
5370 (make_gimple_asm_edges): Likewise.
5371 (main_block_label): Likewise.
5372 (group_case_labels_stmt): Likewise.
5373 (find_taken_edge_computed_goto): Likewise.
5374 (find_taken_edge_switch_expr): Likewise.
5375 (gimple_verify_flow_info): Likewise.
5376 (gimple_redirect_edge_and_branch): Likewise.
5377 (gimple_switch_label_bb): New function.
5378 (gimple_switch_default_bb): Likewise.
5379 (gimple_switch_edge): Likewise.
5380 (gimple_switch_default_edge): Likewise.
5381 * tree-cfg.h (label_to_block_fn): Remove and replace ...
5382 (label_to_block): ... with this.
5383 (gimple_switch_label_bb): New.
5384 (gimple_switch_default_bb): Likewise.
5385 (gimple_switch_edge): Likewise.
5386 (gimple_switch_default_edge): Likewise.
5387 * tree-cfgcleanup.c (convert_single_case_switch): Use
5388 new gimple functions and pass new argument to label_to_block.
5389 (cleanup_control_flow_bb):
5390 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
5391 cfun argument explicitly.
5392 (make_eh_edges): Likewise.
5393 (redirect_eh_dispatch_edge): Likewise.
5394 (lower_resx): Likewise.
5395 (lower_eh_dispatch): Likewise.
5396 (maybe_remove_unreachable_handlers): Likewise.
5397 (unsplit_eh): Likewise.
5398 (cleanup_empty_eh): Likewise.
5399 (verify_eh_edges): Likewise.
5400 (verify_eh_dispatch_edge): Likewise.
5401 * tree-ssa-dom.c (record_edge_info): Likewise.
5402 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
5403 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
5404 (thread_through_normal_block): Likewise.
5405 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
5406 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
5407 * tree-switch-conversion.c (switch_conversion::collect): Use new
5409 (switch_conversion::check_final_bb): Likewise.
5410 (switch_conversion::gather_default_values): Pass new argument
5412 (switch_conversion::build_constructors): Likewise.
5413 (switch_decision_tree::compute_cases_per_edge): Use new
5414 gimple_switch_edge function.
5415 (switch_decision_tree::analyze_switch_statement): Pass new argument
5417 (switch_decision_tree::try_switch_expansion): Use
5418 gimple_switch_default_edge.
5419 * tree-vrp.c (find_switch_asserts): Pass new argument
5421 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
5422 (vr_values::simplify_switch_using_ranges): Likewise.
5424 2018-08-27 Richard Biener <rguenther@suse.de>
5426 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
5427 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
5429 * tree-ssa-sccvn.h (struct vn_pval): New structure.
5430 (struct vn_nary_op_s): Add unwind_to member. Add
5431 predicated_values flag and put result into a union together
5432 with a linked list of vn_pval.
5433 (struct vn_ssa_aux): Add name member to make maintaining
5434 a map of SSA name to vn_ssa_aux possible. Remove no longer
5435 needed info, dfsnum, low, visited, on_sccstack, use_processed
5436 and range_info_anti_range_p members.
5437 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
5438 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
5440 (vn_valueize): New global.
5441 (vn_context_bb): Likewise.
5442 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
5443 VN_INFO_PTR_INFO): Remove.
5444 * tree-ssa-sccvn.c: ... (rewrite)
5445 (pass_fre::execute): For -O2+ initialize loops and run
5446 RPO VN in optimistic mode (iterating). For -O1 and -Og
5447 run RPO VN in non-optimistic mode.
5448 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
5449 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
5450 * doc/invoke.texi (sccvn-max-scc-size): Remove.
5451 (rpo-vn-max-loop-depth): Document.
5452 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
5453 when valuezing the VUSE signals we walked out of the region.
5454 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
5455 (phi_translate): Set VN context block to use for availability
5457 (compute_avail): Likewise.
5458 (pre_valueize): New function.
5459 (pass_pre::execute): Adjust to the RPO VN API.
5461 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
5462 (propagate_constants_for_unrolling): Remove.
5463 (tree_unroll_loops_completely): Perform value-numbering
5464 on the unrolled bodies loop parent.
5466 2018-08-27 Richard Biener <rguenther@suse.de>
5468 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
5469 for partial antic compute.
5471 2018-08-27 Jakub Jelinek <jakub@redhat.com>
5473 PR rtl-optimization/87065
5474 * combine.c (simplify_if_then_else): Formatting fix.
5475 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
5477 (known_cond): Don't return const_true_rtx for vector modes. Use
5478 CONST0_RTX instead of const0_rtx. Formatting fixes.
5480 2018-08-27 Martin Liska <mliska@suse.cz>
5482 PR gcov-profile/87069
5483 * gcov.c (process_file): Record files already processed
5484 and warn about a file being processed multiple times.
5486 2018-08-27 Martin Liska <mliska@suse.cz>
5489 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5490 Set default values for x_aarch64_*_string strings.
5491 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
5492 prefix. For -mabi do not print '=ABI' in help and use
5493 <option_value> format for -msve-vector-bits and -moverride
5496 2018-08-26 Jeff Law <law@redhat.com>
5498 * config/mips/frame-header-opt.c: Include "backend.h" rather than
5501 2018-08-26 Marek Polacek <polacek@redhat.com>
5503 PR c++/87029, Implement -Wredundant-move.
5504 * doc/invoke.texi: Document -Wredundant-move.
5506 2018-08-25 Martin Sebor <msebor@redhat.com>
5508 PR tree-optimization/87059
5509 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
5510 to the same type as the other.
5511 * fold-const.c (fold_binary_loc): Assert expectation.
5513 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
5515 * config/darwin.c (machopic_legitimize_pic_address): Clean up
5516 extraneous parentheses, dead code section and formatting.
5518 2018-08-24 David Malcolm <dmalcolm@redhat.com>
5521 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
5522 wide enough for jumps in the line-numbering to be visible.
5523 (layout::print_gap_in_line_numbering): New member function.
5524 (layout::calculate_line_spans): When using line numbering, merge
5525 line spans that are only 1 line apart.
5526 (diagnostic_show_locus): When printing line numbers, show gaps in
5527 line numbering directly, rather than printing headers.
5528 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
5529 line-numbering with multiple line spans.
5530 (selftest::test_fixit_insert_containing_newline_2): Add test of
5531 line-numbering, in which the spans are close enough to be merged.
5533 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
5535 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
5536 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
5537 * tree-vrp.c (range_is_nonnull): Remove.
5538 (range_includes_zero_p): Accept value_range instead of min/max.
5539 (extract_range_from_binary_expr_1): Do not early bail on
5541 Use range_includes_zero_p instead of range_is_nonnull.
5542 (extract_range_from_unary_expr): Use range_includes_zero_p instead
5543 of range_is_nonnull.
5544 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
5545 special case VR_ANTI_RANGE.
5546 (vrp_finalize): Same.
5547 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
5549 (range_is_nonnull): Remove.
5550 * vr-values.c (vrp_stmt_computes_nonzero): Use
5551 range_includes_zero_p instead of range_is_nonnull.
5552 (extract_range_basic): Pass value_range to range_includes_zero_p
5553 instead of range_is_nonnull.
5555 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
5557 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
5558 * emit-rtl.h (rtl_data): Remove return_bnd.
5559 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
5560 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
5561 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
5562 (POINTER_BOUNDS_MODE): Remove definition.
5563 (make_pointer_bounds_mode): Remove.
5564 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
5565 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
5566 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
5567 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
5568 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
5569 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
5570 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
5572 * config/i386/i386-modes.def (BND32, BND64): Remove.
5573 * config/i386/i386.c (dbx_register_map): Remove bound registers.
5574 (dbx64_register_map): Ditto.
5575 (svr4_dbx_register_map): Ditto.
5576 (indirect_thunk_bnd_needed): Remove.
5577 (indirect_thunks_bnd_used): Ditto.
5578 (indirect_return_bnd_needed): Ditto.
5579 (indirect_return_via_cx_bnd): Ditto.
5580 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
5581 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
5582 (output_indirect_thunk): Ditto. Remove need_prefix argument.
5583 (output_indirect_thunk_function): Remove handling of
5584 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
5585 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
5586 (ix86_save_reg): Remove handling of crtl->return_bnd.
5587 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
5588 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
5589 and UNSPEC_BNDLX_ADDR.
5590 (ix86_output_indirect_branch_via_reg): Remove handling of
5591 indirect_thunk_prefix_bnd.
5592 (ix86_output_indirect_branch_via_push): Ditto.
5593 (ix86_output_function_return): Ditto.
5594 (ix86_output_indirect_function_return): Ditto.
5595 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
5596 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
5597 (CALL_USED_REGISTERS): Ditto.
5598 (REG_ALLOC_ORDER): Update for removal of bound registers.
5599 (HI_REGISTER_NAMES): Ditto.
5600 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
5601 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
5602 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
5603 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
5604 (FIRST_PSEUDO_REG): Update.
5605 (BND): Remove mode iterator.
5606 * config/i386/predicates.md (bnd_mem_operator): Remove.
5608 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
5610 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
5613 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
5615 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
5616 the case in which the permute needs only a single element and
5617 repeats for every vector of the result. Extend that case to
5618 handle variable-length vectors.
5619 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
5621 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
5624 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
5625 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
5627 2018-08-24 Richard Biener <rguenther@suse.de>
5629 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
5630 bb_flags_allocated members.
5631 (auto_flag): New RAII class for allocating flags.
5632 (auto_edge_flag): New RAII class for allocating edge flags.
5633 (auto_bb_flag): New RAII class for allocating bb flags.
5634 * cfgloop.c (verify_loop_structure): Allocate temporary edge
5636 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
5637 in favor of temporarily allocated BB flag.
5638 * hsa-brig.c: Re-order includes.
5639 * hsa-dump.c: Likewise.
5640 * hsa-regalloc.c: Likewise.
5641 * print-rtl.c: Likewise.
5642 * profile-count.c: Likewise.
5644 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
5647 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
5650 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
5653 * wide-int-range.cc (wide_int_range_div): Do not ignore result
5654 from wide_int_range_multiplicative_op.
5656 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5658 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
5661 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
5663 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
5666 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
5668 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
5669 full GENERIC expression used for replacement.
5671 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
5673 * tree-vrp.c (abs_extent_range): Remove.
5674 (extract_range_into_wide_ints): Pass wide ints by reference.
5675 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
5676 Pass wide ints by reference in all calls to
5677 extract_range_into_wide_ints.
5678 * wide-int-range.cc (wide_int_range_div): New.
5679 * wide-int-range.h (wide_int_range_div): New.
5680 (wide_int_range_includes_zero_p): New.
5681 (wide_int_range_zero_p): New.
5683 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
5685 * config/aarch64/aarch64.md (arches): New enum.
5686 (arch): New enum attr.
5687 (arch_enabled): New attr.
5688 (enabled): Now uses arch_enabled only.
5689 (simd, sve, fp16): Removed attribute.
5690 (fp): Attr now defined in terms of 'arch'.
5691 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
5692 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
5693 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
5694 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
5695 attributes into 'arch'.
5696 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
5697 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
5698 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
5699 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
5700 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
5701 'simd' attribute into 'arch'.
5702 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
5703 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
5704 Convert use of 'fp' attribute to 'arch'.
5705 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
5706 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
5708 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
5709 (different modes) Merge 'fp' and 'simd' into 'arch'.
5710 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
5713 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
5715 PR rtl-optimization/87026
5716 * expmed.c (canonicalize_comparison): If we can no longer create
5717 pseudoregisters, don't.
5719 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
5722 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
5724 * config/arm/arm.c (speculation_barrier_libfunc): New static
5726 (arm_init_libfuncs): Initialize it.
5727 (arm_emit_speculation_barrier): New function.
5728 * config/arm/arm.md (speculation_barrier): Call
5729 arm_emit_speculation_barrier for architectures that do not have
5731 (speculation_barrier_insn): Only match on Armv7 or later.
5733 2018-08-23 Richard Biener <rguenther@suse.de>
5736 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
5739 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
5741 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
5742 of single-vector TBLs.
5743 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
5746 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
5749 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
5750 aarch64_evpc_tbl guard.
5752 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
5754 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
5757 2018-08-22 Martin Sebor <msebor@redhat.com>
5760 * tree-pretty-print.c (pretty_print_string): Add argument.
5761 (dump_generic_node): Call to pretty_print_string with string size.
5763 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
5765 PR rtl-optimization/86771
5766 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
5767 of two SETs into those two SETs, one to be placed at i2, if that SETs
5768 destination is modified between i2 and i3.
5770 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
5772 PR tree-optimization/86725
5773 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
5775 (vect_analyze_scalar_cycles_1): Check it.
5777 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
5779 PR tree-optimization/86725
5780 * tree-vect-loop.c (vect_is_simple_reduction): When treating
5781 an outer loop phi as a double reduction, make sure that the
5782 single user of the phi result is an inner loop phi.
5784 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
5786 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
5787 grouped stores with gaps to a strided group.
5789 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
5791 * tree-vect-stmts.c (get_group_load_store_type)
5792 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
5793 first statement in a group.
5795 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5797 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
5798 the sequence used in gcc/gcc.c.
5800 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5803 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
5806 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5808 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
5809 Darwin10-specific unwinder-shim.
5810 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
5811 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
5812 New to cater for Darwin10 Rosetta.
5814 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5816 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
5817 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
5819 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5824 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
5825 (dwarf2out_switch_text_section): Generate a local label for the
5826 second function sub-section and apply it as the second FDE start
5828 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
5829 second sub-section start.
5831 2018-08-22 Richard Biener <rguenther@suse.de>
5833 PR tree-optimization/86988
5834 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
5836 2018-08-22 Richard Biener <rguenther@suse.de>
5838 PR tree-optimization/86945
5839 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
5841 2018-08-22 Alexandre Oliva <oliva@adacore.com>
5843 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
5844 a comment about how uses of r2 for .sdata2 come about.
5846 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
5848 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
5850 2018-08-21 Marek Polacek <polacek@redhat.com>
5852 PR c++/86981, Implement -Wpessimizing-move.
5853 * doc/invoke.texi: Document -Wpessimizing-move.
5855 2018-08-21 Jan Hubicka <jh@suse.cz>
5857 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
5858 * tree.h (is_redundant_typedef): Remove.
5859 * dwarf2out.c (is_redundant_typedef): Turn into static function.
5861 2018-08-21 Jan Hubicka <jh@suse.cz>
5863 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
5866 2018-08-21 Tamar Christina <tamar.christina@arm.com>
5868 * expmed.c (extract_low_bits): Reject invalid subregs early.
5870 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
5873 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
5876 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
5878 * config/vxworks.h: Guard vxworks_asm_out_constructor and
5879 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
5880 * config/vxworks.c: Likewise.
5882 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
5884 * config/vxworks.c: Set targetm.have_ctors_dtors
5885 if HAVE_INITFINI_ARRAY_SUPPORT.
5886 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
5887 if HAVE_INITFINI_ARRAY_SUPPORT.
5889 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
5891 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
5892 default search path for VxWorks < 7.
5894 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
5896 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
5897 (get_format_string): Refer to c_getstr.
5899 2018-08-21 Tom de Vries <tdevries@suse.de>
5901 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
5902 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
5903 (debuginfo_early_stop): Declare.
5904 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
5905 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
5906 (debuginfo_early_stop): New function.
5907 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
5908 and debuginfo_early_stop.
5909 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
5910 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
5911 (general_init): Call debuginfo_early_init.
5912 (finalize): Call debuginfo_fini.
5913 (do_compile): Call debuginfo_init.
5914 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
5916 (@item -fdump-debug, @item -fdump-earlydebug): Add.
5918 2018-08-21 Tom de Vries <tdevries@suse.de>
5920 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
5921 flag_dump_noaddr and flag_dump_unnumbered.
5923 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
5925 * wide-int-range.cc (wide_int_range_abs): New.
5926 (wide_int_range_order_set): Rename from wide_int_range_min_max.
5927 * wide-int-range.h (wide_int_range_abs): New.
5928 (wide_int_range_min_max): New.
5929 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
5930 case to call wide_int_range_abs.
5931 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
5932 (extract_range_from_abs_expr): Delete.
5934 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
5937 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
5938 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
5941 2018-08-20 Nathan Sidwell <nathan@acm.org>
5942 Jeff Law <law@redhat.com>
5944 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
5945 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
5947 2018-08-20 David Malcolm <dmalcolm@redhat.com>
5950 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
5951 (decl_attributes): Likewise.
5952 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
5954 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
5955 * diagnostic-core.h (class auto_diagnostic_group): New class.
5956 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
5957 (diagnostic_report_diagnostic): Handle the first diagnostics within
5959 (emit_diagnostic): Add auto_diagnostic_group instance.
5961 (inform_n): Likewise.
5962 (warning): Likewise.
5963 (warning_at): Likewise.
5964 (warning_n): Likewise.
5965 (pedwarn): Likewise.
5966 (permerror): Likewise.
5968 (error_n): Likewise.
5969 (error_at): Likewise.
5971 (fatal_error): Likewise.
5972 (internal_error): Likewise.
5973 (internal_error_no_backtrace): Likewise.
5974 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
5975 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
5976 * diagnostic.h (struct diagnostic_context): Add fields
5977 "diagnostic_group_nesting_depth",
5978 "diagnostic_group_emission_count", "begin_group_cb",
5980 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
5981 Add auto_diagnostic_group instance(s).
5982 (find_explicit_erroneous_behavior): Likewise.
5983 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
5984 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
5985 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
5986 (gimplify_va_arg_expr): Likewise.
5987 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
5988 (HSA_SORRY_AT): Likewise.
5989 * ipa-devirt.c (compare_virtual_tables): Likewise.
5990 (warn_odr): Likewise.
5991 * multiple_target.c (expand_target_clones): Likewise.
5992 * opts-common.c (cmdline_handle_error): Likewise.
5993 * reginfo.c (globalize_reg): Likewise.
5994 * substring-locations.c (format_warning_n_va): Likewise.
5995 * tree-inline.c (expand_call_inline): Likewise.
5996 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
5997 * tree-ssa-loop-niter.c
5998 (do_warn_aggressive_loop_optimizations): Likewise.
5999 * tree-ssa-uninit.c (warn_uninit): Likewise.
6000 * tree.c (warn_deprecated_use): Likewise.
6002 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
6005 * config/i386/i386.md (eh_return): Always update EH return
6006 address in word_mode.
6008 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
6010 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
6011 TARGET_SPLIT_COMPLEX_ARG is defined.
6013 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
6015 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
6017 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
6020 * expr.c (expand_assignment): Assert that bitpos is positive.
6021 (store_field): Likewise
6022 (expand_expr_real_1): Make sure that bitpos is positive.
6023 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
6026 2018-08-20 Nathan Sidwell <nathan@acm.org>
6028 * Makefile.in (CPP_ID_DATA_H): Delete.
6029 (CPP_INTERNAL_H): Don't add it.
6030 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
6031 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
6033 2018-08-20 Richard Biener <rguenther@suse.de>
6035 PR tree-optimization/78655
6036 * tree-vrp.c (extract_range_from_binary_expr_1): Make
6037 pointer + offset nonnull if either operand is nonnull work.
6039 2018-08-20 Tom de Vries <tdevries@suse.de>
6041 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
6042 unless the referenced die describes the added property using
6043 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
6044 Otherwise, add a DW_AT_location to the referenced die.
6046 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
6049 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
6050 register_operand when calling ix86_set_reg_reg_cost.
6051 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
6052 Set *total to 0 for operands that satisfy x86_64_immediate_operand
6053 predicate and to 1 otherwise.
6055 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
6057 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
6058 emit a diagnostic that it is not supported and reset the option.
6059 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
6060 supported and consume the option. (ASM_FINAL_SPEC): New.
6062 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
6064 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
6067 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
6069 C-SKY port: Documentation
6071 * doc/extend.texi (C-SKY Function Attributes): New section.
6072 * doc/invoke.texi (Option Summary): Add C-SKY options.
6073 (C-SKY Options): New section.
6074 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
6076 2018-08-17 Jojo <jijie_rong@c-sky.com>
6077 Huibin Wang <huibin_wang@c-sky.com>
6078 Sandra Loosemore <sandra@codesourcery.com>
6079 Chung-Lin Tang <cltang@codesourcery.com>
6081 C-SKY port: Backend implementation
6083 * config/csky/*: New.
6084 * common/config/csky/*: New.
6086 2018-08-17 Jojo <jijie_rong@c-sky.com>
6087 Huibin Wang <huibin_wang@c-sky.com>
6088 Sandra Loosemore <sandra@codesourcery.com>
6089 Chung-Lin Tang <cltang@codesourcery.com>
6090 Andrew Jenner <andrew@codesourcery.com>
6092 C-SKY port: Configury
6094 * config.gcc (csky-*-*): New.
6095 * configure.ac: Add csky to targets for dwarf2 debug_line support.
6096 * configure: Regenerated.
6098 2018-08-17 David Malcolm <dmalcolm@redhat.com>
6100 * dump-context.h: Include "dumpfile.h".
6101 (dump_context::dump_printf_va): Convert final param from va_list
6102 to va_list *. Convert from ATTRIBUTE_PRINTF to
6103 ATTRIBUTE_GCC_DUMP_PRINTF.
6104 (dump_context::dump_printf_loc_va): Likewise.
6105 * dumpfile.c: Include "stringpool.h".
6106 (make_item_for_dump_printf_va): Delete.
6107 (make_item_for_dump_printf): Delete.
6108 (class dump_pretty_printer): New class.
6109 (dump_pretty_printer::dump_pretty_printer): New ctor.
6110 (dump_pretty_printer::emit_items): New member function.
6111 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
6113 (dump_pretty_printer::emit_item): New member function.
6114 (dump_pretty_printer::stash_item): New member function.
6115 (dump_pretty_printer::format_decoder_cb): New member function.
6116 (dump_pretty_printer::decode_format): New member function.
6117 (dump_context::dump_printf_va): Reimplement in terms of
6118 dump_pretty_printer.
6119 (dump_context::dump_printf_loc_va): Convert final param from va_list
6121 (dump_context::begin_scope): Reimplement call to
6122 make_item_for_dump_printf.
6123 (dump_printf): Update for change to dump_printf_va.
6124 (dump_printf_loc): Likewise.
6125 (selftest::test_capture_of_dump_calls): Convert "stmt" from
6126 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
6127 with %T, %E, and %G.
6128 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
6129 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
6130 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
6131 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
6132 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
6133 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
6134 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
6135 within a dump_printf_loc call to "%wu".
6136 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
6137 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
6138 missing space after "=".
6139 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
6140 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
6141 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
6142 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
6143 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
6144 duplicate "vectorized" from message.
6146 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
6148 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
6149 polyNxK_t element's TYPE_STRING_FLAG.
6151 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
6153 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
6154 (they were unnamed before). Fix comments.
6156 2018-08-17 Nathan Sidwell <nathan@acm.org>
6158 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
6160 2018-08-17 Richard Biener <rguenther@suse.de>
6162 PR tree-optimization/86841
6163 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
6165 2018-08-17 Martin Liska <mliska@suse.cz>
6167 * common.opt: Remove Warn, Init and Report for options with
6168 Ignore/Deprecated flag. Warning is done automatically for
6170 * config/i386/i386.opt: Likewise.
6171 * config/ia64/ia64.opt: Likewise.
6172 * config/rs6000/rs6000.opt: Likewise.
6173 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
6174 Remove usage of flag_check_pointer_bounds.
6175 * lto-wrapper.c (merge_and_complain): Do not handle
6176 OPT_fcheck_pointer_bounds.
6177 (append_compiler_options): Likewise.
6178 * opt-functions.awk: Do not handle Deprecated.
6179 * optc-gen.awk: Check that Var, Report and Init are not
6180 used for an option with Ignore/Deprecated flag.
6181 * opts-common.c (decode_cmdline_option): Do not report
6183 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
6185 * opts.h (struct cl_option): Remove cl_deprecated flag.
6186 (CL_ERR_DEPRECATED): Remove error enum value.
6188 2018-08-17 Richard Biener <rguenther@suse.de>
6191 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
6192 across a va-arg-pack using call adjust its return value accordingly.
6194 2018-08-16 Martin Sebor <msebor@redhat.com>
6196 PR tree-optimization/86853
6197 * gimple-ssa-sprintf.c (struct format_result): Rename member.
6198 (struct fmtresult): Add member and initialize it in ctors.
6199 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
6200 (format_string): Handle %S the same as %ls. Set MAYFAIL.
6201 (format_directive): Set POSUNDER4K when MAYFAIL is set.
6202 (parse_directive): Handle %C same as %c.
6203 (sprintf_dom_walker::compute_format_length): Adjust.
6204 (is_call_safe): Adjust.
6206 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
6208 * builtins.c (c_strlen): Add new parameter eltsize. Use it
6209 for determining how to count the elements.
6210 * builtins.h (c_strlen): Adjust prototype.
6211 * expr.c (string_constant): Add new parameter mem_size.
6212 Set *mem_size appropriately.
6213 * expr.h (string_constant): Adjust protoype.
6214 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
6215 * gimple-fold.h (get_range_strlen): Adjust prototype.
6216 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
6217 (format_string): Call get_string_length with eltsize.
6219 2018-08-16 David Malcolm <dmalcolm@redhat.com>
6221 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
6222 to emit the span, rather than setting it as the prefix.
6224 2018-08-16 David Malcolm <dmalcolm@redhat.com>
6226 * diagnostic-show-locus.c (layout::start_annotation_line): Add
6227 "margin_char" parameter, defaulting to space. Use it in place
6228 of pp_space for the initial part of the margin.
6229 (layout::print_leading_fixits): Use '+' when filling the margin
6230 of line-insertion fix-it hints.
6232 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6234 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
6237 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6239 * config/rs6000/altivec.md: Don't set length attribute to the default
6241 * config/rs6000/darwin.md: Ditto.
6242 * config/rs6000/dfp.md: Ditto.
6243 * config/rs6000/htm.md: Ditto.
6244 * config/rs6000/rs6000.md: Ditto.
6245 * config/rs6000/sync.md: Ditto.
6246 * config/rs6000/vsx.md: Ditto.
6248 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6250 * config/rs6000/altivec.md: Don't set length attribute to the default
6251 value, for branch instructions.
6252 * config/rs6000/darwin.md: Ditto.
6253 * config/rs6000/rs6000.md: Ditto.
6255 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6257 * config/rs6000/rs6000.md (length): Always define as const_int 4.
6258 (unnamed conditional branch define_insn): Set length to 4 or 8
6259 depending on offset.
6260 (<bd>_<mode>): Similar, for alternative 0.
6261 (<bd>tf_<mode>): Ditto.
6263 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6265 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
6267 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
6269 * doc/rtl.texi: Replace old RTX class names with new names.
6272 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
6274 * expmed.h (canonicalize_comparison): New declaration.
6275 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
6276 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
6277 * optabs.c (prepare_cmp_insn): Likewise.
6278 * rtl.h (unsigned_condition_p): New function which checks if a
6279 comparison operator is unsigned.
6281 2018-08-16 Nathan Sidwell <nathan@acm.org>
6283 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
6284 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
6286 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6289 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
6290 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
6291 (mov<mov>): ..this and enable unconditionally.
6293 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6295 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
6297 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
6299 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
6300 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
6301 "Common" with "Target".
6303 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
6305 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
6306 * doc/invoke.texi (mmitigate-rop): Remove.
6307 * config/i386/i386.c: Do not include "regrename.h".
6308 (ix86_rop_should_change_byte_p, reg_encoded_number)
6309 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
6311 (ix86_reorg): Remove call to ix86_mitigate_rop.
6312 * config/i386/i386.md (attr "modrm_class"): Remove.
6313 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
6314 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
6315 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
6317 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6319 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
6320 allow folding of mergeh() and mergel() for the float and double types.
6321 (fold_mergehl_helper): Rework to handle building a permute tree
6324 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
6326 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
6329 2018-08-15 David Malcolm <dmalcolm@redhat.com>
6331 * common.opt (fdiagnostics-show-labels): New option.
6332 * diagnostic-show-locus.c (class layout_range): Add field
6334 (class layout): Add field "m_show_labels_p".
6335 (layout_range::layout_range): Add param "label" and use it to
6337 (make_range): Pass in NULL for new "label" param of layout_range's
6339 (layout::layout): Initialize m_show_labels_p.
6340 (layout::maybe_add_location_range): Pass in loc_range->m_label
6341 when constructing layout_range instances.
6342 (struct line_label): New struct.
6343 (layout::print_any_labels): New member function.
6344 (layout::print_line): Call it if label-printing is enabled.
6345 (selftest::test_one_liner_labels): New test.
6346 (selftest::test_diagnostic_show_locus_one_liner): Call it.
6347 * diagnostic.c (diagnostic_initialize): Initialize
6348 context->show_labels_p.
6349 * diagnostic.h (struct diagnostic_context): Add field
6351 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
6352 -fno-diagnostics-show-labels.
6353 * dwarf2out.c (gen_producer_string): Add
6354 OPT_fdiagnostics_show_labels to the ignored options.
6355 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
6357 (gcc_rich_location::maybe_add_expr): Likewise.
6358 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
6359 label" param, defaulting to NULL.
6360 (gcc_rich_location::add_expr): Add "label" param.
6361 (gcc_rich_location::maybe_add_expr): Likewise.
6362 (class text_range_label): New class.
6363 (class range_label_for_type_mismatch): New class.
6364 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
6365 of format_warning_va.
6366 (fmtwarn_n): Likewise for new params of format_warning_n_va.
6367 * lto-wrapper.c (merge_and_complain): Add
6368 OPT_fdiagnostics_show_labels to the "pick one setting" options.
6369 (append_compiler_options): Likewise to the dropped options.
6370 (append_diag_options): Likewise to the passed-on options.
6371 * opts.c (common_handle_option): Handle the new option.
6372 * selftest-diagnostic.c
6373 (test_diagnostic_context::test_diagnostic_context): Enable
6375 * substring-locations.c: Include "gcc-rich-location.h".
6376 (format_warning_n_va): Add "fmt_label" and "param_label" params
6377 and use them as appropriate.
6378 (format_warning_va): Add "fmt_label" and "param_label" params,
6379 passing them on to format_warning_n_va.
6380 (format_warning_at_substring): Likewise.
6381 (format_warning_at_substring_n): Likewise.
6382 * substring-locations.h (format_warning_va): Add "fmt_label" and
6383 "param_label" params.
6384 (format_warning_n_va): Likewise.
6385 (format_warning_at_substring): Likewise.
6386 (format_warning_at_substring_n): Likewise.
6387 * toplev.c (general_init): Initialize global_dc->show_labels_p.
6389 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
6392 * builtins.c (expand_builtin_memcmp): Do not expand the call
6393 when overflow is detected.
6395 2018-08-15 Martin Sebor <msebor@redhat.com>
6397 PR tree-optimization/71625
6398 * config/aarch64/aarch64-builtins.c
6399 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
6401 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
6403 * config/s390/s390.c (s390_reorg): Remove loop.
6405 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
6408 (darwin_function_switched_text_sections): Delete.
6409 * gcc/config/darwin.h
6410 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
6412 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
6415 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
6416 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
6417 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
6419 2018-08-15 Martin Liska <mliska@suse.cz>
6421 PR tree-optimization/86925
6422 * predict.c (expr_expected_value_1): When taking
6423 later predictor, assign also probability.
6424 Use fold_build2_initializer_loc in order to fold
6425 the expression in -frounding-math.
6427 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
6429 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
6431 (expand_vec_perm_1): Try the new method.
6433 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
6436 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
6437 Check whether lra_live_max_point is 0 before dividing.
6439 2018-08-14 Martin Sebor <msebor@redhat.com>
6441 PR tree-optimization/86650
6442 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
6443 (vrp_prop::check_mem_ref): Same.
6445 2018-08-13 Liu Hao <lh_mouse@126.com>
6447 * pretty-print.c (eat_esc_sequence): Swap the foreground and
6448 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
6449 and clear it thereafter, as it only works for DBCS.
6451 2018-08-13 Liu Hao <lh_mouse@126.com>
6453 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
6454 handle returned by _get_osf_handle().
6456 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6458 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
6459 for folding vec_perm.
6461 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6463 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
6464 Add support for gimple-folding of vec_pack() and vec_unpack()
6467 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6469 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
6470 vec_xst variants to the list.
6471 (rs6000_gimple_fold_builtin): Add support for folding unaligned
6472 vector loads and stores.
6474 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
6476 * config.gcc (rs6000-ibm-aix4.x): Delete.
6477 (rs6000-ibm-aix5.1): Delete.
6478 (rs6000-ibm-aix5.2): Delete.
6479 (rs6000-ibm-aix5.3): Delete.
6480 * config/rs6000/aix43.h: Delete.
6481 * config/rs6000/aix51.h: Delete.
6482 * config/rs6000/aix52.h: Delete.
6483 * config/rs6000/t-aix43: Delete.
6485 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
6487 * config/s390/s390.c (s390_decompose_constant_pool_ref):
6489 (s390_decompose_address): Factor out constant pool ref
6492 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6494 * config/nds32/nds32-predicates.c
6495 (nds32_can_use_bclr_p): Change return type as bool.
6496 (nds32_can_use_bset_p): Ditto.
6497 (nds32_can_use_btgl_p): Ditto.
6498 (nds32_can_use_bitci_p): Ditto.
6499 * config/nds32/nds32-protos.h
6500 (nds32_can_use_bclr_p): Change declaration.
6501 (nds32_can_use_bset_p): Ditto.
6502 (nds32_can_use_btgl_p): Ditto.
6503 (nds32_can_use_bitci_p): Ditto.
6505 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6507 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
6508 Support -msched-prolog-epilog option.
6509 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
6511 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6513 * common/config/nds32/nds32-common.c
6514 (nds32_option_optimization_table): Enalbe -malways-align.
6516 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6518 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
6520 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
6521 OPT_misr_secure_ case.
6522 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
6523 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
6524 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
6526 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
6527 (nds32_isr_info): New field security_level.
6528 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
6529 * config/nds32/nds32.md (return_internal): Consider critical attribute.
6530 * config/nds32/nds32.opt (misr-secure): New option.
6531 * config/nds32/nds32_init.inc: New file.
6532 * config/nds32/nds32_isr.h: New file.
6534 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
6536 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
6537 Update comment for atomic instructions.
6538 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
6539 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
6541 (atomic_loaddi): Revise fence expansion to only emit fence prior to
6542 load for __ATOMIC_SEQ_CST model.
6543 (atomic_loaddi_1): Remove float register target.
6544 (atomic_storedi): Handle CONST_INT values.
6545 (atomic_storedi_1): Remove float register source. Add special case
6547 (memory_barrier): New expander and insn.
6549 2018-08-11 Jakub Jelinek <jakub@redhat.com>
6551 PR tree-optimization/86835
6552 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
6553 new_stmt after def_gsi, make sure to insert new_square_stmt after
6554 that stmt, not 2 stmts before it.
6556 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
6559 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
6562 2018-08-10 Martin Liska <mliska@suse.cz>
6565 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
6567 * builtins.c (expand_builtin_expect_with_probability):
6569 (expand_builtin_expect_with_probability): New function.
6570 (build_builtin_expect_predicate): Add new argumnet probability
6571 for BUILT_IN_EXPECT_WITH_PROBABILITY.
6572 (fold_builtin_expect):
6575 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
6576 * builtins.h (fold_builtin_expect): Set new argument.
6577 * doc/extend.texi: Document __builtin_expect_with_probability.
6578 * doc/invoke.texi: Likewise.
6579 * gimple-fold.c (gimple_fold_call): Pass new argument.
6580 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
6581 also BUILT_IN_EXPECT_WITH_PROBABILITY.
6582 * predict.c (get_predictor_value): New function.
6583 (expr_expected_value): Add new argument probability. Assume
6584 that predictor and probability are always non-null.
6585 (expr_expected_value_1): Likewise. For __builtin_expect and
6586 __builtin_expect_with_probability set probability. Handle
6587 combination in binary expressions.
6588 (tree_predict_by_opcode): Simplify code by simply calling
6589 get_predictor_value.
6590 (pass_strip_predict_hints::execute): Add handling of
6591 BUILT_IN_EXPECT_WITH_PROBABILITY.
6592 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
6594 * tree.h (DECL_BUILT_IN_P): New function.
6596 2018-08-10 Martin Liska <mliska@suse.cz>
6598 PR tree-optimization/85799
6599 * passes.def: Add argument for pass_strip_predict_hints.
6600 * predict.c (class pass_strip_predict_hints): Add new argument
6602 (strip_predictor_early): New function.
6603 (pass_strip_predict_hints::execute): Call the function to
6605 (strip_predict_hints): New function.
6606 * predict.def: Fix comment.
6608 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
6610 * Makefile.in: Clarify which tm.texi to copy over to assert the
6611 right to grant a GFDL license for all.
6613 2018-08-09 Jeff Law <law@redhat.com>
6615 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
6618 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
6620 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
6623 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
6625 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
6626 reductions for variable-length vectors.
6628 2018-08-09 David Malcolm <dmalcolm@redhat.com>
6631 * common.opt (fdiagnostics-show-line-numbers): New option.
6632 * diagnostic-show-locus.c (class layout): Add fields
6633 "m_show_line_numbers_p" and "m_linenum_width";
6634 (num_digits): New function.
6635 (test_num_digits): New function.
6636 (layout::layout): Initialize new fields. Update m_x_offset
6637 logic to handle any left margin.
6638 (layout::print_source_line): Print line number when requested.
6639 (layout::start_annotation_line): New member function.
6640 (layout::print_annotation_line): Call it.
6641 (layout::print_leading_fixits): Likewise.
6642 (layout::print_trailing_fixits): Likewise. Update calls to
6643 move_to_column for new parameter.
6644 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
6645 it to potentially call start_annotation_line.
6646 (layout::show_ruler): Call start_annotation_line.
6647 (selftest::test_line_numbers_multiline_range): New selftest.
6648 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
6649 and selftest::test_line_numbers_multiline_range.
6650 * diagnostic.c (diagnostic_initialize): Initialize
6651 show_line_numbers_p.
6652 * diagnostic.h (struct diagnostic_context): Add field
6653 "show_line_numbers_p".
6654 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
6655 -fno-diagnostics-show-line-numbers.
6656 * dwarf2out.c (gen_producer_string): Add
6657 OPT_fdiagnostics_show_line_numbers to the ignored options.
6658 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
6659 one setting" options.
6660 (append_compiler_options): Likewise to the dropped options.
6661 (append_diag_options): Likewise to the passed-on options.
6662 * opts.c (common_handle_option): Handle the new option.
6663 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
6665 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
6667 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
6668 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
6669 third argument of type "const signed char" to descriptions of
6670 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
6671 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
6672 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
6673 __builtin_bcdsub_ov functions.
6675 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
6677 PR tree-optimization/86858
6678 * tree-vect-loop.c (vect_is_simple_reduction): Restore
6679 flow_bb_inside_loop_p calls.
6681 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
6683 PR tree-optimization/86871
6684 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
6685 instead of gimple_assign_lhs.
6687 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
6690 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
6691 register constraint to operand 0.
6692 (add<mode>3_carryinC): Likewise.
6693 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
6695 2018-08-09 Martin Liska <mliska@suse.cz>
6698 * common.opt: Remove extra line.
6700 2018-08-09 Martin Liska <mliska@suse.cz>
6702 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
6703 at the end of a line, make first letter capital and end up
6704 a sentence with a dot.
6705 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
6706 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
6707 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
6708 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
6709 (PARAM_MAX_ISL_OPERATIONS): Likewise.
6710 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
6711 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
6712 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
6713 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
6714 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
6715 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
6716 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
6717 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
6718 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
6719 (PARAM_TREE_REASSOC_WIDTH): Likewise.
6720 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
6721 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
6722 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
6724 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
6727 * config/s390/s390.c (s390_option_override_internal): Reduce the
6728 stack-clash-protection-probe-interval param if it would be too big
6731 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
6734 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
6735 * config/m68k/m68k.c (handle_move_double): Don't call
6736 m68k_final_prescan_insn.
6737 (m68k_adjust_decorated_operand): Renamed from
6738 m68k_final_prescan_insn, remove first and third operand and
6740 (print_operand): Call it.
6741 (print_operand_address): Call it.
6743 2018-08-08 Nathan Sidwell <nathan@acm.org>
6745 * diagnostic.c (diagnostic_report_current_module): Use
6746 linemap_included_from & linemap_included_from_linemap.
6748 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
6750 * config/aarch64/aarch64-cores.def: Add phecda core.
6751 * config/aarch64/aarch64-tune.md: Regenerate.
6752 * doc/invoke.texi: Add phecda core.
6754 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
6757 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
6759 * config/s390/s390.md ("movti"): Add more alternatives for
6760 constant to GPR copies.
6762 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
6764 * config/s390/s390.c: Fix whitespace damage throughout the file.
6765 * config/s390/s390.h: Likewise.
6766 * config/s390/tpf.h: Likewise.
6768 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
6770 * config/s390/s390.c (s390_loadrelative_operand_p):
6771 Remove TARGET_CPU_ZARCH usages.
6772 (s390_rtx_costs): Likewise.
6773 (s390_legitimate_constant_p): Likewise.
6774 (s390_cannot_force_const_mem): Likewise.
6775 (legitimate_reload_constant_p): Likewise.
6776 (s390_preferred_reload_class): Likewise.
6777 (legitimize_pic_address): Likewise.
6778 (legitimize_tls_address): Likewise.
6779 (s390_split_branches): Removed.
6780 (s390_add_execute): Removed.
6781 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
6782 (s390_mainpool_start): Likewise.
6783 (s390_mainpool_finish): Likewise.
6784 (s390_mainpool_cancel): Removed.
6785 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
6786 (s390_chunkify_cancel): Likewise.
6787 (s390_return_addr_rtx): Likewise.
6788 (s390_register_info): Remove split_branches_pending_p uages.
6789 (s390_optimize_register_info): Likewise.
6790 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
6791 split_branches_pending_p usages.
6792 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
6793 (s390_load_got): Likewise.
6794 (s390_expand_split_stack_prologue): Likewise.
6795 (output_asm_nops): Likewise.
6796 (s390_function_profiler): Likewise.
6797 (s390_emit_call): Likewise.
6798 (s390_conditional_register_usage): Likewise.
6799 (s390_optimize_prologue): Likewise.
6800 (s390_reorg): Remove TARGET_CPU_ZARCH and
6801 split_branches_pending_p usages.
6802 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
6804 (s390_output_indirect_thunk_function): Likewise.
6805 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
6806 (TARGET_CPU_ZARCH_P): Removed.
6807 (struct machine_function): Remove split_branches_pending_p.
6808 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
6810 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
6812 * common/config/s390/s390-common.c (processor_flags_table):
6814 * config.gcc: Remove with_arch/with_tune support.
6815 * config/s390/2064.md: Remove cpu attribute comparisons.
6816 * config/s390/driver-native.c (s390_host_detect_local_cpu):
6818 * config/s390/linux.h (ASM_SPEC):
6819 Remove -march support.
6820 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
6821 Use a table to get an arch level.
6822 * config/s390/s390-opts.h (enum processor_type):
6824 * config/s390/s390.c
6825 (processor_table): Remove entries, add arch_level values.
6826 (s390_issue_rate): Remove cases.
6827 (s390_option_override): Adjust
6828 s390_option_override_internal() call.
6829 (s390_option_override_internal): Remove deprecation warning.
6830 (s390_valid_target_attribute_tree): Adjust
6831 s390_option_override_internal() call.
6832 * config/s390/s390.h (struct s390_processor):
6833 Share with s390-c.c, add arch_level field.
6834 * config/s390/s390.md:
6835 Remove occurrences in cpu attribute.
6836 * config/s390/s390.opt: Remove -march/-mtune support.
6837 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
6838 * doc/invoke.texi: Remove deprecation warning.
6840 2018-08-08 Luis Machado <luis.machado@linaro.org>
6842 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
6844 (qdf24xx_tunings): Set vector cost structure to
6845 qdf24xx_vector_cost.
6847 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
6848 <register_sextend>: Set to 3.
6850 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
6853 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
6854 * config/aarch64/aarch64-simd.md
6855 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
6856 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
6857 (aarch64_frecpx<mode>): ...this new pattern.
6858 * config/aarch64/aarch64-simd-builtins.def: Remove comment
6859 about aarch64_frecp<FRECP:frecp_suffix><mode>.
6861 2018-08-07 Martin Liska <mliska@suse.cz>
6864 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
6865 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
6866 * predict.def (PRED_MALLOC_NONNULL): New predictor.
6867 * doc/extend.texi: Document that malloc attribute adds
6870 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
6873 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6874 Define to speculation_safe_value_not_needed.
6876 2018-08-06 Jeff Law <law@redhat.com>
6878 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
6879 the vr_values instance to cprop_into_stmt.
6880 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
6881 (cprop_operand): Also query EVRP to determine if OP is a constant.
6883 2018-08-06 Nathan Sidwell <nathan@acm.org>
6885 * diagnostic.c (diagnostic_report_current_module): Reroll
6886 included-at loop. Translate text.
6888 2018-08-06 David Malcolm <dmalcolm@redhat.com>
6890 * function-tests.c (selftest::test_expansion_to_rtl): Call
6891 free_after_compilation.
6893 2018-08-06 Alan Hayward <alan.hayward@arm.com>
6895 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
6897 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
6899 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
6900 loops with memory block operations from getting unrolled.
6902 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
6905 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6906 Define to speculation_safe_value_not_needed.
6908 2018-08-06 Jeff Law <law@redhat.com>
6910 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
6913 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
6916 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
6917 with all enabled __intN types.
6919 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
6921 2018-08-06 Alan Hayward <alan.hayward@arm.com>
6923 * alias.c (record_set): Check for clobber high.
6924 * cfgexpand.c (expand_gimple_stmt): Likewise.
6925 * combine-stack-adj.c (single_set_for_csa): Likewise.
6926 * combine.c (find_single_use_1): Likewise.
6927 (set_nonzero_bits_and_sign_copies): Likewise.
6928 (get_combine_src_dest): Likewise.
6929 (is_parallel_of_n_reg_sets): Likewise.
6930 (try_combine): Likewise.
6931 (record_dead_and_set_regs_1): Likewise.
6932 (reg_dead_at_p_1): Likewise.
6933 (reg_dead_at_p): Likewise.
6934 * dce.c (deletable_insn_p): Likewise.
6935 (mark_nonreg_stores_1): Likewise.
6936 (mark_nonreg_stores_2): Likewise.
6937 * df-scan.c (df_find_hard_reg_defs): Likewise.
6938 (df_uses_record): Likewise.
6939 (df_get_call_refs): Likewise.
6940 * dwarf2out.c (mem_loc_descriptor): Likewise.
6941 * haifa-sched.c (haifa_classify_rtx): Likewise.
6942 * ira-build.c (create_insn_allocnos): Likewise.
6943 * ira-costs.c (scan_one_insn): Likewise.
6944 * ira.c (equiv_init_movable_p): Likewise.
6945 (rtx_moveable_p): Likewise.
6946 (interesting_dest_for_shprep): Likewise.
6947 * jump.c (mark_jump_label_1): Likewise.
6948 * postreload-gcse.c (record_opr_changes): Likewise.
6949 * postreload.c (reload_cse_simplify): Likewise.
6950 (struct reg_use): Add source expr.
6951 (reload_combine): Check for clobber high.
6952 (reload_combine_note_use): Likewise.
6953 (reload_cse_move2add): Likewise.
6954 (move2add_note_store): Likewise.
6955 * print-rtl.c (print_pattern): Likewise.
6956 * recog.c (decode_asm_operands): Likewise.
6957 (store_data_bypass_p): Likewise.
6958 (if_test_bypass_p): Likewise.
6959 * regcprop.c (kill_clobbered_value): Likewise.
6960 (kill_set_value): Likewise.
6961 * reginfo.c (reg_scan_mark_refs): Likewise.
6962 * reload1.c (maybe_fix_stack_asms): Likewise.
6963 (eliminate_regs_1): Likewise.
6964 (elimination_effects): Likewise.
6965 (mark_not_eliminable): Likewise.
6966 (scan_paradoxical_subregs): Likewise.
6967 (forget_old_reloads_1): Likewise.
6968 * reorg.c (find_end_label): Likewise.
6969 (try_merge_delay_insns): Likewise.
6970 (redundant_insn): Likewise.
6971 (own_thread_p): Likewise.
6972 (fill_simple_delay_slots): Likewise.
6973 (fill_slots_from_thread): Likewise.
6974 (dbr_schedule): Likewise.
6975 * resource.c (update_live_status): Likewise.
6976 (mark_referenced_resources): Likewise.
6977 (mark_set_resources): Likewise.
6978 * rtl.c (copy_rtx): Likewise.
6979 * rtlanal.c (reg_referenced_p): Likewise.
6980 (single_set_2): Likewise.
6981 (noop_move_p): Likewise.
6982 (note_stores): Likewise.
6983 * sched-deps.c (sched_analyze_reg): Likewise.
6984 (sched_analyze_insn): Likewise.
6986 2018-08-06 Alan Hayward <alan.hayward@arm.com>
6988 * cse.c (invalidate_reg): New function extracted from...
6989 (invalidate): ...here.
6990 (canonicalize_insn): Check for clobber high.
6991 (invalidate_from_clobbers): invalidate clobber highs.
6992 (invalidate_from_sets_and_clobbers): Likewise.
6993 (count_reg_usage): Check for clobber high.
6994 (insn_live_p): Likewise.
6995 * cselib.c (cselib_expand_value_rtx_1):Likewise.
6996 (cselib_invalidate_regno): Check for clobber in setter.
6997 (cselib_invalidate_rtx): Pass through setter.
6998 (cselib_invalidate_rtx_note_stores):
6999 (cselib_process_insn): Check for clobber high.
7000 * cselib.h (cselib_invalidate_rtx): Add operand.
7002 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7004 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
7005 (mark_not_eliminable): Likewise.
7006 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
7007 * lra-lives.c (process_bb_lives): Check for clobber high.
7008 * lra.c (new_insn_reg): Remember clobber highs.
7009 (collect_non_operand_hard_regs): Check for clobber high.
7010 (lra_set_insn_recog_data): Likewise.
7011 (add_regs_to_insn_regno_info): Likewise.
7012 (lra_update_insn_regno_info): Likewise.
7014 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7016 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
7017 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
7019 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7021 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
7022 (copy_insn_1): Likewise.
7023 (gen_hard_reg_clobber_high): New gen function.
7024 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
7025 * genemit.c (gen_exp): Likewise.
7026 (gen_emit_seq): Pass through info.
7027 (gen_insn): Check for CLOBBER_HIGH.
7028 (gen_expand): Pass through info.
7029 (gen_split): Likewise.
7030 (output_add_clobbers): Likewise.
7031 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
7032 (remove_clobbers): Likewise.
7033 * rtl.h (gen_hard_reg_clobber_high): New declaration.
7035 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7037 * doc/rtl.texi (clobber_high): Add.
7038 (parallel): Add in clobber high
7039 * rtl.c (rtl_check_failed_code3): Add function.
7040 * rtl.def (CLOBBER_HIGH): Add expression.
7041 * rtl.h (RTL_CHECKC3): Add macro.
7042 (rtl_check_failed_code3): Add declaration.
7043 (XC3EXP): Add macro.
7045 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
7048 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
7049 cfun->machine->max_used_stack_alignment if needed.
7051 2018-08-04 Martin Sebor <msebor@redhat.com>
7053 PR tree-optimization/86571
7054 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
7057 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
7060 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7063 2018-08-03 Jeff Law <law@redhat.com>
7066 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7067 Define to speculation_safe_value_not_needed.
7069 2018-08-03 David Malcolm <dmalcolm@redhat.com>
7071 * doc/gcov.texi (-x): Remove duplicate "to".
7072 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
7073 (-Wif-not-aligned): Remove duplicate "is".
7074 (-flto): Remove duplicate "the".
7075 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
7076 duplicate "v5.00.b".
7077 (MSP430 Options): Remove duplicate "and" from the description
7078 of "-mgprel-sec=regexp".
7079 (x86 Options): Remove duplicate copies of "vmldLog102" and
7080 vmlsLog104 from description of "-mveclibabi=type".
7082 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
7084 * internal-fn.h (first_commutative_argument): Declare.
7085 * internal-fn.c (first_commutative_argument): New function.
7086 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
7087 restrictions for pattern statements. Use first_commutative_argument
7088 to look for commutative operands in calls to internal functions.
7090 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
7092 * Makefile.in (wide-int-range.o): New.
7093 * tree-vrp.c: Move all the wide_int_* functions to...
7094 * wide-int-range.cc: ...here.
7095 * tree-vrp.h: Move all the wide_int_* prototypes to...
7096 * wide-int-range.h: ...here.
7098 2018-08-03 Tom de Vries <tdevries@suse.de>
7100 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
7102 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
7103 * except.c (output_function_exception_table): Do early exit if
7104 targetm_common.except_unwind_info (&global_options) == UI_NONE.
7106 2018-08-03 Martin Liska <mliska@suse.cz>
7108 * predict.c (dump_prediction): Change to 2 digits
7111 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
7113 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
7115 (falkor_am_1_gtov_gtov): ... a new insn reservation.
7117 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
7119 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
7120 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
7121 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
7122 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
7123 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
7125 2018-08-02 David Malcolm <dmalcolm@redhat.com>
7127 * diagnostic-show-locus.c (diagnostic_show_locus): Use
7128 pp_take_prefix when saving the existing prefix.
7129 * diagnostic.c (diagnostic_append_note): Likewise.
7130 * langhooks.c (lhd_print_error_function): Likewise.
7131 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
7132 param's type. Free the existing prefix.
7133 (pp_take_prefix): New function.
7134 (pretty_printer::pretty_printer): Drop the prefix parameter.
7135 Rename the length parameter to match the comment.
7136 (pretty_printer::~pretty_printer): Free the prefix.
7137 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
7139 (struct pretty_printer): Drop the "const" from "prefix" field's
7140 type and clarify memory management.
7141 (pp_set_prefix): Drop the "const" from the 2nd param.
7142 (pp_take_prefix): New decl.
7144 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
7146 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
7147 for word_mode_ok here instead of passing as argument.
7148 (expand_block_compare): Change select_block_compare_mode() call.
7149 (expand_strncmp_gpr_sequence): New function.
7150 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
7152 2018-08-02 Jeff Law <law@redhat.com>
7155 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7156 Define to speculation_safe_value_not_needed.
7159 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7160 Define to speculation_safe_value_not_needed.
7162 2018-08-02 Tom de Vries <tdevries@suse.de>
7165 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
7166 function. Return UI_TARGET unconditionally.
7167 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
7168 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
7170 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7172 * genemit.c (print_overload_test): Fix typo.
7174 2018-08-02 Richard Biener <rguenther@suse.de>
7176 PR tree-optimization/86816
7177 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
7178 which checks for value availability before querying it.
7179 (gvn_uses_equal): Use it.
7180 (same_succ_hash): Likewise.
7181 (gimple_equal_p): Likewise.
7183 2018-08-02 Nick Clifton <nickc@redhat.com>
7186 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7187 Define to speculation_safe_value_not_needed.
7190 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7191 Define to speculation_safe_value_not_needed.
7194 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7195 Define to speculation_safe_value_not_needed.
7198 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7199 Define to speculation_safe_value_not_needed.
7202 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7203 Define to speculation_safe_value_not_needed.
7206 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7207 Define to speculation_safe_value_not_needed.
7210 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7211 Define to speculation_safe_value_not_needed.
7214 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7215 Define to speculation_safe_value_not_needed.
7218 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
7219 speculation_safe_value_not_needed.
7222 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
7223 to speculation_safe_value_not_needed.
7225 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7227 * doc/md.texi: Expand the documentation of instruction names
7228 to mention port-local uses. Document '@' in pattern names.
7229 * read-md.h (overloaded_instance, overloaded_name): New structs.
7231 (md_reader::handle_overloaded_name): New member function.
7232 (md_reader::get_overloads): Likewise.
7233 (md_reader::m_first_overload): New member variable.
7234 (md_reader::m_next_overload_ptr): Likewise.
7235 (md_reader::m_overloads_htab): Likewise.
7236 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
7237 m_next_overload_ptr and m_overloads_htab.
7238 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
7239 (get_mode_token, get_code_token, get_int_token): New functions.
7240 (map_attr_string): Add an optional argument that passes back
7241 the associated iterator.
7242 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
7243 (md_reader::handle_overloaded_name, add_overload_instance): New
7245 (apply_iterators): Handle '@' names. Report an error if '@'
7246 is used without iterators.
7247 (initialize_iterators): Initialize the new iterator_group fields.
7248 * genopinit.c (handle_overloaded_code_for)
7249 (handle_overloaded_gen): New functions.
7250 (main): Use them to print declarations of maybe_code_for_* and
7251 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
7252 * genemit.c (print_overload_arguments, print_overload_test)
7253 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
7254 (main): Use it to print definitions of maybe_code_for_* and
7255 maybe_gen_* functions.
7256 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
7257 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
7258 instead of explicit mode checks.
7259 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
7260 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
7261 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
7262 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
7263 (aarch64_expand_compare_and_swap): Likewise
7264 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
7265 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
7266 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
7267 (aarch64_constant_pool_reload_icode): Delete.
7268 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
7269 instead of aarch64_constant_pool_reload_icode. Use
7270 code_for_aarch64_reload_mov instead of explicit mode checks.
7271 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
7272 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
7273 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
7274 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
7275 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
7276 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
7277 (aarch64_atomic_load_op_code): Delete.
7278 (aarch64_emit_atomic_load_op): Likewise.
7279 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
7280 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
7281 instead of aarch64_emit_atomic_load_op.
7282 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
7283 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
7284 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
7285 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
7286 character before the pattern name.
7287 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7288 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
7289 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
7290 (aarch64_frecps<mode>): Likewise.
7291 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
7292 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
7293 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
7294 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
7295 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
7297 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7299 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
7300 Allow HFmode constants if TARGET_FP_F16INST.
7302 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
7305 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
7306 No longer check last store for clobber of address register.
7308 2018-08-02 Martin Liska <mliska@suse.cz>
7310 PR gcov-profile/86817
7311 * gcov.c (process_all_functions): New function.
7313 (process_file): Move functions processing to
7314 process_all_functions.
7316 2018-08-02 David Malcolm <dmalcolm@redhat.com>
7318 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
7319 "const" to the "gimple *" and "rtx_insn *" parameters.
7320 * dumpfile.h (dump_user_location_t::dump_user_location_t):
7322 (dump_location_t::dump_location_t): Likewise.
7324 2018-08-01 Martin Sebor <msebor@redhat.com>
7326 PR tree-optimization/86650
7327 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
7328 rather than a "gcall *". Directly pass the data of interest
7329 to percent_K_format, rather than building a temporary CALL_EXPR
7331 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
7332 (gimple_fold_builtin_strncat): Adjust.
7333 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
7334 gcall* argument with gimple*.
7335 * gimple-ssa-warn-restrict.c (check_call): Same.
7336 (wrestrict_dom_walker::before_dom_children): Same.
7337 (builtin_access::builtin_access): Same.
7338 (check_bounds_or_overlap): Same
7339 (maybe_diag_overlap): Same.
7340 (maybe_diag_offset_bounds): Same.
7341 * tree-diagnostic.c (default_tree_printer): Move usage of
7342 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
7344 * tree-pretty-print.c (percent_K_format): Add argument.
7345 * tree-pretty-print.h: Add argument.
7346 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
7347 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
7348 (maybe_diag_stxncpy_trunc): Same.
7349 (handle_builtin_stxncpy): Same.
7350 (handle_builtin_strcat): Same.
7352 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7354 * match.pd: Optimise pointer range checks.
7356 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7358 PR tree-optimization/86758
7359 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
7360 to remove pattern statements.
7362 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7364 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
7365 result of dfs_enumerate_from when constructing stmt_vec_infos,
7366 instead of additionally calling get_loop_body.
7368 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7370 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
7372 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
7373 When creating an iv, assert that the step is not known to be zero.
7374 (vect_setup_realignment): Update call accordingly.
7375 * tree-vect-stmts.c (vectorizable_store): Likewise.
7376 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
7378 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7380 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
7381 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
7382 (vectorizable_reduction): Likewise.
7383 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
7384 (vect_detect_hybrid_slp_stmts): Likewise.
7385 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
7387 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
7389 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
7390 (wide_int_set_zero_nonzero_bits): ...this.
7391 (zero_nonzero_bits_from_vr): Rename to...
7392 (vrp_set_zero_nonzero_bits): ...this.
7393 (extract_range_from_multiplicative_op_1): Abstract wide int
7395 (wide_int_range_multiplicative_op): ...here.
7396 (extract_range_from_binary_expr_1): Extract wide int binary
7397 operations into their own functions.
7398 (wide_int_range_lshift): New.
7399 (wide_int_range_can_optimize_bit_op): New.
7400 (wide_int_range_shift_undefined_p): New.
7401 (wide_int_range_bit_xor): New.
7402 (wide_int_range_bit_ior): New.
7403 (wide_int_range_bit_and): New.
7404 (wide_int_range_trunc_mod): New.
7405 (extract_range_into_wide_ints): New.
7406 (vrp_shift_undefined_p): New.
7407 (extract_range_from_multiplicative_op): New.
7408 (vrp_can_optimize_bit_op): New.
7409 * tree-vrp.h (value_range::dump): New.
7410 (wide_int_range_multiplicative_op): New.
7411 (wide_int_range_lshift):New.
7412 (wide_int_range_shift_undefined_p): New.
7413 (wide_int_range_bit_xor): New.
7414 (wide_int_range_bit_ior): New.
7415 (wide_int_range_bit_and): New.
7416 (wide_int_range_trunc_mod): New.
7417 (zero_nonzero_bits_from_bounds): Rename to...
7418 (wide_int_set_zero_nonzero_bits): ...this.
7419 (zero_nonzero_bits_from_vr): Rename to...
7420 (vrp_set_zero_nonzero_bits): ...this.
7421 (range_easy_mask_min_max): Rename to...
7422 (wide_int_range_can_optimize_bit_op): this.
7423 * vr-values.c (simplify_bit_ops_using_ranges): Rename
7424 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
7426 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7428 * tree-vectorizer.h (vect_orig_stmt): New function.
7429 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
7430 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
7431 (vect_create_epilog_for_reduction): Likewise.
7432 (vectorizable_live_operation): Likewise.
7433 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
7434 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
7435 * tree-vect-stmts.c (vectorizable_call): Likewise.
7436 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
7438 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7440 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
7442 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
7443 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
7444 (vect_transform_loop): Likewise.
7445 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7447 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7449 * tree-vectorizer.h (vect_schedule_slp): Return void.
7450 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7451 (vect_schedule_slp): Likewise.
7453 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7455 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
7457 (vect_transform_loop): Update calls accordingly. Schedule SLP
7458 instances before the main loop, if any exist.
7460 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7462 PR tree-optimization/86749
7463 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
7464 If the lhs is used in a COND_EXPR, check that it is being used
7465 as the "then" or "else" value.
7467 2018-08-01 Tom de Vries <tdevries@suse.de>
7470 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
7471 speculation_safe_value_not_needed.
7473 2018-08-01 Richard Biener <rguenther@suse.de>
7475 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
7478 2018-08-01 Martin Liska <mliska@suse.cz>
7480 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
7481 format how successful transformation is dumped.
7482 (gimple_mod_pow2_value_transform): Likewise.
7483 (gimple_mod_subtract_transform): Likewise.
7484 (gimple_stringops_transform): Likewise.
7486 2018-08-01 Martin Liska <mliska@suse.cz>
7489 * value-prof.c (interesting_stringop_to_profile_p):
7490 Simplify the code and add BUILT_IN_MEMMOVE.
7491 (gimple_stringops_transform): Likewise.
7493 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
7495 * config/aarch64/aarch64-simd.md
7496 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
7497 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
7498 use GPI iterator instead of SI mode.
7500 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
7502 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
7503 rs6000_speculation_barrier.
7504 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
7505 new barrier pattern name.
7507 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
7509 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
7510 (speculation_barrier): New insn.
7512 2018-08-01 Richard Biener <rguenther@suse.de>
7515 * graphite.h: Include isl/id.h and isl/space.h to allow build
7518 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
7521 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
7522 mode for COFF targets.
7523 * defaults.h (TARGET_COFF): Define.
7524 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
7525 TARGET_COFF): Define.
7526 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
7527 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
7528 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
7530 2018-07-31 Alexandre Oliva <oliva@adacore.com>
7532 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
7533 * gimple-streamer-out.c (output_bb): Save it.
7534 * lto-streamer-in.c (input_struct_function_base): Restore
7535 instance discriminator if available. Create map on demand.
7536 * lto-streamer-out.c (output_struct_function_base): Save it if
7538 * final.c (decl_to_instance_map): Document LTO strategy.
7540 2018-07-31 Alexandre Oliva <oliva@adacore.com>
7541 Olivier Hainque <hainque@adacore.com>
7543 * debug.h (decl_to_instance_map_t): New type.
7544 (decl_to_instance_map): Declare.
7545 (maybe_create_decl_to_instance_map): New inline function.
7546 * final.c (bb_discriminator, last_bb_discriminator): New statics,
7547 to track basic block discriminators.
7548 (final_start_function_1): Initialize them.
7549 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
7551 (decl_to_instance_map): New variable.
7552 (map_decl_to_instance, maybe_set_discriminator): New functions.
7553 (notice_source_line): Set discriminator.
7555 2018-07-31 Ian Lance Taylor <iant@golang.org>
7557 * targhooks.c (default_have_speculation_safe_value): Add
7560 2018-07-31 David Malcolm <dmalcolm@redhat.com>
7562 * dump-context.h: Include "pretty-print.h".
7563 (dump_context::refresh_dumps_are_enabled): New decl.
7564 (dump_context::emit_item): New decl.
7565 (class dump_context): Add fields "m_test_pp" and
7567 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
7568 (temp_dump_context::get_dumped_text): New decl.
7569 (class temp_dump_context): Add field "m_pp".
7570 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
7571 (dump_context::refresh_dumps_are_enabled): ...and add a test for
7573 (set_dump_file): Update for above change.
7574 (set_alt_dump_file): Likewise.
7575 (dump_loc): New overload, taking a pretty_printer *.
7576 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
7577 to any test pretty-printer.
7578 (make_item_for_dump_gimple_stmt): New function, adapted from
7579 optinfo::add_gimple_stmt.
7580 (dump_context::dump_gimple_stmt): Call it, and use the result,
7581 eliminating the direct usage of dump_file and alt_dump_file in
7582 favor of indirectly using them via emit_item.
7583 (make_item_for_dump_gimple_expr): New function, adapted from
7584 optinfo::add_gimple_expr.
7585 (dump_context::dump_gimple_expr): Call it, and use the result,
7586 eliminating the direct usage of dump_file and alt_dump_file in
7587 favor of indirectly using them via emit_item.
7588 (make_item_for_dump_generic_expr): New function, adapted from
7590 (dump_context::dump_generic_expr): Call it, and use the result,
7591 eliminating the direct usage of dump_file and alt_dump_file in
7592 favor of indirectly using them via emit_item.
7593 (make_item_for_dump_printf_va): New function, adapted from
7594 optinfo::add_printf_va.
7595 (make_item_for_dump_printf): New function.
7596 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
7597 and use the result, eliminating the direct usage of dump_file and
7598 alt_dump_file in favor of indirectly using them via emit_item.
7599 (make_item_for_dump_dec): New function.
7600 (dump_context::dump_dec): Call it, and use the result,
7601 eliminating the direct usage of dump_file and alt_dump_file in
7602 favor of indirectly using them via emit_item.
7603 (make_item_for_dump_symtab_node): New function, adapted from
7604 optinfo::add_symtab_node.
7605 (dump_context::dump_symtab_node): Call it, and use the result,
7606 eliminating the direct usage of dump_file and alt_dump_file in
7607 favor of indirectly using them via emit_item.
7608 (dump_context::begin_scope): Reimplement, avoiding direct usage
7609 of dump_file and alt_dump_file in favor of indirectly using them
7611 (dump_context::emit_item): New member function.
7612 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
7613 Set up test pretty-printer on the underlying context. Call
7614 refresh_dumps_are_enabled.
7615 (temp_dump_context::~temp_dump_context): Call
7616 refresh_dumps_are_enabled.
7617 (temp_dump_context::get_dumped_text): New member function.
7618 (selftest::verify_dumped_text): New function.
7619 (ASSERT_DUMPED_TEXT_EQ): New macro.
7620 (selftest::test_capture_of_dump_calls): Run all tests twice, with
7621 and then without optinfo enabled. Add uses of
7622 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
7623 * dumpfile.h: Update comment for the dump_* API.
7624 * optinfo-emit-json.cc
7625 (selftest::test_building_json_from_dump_calls): Update for new
7626 param for temp_dump_context ctor.
7627 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
7628 and "m_owned" field.
7629 (optinfo_item::~optinfo_item): Likewise.
7630 (optinfo::add_item): New member function.
7631 (optinfo::emit): Update comment.
7632 (optinfo::add_string): Delete.
7633 (optinfo::add_printf): Delete.
7634 (optinfo::add_printf_va): Delete.
7635 (optinfo::add_gimple_stmt): Delete.
7636 (optinfo::add_gimple_expr): Delete.
7637 (optinfo::add_tree): Delete.
7638 (optinfo::add_symtab_node): Delete.
7639 (optinfo::add_dec): Delete.
7640 * optinfo.h (class dump_context): New forward decl.
7641 (optinfo::add_item): New decl.
7642 (optinfo::add_string): Delete.
7643 (optinfo::add_printf): Delete.
7644 (optinfo::add_printf_va): Delete.
7645 (optinfo::add_gimple_stmt): Delete.
7646 (optinfo::add_gimple_expr): Delete.
7647 (optinfo::add_tree): Delete.
7648 (optinfo::add_symtab_node): Delete.
7649 (optinfo::add_dec): Delete.
7650 (optinfo::add_poly_int): Delete.
7651 (optinfo_item::optinfo_item): Remove "owned" param.
7652 (class optinfo_item): Remove field "m_owned".
7654 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
7657 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
7658 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
7659 requested variable alignment.
7660 (expand_one_ssa_partition): Likewise.
7661 (expand_one_var): Likewise.
7663 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7665 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
7666 to speculation_safe_value_not_needed.
7668 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7670 * targhooks.h (speculation_safe_value_not_needed): New prototype.
7671 * targhooks.c (speculation_safe_value_not_needed): New function.
7672 * target.def (have_speculation_safe_value): Update documentation.
7673 * doc/tm.texi: Regenerated.
7675 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7677 * config/aarch64/iterators.md (ALLI_TI): New iterator.
7678 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
7680 (despeculate_copy<ALLI:mode>_insn): New insn.
7681 (despeculate_copyti_insn): New insn.
7682 (despeculate_simple<ALLI:mode>): New insn
7683 (despeculate_simpleti): New insn.
7684 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
7686 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
7687 aarch64_speculation_safe_value.
7688 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
7690 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7692 * config/aarch64/aarch64-speculation.cc: New file.
7693 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
7694 before pass_reorder_blocks.
7695 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
7697 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
7698 X14 and X15 when tracking speculation.
7699 * config/aarch64/aarch64.md (register name constants): Add
7700 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
7701 (unspec): Add UNSPEC_SPECULATION_TRACKER.
7702 (speculation_barrier): New insn attribute.
7703 (cmp<mode>): Allow SP in comparisons.
7704 (speculation_tracker): New insn.
7705 (speculation_barrier): Add speculation_barrier attribute.
7706 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
7707 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
7708 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
7710 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7712 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
7713 aarch64_track_speculation is true.
7714 (tb<optab><mode>1): Likewise.
7715 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
7716 generate CB[N]Z when tracking speculation.
7717 (aarch64_split_compare_and_swap): Likewise.
7718 (aarch64_split_atomic_op): Likewise.
7720 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7722 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
7724 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7726 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
7727 (speculation_barrier): New insn.
7729 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7731 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
7732 * config/arm/arm.md (speculation_barrier): New expand.
7733 (speculation_barrier_insn): New pattern.
7735 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7737 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
7738 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
7739 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
7740 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
7742 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
7743 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
7744 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
7745 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
7746 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
7747 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
7748 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
7749 * builtins.c (expand_speculation_safe_value): New function.
7750 (expand_builtin): Call it.
7751 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
7752 * doc/extend.texi: Document __builtin_speculation_safe_value.
7753 * doc/md.texi: Document "speculation_barrier" pattern.
7754 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
7755 TARGET_HAVE_SPECULATION_SAFE_VALUE.
7756 * doc/tm.texi: Regenerated.
7757 * target.def (have_speculation_safe_value, speculation_safe_value): New
7759 * targhooks.c (default_have_speculation_safe_value): New function.
7760 (default_speculation_safe_value): New function.
7761 * targhooks.h (default_have_speculation_safe_value): Add prototype.
7762 (default_speculation_safe_value): Add prototype.
7764 2018-07-31 David Malcolm <dmalcolm@redhat.com>
7766 * dump-context.h (dump_context::dump_loc): New decl.
7767 * dumpfile.c (dump_context::dump_loc): New member function.
7768 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
7769 and dump_gimple_stmt.
7770 (dump_context::dump_gimple_expr_loc): Likewise, using
7772 (dump_context::dump_generic_expr_loc): Likewise, using
7774 (dump_context::dump_printf_loc_va): Likewise, using
7776 (dump_context::begin_scope): Explicitly using the global function
7777 "dump_loc", rather than the member function.
7779 2018-07-31 Martin Sebor <msebor@redhat.com>
7781 PR tree-optimization/86741
7782 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
7784 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
7786 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
7787 depend on whether prefetch instructions will be emitted or not.
7788 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
7789 will be emitted or not.
7790 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
7791 (TARGET_SETMEM_PFD): New macros.
7793 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7795 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
7796 (NULL_STMT_VEC_INFO): Delete.
7797 (stmt_vec_info::operator*): Likewise.
7798 (stmt_vec_info::operator gimple *): Likewise.
7799 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
7800 of NULL_STMT_VEC_INFO.
7801 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
7802 (vect_reassociating_reduction_p): Likewise.
7803 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
7804 (vectorizable_store): Likewise.
7805 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
7806 (vec_info::free_stmt_vec_infos): Likewise.
7808 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7810 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
7811 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
7813 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7815 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
7816 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
7817 (vec_info::free_stmt_vec_info): New private member functions.
7818 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
7819 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
7820 * tree-parloops.c (gather_scalar_reductions): Remove calls to
7821 set_stmt_vec_info_vec and free_stmt_vec_infos.
7822 * tree-vect-loop.c (_loop_vec_info): Remove call to
7823 set_stmt_vec_info_vec.
7824 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
7825 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
7826 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
7827 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
7828 (vec_info::free_stmt_vec_info): ...these new functions. Remove
7829 assignments in {vec_info::,}new_stmt_vec_info that are redundant
7830 with the clearing in the xcalloc.
7831 (stmt_vec_info_vec): Delete.
7832 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
7833 (vectorize_loops): Likewise.
7834 (vec_info::~vec_info): Remove argument from call to
7835 free_stmt_vec_infos.
7836 (vec_info::add_stmt): Remove vinfo argument from call to
7839 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7841 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
7842 rather than a gimple stmt.
7843 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
7844 information for pattern statements when passed the original
7845 statement; instead wait to be passed the pattern statement itself.
7846 Don't call set_vinfo_for_stmt here.
7847 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
7848 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
7849 stmt_vec_infos here.
7850 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
7851 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
7852 stmt_vec_infos entry.
7854 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7856 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
7857 * tree-vectorizer.c (vec_info::replace_stmt): New function.
7858 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
7859 * tree-vect-stmts.c (vectorizable_call): Likewise.
7860 (vectorizable_simd_clone_call): Likewise.
7862 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7864 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
7865 * tree-vectorizer.c (vec_info::remove_stmt): New function.
7866 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
7867 * tree-vect-loop.c (vect_transform_loop): Likewise.
7868 * tree-vect-slp.c (vect_schedule_slp): Likewise.
7869 * tree-vect-stmts.c (vect_remove_stores): Likewise.
7871 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7873 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
7874 (vect_dr_stmt): Delete.
7875 * tree-vectorizer.c (vec_info::lookup_dr): New function.
7876 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
7878 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
7879 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
7880 (vect_verify_datarefs_alignment, vect_peeling_supportable)
7881 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
7882 (vect_analyze_data_refs): Likewise.
7883 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
7885 (vect_find_same_alignment_drs): Likewise.
7886 (vect_slp_analyze_node_dependences): Update calls accordingly.
7887 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
7888 instead of DR_VECT_AUX.
7889 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
7890 of a vector data references. Use vec_info::lookup_dr instead of
7892 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
7893 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
7894 instead of DR_VECT_AUX.
7896 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7898 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
7900 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
7902 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
7903 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
7904 (vect_gen_prolog_loop_niters): Likewise.
7906 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7908 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
7909 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
7910 (vect_known_alignment_in_bytes, vect_dr_behavior)
7911 (vect_get_scalar_dr_size): Take references as dr_vec_infos
7912 instead of data_references. Update calls to other routines for
7913 which the same change has been made.
7914 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
7915 dr_vec_infos instead of stmt_vec_infos.
7916 (vect_analyze_data_ref_dependence): Update call accordingly.
7917 (vect_slp_analyze_data_ref_dependence)
7918 (vect_record_base_alignments): Use DR_VECT_AUX.
7919 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
7920 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
7921 (vector_alignment_reachable_p, vect_get_data_access_cost)
7922 (vect_peeling_supportable, vect_analyze_group_access_1)
7923 (vect_analyze_group_access, vect_analyze_data_ref_access)
7924 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
7925 (vect_compile_time_alias, vect_small_gap_p)
7926 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
7927 (vect_supportable_dr_alignment): Take references as dr_vec_infos
7928 instead of data_references. Update calls to other routines for
7929 which the same change has been made.
7930 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
7931 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
7932 (vect_slp_analyze_and_verify_node_alignment)
7933 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
7934 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
7935 (vect_setup_realignment): Use dr_vec_infos. Update calls after
7937 (_vect_peel_info::dr): Replace with...
7938 (_vect_peel_info::dr_info): ...this new field.
7939 (vect_peeling_hash_get_most_frequent)
7940 (vect_peeling_hash_choose_best_peeling): Update accordingly.
7941 (vect_peeling_hash_get_lowest_cost):
7942 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
7943 routines for which the same change has been made.
7944 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
7946 * tree-vect-loop-manip.c (get_misalign_in_elems)
7947 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
7949 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
7950 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
7951 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
7952 (get_group_load_store_type, get_negative_load_store_type)
7953 (vect_get_data_ptr_increment, vectorizable_store)
7954 (vectorizable_load): Likewise.
7955 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
7956 Update calls to other routines for which the same change has been made.
7958 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7960 * tree-vectorizer.h (vec_info::move_dr): New member function.
7961 (dataref_aux): Rename to...
7962 (dr_vec_info): ...this and add "dr" and "stmt" fields.
7963 (_stmt_vec_info::dr_aux): Update accordingly.
7964 (_stmt_vec_info::data_ref_info): Delete.
7965 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
7966 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
7967 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
7968 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
7970 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
7971 (STMT_VINFO_DR_INFO): New macro.
7972 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
7973 (set_dr_misalignment): Update after rename of dataref_aux.
7974 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
7975 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
7976 initialization of STMT_VINFO_DATA_REF.
7977 * tree-vectorizer.c (vec_info::move_dr): New function.
7978 * tree-vect-patterns.c (vect_recog_bool_pattern)
7979 (vect_recog_mask_conversion_pattern)
7980 (vect_recog_gather_scatter_pattern): Use it.
7981 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
7982 the "dr" and "stmt" fields of dr_vec_info instead of
7983 STMT_VINFO_DATA_REF.
7985 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7987 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
7988 (is_pattern_stmt_p): Use it.
7989 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
7990 on pattern statements.
7992 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7994 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
7995 original stmt as a stmt_vec_info rather than a gimple stmt.
7996 (vect_pattern_recog_1): Take the statement directly as a
7997 stmt_vec_info, rather than via a gimple_stmt_iterator.
7998 Update call to vect_mark_pattern_stmts.
7999 (vect_pattern_recog): Update calls accordingly.
8001 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8003 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
8004 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
8005 a vect_def_type for the first argument.
8006 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
8007 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
8008 operand if it isn't defined by a vectorized statement.
8009 (vect_build_gather_load_calls): Remove the mask_dt argument and
8010 update calls to vect_get_vec_def_for_stmt_copy.
8011 (vectorizable_bswap): Likewise the dt argument.
8012 (vectorizable_call): Update calls to vectorizable_bswap and
8013 vect_get_vec_def_for_stmt_copy.
8014 (vectorizable_simd_clone_call, vectorizable_assignment)
8015 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
8016 (vectorizable_comparison): Update calls to
8017 vect_get_vec_def_for_stmt_copy.
8018 (vectorizable_store): Likewise. Remove now-unnecessary calls to
8020 (vect_get_loop_based_defs): Remove dt argument and update call
8021 to vect_get_vec_def_for_stmt_copy.
8022 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
8023 and vect_get_vec_def_for_stmt_copy.
8024 (vectorizable_load): Update calls to vect_build_gather_load_calls
8025 and vect_get_vec_def_for_stmt_copy.
8026 * tree-vect-loop.c (vect_create_epilog_for_reduction)
8027 (vectorizable_reduction, vectorizable_live_operation): Update calls
8028 to vect_get_vec_def_for_stmt_copy.
8030 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8032 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
8033 and gimple stmt arguments with a stmt_vec_info.
8034 (vect_record_base_alignments): Update calls accordingly.
8035 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
8036 and gimple stmt arguments with a stmt_vec_info.
8037 (vect_build_slp_tree_1): Remove vinfo argument and update call
8038 to vect_record_max_nunits.
8039 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
8040 and vect_record_max_nunits.
8042 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8044 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
8045 file and take a stmt_vec_info instead of a gimple stmt.
8046 (supportable_widening_operation, vect_finish_replace_stmt)
8047 (vect_finish_stmt_generation, vect_get_store_rhs)
8048 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
8049 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
8050 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
8051 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
8052 (vect_create_data_ref_ptr, bump_vector_ptr)
8053 (vect_permute_store_chain, vect_setup_realignment)
8054 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
8055 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
8056 (vectorizable_reduction, vectorizable_induction)
8057 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
8058 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
8059 than gimple stmts as arguments.
8060 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
8061 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
8062 (can_group_stmts_p, vect_check_gather_scatter)
8063 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
8064 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
8065 (vect_permute_load_chain, vect_shift_permute_load_chain)
8066 (vect_transform_grouped_load)
8067 (vect_record_grouped_load_vectors): Likewise.
8068 * tree-vect-loop.c (vect_fixup_reduc_chain)
8069 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
8070 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
8071 (vectorizable_reduction, vectorizable_induction)
8072 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
8073 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
8074 (vect_get_load_store_mask): Likewise.
8075 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
8076 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
8077 * tree-vect-stmts.c (vect_mark_relevant)
8078 (is_simple_and_all_uses_invariant)
8079 (exist_non_indexing_operands_for_use_p, process_use)
8080 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
8081 (vect_get_vec_def_for_operand, vect_get_vec_defs)
8082 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
8083 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
8084 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
8085 (get_negative_load_store_type, get_load_store_type)
8086 (vect_check_load_store_mask, vect_check_store_rhs)
8087 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
8088 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
8089 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
8090 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8091 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
8092 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
8093 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
8094 (supportable_widening_operation): Likewise.
8096 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8098 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
8099 a stmt_vec_info instead of a gcall.
8100 (vect_check_gather_scatter): Update call accordingly.
8101 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
8103 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
8104 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
8105 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
8106 instead of a gimple stmt.
8107 (vect_transform_loop): Update calls accordingly.
8108 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
8109 stmt_vec_infos instead of gimple stmts.
8110 (vect_analyze_slp_instance): Update use accordingly.
8111 * tree-vect-stmts.c (read_vector_array, write_vector_array)
8112 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
8113 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
8114 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
8115 (vect_gen_widened_results_half, vect_get_loop_based_defs)
8116 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
8117 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
8118 down to subroutines.
8120 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8122 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
8123 of the worklist from a vector of gimple stmts to a vector of
8125 * tree-vect-stmts.c (vect_mark_relevant, process_use)
8126 (vect_mark_stmts_to_be_vectorized): Likewise
8128 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8130 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
8131 statement before passing it to vect_analyze_stmt.
8132 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
8133 the chain of phi vector definitions. Track the exit phi via its
8135 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
8136 STMT_VINFO_REDUC_DEF.
8137 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
8138 stmt_vec_infos to handle the statement chains.
8139 (vect_get_slp_defs): Record the first statement in the node
8140 using a stmt_vec_info.
8141 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
8142 statements here and pass their stmt_vec_info down to subroutines.
8143 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
8144 down to vect_finish_stmt_generation.
8145 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
8146 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
8147 stmt_vec_infos to subroutines.
8148 (vect_remove_stores): Use stmt_vec_infos to handle the statement
8151 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8153 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
8154 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
8155 (vect_permute_store_chain, vect_setup_realignment)
8156 (vect_permute_load_chain, vect_shift_permute_load_chain)
8157 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
8158 stmts internally, and when passing values to other vectorizer routines.
8159 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
8160 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
8161 (vect_analyze_loop_operations, get_initial_def_for_reduction)
8162 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
8163 (vectorizable_reduction, vectorizable_induction)
8164 (vectorizable_live_operation, vect_transform_loop_stmt)
8165 (vect_transform_loop): Likewise.
8166 * tree-vect-patterns.c (vect_reassociating_reduction_p)
8167 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
8168 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
8169 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8170 (vect_slp_analyze_node_operations_1): Likewise.
8171 * tree-vect-stmts.c (vect_mark_relevant, process_use)
8172 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
8173 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
8174 (vect_finish_stmt_generation_1, get_group_load_store_type)
8175 (get_load_store_type, vect_build_gather_load_calls)
8176 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
8177 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
8178 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8179 (vectorizable_store, vectorizable_load, vectorizable_condition)
8180 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
8181 (supportable_widening_operation): Likewise.
8182 (vect_get_vector_types_for_stmt): Likewise.
8183 * tree-vectorizer.h (vect_dr_behavior): Likewise.
8185 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8187 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
8188 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
8189 (vect_permute_store_chain, vect_permute_load_chain)
8190 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
8191 repeated stmt_vec_info lookups.
8192 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
8193 (vect_update_ivs_after_vectorizer): Likewise.
8194 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
8195 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
8196 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
8197 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8198 (vect_bb_slp_scalar_cost): Likewise.
8199 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
8201 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8203 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
8204 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
8205 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
8206 to get gassigns and gcalls, rather than operating on generc gimple
8208 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
8209 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
8210 (vectorizable_load, vect_analyze_stmt): Likewise.
8211 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
8213 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8215 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
8216 return stmt_vec_infos rather than gimple stmts. Do not accept
8218 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
8220 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
8221 Update use of get_later_stmt.
8222 (vect_get_constant_vectors): Update call accordingly.
8223 (vect_schedule_slp_instance): Likewise
8224 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
8225 (vect_slp_analyze_instance_dependence): Likewise.
8226 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
8228 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8230 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
8231 (stmt_info_for_cost::stmt_info): ...this new field.
8232 (add_stmt_costs): Update accordingly.
8233 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
8234 (vect_get_known_peeling_cost): Likewise.
8235 (vect_estimate_min_profitable_iters): Likewise.
8236 * tree-vect-stmts.c (record_stmt_cost): Likewise.
8238 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8240 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
8241 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
8242 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
8244 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
8246 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8248 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
8249 a gimple stmt to a stmt_vec_info.
8250 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
8252 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8254 * tree-vectorizer.h (vec_info::grouped_stores): Change from
8255 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
8256 (_loop_vec_info::reduction_chains): Likewise.
8257 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
8259 * tree-vect-slp.c (vect_analyze_slp): Likewise.
8261 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8263 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
8264 a gimple stmt to a stmt_vec_info.
8265 (_stmt_vec_info::next_element): Likewise.
8266 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
8267 (vect_slp_analyze_and_verify_node_alignment)
8268 (vect_analyze_group_access_1, vect_analyze_group_access)
8269 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
8270 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
8271 (vect_supportable_dr_alignment): Update accordingly.
8272 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
8273 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
8274 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
8275 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
8276 * tree-vect-slp.c (vect_build_slp_tree_1)
8277 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
8278 (vect_split_slp_store_group, vect_analyze_slp_instance)
8279 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
8280 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
8281 (get_group_load_store_type, get_load_store_type)
8282 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
8283 (vect_transform_stmt, vect_remove_stores): Likewise.
8285 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8287 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
8289 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
8290 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
8291 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
8292 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
8293 (vector_alignment_reachable_p, vect_get_data_access_cost)
8294 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
8295 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
8296 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
8297 (vect_analyze_group_access_1, vect_analyze_group_access)
8298 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
8299 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
8300 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
8301 result of vect_dr_stmt and use the stmt_vec_info instead of
8302 the associated gimple stmt.
8303 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
8304 (vect_gen_prolog_loop_niters): Likewise.
8305 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8307 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8309 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
8310 to a vec<stmt_vec_info>.
8311 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
8312 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
8314 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
8315 to a vec<stmt_vec_info>.
8316 (bst_traits::value_type, bst_traits::value_type): Likewise.
8317 (bst_traits::hash): Update accordingly.
8318 (vect_get_and_check_slp_defs): Change the stmts parameter from
8319 a vec<gimple *> to a vec<stmt_vec_info>.
8320 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
8321 (vect_build_slp_tree): Likewise.
8322 (vect_build_slp_tree_2): Likewise. Update uses of
8323 SLP_TREE_SCALAR_STMTS.
8324 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
8325 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
8326 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
8327 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
8328 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
8329 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
8330 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
8331 (vect_get_constant_vectors, vect_get_slp_defs)
8332 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
8333 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
8334 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
8335 instead of gimple stmts.
8336 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
8337 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
8338 (vect_slp_analyze_instance_dependence): Update uses of
8339 SLP_TREE_SCALAR_STMTS.
8340 (vect_slp_analyze_and_verify_node_alignment): Likewise.
8341 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
8342 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
8343 (get_initial_defs_for_reduction): Likewise.
8344 (vect_create_epilog_for_reduction): Likewise.
8345 (vectorize_fold_left_reduction): Likewise.
8346 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
8347 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
8348 (can_vectorize_live_stmts): Likewise.
8350 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8352 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
8353 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
8354 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
8356 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
8358 (gather_scalar_reductions): Update after above interface changes.
8359 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
8360 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
8362 (vect_force_simple_reduction): Likewise.
8363 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
8364 LOOP_VINFO_REDUCTIONS.
8365 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8367 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8369 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
8370 a gimple stmt to a stmt_vec_info.
8371 * tree-vect-loop.c (vect_active_double_reduction_p)
8372 (vect_force_simple_reduction, vectorizable_reduction): Update
8375 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8377 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
8378 vec<gimple *> to a vec<stmt_vec_info>.
8379 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
8380 the reduction_phis argument from a vec<gimple *> to a
8382 (vectorizable_reduction): Likewise the phis local variable that
8383 is passed to vect_create_epilog_for_reduction. Update for new type
8384 of SLP_TREE_VEC_STMTS.
8385 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
8386 (vectorizable_live_operation): Likewise.
8387 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
8388 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
8390 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8392 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
8393 a gimple stmt to a stmt_vec_info.
8394 (vectorizable_condition, vectorizable_live_operation)
8395 (vectorizable_reduction, vectorizable_induction): Pass back the
8396 vectorized statement as a stmt_vec_info.
8397 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
8398 use of STMT_VINFO_VEC_STMT.
8399 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
8400 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
8401 as stmt_vec_infos rather than gimple stmts.
8402 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
8404 (vectorizable_live_operation): Likewise.
8405 (vectorizable_reduction, vectorizable_induction): Likewise,
8406 updating use of STMT_VINFO_VEC_STMT.
8407 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
8408 of STMT_VINFO_VEC_STMT.
8409 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
8410 (vectorizable_simd_clone_call, vectorizable_conversion)
8411 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8412 (vectorizable_store, vectorizable_load, vectorizable_condition)
8413 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
8414 from a gimple stmt to a stmt_vec_info.
8415 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
8416 pointer to a stmt_vec_info to the vectorizable_* routines.
8418 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8420 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
8421 a gimple stmt to a stmt_vec_info.
8422 (is_pattern_stmt_p): Update accordingly.
8423 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
8424 (vect_record_grouped_load_vectors): Likewise.
8425 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
8426 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
8427 (vect_model_reduction_cost): Likewise.
8428 (vect_create_epilog_for_reduction): Likewise.
8429 (vectorizable_reduction, vectorizable_induction): Likewise.
8430 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8431 Return the stmt_vec_info for the pattern statement.
8432 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
8433 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
8434 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
8435 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
8436 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
8437 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
8438 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
8439 (free_stmt_vec_info, vect_is_simple_use): Likewise.
8441 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8443 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
8444 (vect_finish_stmt_generation): Likewise.
8445 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
8446 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
8447 (vect_build_gather_load_calls): Use the return value of the above
8448 functions instead of a separate call to vinfo_for_stmt. Use narrow
8449 scopes for the input gimple stmt and wider scopes for the associated
8450 stmt_vec_info. Use vec_info::lookup_def when setting these
8451 stmt_vec_infos from an SSA_NAME definition.
8452 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
8453 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
8454 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8455 (vectorizable_store, vectorizable_load, vectorizable_condition)
8456 (vectorizable_comparison): Likewise.
8457 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
8458 (vectorizable_reduction): Likewise.
8460 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8462 * tree-vectorizer.h (vect_is_simple_use): Add an optional
8463 stmt_vec_info * parameter before the optional gimple **.
8464 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
8465 (process_use, vect_get_vec_def_for_operand_1): Update callers.
8466 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
8467 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8468 (vectorizable_live_operation): Likewise.
8469 * tree-vect-patterns.c (type_conversion_p): Likewise.
8470 (vect_look_through_possible_promotion): Likewise.
8471 (vect_recog_rotate_pattern): Likewise.
8472 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
8474 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8476 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
8477 a typedef to a wrapper class.
8478 (NULL_STMT_VEC_INFO): New macro.
8479 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
8480 (stmt_vec_info::operator*): New function.
8481 (stmt_vec_info::operator gimple *): Likewise.
8482 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
8483 (add_stmt_costs): Likewise.
8484 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
8485 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
8486 (vect_get_known_peeling_cost): Likewise.
8487 (vect_estimate_min_profitable_iters): Likewise.
8488 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8489 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
8490 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
8491 (vectorizable_store, free_stmt_vec_infos): Likewise.
8492 (new_stmt_vec_info): Change return type of xcalloc to
8495 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8497 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
8498 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
8499 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
8500 a single_imm_use-based sequence.
8501 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
8503 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8505 * tree-vectorizer.h (vec_info::lookup_def): Declare.
8506 * tree-vectorizer.c (vec_info::lookup_def): New function.
8507 * tree-vect-patterns.c (vect_get_internal_def): Use it.
8508 (vect_widened_op_tree): Likewise.
8509 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
8510 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
8511 (vectorizable_reduction): Likewise.
8512 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
8514 (vect_is_slp_reduction): Update calls accordingly. Use
8515 vec_info::lookup_def.
8516 (vect_is_simple_reduction): Likewise
8517 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
8519 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8521 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
8522 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
8523 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
8525 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
8526 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
8527 (vect_update_vf_for_slp, vect_analyze_loop_operations)
8528 (vect_is_slp_reduction, vectorizable_induction)
8529 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
8530 * tree-vect-patterns.c (vect_init_pattern_stmt):
8531 (vect_determine_min_output_precision_1, vect_determine_precisions)
8532 (vect_pattern_recog): Likewise.
8533 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
8534 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
8535 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
8536 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
8537 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
8538 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
8539 info field from a loop to a loop_vec_info.
8541 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8543 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
8544 (vec_info::add_stmt): Declare.
8545 * tree-vectorizer.c (vec_info::add_stmt): New function.
8546 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
8547 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
8548 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
8549 (vectorizable_induction): Likewise.
8550 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
8551 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
8552 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
8553 (vectorizable_load): Likewise.
8554 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8555 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
8556 (vect_recog_gather_scatter_pattern): Likewise.
8557 (append_pattern_def_seq): Likewise. Remove a check that is
8558 performed by add_stmt itself.
8560 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8562 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
8563 which make_ssa_name was called with new_stmt before new_stmt
8566 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8568 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
8570 (vect_is_slp_reduction): ...here...
8571 (vect_is_simple_reduction): ...and here. Remove repetition of tests
8572 that are already known to be false.
8574 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8576 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
8577 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
8578 STMT_VINFO_NUM_SLP_USES when it's true.
8579 (vect_free_slp_instance): Add a final_p parameter and pass it to
8581 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
8582 (vect_analyze_slp_instance): Likewise.
8583 (vect_slp_analyze_operations): Likewise.
8584 (vect_slp_analyze_bb_1): Likewise.
8585 * tree-vectorizer.c (vec_info): Likewise.
8586 * tree-vect-loop.c (vect_transform_loop): Likewise.
8588 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8590 * tree-vect-loop.c (vectorizable_reduction): Assert that the
8591 function is not called for second and subsequent members of
8594 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8596 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
8597 cases for nested loops from here to ...
8598 (vect_create_epilog_for_reduction): ...here. Only call
8599 vect_is_simple_use for inner-loop reductions.
8601 2018-07-31 Martin Liska <mliska@suse.cz>
8603 PR gcov-profile/85338
8604 PR gcov-profile/85350
8605 PR gcov-profile/85372
8606 * profile.c (struct location_triplet): New.
8607 (struct location_triplet_hash): Likewise.
8608 (output_location): Do not output a BB that
8609 is already recorded for a line.
8610 (branch_prob): Use streamed_locations.
8612 2018-07-31 Martin Liska <mliska@suse.cz>
8614 PR gcov-profile/85370
8615 * coverage.c (coverage_begin_function): Do not mark target
8616 clones as artificial functions.
8618 2018-07-31 Martin Liska <mliska@suse.cz>
8620 PR gcov-profile/83813
8621 PR gcov-profile/84758
8622 PR gcov-profile/85217
8623 PR gcov-profile/85332
8624 * profile.c (branch_prob): Do not record GOTO expressions
8625 for GIMPLE statements which locations are already streamed.
8627 2018-07-31 Olivier Hainque <hainque@adacore.com>
8629 * gcc.c (handle_spec_function): Accept a soft_matched_part
8630 argument, as do_spec_1. Pass it down to ...
8631 (eval_spec_function): Accept a soft_matched_part argument,
8632 and pass it down to ...
8633 (do_spec_2): Accept a soft_matched_part argument, and pass
8634 it down to do_spec_1.
8635 (do_spec_1): Pass soft_matched_part to handle_spec_function.
8636 (handle_braces): Update call to handle_spec_function.
8637 (driver::set_up_specs): Update calls to do_spec_2.
8638 (compare_debug_dump_opt_spec_function): Likewise.
8639 (compare_debug_self_opt_spec_function): Likewise.
8641 2018-07-31 Olivier Hainque <hainque@adacore.com>
8643 * common.opt (nolibc): New option.
8644 * doc/invoke.texi (Link Options): Document it.
8645 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
8646 * config/alpha/linux.h: Likewise.
8647 * config/arc/elf.h: Likewise.
8648 * config/arm/uclinux-elf.h: Likewise.
8649 * config/arm/unknown-elf.h: Likewise.
8650 * config/avr/avrlibc.h: Likewise.
8651 * config/bfin/bfin.h: Likewise.
8652 * config/bfin/linux.h: Likewise.
8653 * config/bfin/uclinux.h: Likewise.
8654 * config/darwin.h: Likewise.
8655 * config/darwin10.h: Likewise.
8656 * config/darwin12.h: Likewise.
8657 * config/gnu-user.h: Likewise.
8658 * config/lm32/uclinux-elf.h: Likewise.
8659 * config/pa/pa-hpux11.h: Likewise.
8660 * config/pa/pa64-hpux.h: Likewise.
8661 * config/sparc/sparc.h: Likewise.
8663 2018-07-31 Olivier Hainque <hainque@adacore.com>
8665 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
8666 undefined variables.
8668 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
8671 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
8674 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
8676 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
8677 terminated string literal.
8679 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
8681 PR rtl-optimization/85160
8682 * combine.c (is_just_move): New function.
8683 (try_combine): Allow combining two instructions into two if neither of
8684 the original instructions was a move.
8686 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
8689 * doc/extend.texi (Global Register Variables): Discourage use of type
8691 (Local Register Variables): Likewise.
8693 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
8695 PR tree-optimization/86506
8696 * hwint.h (ceil_log2): Resync with hwint.c implementation.
8698 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
8701 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
8702 hard_regno, make sure no insn between `from` and `to` clobbers it.
8704 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
8705 Tom de Vries <tdevries@suse.de>
8707 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
8708 (PTX_DEFAULT_RUNTIME_DIM): ... this.
8709 (nvptx_goacc_validate_dims): Set default worker and gang dims to
8710 PTX_DEFAULT_RUNTIME_DIM.
8711 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
8713 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
8715 * config/pa/pa.c (pa_output_addr_vec): Align address table.
8716 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
8717 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
8719 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
8721 * config/rs6000/constraints.md (wG constraint): Delete, no longer
8723 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
8724 predicate to reflect toc fusion has been deleted.
8725 (toc_fusion_mem_raw): Delete, no longer used.
8726 (toc_fusion_mem_wrapped): Likewise.
8727 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
8729 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
8730 Delete, no longer used.
8731 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
8732 meant to be used for toc fusion.
8733 (rs6000_debug_print_mode): Delete toc fusion debugging.
8734 (rs6000_debug_reg_global): Likewise.
8735 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
8736 fusion and secondary reload support that were never used.
8737 (rs6000_option_override_internal): Delete TOC fusion, that was only
8738 partially defined, and it did not work unless you also used the
8740 (rs6000_legitimate_address_p): Delete TOC fusion support.
8741 (rs6000_opt_masks): Likewise.
8742 (fusion_wrap_memory_address): Delete function, no longer used.
8743 (fusion_split_address); Delete TOC fusion support.
8744 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
8745 longer used with toc fusion being deleted.
8746 (TARGET_TOC_FUSION_FP): Likewise.
8747 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
8749 (toc fusion spliter): Delete TOC fusion support.
8750 (toc_fusionload_<mode>): Likewise.
8751 (toc_fusionload_di): Likewise.
8752 (fusion_gpr_load_<mode>): Delete generator function, this insn no
8753 longer needs to be named. Rename predicate to delete TOC fusion.
8754 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
8755 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
8756 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
8757 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
8758 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
8760 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
8762 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
8763 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
8764 __int128 in built-in function prototypes.
8765 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
8766 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
8768 2018-07-27 Martin Sebor <msebor@redhat.com>
8770 PR tree-optimization/86696
8771 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
8772 types, including enums.
8773 (handle_char_store): Be prepared for the above function to fail.
8775 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
8777 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
8778 when optimization level is lower than 2 or optimize for size.
8780 2018-07-26 Martin Sebor <msebor@redhat.com>
8782 PR tree-optimization/86043
8783 PR tree-optimization/86042
8784 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
8785 (get_string_cst_length): Rename...
8786 (get_min_string_length): ...to this. Add argument.
8787 (handle_char_store): Extend to handle multi-character stores by
8789 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
8790 * tree.h (initializer_zerop): Add argument.
8792 2018-07-26 Jakub Jelinek <jakub@redhat.com>
8795 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
8796 declare target to variables if they have always,{to,from,tofrom} map
8799 2018-07-26 Martin Liska <mliska@suse.cz>
8802 * lto-wrapper.c: Add linker_output as prefix
8803 for ltrans_output_file.
8805 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
8807 PR rtl-optimization/85805
8808 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
8809 value for hard registers if that was written in the same mode.
8811 2018-07-26 Martin Liska <mliska@suse.cz>
8813 PR gcov-profile/86536
8814 * gcov.c (format_gcov): Use printf format %.*f directly
8815 and do not handle special values.
8817 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
8819 * common/config/arc/arc-common.c (arc_option_optimization_table):
8820 Update default optimizations for size.
8822 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
8824 * config/arc/arc.md (movsf_insn): Add short instruction selection.
8825 * config/arc/constraints.md (CfZ): New constraint.
8826 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
8827 (subsf3_fpu): Likewise.
8828 (cmpsf_fpu): Likewise.
8829 (cmpsf_fpu_uneq): Likewise.
8831 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
8833 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
8835 (arc_is_uncached_mem_p): uncached applies to both the variable and
8838 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
8840 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
8843 2018-07-25 David Malcolm <dmalcolm@redhat.com>
8845 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
8846 field "m_scopes" from vec to auto_vec.
8848 2018-07-25 Martin Liska <mliska@suse.cz>
8850 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
8853 2018-07-25 Richard Biener <rguenther@suse.de>
8856 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
8857 special wrt context_die late.
8858 (gen_subprogram_die): Re-use DIEs in local scope.
8860 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
8862 PR tree-optimization/86644
8863 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
8865 2018-07-25 Martin Liska <mliska@suse.cz>
8868 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
8869 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
8871 2018-07-25 Martin Liska <mliska@suse.cz>
8874 * params.def: Explain ASan abbreviation and provide
8875 a documentation link.
8877 2018-07-24 Martin Sebor <msebor@redhat.com>
8879 PR tree-optimization/86622
8880 PR tree-optimization/86532
8881 * builtins.h (string_length): Declare.
8882 * builtins.c (c_strlen): Correct handling of non-constant offsets.
8883 (check_access): Be prepared for non-constant length ranges.
8884 (string_length): Make extern.
8885 * expr.c (string_constant): Only handle the minor non-constant
8886 array index. Use string_constant to compute the length of
8887 a generic string constant.
8889 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
8891 PR tree-optimization/86618
8892 * tree-vect-stmts.c (vectorizable_call): Don't take the address
8893 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
8895 2018-07-24 David Malcolm <dmalcolm@redhat.com>
8897 PR tree-optimization/86636
8898 * json.cc (json::object::set): Fix comment. Add assertions.
8899 (json::array::append): Move here from json.h. Add comment and an
8901 (json::string::string): Likewise.
8902 * json.h (json::array::append): Move to json.cc.
8903 (json::string::string): Likewise.
8904 * optinfo-emit-json.cc
8905 (optrecord_json_writer::impl_location_to_json): Assert that we
8906 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
8907 wrapper around it. Expand the location once, rather than three
8909 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
8910 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
8912 (optrecord_json_writer::optinfo_to_json): Likewise, in four
8913 places. Fix some overlong lines.
8915 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
8917 * config/aarch64/aarch64-simd.md
8918 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
8919 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
8920 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
8921 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
8922 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
8923 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
8924 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
8925 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
8926 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
8928 2018-07-24 Jakub Jelinek <jakub@redhat.com>
8931 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
8932 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
8933 and abs_d == d, do the power of two handling if profitable.
8935 2018-07-24 Richard Biener <rguenther@suse.de>
8937 * match.pd: Add BIT_FIELD_REF canonicalizations.
8939 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
8942 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
8944 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
8946 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
8947 terminated STRING_CST object.
8949 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
8951 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
8953 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
8955 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
8956 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
8957 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
8958 the elements into a register.
8959 (rs6000_split_v4si_init_di_reg): Delete.
8960 (rs6000_split_v4si_init): Delete.
8961 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
8962 (vsx_init_v4si): Rewrite as a define_expand.
8964 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
8966 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
8967 zero_extend argument from memory): New.
8969 2018-07-22 Martin Sebor <msebor@redhat.com>
8972 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
8973 diagnosing calls with unknown arguments unless -Walloca-larger-than
8974 is restricted to less than PTRDIFF_MAX bytes.
8976 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
8978 * doc/gcov.texi (Invoking Gcov): Editorial changes.
8980 2018-07-20 David Malcolm <dmalcolm@redhat.com>
8982 * pretty-print.c (text_info::set_location): Remove redundant
8983 "line_table" parameter from call to rich_location::set_range.
8985 2018-07-20 Martin Sebor <msebor@redhat.com>
8988 * builtins.c (expand_builtin_alloca): Adjust.
8989 * calls.c (alloc_max_size): Simplify.
8990 * cgraphunit.c (cgraph_node::expand): Adjust.
8991 * common.opt (larger_than_size, warn_frame_larger_than): Remove
8993 (frame_larger_than_size): Same.
8994 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
8995 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
8997 * doc/invoke.texi (GCC Command Options): Document option arguments.
8998 Explain byte-size arguments and suffixes.
8999 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
9000 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
9001 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
9002 * doc/options.texi (UInteger): Expand.
9003 (Host_Wide_Int, ByteSize): Document new properties.
9004 * final.c (final_start_function_1): Include sizes in an error message.
9005 * function.c (frame_offset_overflow): Same.
9006 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
9007 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
9008 Diagnose unbounded alloca calls only for limits of less than
9010 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
9011 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
9013 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
9014 only for limits of less than PTRDIFF_MAX.
9015 * langhooks-def.h (lhd_handle_option): Change function argument
9017 * langhooks.c (lhd_handle_option): Same.
9018 * langhooks.h (handle_option): Same.
9019 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
9021 (var_type, var_type_struct): Same.
9022 (var_set): Handle ByteSize flag.
9023 * optc-gen.awk: Add comments to output to ease debugging. Make
9024 use of HOST_WIDE_INT where appropriate.
9025 * opts-gen-save.awk: Use %lx to format unsigned long.
9026 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
9027 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
9028 arguments. Parse bytes-size suffixes.
9029 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
9030 (enum_value_to_arg): Same.
9031 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
9032 (handle_option): Adjust.
9033 (generate_option): Change function argument to HOST_WIDE_INT.
9034 (cmdline_handle_error): Adjust.
9035 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
9036 (set_option): Change function argument to HOST_WIDE_INT.
9037 (option_enabled): Handle cl_host_wide_int.
9038 (get_option_state): Handle CLVC_SIZE.
9039 (control_warning_option): Same.
9040 * opts.c (common_handle_option): Change function argument to
9041 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
9042 OPT_Wvla_larger_than_.
9043 * opts.h (enum cl_var_type): Add an enumerator.
9044 * stor-layout.c (layout_decl): Print a more meaningful warning.
9045 * toplev.c (output_stack_usage): Adjust.
9047 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
9049 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
9050 call to inline_expand_builtin_string_cmp.
9051 (expand_builtin_strcmp): Likewise.
9052 (expand_builtin_strncmp): Likewise.
9053 (inline_string_cmp): Delete the last parameter, change char_type_node
9054 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
9056 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
9057 the inlining expansion on target where the type of the call has same or
9058 narrower precision than unsigned char.
9060 2018-07-20 David Malcolm <dmalcolm@redhat.com>
9062 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
9063 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
9064 * common.opt (fsave-optimization-record): New option.
9065 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
9066 * doc/invoke.texi (-fsave-optimization-record): New option.
9067 * dumpfile.c: Include "optinfo-emit-json.h".
9068 (struct kv_pair): Move to coretypes.h.
9069 (optgroup_options): Make non-static.
9070 (dump_context::end_scope): Call
9071 optimization_records_maybe_pop_dump_scope.
9072 * dumpfile.h (optgroup_options): New decl.
9073 * json.cc: New file.
9075 * optinfo-emit-json.cc: New file.
9076 * optinfo-emit-json.h: New file.
9077 * optinfo.cc: Include "optinfo-emit-json.h".
9078 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
9079 (optinfo_enabled_p): Check optimization_records_enabled_p.
9080 (optinfo_wants_inlining_info_p): Likewise.
9081 * optinfo.h: Update comment.
9082 * profile-count.c (profile_quality_as_string): New function.
9083 * profile-count.h (profile_quality_as_string): New decl.
9084 (profile_count::quality): New accessor.
9085 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
9086 and optinfo_emit_json_cc_tests.
9087 * selftest.h (selftest::json_cc_tests): New decl.
9088 (selftest::optinfo_emit_json_cc_tests): New decl.
9089 * toplev.c: Include "optinfo-emit-json.h".
9090 (compile_file): Call optimization_records_finish.
9091 (do_compile): Call optimization_records_start.
9092 * tree-ssa-live.c: Include optinfo.h.
9093 (remove_unused_scope_block_p): Retain inlining information if
9094 optinfo_wants_inlining_info_p returns true.
9096 2018-07-20 Richard Biener <rguenther@suse.de>
9099 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
9100 to cover -flto-partition=none.
9102 2018-07-20 Martin Liska <mliska@suse.cz>
9104 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
9105 (get_decl_source_range): Remove unused function.
9107 2018-07-20 Richard Biener <rguenther@suse.de>
9109 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
9110 (struct vn_phi_s): Likewise.
9111 (struct vn_reference_s): Likewise.
9112 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
9113 for searching the slot of an entry known to be in the hash itself.
9114 (vn_phi_hasher::equal): Likewise.
9115 (vn_reference_hasher::equal): Likewise.
9116 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
9118 (optimistic_info, current_info): Remove, keeping only valid_info.
9119 (vn_reference_lookup_1): Remove fallback lookup.
9120 (vn_reference_lookup_2): Likewise.
9121 (vn_nary_op_lookup_1): Likewise.
9122 (vn_phi_lookup): Likewise.
9123 (vn_nary_build_or_lookup_1): Make sure to not chain the built
9125 (vn_reference_insert): Adjust, chain the inserted hash element
9126 at last_inserted_ref.
9127 (vn_reference_insert_pieces): Likewise.
9128 (visit_reference_op_call): Likewise.
9129 (vn_nary_op_insert_into): Chain the inserted hash element at
9131 (vn_nary_op_insert_pieces): Adjust.
9132 (vn_nary_op_insert): Likewise.
9133 (vn_nary_op_insert_stmt): Likewise.
9134 (vn_phi_insert): Adjust, chain the inserted hash element at
9136 (process_scc): Remove clearing and copying the optimistic
9137 table. Instead remove elements inserted during an optimistic
9138 iteration from the single table we maintain.
9139 (init_scc_vn): Adjust.
9140 (free_scc_vn): Likewise.
9141 (sccvn_dom_walker::record_cond): Likewise.
9142 (sccvn_dom_walker::after_dom_children): Likewise.
9144 2018-07-19 Martin Sebor <msebor@redhat.com>
9146 PR tree-optimization/84047
9147 PR tree-optimization/83776
9148 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
9149 (check_array_bounds): Call it.
9151 2018-07-19 Martin Sebor <msebor@redhat.com>
9153 * align.h (align_flags): Use member initialization.
9155 2018-07-19 David Malcolm <dmalcolm@redhat.com>
9157 * Makefile.in (OBJS): Add optinfo.o.
9158 * coretypes.h (class symtab_node): New forward decl.
9159 (struct cgraph_node): New forward decl.
9160 (class varpool_node): New forward decl.
9161 * dump-context.h: New file.
9162 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
9164 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
9165 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
9166 (set_alt_dump_file): Likewise.
9167 (dump_context::~dump_context): New dtor.
9168 (dump_gimple_stmt): Move implementation to...
9169 (dump_context::dump_gimple_stmt): ...this new member function.
9170 Add the stmt to any pending optinfo, creating one if need be.
9171 (dump_gimple_stmt_loc): Move implementation to...
9172 (dump_context::dump_gimple_stmt_loc): ...this new member function.
9173 Start a new optinfo and add the stmt to it.
9174 (dump_gimple_expr): Move implementation to...
9175 (dump_context::dump_gimple_expr): ...this new member function.
9176 Add the stmt to any pending optinfo, creating one if need be.
9177 (dump_gimple_expr_loc): Move implementation to...
9178 (dump_context::dump_gimple_expr_loc): ...this new member function.
9179 Start a new optinfo and add the stmt to it.
9180 (dump_generic_expr): Move implementation to...
9181 (dump_context::dump_generic_expr): ...this new member function.
9182 Add the tree to any pending optinfo, creating one if need be.
9183 (dump_generic_expr_loc): Move implementation to...
9184 (dump_context::dump_generic_expr_loc): ...this new member
9185 function. Add the tree to any pending optinfo, creating one if
9187 (dump_printf): Move implementation to...
9188 (dump_context::dump_printf_va): ...this new member function. Add
9189 the text to any pending optinfo, creating one if need be.
9190 (dump_printf_loc): Move implementation to...
9191 (dump_context::dump_printf_loc_va): ...this new member function.
9192 Start a new optinfo and add the stmt to it.
9193 (dump_dec): Move implementation to...
9194 (dump_context::dump_dec): ...this new member function. Add the
9195 value to any pending optinfo, creating one if need be.
9196 (dump_context::dump_symtab_node): New member function.
9197 (dump_context::get_scope_depth): New member function.
9198 (dump_context::begin_scope): New member function.
9199 (dump_context::end_scope): New member function.
9200 (dump_context::ensure_pending_optinfo): New member function.
9201 (dump_context::begin_next_optinfo): New member function.
9202 (dump_context::end_any_optinfo): New member function.
9203 (dump_context::s_current): New global.
9204 (dump_context::s_default): New global.
9205 (dump_scope_depth): Delete global.
9206 (dumpfile_ensure_any_optinfo_are_flushed): New function.
9207 (dump_symtab_node): New function.
9208 (get_dump_scope_depth): Reimplement in terms of dump_context.
9209 (dump_begin_scope): Likewise.
9210 (dump_end_scope): Likewise.
9211 (selftest::temp_dump_context::temp_dump_context): New ctor.
9212 (selftest::temp_dump_context::~temp_dump_context): New dtor.
9213 (selftest::verify_item): New function.
9214 (ASSERT_IS_TEXT): New macro.
9215 (ASSERT_IS_TREE): New macro.
9216 (ASSERT_IS_GIMPLE): New macro.
9217 (selftest::test_capture_of_dump_calls): New test.
9218 (selftest::dumpfile_c_tests): Call it.
9219 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
9220 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
9221 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
9222 descriptive comment.
9223 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
9224 (dump_node, dump_bb): Move these unrelated decls.
9225 (class dump_manager): Add leading comment.
9226 * optinfo.cc: New file.
9227 * optinfo.h: New file.
9229 2018-07-19 Michael Collison <michael.collison@arm.com>
9230 Richard Henderson <rth@redhat.com>
9232 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
9233 (subti): Handle op1 zero.
9234 (subvti4, usub4ti4): New.
9235 (*sub<GPI>3_compare1_imm): New.
9236 (sub<GPI>3_carryinCV): New.
9237 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
9238 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
9240 2018-07-19 Michael Collison <michael.collison@arm.com>
9241 Richard Henderson <rth@redhat.com>
9243 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
9244 (addti3): Create simpler code if low part is already known to be 0.
9245 (addvti4, uaddvti4): New.
9246 (*add<GPI>3_compareC_cconly_imm): New.
9247 (*add<GPI>3_compareC_cconly): New.
9248 (*add<GPI>3_compareC_imm): New.
9249 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
9250 handle constants within this pattern..
9251 (*add<GPI>3_compareV_cconly_imm): New.
9252 (*add<GPI>3_compareV_cconly): New.
9253 (*add<GPI>3_compareV_imm): New.
9254 (add<GPI>3_compareV): New.
9255 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
9256 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
9257 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
9258 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
9259 with 'comparison' operator.
9260 (*add<GPI>3_compareV_cconly_imm): Ditto.
9261 (*add<GPI>3_compareV_cconly): Ditto.
9262 (*add<GPI>3_compareV_imm): Ditto.
9263 (add<GPI>3_compareV): Ditto.
9264 (add<mode>3_carryinC): Ditto.
9265 (*add<mode>3_carryinC_zero): Ditto.
9266 (*add<mode>3_carryinC): Ditto.
9267 (add<mode>3_carryinV): Ditto.
9268 (*add<mode>3_carryinV_zero): Ditto.
9269 (*add<mode>3_carryinV): Ditto.
9271 2018-07-19 Michael Collison <michael.collison@arm.com>
9272 Richard Henderson <rth@redhat.com>
9274 * config/aarch64/aarch64-modes.def (CC_V): New.
9275 * config/aarch64/aarch64-protos.h
9276 (aarch64_addti_scratch_regs): Declare
9277 (aarch64_subvti_scratch_regs): Declare.
9278 (aarch64_expand_subvti): Declare.
9279 (aarch64_gen_unlikely_cbranch): Declare
9280 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
9281 for signed overflow using CC_Vmode.
9282 (aarch64_get_condition_code_1): Handle CC_Vmode.
9283 (aarch64_gen_unlikely_cbranch): New function.
9284 (aarch64_addti_scratch_regs): New function.
9285 (aarch64_subvti_scratch_regs): New function.
9286 (aarch64_expand_subvti): New function.
9288 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9290 * config/aarch64/aarch64-option-extensions.def: New entry for profile
9292 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
9293 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
9296 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9299 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
9300 address check not strict.
9302 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9304 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
9306 (store_pair_lanes<mode>): Likewise.
9307 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
9308 enum value 'ADDR_QUERY_LDP_STP_N'.
9309 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
9310 (aarch64_print_address_internal): Add declaration.
9311 (aarch64_print_ldpstp_address): Remove.
9312 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
9313 (aarch64_print_operand): Change printing of 'y'.
9314 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
9315 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
9316 'true' rather than '1'.
9317 * config/aarch64/constraints.md (Uml): Likewise.
9318 (Uml): Rename to Umn.
9321 2018-07-19 Richard Biener <rguenther@suse.de>
9323 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
9325 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
9326 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
9327 (vn_reference_hasher): Likewise.
9328 (struct vn_tables_s): Remove obstack and alloc-pool members.
9329 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
9330 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
9331 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
9332 (vn_reference_insert_pieces): Likewise.
9333 (alloc_vn_nary_op_noinit): Adjust.
9334 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
9335 (vn_phi_eq): Adjust.
9336 (shared_lookup_phiargs): Remove.
9337 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
9338 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
9339 (visit_reference_op_call): Likewise.
9340 (copy_nary, copy_phi, copy_reference): Remove.
9341 (process_scc): Rewind the obstack when iterating. Do not
9342 copy the elements to valid_info but just move them from one
9343 hashtable to the other.
9344 (allocate_vn_table): Adjust.
9345 (free_vn_table): Likewise.
9346 (init_scc_vn): Likewise.
9347 (free_scc_vn): Likewise.
9349 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
9352 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
9353 indirect_return as function type attribute.
9354 (ix86_attribute_table): Change indirect_return to function
9356 * doc/extend.texi: Update indirect_return attribute.
9358 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
9360 * wide-int.h (widest2_int): New.
9361 * gimple-fold.c (arith_overflowed_p): Use it.
9362 * tree.h (widest2_int_cst): New.
9363 * tree-vrp.c (wide_int_binop_overflow): Rename from
9364 vrp_int_const_binop.
9365 Rewrite to work on trees.
9366 (extract_range_from_multiplicative_op_1): Abstract code to...
9367 (wide_int_range_min_max): ...here.
9368 (wide_int_range_cross_product): ...and here.
9369 (extract_range_from_binary_expr_1): Abstract overflow code to...
9370 (wide_int_range_mult_wrapping): ...here.
9371 * tree-vrp.h (wide_int_range_cross_product): New.
9372 (wide_int_range_mult_wrapping): New.
9374 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
9375 Julia Koval <julia.koval@intel.com>
9377 * config/i386/x86-tune-costs.h (skylake_memcpy,
9378 skylake_memset): Replace rep_prefix with unrolling for size 512.
9380 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
9383 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
9384 comparision with EQ_EXPR in last stmt.
9386 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
9388 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
9389 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
9390 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
9391 previously known as "PowerPC AltiVec Built-in Functions". Move
9392 some material to new subsubsections "PowerPC AltiVec Built-in
9393 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
9395 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
9396 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
9397 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
9398 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
9400 2018-07-18 Richard Biener <rguenther@suse.de>
9402 PR tree-optimization/86557
9403 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
9406 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
9408 * config/s390/s390.c (s390_function_profiler): Generate CFI.
9410 2018-07-17 Jeff Law <law@redhat.com>
9412 * config/arm/arm.c (get_label_padding): Update for recent
9413 changes to label_to_alignment.
9415 PR tree-optimization/86010
9416 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
9418 * config/mips/mips.c (vr4130_align_insns): Update for recent
9419 changes to label_to_alignment.
9421 * config/frv/frv.c (frv_label_align): Update for recent changes
9422 to label_to_alignment.
9424 * config/nios2/nios2.c (nios2_label_align): Update for recent
9425 changes which dropped ALIGN_LABELS_LOG.
9427 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
9429 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
9430 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
9432 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9434 * config/arc/arc.c (arc_label_align): Use align_labels instead of
9435 deprecated align_labels_log.
9437 2018-07-17 Richard Biener <rguenther@suse.de>
9440 * dwarf2out.c (init_sections_and_labels): Always generate
9441 a debug_line_str_section for early LTO debug.
9442 (dwarf2out_finish): Reset debug_line_str_hash output early.
9443 Bump counter for extra dwarf5 .debug_loc labels to not conflict
9445 (dwarf2out_early_finish): Output debug_line_str.
9447 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
9449 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
9450 index register on z196 or later.
9452 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
9454 * config/s390/s390.c (s390_default_align): Set default function
9456 (s390_override_options_after_change): Call s390_default align.
9457 (s390_option_override_internal): Call s390_default_align.
9458 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
9460 2018-07-17 Jakub Jelinek <jakub@redhat.com>
9463 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
9464 to _looptemp_ clauses, other than the first two.
9466 2018-07-17 Martin Liska <mliska@suse.cz>
9468 * opts.c: Do not enable OPT_falign_* for -Os.
9470 2018-07-17 Martin Liska <mliska@suse.cz>
9472 * align.h (MAX_CODE_ALIGN): New.
9473 (MAX_CODE_ALIGN_VALUE): New.
9474 * common/config/i386/i386-common.c (ix86_handle_option):
9475 (MAX_CODE_ALIGN): Moved to align.h.
9476 * final.c (MAX_CODE_ALIGN): Likewise.
9477 * opts.c (parse_and_check_align_values):
9478 (MAX_CODE_ALIGN): Likewise.
9479 (MAX_CODE_ALIGN_VALUE): Likewise.
9481 2018-07-17 Martin Liska <mliska@suse.cz>
9483 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
9484 in order to fulfil coding style.
9485 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
9486 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9487 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
9488 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
9489 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
9490 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
9491 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
9492 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
9493 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9495 2018-07-17 Martin Liska <mliska@suse.cz>
9497 * align.h: New file.
9498 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
9500 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
9501 align_flags of label_to_alignment.
9502 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
9504 * config/m68k/m68k.c: Do not use removed align_labels_value and
9506 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
9507 (LOOP_ALIGN): Likewise.
9508 (LABEL_ALIGN): Likewise.
9509 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
9510 Remove not used macro.
9511 (rs6000_loop_align): Change return type to align_flags.
9512 (rs6000_loop_align_max_skip): Remove.
9513 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
9514 Change return type to align_flags.
9515 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
9516 Remove not used macro.
9517 (rs6000_loop_align): Change return type to align_flags.
9518 (rs6000_loop_align_max_skip): Remove.
9519 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
9520 * config/rx/rx-protos.h (rx_align_for_label): Make it
9522 * config/rx/rx.c (rx_align_for_label): Change return type
9524 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
9526 into align_flags class.
9527 (LABEL_ALIGN): Likewise.
9528 (LOOP_ALIGN): Likewise.
9529 * config/s390/s390.c (s390_label_align): Use align_flags
9531 (s390_asm_output_function_label): Likewise.
9532 * config/sh/sh.c (sh_override_options_after_change):
9533 Use align_flags class directly without macros.
9534 (find_barrier): Likewise.
9535 (barrier_align): Likewise.
9536 (sh_loop_align): Likewise.
9537 * config/spu/spu.c (spu_option_override):
9538 Use align_flags_tuple::get_value instead of removed macros.
9539 (spu_sched_init): Likewise.
9540 * config/spu/spu.h (GTY): Likewise.
9541 * config/visium/visium.c (visium_option_override):
9542 Set "8" as default secondary alignment.
9543 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
9544 in order to guarantee secondary alignment of 8.
9545 * coretypes.h: Include align.h header file.
9546 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
9547 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
9548 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
9549 * doc/tm.texi.in: Likewise.
9550 * final.c (struct label_alignment): Remove not used structure.
9551 (LABEL_ALIGN): Change type to align_flags.
9552 (LOOP_ALIGN): Likewise.
9553 (JUMP_ALIGN): Likewise.
9554 (default_loop_align_max_skip): Remove.
9555 (default_label_align_max_skip): Likewise.
9556 (default_jump_align_max_skip): Likewise.
9557 (default_label_align_after_barrier_max_skip):
9558 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
9559 (LABEL_TO_MAX_SKIP): Remove.
9560 (label_to_alignment): Return align_flags type instead of integer.
9561 (label_to_max_skip): Remove.
9562 (align_fuzz): Use align_flags type.
9563 (compute_alignments): Use align_flags type and use align_flags::max
9564 to combine multiple alignments.
9565 (grow_label_align): Grow vec instead of C array.
9566 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
9567 (shorten_branches): Use align_flags type and use align_flags::max
9568 to combine multiple alignments.
9569 (final_scan_insn_1): Remove usage of secondary alignment that comes
9570 from label alignment, but instead use proper secondary alignment
9571 which is computed in grow_label_align.
9572 * flags.h (struct align_flags_tuple): Move to align.h.
9573 (struct align_flags): Likewise.
9574 (state_align_loops): Rename to align_loops.
9575 (state_align_jumps): Rename to align_jumps.
9576 (state_align_labels): Rename to align_labels.
9577 (state_align_functions): Rename to align_functions.
9578 (align_loops_log): Remove.
9579 (align_jumps_log): Remove.
9580 (align_labels_log): Remove.
9581 (align_functions_log): Remove.
9582 (align_loops_max_skip): Remove.
9583 (align_jumps_max_skip): Remove.
9584 (align_labels_max_skip): Remove.
9585 (align_functions_max_skip): Remove.
9586 (align_loops_value): Remove.
9587 (align_jumps_value): Remove.
9588 (align_labels_value): Remove.
9589 (align_functions_value): Remove.
9590 * output.h (label_to_alignment): Change return type to align_flags.
9591 (label_to_max_skip): Remove.
9592 * target.def: Remove loop_align_max_skip, label_align_max_skip,
9593 jump_align_max_skip macros.
9594 * targhooks.h (default_loop_align_max_skip): Remove.
9595 (default_label_align_max_skip): Likewise.
9596 (default_jump_align_max_skip): Likewise.
9597 (default_label_align_after_barrier_max_skip): Remove.
9598 * toplev.c (read_log_maxskip): Use ::normalize function.
9599 (parse_N_M): Remove not used argument and also call ::normalize.
9600 (parse_alignment_opts): Do not pass unused arguments.
9601 * varasm.c (assemble_start_function): Use directly align_functions
9602 instead of removed macros.
9603 * system.h: Do not poison removed macros.
9605 2018-07-17 Jakub Jelinek <jakub@redhat.com>
9608 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
9609 and cond temporaries don't have reference type if iterator has
9610 pointer type. For init use &for_pre_body instead of pre_p if
9611 for_pre_body is non-empty.
9613 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
9615 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
9616 double-double modes to SFmode directly directly.
9617 (trunc<mode>sf2_fprs): Delete.
9619 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
9621 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
9622 for conversions between IFmode and the decimal floating point modes.
9623 (init_float128_ieee): Use the correct names for conversions between
9624 KFmode and the decimal floating point modes.
9626 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
9628 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
9629 for the conversions between TDmode and IFmode.
9630 (init_float128_ieee): Use more correct names for the conversions
9631 between TDmode and KFmode.
9633 2018-07-16 Jakub Jelinek <jakub@redhat.com>
9635 PR tree-optimization/86526
9636 * builtins.c (expand_builtin_memcmp): Formatting fixes.
9637 (inline_expand_builtin_string_cmp): Likewise.
9638 (inline_string_cmp): Likewise. Use c_readstr instead of
9639 builtin_memcpy_read_str. Add unit_mode temporary.
9641 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
9644 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
9645 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
9647 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
9649 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
9650 Alphabetize prototypes of built-in functions, separating out
9651 built-in functions that are listed in this section but should be
9652 described elsewhere.
9654 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
9657 * expmed.c (emit_store_flag): Do not emit setcc followed by a
9658 conditional move when trapping comparison was split to a
9659 non-trapping one (and vice versa).
9661 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
9663 * config/s390/s390.c (s390_function_profiler): Generate nops
9664 instead of profiler call sequences.
9665 * config/s390/s390.opt: Add the new option.
9667 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
9669 * config/s390/s390.c (s390_function_profiler): Generate
9670 __mcount_loc section.
9671 * config/s390/s390.opt: Add the new option.
9673 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
9675 * common.opt: Add the new warning.
9676 * config/s390/s390.c (s390_function_profiler): Emit "brasl
9677 %r0,__fentry__" when -mfentry is specified.
9678 (s390_option_override_internal): Disallow -mfentry for 31-bit
9680 * config/s390/s390.opt: Add the new option.
9682 2018-07-16 Richard Biener <rguenther@suse.de>
9685 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
9686 for function-local FUNCTION_DECL and RESULT_DECL immediately.
9688 2018-07-16 Martin Liska <mliska@suse.cz>
9691 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
9694 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
9696 * config/arc/arcHS.md: Update ARCHS scheduling rules.
9698 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
9700 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
9702 * config/arc/arc-cpus.def (hs4x): New cpu.
9704 * config/arc/arc-tables.opt: Regenerate.
9705 * config/arc/arc.c (arc_sched_issue_rate): New function.
9706 (TARGET_SCHED_ISSUE_RATE): Define.
9707 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
9708 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
9710 (attr tune): Add ARCHS4x tune values.
9711 (attr tune_dspmpy): Define.
9712 (*tst): Correct instruction type.
9713 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
9714 * config/arc/arcHS4x.md: New file.
9715 * config/arc/fpu.md: Update instruction type attributes.
9716 * config/arc/t-multilib: Regenerate.
9718 2018-07-16 Tom de Vries <tdevries@suse.de>
9721 * var-tracking.c (vt_initialize): Fix pre_dec handling.
9723 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9725 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
9728 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
9730 PR tree-optimization/86514
9731 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
9732 conversion to a boolean type from a type with greater precision.
9734 2018-07-16 Tom de Vries <tdevries@suse.de>
9736 * var-tracking.c (vt_initialize): Print adjusted insn slim if
9737 dump_flags request TDF_SLIM.
9739 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
9741 * fold-const.c (int_const_binop_1): Abstract...
9742 (wide_int_binop): ...wide int code here.
9743 (poly_int_binop): ...poly int code here.
9744 Abstract the rest of int_const_binop_1 into int_const_binop.
9745 * fold-const.h (wide_int_binop): New.
9746 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
9747 Remove useless PLUS/MINUS_EXPR case.
9748 (zero_nonzero_bits_from_vr): Move wide int code...
9749 (zero_nonzero_bits_from_bounds): ...here.
9750 (extract_range_from_binary_expr_1): Move mask optimization code...
9751 (range_easy_mask_min_max): ...here.
9752 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
9753 (range_easy_mask_min_max): New.
9755 2018-07-15 Jeff Law <law@redhat.com>
9758 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
9761 2018-07-14 Jim Wilson <jimw@sifive.com>
9763 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
9765 2018-07-14 Paul Koning <ni1d@arrl.net>
9767 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
9769 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
9771 * lto-streamer-out.c (copy_function_or_variable): Dump info about
9774 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
9775 Steve Munroe <munroesj52@gmail.com>
9777 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
9778 (_mm_andnot_si128): Likewise.
9779 (_mm_or_si128): Likewise.
9780 (_mm_xor_si128): Likewise.
9782 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
9785 * builtins.c (expand_builtin_memcmp): Inline the calls first
9786 when result_eq is false.
9787 (expand_builtin_strcmp): Inline the calls first.
9788 (expand_builtin_strncmp): Likewise.
9789 (inline_string_cmp): New routine. Expand a string compare
9790 call by using a sequence of char comparison.
9791 (inline_expand_builtin_string_cmp): New routine. Inline expansion
9792 a call to str(n)cmp/memcmp.
9793 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
9795 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
9797 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
9799 * config/arm/driver-arm.c: Include arm-native.h.
9800 (host_detect_local_cpu): Use auto-generated data tables.
9801 (vendors, arm_cpu_table): Delete. Move part information to ...
9802 * config/arm/arm-cpus.in: ... here.
9803 * config/arm/parsecpu.awk (gen_native): New function.
9804 (vendor, part): New CPU fields.
9805 (END): Add support for building the native CPU detection tables.
9806 * config/arm/t-arm (arm-native.h): Add build rule.
9807 (driver-arm.o): Add dependency on arm-native.h.
9809 2018-07-13 Richard Biener <rguenther@suse.de>
9812 * match.pd (addr1 - addr2): Allow either of the operand to
9815 2018-07-13 Tom de Vries <tdevries@suse.de>
9817 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
9820 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
9822 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
9823 arrays instead of numbered variables.
9825 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
9827 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
9828 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
9830 2018-07-13 Richard Biener <rguenther@suse.de>
9833 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
9834 instead of get_context_die.
9836 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9837 Richard Biener <rguenther@suse.de>
9840 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
9841 that the loop latch destination where phi is defined.
9843 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
9845 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
9846 (riscv_expand_epilogue): Add assertion to check interrupt mode.
9847 (riscv_set_current_function): Extract getting interrupt type to new
9849 (riscv_get_interrupt_type): New function.
9850 (riscv_merge_decl_attributes): New function, checking interrupt type is
9852 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
9854 2018-07-12 Paul Koning <ni1d@arrl.net>
9856 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
9859 2018-07-12 Paul Koning <ni1d@arrl.net>
9861 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
9862 zero reference, add doloop_end instead.
9863 * doc/md.texi (decrement_and_branch_until_zero): Remove.
9864 (Looping patterns): Remove decrement_and_branch_until_zero. Add
9865 detail for doloop_end.
9867 2018-07-12 Martin Sebor <msebor@redhat.com>
9870 * attribs.c (decl_attributes): Reject conflicting attributes before
9871 calling attribute handlers.
9873 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
9875 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
9877 (gcc::dump_manager::get_dump_file_name): likewise.
9878 (dump_begin): Likewise.
9879 * dumpfile.h (dump_begin): Update prototype.
9880 (gcc::dump_manager::get_dump_file_name,
9881 gcc::dump_manager::get_dump_file_name): Update prototype.
9883 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
9885 * internal-fn.h (vectorizable_internal_fn_p): New function.
9886 * tree-vect-slp.c (compatible_calls_p): Likewise.
9887 (vect_build_slp_tree_1): Remove nops argument. Handle calls
9888 to internal functions.
9889 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
9891 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
9893 * fold-const.h (inverse_conditions_p): Declare.
9894 * fold-const.c (inverse_conditions_p): New function.
9895 * match.pd: Use inverse_conditions_p. Add folds of view_converts
9896 that test the inverse condition of a conditional internal function.
9897 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
9898 * internal-fn.c (internal_fn_mask_index): Handle conditional
9900 (vectorized_internal_fn_supported_p): New function.
9901 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
9902 (any_pred_load_store): Replace with...
9903 (need_to_predicate): ...this new variable.
9904 (redundant_ssa_names): New variable.
9905 (ifcvt_can_use_mask_load_store): Move initial checks to...
9906 (ifcvt_can_predicate): ...this new function. Handle tree codes
9907 for which a conditional internal function exists.
9908 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
9909 instead of ifcvt_can_use_mask_load_store. Update after variable
9911 (predicate_load_or_store): New function, split out from
9912 predicate_mem_writes.
9913 (check_redundant_cond_expr): New function.
9914 (value_available_p): Likewise.
9915 (predicate_rhs_code): Likewise.
9916 (predicate_mem_writes): Rename to...
9917 (predicate_statements): ...this. Use predicate_load_or_store
9918 and predicate_rhs_code.
9919 (combine_blocks, tree_if_conversion): Update after above name changes.
9920 (ifcvt_local_dce): Handle redundant_ssa_names.
9921 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
9922 general conditional functions.
9923 * tree-vect-stmts.c (vectorizable_call): Likewise.
9925 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
9926 Alan Hayward <alan.hayward@arm.com>
9927 David Sherwood <david.sherwood@arm.com>
9929 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
9930 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
9931 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
9932 plus and minus and convert them into IFN_COND_FMA-based sequences.
9933 (convert_mult_to_fma): Handle conditional plus and minus.
9935 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
9937 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
9938 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
9939 (cond_fnms_optab): New optabs.
9940 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
9942 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
9943 * internal-fn.h (get_conditional_internal_fn): Declare.
9944 (get_unconditional_internal_fn): Likewise.
9945 * internal-fn.c (cond_ternary_direct): New macro.
9946 (expand_cond_ternary_optab_fn): Likewise.
9947 (direct_cond_ternary_optab_supported_p): Likewise.
9948 (FOR_EACH_COND_FN_PAIR): Likewise.
9949 (get_conditional_internal_fn): New function.
9950 (get_unconditional_internal_fn): Likewise.
9951 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
9952 (gimple_match_op::gimple_match_op): Add a new overload for 5
9954 (gimple_match_op::set_op): Likewise.
9955 (gimple_resimplify5): Declare.
9956 * genmatch.c (decision_tree::gen): Generate simplifications for
9958 * gimple-match-head.c (gimple_simplify): Define an overload for
9959 5 operands. Handle calls with 5 arguments in the top-level overload.
9960 (convert_conditional_op): Handle conversions from unconditional
9961 internal functions to conditional ones.
9962 (gimple_resimplify5): New function.
9963 (build_call_internal): Pass a fifth operand.
9964 (maybe_push_res_to_seq): Likewise.
9965 (try_conditional_simplification): Try converting conditional
9966 internal functions to unconditional internal functions.
9967 Handle 3-operand unconditional forms.
9968 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
9969 Define ternary equivalents of the current rules for binary conditional
9971 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
9973 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
9974 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
9975 (optab): Handle them.
9976 (SVE_COND_FP_TERNARY): New int iterator.
9977 (sve_fmla_op, sve_fmad_op): New int attributes.
9978 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
9979 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
9980 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
9982 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
9984 * target.def (preferred_else_value): New target hook.
9985 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
9986 * doc/tm.texi: Regenerate.
9987 * targhooks.h (default_preferred_else_value): Declare.
9988 * targhooks.c (default_preferred_else_value): New function.
9989 * internal-fn.h (conditional_internal_fn_code): Declare.
9990 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
9991 (get_conditional_internal_fn): Use it.
9992 (conditional_internal_fn_code): New function.
9993 * gimple-match.h (gimple_match_cond): New struct.
9994 (gimple_match_op): Add a cond member function.
9995 (gimple_match_op::gimple_match_op): Update all forms to take a
9997 * genmatch.c (expr::gen_transform): Use the same condition as res_op
9998 for the suboperation, but don't specify a particular else_value.
9999 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
10000 (visit_nary_op, visit_reference_op_load): Pass
10001 gimple_match_cond::UNCOND to the gimple_match_op constructor.
10002 * gimple-match-head.c: Include tree-eh.h
10003 (convert_conditional_op): New function.
10004 (maybe_resimplify_conditional_op): Likewise.
10005 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
10006 (gimple_resimplify2): Likewise.
10007 (gimple_resimplify3): Likewise.
10008 (gimple_resimplify4): Likewise.
10009 (maybe_push_res_to_seq): Return null for conditional operations.
10010 (try_conditional_simplification): New function.
10011 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
10013 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
10015 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
10017 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
10019 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
10021 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
10023 (hash_tree): Do not hash DECL_FCONTEXT
10024 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
10025 Do not stream DECL_FCONTEXT.
10026 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
10027 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
10029 2018-07-12 Richard Biener <rguenther@suse.de>
10032 * dwarf2out.c (gen_block_die): Only output blocks when they have
10033 at least one !DECL_IGNORED_P variable.
10035 2018-07-12 Richard Biener <rguenther@suse.de>
10038 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
10039 Remove -mieee-fp handling.
10041 2018-07-12 Richard Biener <rguenther@suse.de>
10043 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
10044 left-over from last patch.
10046 2018-07-12 Jakub Jelinek <jakub@redhat.com>
10048 PR tree-optimization/86492
10049 * gimple-ssa-store-merging.c
10050 (imm_store_chain_info::coalesce_immediate_stores): Call
10051 check_no_overlap even for the merge_overlapping case. Formatting fix.
10053 2018-07-12 Richard Biener <rguenther@suse.de>
10055 PR middle-end/86479
10056 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
10057 move possibly trapping operations into the conditional.
10059 2018-07-12 Richard Biener <rguenther@suse.de>
10061 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
10062 (vn_lookup_simplify_result): Remove recursion limit applied
10064 (vn_nary_build_or_lookup_1): Adjust.
10065 (try_to_simplify): Likewise.
10066 * gimple-match-head.c (gimple_resimplify1): Instead apply one
10068 (gimple_resimplify2): Likewise.
10069 (gimple_resimplify3): Likewise.
10070 (gimple_resimplify4): Likewise.
10072 2018-07-11 Jakub Jelinek <jakub@redhat.com>
10074 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
10075 Use __mmask64 type instead of __mmask8 for __M argument.
10076 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
10077 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
10079 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
10080 __mmask16 for __M argument.
10081 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
10082 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
10083 to __mmask16 instead of __mmask8.
10084 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
10085 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
10086 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
10087 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
10088 instead of __mmask16 for __U argument.
10089 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
10090 __mmask16 instead of __mmask8 for __U argument.
10091 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
10093 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
10095 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
10097 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
10099 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
10101 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
10103 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
10105 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
10107 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
10109 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
10111 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
10112 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
10113 return type as well as __M argument type and all casts from __mmask8
10115 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
10116 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
10117 return type as well as __M argument type and all casts from __mmask8
10119 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
10120 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
10121 return type as well as __M argument type and all casts from __mmask8
10123 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
10124 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
10125 return type as well as __M argument type and all casts from __mmask8
10127 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
10128 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
10131 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
10133 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
10134 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
10137 2018-07-11 Paul Koning <ni1d@arrl.net>
10139 * doc/md.texi (define_subst): Document how multiple occurrences of
10140 the same argument in the replacement pattern are handled.
10142 2018-07-11 Paul Koning <ni1d@arrl.net>
10144 * doc/extend.texi (Common Variable Attributes): Move "mode" into
10145 alphabetical order.
10146 (Common Type Attributes): Add "mode" attribute.
10148 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
10150 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
10151 stream DECL_ORIGINAL_TYPE.
10152 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
10153 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
10154 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
10155 Do not walk original type.
10156 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
10158 (write_ts_decl_non_common_tree_pointers): Do not stream
10160 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
10161 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
10163 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
10165 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
10166 thread twice from the same starting edge.
10168 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
10170 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
10171 * gimple.c (gimple_call_nonnull_result_p): ...here...
10172 (gimple_call_nonnull_arg): ...and here.
10173 * gimple.h (gimple_call_nonnull_result_p): New.
10174 (gimple_call_nonnull_arg): New.
10176 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
10178 * config/arm/arm-cpus.in: Move information from fpu field of each
10179 cpu definition to the isa field.
10180 * config/arm/parsecpu.awk (fpu): Delete match rule.
10181 (gen_comm_data): Don't add bits from the CPU's FPU entry.
10183 2018-07-11 Richard Biener <rguenther@suse.de>
10186 * dwarf2out.c (init_sections_and_labels): Use
10187 output_asm_line_debug_info consistently.
10188 (dwarf2out_early_finish): Likewise.
10189 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
10192 2018-07-11 Richard Biener <rguenther@suse.de>
10194 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
10195 Rework father_bb setting in a way to avoid propagating constants
10196 multiple times on a loop body.
10198 2018-07-10 Mark Wielaard <mark@klomp.org>
10201 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
10203 2018-07-10 Richard Biener <rguenther@suse.de>
10205 * hash-map.h (hash_map::iterator::operator*): Return
10206 references to key and value.
10208 2018-07-10 Jakub Jelinek <jakub@redhat.com>
10211 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
10212 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
10213 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
10214 (gimplify_omp_for): For composite loops, move outer
10215 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
10216 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
10217 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
10218 TREE_LIST for both the original class iterator and the "last" helper
10219 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
10220 loop, remember has_decl_expr from outer composite loops for the
10221 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
10223 2018-07-09 Martin Sebor <msebor@redhat.com>
10225 PR middle-end/77357
10226 PR middle-end/86428
10227 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
10228 accessing implicitly initialized array elements.
10229 * expr.c (string_constant): Handle string initializers of
10230 character arrays within aggregates.
10231 * gimple-fold.c (fold_array_ctor_reference): Add argument.
10232 Store element offset. As a special case, handle zero size.
10233 (fold_nonarray_ctor_reference): Same.
10234 (fold_ctor_reference): Add argument. Store subobject offset.
10235 * gimple-fold.h (fold_ctor_reference): Add argument.
10237 2018-07-09 Paul Koning <ni1d@arrl.net>
10239 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
10240 (pdp11_insn_cost): New function.
10241 (pdp11_md_asm_adjust): New function.
10242 (TARGET_INVALID_WITHIN_DOLOOP): Define.
10243 (pdp11_rtx_costs): Update to match machine better.
10244 (output_addr_const_pdp11): Correct format mismatch warnings.
10245 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
10246 * config/pdp11/pdp11.md: General change to add base_cost and/or
10247 length attributes for use by new pdp11_insn_cost function.
10248 (MIN_BRANCH): Correct definition.
10250 (doloop_end): Use standard pattern name for looping pattern.
10251 (doloop_end_nocc): New.
10252 (movsf): Add another constraint alternative.
10253 (zero_extendqihi2): Add constraint alternatives for not in place
10255 (zero_extendhisi2): Remove.
10256 (shift patterns): Add CC handling variants.
10260 (define_peephole2): New peephole to recognize mov that sets CC for
10263 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10265 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
10266 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
10269 2018-07-09 Jakub Jelinek <jakub@redhat.com>
10272 * real.c (real_nextafter): Return true if result is denormal.
10274 2018-07-09 Martin Liska <mliska@suse.cz>
10276 * common.opt: Add back wrongly removed attribute.
10278 2018-07-09 Richard Biener <rguenther@suse.de>
10281 * dwarf2out.c (gen_block_die): For an early generated DIE
10282 always output high/low PC attributes.
10284 2018-07-09 Tom de Vries <tdevries@suse.de>
10286 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
10287 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
10288 onto VAR_DECL with abstract origin.
10290 2018-07-07 Jim Wilson <jimw@sifive.com>
10292 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
10294 2018-07-07 Tom de Vries <tdevries@suse.de>
10296 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
10298 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
10300 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
10302 (combine_bound): Use wide-int overflow calculation instead of
10304 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
10306 * fold-const.c (int_const_binop_2): Same.
10307 (extract_muldiv_1): Same.
10308 (fold_div_compare): Same.
10309 (fold_abs_const): Same.
10311 * poly-int.h (add): Same.
10315 * predict.c (predict_iv_comparison): Same.
10316 * profile-count.c (slow_safe_scale_64bit): Same.
10317 * simplify-rtx.c (simplify_const_binary_operation): Same.
10318 * tree-chrec.c (tree_fold_binomial): Same.
10319 * tree-data-ref.c (split_constant_offset_1): Same.
10320 * tree-if-conv.c (idx_within_array_bound): Same.
10321 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
10322 * tree-ssa-phiopt.c (minmax_replacement): Same.
10323 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
10324 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
10325 * vr-values.c (vr_values::adjust_range_with_scev): Same.
10326 * wide-int.cc (wi::add_large): Same.
10327 (wi::mul_internal): Same.
10328 (wi::sub_large): Same.
10329 (wi::divmod_internal): Same.
10330 * wide-int.h: Change overflow type to overflow_type for neg, add,
10331 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
10332 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
10333 mul_internal, divmod_internal.
10334 (overflow_type): New enum.
10335 (accumulate_overflow): New.
10337 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
10339 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
10340 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
10342 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
10344 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
10345 argument is checked for zero before entering loop, avoid checking again.
10347 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
10349 * gimplify.h (generic_expr_could_trap_p): Set as global function.
10350 * gimplify.h (generic_expr_could_trap_p): Likwise.
10351 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
10353 2018-07-06 Jakub Jelinek <jakub@redhat.com>
10355 PR tree-optimization/86401
10356 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
10357 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
10358 (fold_bit_and_mask): ... here. New helper function for match.pd.
10359 * fold-const.h (fold_bit_and_mask): Declare.
10360 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
10362 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
10365 * target.def (translate_mode_attribute): New hook.
10366 * targhooks.h (default_translate_mode_attribute): Declare.
10367 * targhooks.c (default_translate_mode_attribute): New function.
10368 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
10369 * doc/tm.texi: Regenerate.
10370 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
10371 (rs6000_translate_mode_attribute): New function.
10373 2018-07-06 Paul Koning <ni1d@arrl.net>
10375 * doc/md.texi (define_split): Document DONE and FAIL.
10376 (define_peephole2): Ditto.
10378 2018-07-05 Jeff Law <law@redhat.com>
10380 PR tree-optimization/86010
10381 * tree-ssa-dse.c (compute_trims): More aggressively trim at
10382 both the head and tail of mem* and str* calls.
10384 2018-07-05 Jim Wilson <jimw@sifive.com>
10386 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
10388 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
10390 * config/aarch64/aarch64-simd.md: correct flags text for
10391 MIN_EXPR replacement.
10393 2018-07-05 James Clarke <jrtc27@jrtc27.com>
10395 * configure: Regenerated.
10397 2018-07-05 Carl Love <cel@us.ibm.com>
10399 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
10400 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
10401 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
10402 VSX_BUILTIN_DOUBLEL_V4SF.
10404 2018-07-05 Martin Sebor <msebor@redhat.com>
10407 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
10408 than its domain to compute its the upper bound of a char array.
10410 2018-07-05 Nathan Sidwell <nathan@acm.org>
10412 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
10413 * doc/cpp.texi: Update comment.
10414 * doc/tm.texi: Rebuilt.
10415 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
10416 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
10417 * doc/extend.texi (Backwards Compatibility): Clarify it is system
10418 headers affected by extern "C".
10419 * system.h: Poison NO_IMPLICIT_EXTERN_C.
10420 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
10421 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
10422 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
10423 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
10424 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
10425 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
10426 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
10427 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
10428 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
10429 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
10430 config/sparc/sp64-elf.h, config/spu/spu.h,
10431 config/stormy16/stormy16.h, config/v850/v850.h,
10432 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
10433 define NO_IMPLICIT_EXTERN_C.
10434 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
10436 2018-07-05 Tamar Christina <tamar.christina@arm.com>
10439 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
10440 instead of GET_MODE_SIZE when comparing Units.
10442 2018-07-05 Tamar Christina <tamar.christina@arm.com>
10445 * rtlanal.c (set_noop_p): Constrain on mode change,
10446 include hard-reg-set.h
10448 2018-07-05 Tamar Christina <tamar.christina@arm.com>
10450 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
10452 2018-07-05 Jakub Jelinek <jakub@redhat.com>
10455 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
10458 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
10460 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
10462 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
10465 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
10467 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
10469 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
10471 PR middle-end/86380
10472 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
10474 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
10476 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
10477 neg_*_op* variables.
10479 2018-07-04 Martin Liska <mliska@suse.cz>
10481 * tree-switch-conversion.c: Define
10482 max_ratio_for_speed and max_ratio_for_size constants.
10484 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
10485 Martin Liska <mliska@suse.cz>
10487 PR middle-end/66240
10490 * common.opt: Rename align options with 'str_' prefix.
10491 * common/config/i386/i386-common.c (set_malign_value): New
10493 (ix86_handle_option): Use it to set -falign-* options/
10494 * config/aarch64/aarch64-protos.h (struct tune_params): Change
10495 type from int to string.
10496 * config/aarch64/aarch64.c: Update default values from int
10498 * config/alpha/alpha.c (alpha_override_options_after_change):
10500 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
10501 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10502 max skip conditionally.
10503 * config/i386/freebsd.h (SUBALIGN_LOG): New.
10504 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10505 max skip conditionally.
10506 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10507 max skip conditionally.
10508 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
10509 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10510 max skip conditionally.
10511 * config/i386/i386.c (struct ptt): Change type from int to
10513 (ix86_default_align): Set default values.
10514 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
10515 max skip conditionally.
10516 * config/i386/iamcu.h (SUBALIGN_LOG): New.
10517 (ASM_OUTPUT_MAX_SKIP_ALIGN):
10518 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
10519 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10520 max skip conditionally.
10521 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
10522 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
10523 * config/i386/x86-64.h (SUBALIGN_LOG): New.
10524 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10525 max skip conditionally.
10526 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
10527 * config/ia64/ia64.c (ia64_option_override): Set default values
10528 for alignment options.
10529 * config/m68k/m68k.c: Handle new str_align_* options.
10530 * config/mips/mips.c (mips_set_compression_mode): Change
10532 (mips_option_override): Set default values for options.
10533 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
10535 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10537 * config/rx/rx.c (rx_option_override): Likewise.
10538 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
10539 (LABEL_ALIGN): Use align_labels_log.
10540 (LOOP_ALIGN): Use align_loops_align.
10541 * config/s390/s390.c (s390_asm_output_function_label): Use new
10543 * config/sh/sh.c (sh_override_options_after_change):
10544 Change type of constants.
10545 * config/spu/spu.c (spu_sched_init): Likewise.
10546 * config/sparc/sparc.c (sparc_option_override): Set default
10547 values for options.
10548 * config/visium/visium.c (visium_option_override): Likewise.
10549 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
10550 emit p2align format with last argument if it's not needed.
10551 * doc/invoke.texi: Document extended format of -falign-*.
10552 * final.c: Use align_labels alignment.
10553 * flags.h (struct target_flag_state): Change type to use
10555 (struct align_flags_tuple): New.
10556 (struct align_flags): Likewise.
10557 (align_loops_log): Redefine macro to use new types.
10558 (align_loops_max_skip): Redefine macro to use new types.
10559 (align_jumps_log): Redefine macro to use new types.
10560 (align_jumps_max_skip): Redefine macro to use new types.
10561 (align_labels_log): Redefine macro to use new types.
10562 (align_labels_max_skip): Redefine macro to use new types.
10563 (align_functions_log): Redefine macro to use new types.
10564 (align_loops): Redefine macro to use new types.
10565 (align_jumps): Redefine macro to use new types.
10566 (align_labels): Redefine macro to use new types.
10567 (align_functions): Redefine macro to use new types.
10568 (align_functions_max_skip): Redefine macro to use new types.
10569 (align_loops_value): New macro.
10570 (align_jumps_value): New macro.
10571 (align_labels_value): New macro.
10572 (align_functions_value): New macro.
10573 * function.c (invoke_set_current_function_hook): Propagate
10574 alignment values from flags to global variables default in
10576 * ipa-icf.c (sem_function::equals_wpa): Use
10577 cl_optimization_option_eq instead of memcmp.
10578 * lto-streamer.h (cl_optimization_stream_out): Support streaming
10580 (cl_optimization_stream_in): Likewise.
10581 * optc-save-gen.awk: Support strings in cl_optimization.
10582 * opth-gen.awk: Likewise.
10583 * opts.c (finish_options): Remove error checking of invalid
10585 (MAX_CODE_ALIGN): Remove.
10586 (MAX_CODE_ALIGN_VALUE): Likewise.
10587 (parse_and_check_align_values): New function.
10588 (check_alignment_argument): Likewise.
10589 (common_handle_option): Use check_alignment_argument.
10590 * opts.h (parse_and_check_align_values): Declare.
10591 * toplev.c (init_alignments): Remove.
10592 (read_log_maxskip): New.
10593 (parse_N_M): Likewise.
10594 (parse_alignment_opts): Likewise.
10595 (backend_init_target): Remove usage of init_alignments.
10596 * toplev.h (parse_alignment_opts): Declare.
10597 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
10599 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
10600 * tree.c (cl_option_hasher::equal): New.
10601 * varasm.c: Use new global macros.
10603 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
10605 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
10606 Use a simpler align directive also if MAXSKIP = ALIGN-1.
10607 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10608 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10609 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10610 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
10611 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
10612 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
10613 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
10614 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10615 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10616 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10617 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10619 2018-07-04 Martin Liska <mliska@suse.cz>
10620 Jonathan Wakely <jwakely@redhat.com>
10622 * coverage.c: Use correct type.
10623 * doc/invoke.texi: Language correction.
10625 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
10628 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
10629 ENDBRANCH for non-tail call which may return via indirect branch.
10630 * doc/extend.texi: Document indirect_return attribute.
10632 2018-07-03 Martin Sebor <msebor@redhat.com>
10634 PR tree-optimization/86274
10635 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
10637 (format_floating): Correct handling of infinities and NaNs.
10639 2018-07-03 Martin Sebor <msebor@redhat.com>
10641 * print-tree.c (print_real_cst): New function.
10642 (print_node_brief): Call it.
10643 (print_node): Ditto.
10645 2018-07-03 Jeff Law <law@redhat.com>
10647 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
10648 into a single pattern.
10650 * config/h8300/h8300.md (ors code_iterator): New.
10651 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
10652 a single pattern and single splitter.
10653 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
10654 (iorqi3_1, xorqi3_1): Likewise.
10655 (iorqi3, xorqi3 expanders): Similarly.
10657 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
10658 (movmd_internal) into a single pattern using the P mode iterator.
10659 (movmd splitters): Similarly.
10660 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
10661 (movsd splitters): Similarly.
10663 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
10664 ADDB, ADDW and ADDL into a single ADD attribute which selects the
10665 right table based on the size of the operand.
10666 * config/h8300/h8300.md (length_table): Corresponding changes. All
10667 references to "addb", "addw" and "addl" changed to "add".
10668 (btst patterns): Merge two variants into a single pattern.
10669 (tstqi, tsthi): Likewise.
10670 (addhi3_incdec, addsi3_incdec): Likewise.
10671 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
10672 (mulhi3, mulsi3): Likewise.
10673 (udivhi3, udivsi3): Likewise.
10674 (divhi3, divsi3): Likewise.
10675 (andorqi3, andorhi3, andorsi3): Likewise.
10677 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
10680 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
10681 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
10683 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10685 PR tree-optimization/85694
10686 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
10687 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
10689 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
10690 (<u>avg<mode>3_ceil): New patterns.
10692 2018-07-03 David Malcolm <dmalcolm@redhat.com>
10694 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
10695 scan-tree-dump directive.
10696 * gcc.dg/vect/slp-perm-2.c: Likewise.
10697 * gcc.dg/vect/slp-perm-3.c: Likewise.
10698 * gcc.dg/vect/slp-perm-5.c: Likewise.
10699 * gcc.dg/vect/slp-perm-6.c: Likewise.
10700 * gcc.dg/vect/slp-perm-7.c: Likewise.
10701 * gcc.dg/vect/slp-perm-8.c: Likewise.
10703 2018-07-03 Marek Polacek <polacek@redhat.com>
10705 PR middle-end/86202
10706 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
10708 2018-07-03 Richard Biener <rguenther@suse.de>
10711 * tree-ssa-structalias.c (find_func_clobbers): Properly
10712 handle indirect calls.
10714 2018-07-03 Jeff Law <law@redhat.com>
10716 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
10717 (shifts): New code iterator.
10718 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
10719 expander. Fix HImode handling on H8/SX.
10720 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
10721 (subqi3, subhi3, subsi3 expanders): Likewise.
10722 (andqi3, andhi3, andsi3 expanders): Likewise.
10723 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
10724 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
10725 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
10726 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
10727 (zero_extendqihi2, zero_extendqisi2): Likewise.
10728 (extendqihi2, extendqisi2): Likewise.
10729 (rotlqi3, rotlhi3, rotlsi3): Likewise.
10730 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
10731 (rotlqi3_1, rotlhi3_1): Likewise.
10732 (logicalhi3_sn, logicalsi3_sn): Likewise.
10733 (logicalhi3, logicalsi3): Likewise.
10735 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10737 * tree-vect-patterns.c (vect_recog_rotate_pattern)
10738 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
10739 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
10740 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
10741 type to append_pattern_def_seq instead of creating a stmt_vec_info
10743 (build_mask_conversion): Likewise. Remove vinfo argument.
10744 (vect_add_conversion_to_patterm): Likewise, renaming to...
10745 (vect_add_conversion_to_pattern): ...this.
10746 (vect_recog_mask_conversion_pattern): Update call to
10747 build_mask_conversion. Pass the vector type to
10748 append_pattern_def_seq here too.
10749 (vect_recog_gather_scatter_pattern): Update call to
10750 vect_add_conversion_to_pattern.
10752 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10754 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
10755 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
10756 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
10757 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
10758 STMT_VINFO_PATTERN_DEF_SEQ to null here.
10759 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
10760 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
10761 append_pattern_def_seq instead of new_pattern_def_seq.
10762 (vect_recog_divmod_pattern): Do both of the above.
10763 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
10766 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10768 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
10769 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
10770 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
10771 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
10772 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
10773 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
10774 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
10775 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
10776 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
10777 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
10778 parameter with a single stmt_vec_info.
10779 (vect_recog_func_ptr): Likewise.
10780 (vect_recog_gather_scatter_pattern): Likewise, folding in...
10781 (vect_try_gather_scatter_pattern): ...this.
10782 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
10783 the stmt_vec_info of the statement to be matched. Don't clear
10784 STMT_VINFO_RELATED_STMT.
10785 (vect_pattern_recog): Update call accordingly.
10787 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10789 PR tree-optimization/85694
10790 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
10791 (uavgM3_ceil): Document new optabs.
10792 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
10793 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
10795 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
10796 (savg_ceil_optab): New optabs.
10797 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
10798 (vect_vect_recog_func_ptrs): Add it.
10799 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
10800 constant directly from the associated lhs.
10802 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10804 * tree-vect-patterns.c (vect_split_statement): New function.
10805 (vect_convert_input): Use it to try to split an existing cast.
10807 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10809 * poly-int.h (print_hex): New function.
10810 * dumpfile.h (dump_dec, dump_hex): Declare.
10811 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
10812 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
10813 min_input_precision, operation_precision and operation_sign.
10814 * tree-vect-patterns.c (vect_get_range_info): New function.
10815 (vect_same_loop_or_bb_p, vect_single_imm_use)
10816 (vect_operation_fits_smaller_type): Delete.
10817 (vect_look_through_possible_promotion): Add an optional
10818 single_use_p parameter.
10819 (vect_recog_over_widening_pattern): Rewrite to use new
10820 stmt_vec_info infomration. Handle one operation at a time.
10821 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
10822 (vect_truncatable_operation_p, vect_set_operation_type)
10823 (vect_set_min_input_precision): New functions.
10824 (vect_determine_min_output_precision_1): Likewise.
10825 (vect_determine_min_output_precision): Likewise.
10826 (vect_determine_precisions_from_range): Likewise.
10827 (vect_determine_precisions_from_users): Likewise.
10828 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
10829 (vect_vect_recog_func_ptrs): Put over_widening first.
10831 (vect_pattern_recog): Call vect_determine_precisions.
10833 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10835 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
10836 statements that have been replaced by further pattern statements.
10837 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
10839 2018-07-03 Richard Biener <rguenther@suse.de>
10841 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
10842 always set *dt. Dump vectype in vectype overload.
10843 * dumpfile.h (dump_gimple_expr): New function.
10844 (dump_gimple_expr_loc): Likewise.
10845 * dumpfile.c (dump_gimple_expr): New function.
10846 (dump_gimple_expr_loc): Likewise.
10848 2018-07-02 Jeff Law <law@redhat.com>
10850 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
10851 the H8/300, H8/300H and H8/S variants into a single pattern.
10852 (movhi_h8300, movqi_h8300hs): Similarly.
10853 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
10854 (QHI mode iterator): New.
10856 * config/h8300/h8300.md: Remove trailing whitespace.
10858 2018-07-02 Jim Wilson <jimw@sifive.com>
10860 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
10861 instead of emit_insn for interrupt returns.
10862 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
10863 (riscv_sret, riscv_uret): Likewise.
10865 2018-07-02 David Malcolm <dmalcolm@redhat.com>
10867 * pretty-print.c (selftest::test_pp_format): Move save and restore
10868 of quotes to class auto_fix_quotes, and add an instance.
10869 * selftest.c: Include "intl.h".
10870 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
10871 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
10872 * selftest.h (selftest::auto_fix_quotes): New class.
10874 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
10876 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
10877 (aarch64_sve_prepare_conditional_op): Remove.
10878 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
10879 Allow aarch64_simd_reg_or_zero as select operand; remove
10880 the aarch64_sve_prepare_conditional_op call.
10881 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
10882 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
10883 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
10884 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
10885 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
10886 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
10887 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
10888 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
10889 and a splitters to match all of the *_any patterns.
10890 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
10892 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
10893 (SVE_COND_FP_BINARY_REV): Remove.
10894 (sve_int_op_rev, sve_fp_op_rev): New.
10895 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
10896 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
10897 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
10898 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
10899 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
10900 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
10901 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
10902 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
10903 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
10905 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
10906 Remove match_dup 1 from the inner unspec.
10907 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
10909 * config/aarch64/aarch64.md (movprfx): New attr.
10910 (length): Default movprfx to 8.
10911 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
10912 (*madd<SVE_I>, *msub<SVE_I): Likewise.
10913 (*<su>mul<SVE_I>3_highpart): Likewise.
10914 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
10915 (*v<ASHIFT><SVE_I>3): Likewise.
10916 (*<su><MAXMIN><SVE_I>3): Likewise.
10917 (*<su><MAXMIN><SVE_F>3): Likewise.
10918 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
10919 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
10920 (*div<SVE_F>4): Likewise.
10922 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
10924 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
10927 2018-07-02 Richard Biener <rguenther@suse.de>
10929 PR tree-optimization/86363
10930 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
10931 memset argument refers to a non-variable address.
10933 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
10935 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
10936 {PLUS,MINUS}_EXPR code to...
10937 (adjust_symbolic_bound): ...here,
10938 (combine_bound): ...here,
10939 (set_value_range_with_overflow): ...and here.
10941 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
10943 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
10945 (extract_range_from_abs_expr): ...here.
10947 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
10949 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
10950 -fno-omit-frame-pointer when not optimizing.
10952 2018-07-02 Martin Liska <mliska@suse.cz>
10955 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
10956 (propagate_nothrow): Likewise.
10958 2018-07-02 Martin Liska <mliska@suse.cz>
10961 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
10963 2018-07-02 David Malcolm <dmalcolm@redhat.com>
10965 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
10966 function in r262149, changing "loc" param from source_location to
10967 const dump_location_t &.
10968 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
10969 declaration, as above.
10971 2018-07-01 Paul Koning <ni1d@arrl.net>
10973 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
10974 -munit-asm, -mgnu-asm, -mdec-asm.
10975 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
10976 (pdp11_output_labelref): New.
10977 (pdp11_output_def): New.
10978 (pdp11_output_addr_vec_elt): New.
10979 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
10980 %# and %@ format codes.
10981 (pdp11_option_override): New.
10982 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
10983 (pdp11_output_ident): New.
10984 (pdp11_asm_named_section): New.
10985 (pdp11_asm_init_sections): New.
10986 (pdp11_file_start): New.
10987 (pdp11_file_end): New.
10988 (output_ascii): Use .ascii/.asciz for -mdec-asm.
10989 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
10990 %o, like %c but octal.
10991 (pdp11_option_override): New.
10992 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
10994 (DATA_SECTION_ASM_OP): Ditto.
10995 (READONLY_DATA_SECTION_ASM_OP): New.
10996 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
10997 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
10998 (ASM_OUTPUT_LABELREF): Ditto.
10999 (ASM_OUTPUT_DEF): Ditto.
11000 (ASM_OUTPUT_EXTERNAL): New.
11001 (ASM_OUTPUT_SOURCE_FILENAME): New.
11002 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
11003 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
11004 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
11005 %# and %@ format codes.
11006 * config/pdp11/pdp11.opt (mgnu-asm): New.
11007 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
11008 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
11009 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
11011 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
11013 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
11014 dereferencing path[] beyond its length.
11016 (debug_all_paths): New.
11017 (rewire_first_differing_edge): New.
11018 (adjust_paths_after_duplication): New.
11019 (duplicate_thread_path): Call adjust_paths_after_duplication.
11021 (thread_through_all_blocks): Add new argument to
11022 duplicate_thread_path.
11024 2018-06-30 Jim Wilson <jimw@sifive.com>
11026 * config/riscv/predicates.md (p2m1_shift_operand): New.
11027 (high_mask_shift_operand): New.
11028 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
11029 pattern using p2m1_shift_operand.
11030 (lshsi3_zero_extend_3+2): New combiner pattern using
11031 high_mask_shift_operand.
11033 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11035 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
11037 (vect_recog_rotate_pattern): ...here.
11038 (vect_convert_input): Try to insert casts of invariants in the
11040 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
11041 preheader to be empty.
11043 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11045 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
11046 vector type. If given, install it in the new statement's
11047 STMT_VINFO_VECTYPE.
11048 (vect_element_precision): New function.
11049 (vect_unpromoted_value): New struct.
11050 (vect_unpromoted_value::vect_unpromoted_value): New function.
11051 (vect_unpromoted_value::set_op): Likewise.
11052 (vect_look_through_possible_promotion): Likewise.
11053 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
11054 (vect_widened_op_tree, vect_convert_input): Likewise.
11055 (vect_convert_inputs, vect_convert_output): Likewise.
11056 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
11057 to handle the optional cast of the multiplication result and
11058 vect_widened_op_tree to detect the widened multiplication itself.
11059 Do not require the input and output of promotion casts to have
11060 the same sign, but base the signedness of the operation on the
11061 input rather than the result. If the pattern includes two
11062 promotions, check that those promotions have the same sign.
11063 Do not restrict the MULT_EXPR handling to a double-width result;
11064 handle quadruple-width results and wider. Use vect_convert_inputs
11065 to convert the inputs to the common type.
11066 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
11067 to handle the optional cast of the ABS result. Also allow a sign
11068 change or a sign extension between the ABS and MINUS.
11069 Use vect_widened_op_tree to detect the widened subtraction and use
11070 vect_convert_inputs to convert the inputs to the common type.
11071 (vect_handle_widen_op_by_const): Delete.
11072 (vect_recog_widen_op_pattern): New function.
11073 (vect_recog_widen_mult_pattern): Use it.
11074 (vect_recog_widen_shift_pattern): Likewise.
11075 (vect_recog_widen_sum_pattern): Use
11076 vect_look_through_possible_promotion to handle the promoted
11079 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11081 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
11082 the containing gimple_seq *.
11083 * gimple-iterator.h (gsi_for_stmt): Declare it.
11084 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
11085 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
11086 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
11087 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
11088 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
11090 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
11092 (vect_mark_pattern_stmts): ...here. Handle cases in which the
11093 statement being replaced is part of an existing pattern
11094 definition sequence, inserting the new pattern statements before
11096 (vect_pattern_recog_1): Don't return a bool. If the statement
11097 is already part of a pattern, instead apply pattern matching
11098 to the pattern definition statements. Don't clear the
11099 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
11100 (vect_pattern_recog): Don't break after the first match;
11101 continue processing the pattern definition statements instead.
11102 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
11104 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11106 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
11107 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
11108 (vect_recog_widen_sum_pattern): Use it.
11110 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11112 * tree-vect-loop.c (vectorizable_reduction): Assert that the
11113 phi is not a pattern statement and has not been replaced by
11114 a pattern statement.
11115 * tree-vect-patterns.c (type_conversion_p): Don't check
11116 STMT_VINFO_IN_PATTERN_P.
11117 (vect_recog_vector_vector_shift_pattern): Likewise.
11118 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
11119 the pattern statement rather than the original statement; check
11120 directly for a WIDEN_MULT_EXPR here.
11121 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
11122 vect_is_simple_use to return the pattern statement rather
11123 than the original statement; use is_pattern_stmt_p to check
11124 for such a pattern statement.
11125 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
11126 to return the pattern statement rather than the original statement;
11127 don't do the same transformation here.
11128 (vect_is_simple_use): If the defining statement has been replaced
11129 by a pattern statement, return the pattern statement instead.
11130 Remove the corresponding (local) transformation from the vectype
11133 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11135 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
11136 end and default to null.
11137 * tree-vect-loop.c (vect_create_epilog_for_reduction)
11138 (vectorizable_reduction): Update calls accordingly, dropping the
11139 gimple ** argument if the passed-back statement isn't needed.
11140 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
11141 (vect_recog_rotate_pattern): Likewise.
11142 (vect_recog_mask_conversion_pattern): Likewise.
11143 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
11144 (vect_mask_constant_operand_p): Likewise.
11145 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
11146 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
11147 (get_group_load_store_type, get_load_store_type): Likewise.
11148 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
11149 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
11150 (vectorizable_conversion, vectorizable_assignment): Likewise.
11151 (vectorizable_shift, vectorizable_operation): Likewise.
11152 (vectorizable_store, vect_is_simple_cond): Likewise.
11153 (vectorizable_condition, vectorizable_comparison): Likewise.
11154 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
11155 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
11156 and move it to the end. Cope with null def_stmt_outs.
11158 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
11160 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
11162 2018-06-29 Jeff Law <law@redhat.com>
11164 * config/v850/v850.c (v850_legitimate_address_p): Handle large
11165 displacements for TARGET_V850E2V3 and newer.
11166 (TARGET_LRA_P): Remove. Defaults to LRA now.
11167 * config/v850/v850.md (sign23byte_load): Remove.
11168 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
11169 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
11171 2018-06-29 Martin Liska <mliska@suse.cz>
11174 * coverage.c (coverage_init): Mangle full path name.
11175 * doc/invoke.texi: Document the change.
11176 * gcov-io.c (mangle_path): New.
11177 * gcov-io.h (mangle_path): Likewise.
11178 * gcov.c (mangle_name): Use mangle_path for path mangling.
11180 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11182 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
11183 if starting source register is not even.
11185 2018-06-29 Martin Liska <mliska@suse.cz>
11187 PR tree-optimization/86263
11188 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
11189 Make edge redirection.
11191 2018-06-29 David Malcolm <dmalcolm@redhat.com>
11193 * dumpfile.c (dump_loc): Add indentation based on scope depth.
11194 (dump_scope_depth): New variable.
11195 (get_dump_scope_depth): New function.
11196 (dump_begin_scope): New function.
11197 (dump_end_scope): New function.
11198 * dumpfile.h (get_dump_scope_depth): New declaration.
11199 (dump_begin_scope): New declaration.
11200 (dump_end_scope): New declaration.
11201 (class auto_dump_scope): New class.
11202 (AUTO_DUMP_SCOPE): New macro.
11203 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
11206 2018-06-29 Richard Biener <rguenther@suse.de>
11208 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
11209 compute_all_dependences succeeds.
11210 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
11211 exceed --param loop-max-datarefs-for-datadeps.
11213 2018-06-29 Jakub Jelinek <jakub@redhat.com>
11215 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
11217 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
11220 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
11221 alternative 0 in preferred_for_speed attribute.
11223 2018-06-28 Paul Koning <ni1d@arrl.net>
11225 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
11226 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
11227 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
11228 * config/pdp11/pdp11.md: Correct "length" attribute calculation
11229 for shift insn patterns.
11231 2018-06-28 David Malcolm <dmalcolm@redhat.com>
11233 * cgraph.c (cgraph_node::get_body): Replace assignments to
11234 "dump_file" with calls to set_dump_file.
11235 * dumpfile.c (alt_dump_file): Make static, and group with...
11236 (alt_flags): ...this definition.
11237 (dumps_are_enabled): New variable.
11238 (refresh_dumps_are_enabled): New function.
11239 (set_dump_file): New function.
11240 (set_alt_dump_file): New function.
11241 (gcc::dump_manager::dump_start): Replace assignments to
11242 "dump_file" and "alt_dump_file" with calls to set_dump_file and
11244 (gcc::dump_manager::dump_finish): Likewise.
11245 * dumpfile.h (alt_dump_file): Delete decl.
11246 (dumps_are_enabled): New variable decl.
11247 (set_dump_file): New function decl.
11248 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
11250 * tree-nested.c (lower_nested_functions): Replace assignments to
11251 "dump_file" with calls to set_dump_file.
11253 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
11255 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
11256 goto_locus of each outgoing edge of each basic block.
11258 2018-06-28 Richard Biener <rguenther@suse.de>
11260 * dwarf2out.c (decl_scope_table): Remove.
11261 (push_decl_scope): Likewise.
11262 (pop_decl_scope): Likewise.
11263 (gen_type_die_for_member): Do not call push/pop_decl_scope.
11264 (gen_struct_or_union_type_die): Likewise.
11265 (gen_tagged_type_die): Likewise.
11266 (dwarf2out_init): Do not initialize decl_scope_table.
11267 (dwarf2out_c_finalize): Do not free it.
11269 2018-06-28 Richard Biener <rguenther@suse.de>
11271 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
11272 deciding whether to not re-use a DIE.
11274 2018-06-28 Richard Biener <rguenther@suse.de>
11276 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
11277 DW_AT_abstract_origin attribute.
11279 2018-06-28 Martin Liska <mliska@suse.cz>
11281 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
11282 Use newly introduced constants.
11283 * tree-switch-conversion.h (struct jump_table_cluster):
11284 Define max_ratio_for_size and max_ratio_for_speed.
11286 2018-06-28 Martin Liska <mliska@suse.cz>
11288 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
11289 Add new checking assert to catch invalid state.
11290 (jump_table_cluster::can_be_handled): Handle single case
11292 (jump_table_cluster::is_beneficial): Bail out for such case.
11293 (bit_test_cluster::find_bit_tests):
11294 Add new checking assert to catch invalid state.
11295 (bit_test_cluster::can_be_handled): Handle single case
11297 (bit_test_cluster::is_beneficial): Bail out for such case.
11298 (switch_decision_tree::analyze_switch_statement):
11301 2018-06-28 Martin Liska <mliska@suse.cz>
11303 * common.opt: Introduce -completion option.
11304 * gcc.c (driver_handle_option): Handle it.
11305 (driver::main): Print completions if completion
11307 * opt-suggestions.c (option_proposer::get_completions):
11309 (option_proposer::suggest_completion): Likewise.
11310 (option_proposer::find_param_completions): Likewise.
11311 (verify_autocompletions): Likewise.
11312 (test_completion_valid_options): Likewise.
11313 (test_completion_valid_params): Likewise.
11314 (in_completion_p): Likewise.
11315 (empty_completion_p): Likewise.
11316 (test_completion_partial_match): Likewise.
11317 (test_completion_garbage): Likewise.
11318 (opt_proposer_c_tests): Likewise.
11319 * opt-suggestions.h: Declare new functions.
11320 * opts.c (common_handle_option): Handle OPT__completion_.
11321 * selftest-run-tests.c (selftest::run_tests): Add
11322 opt_proposer_c_tests.
11323 * selftest.c (assert_str_startswith): New.
11324 * selftest.h (assert_str_startswith): Likewise.
11325 (opt_proposer_c_tests): New.
11326 (ASSERT_STR_STARTSWITH): Likewise.
11328 2018-06-28 Martin Liska <mliska@suse.cz>
11330 * Makefile.in: Add opt-suggestions.o.
11331 * gcc-main.c: Include opt-suggestions.h.
11332 * gcc.c (driver::driver): Likewise.
11333 (driver::~driver): Remove m_option_suggestions.
11334 (driver::build_option_suggestions): Moved to option_proposer.
11335 (driver::suggest_option): Likewise.
11336 (driver::handle_unrecognized_options): Use option_proposer.
11337 * gcc.h (class driver): Add new memver m_option_proposer.
11338 * opt-suggestions.c: New file.
11339 * opt-suggestions.h: New file.
11341 2018-06-28 Martin Liska <mliska@suse.cz>
11343 * vec.h (class auto_string_vec): New (moved from auto_argvec).
11344 (auto_string_vec::~auto_string_vec): Likewise.
11346 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
11348 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
11349 prevent_decl_creation_for_types fields up and add reset_location field.
11350 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
11351 statement if id->reset_location is true.
11352 (copy_edges_for_bb): Do not set goto_locus on the new edges if
11353 id->reset_location is true.
11354 (copy_phis_for_bb): Force input_location on the arguments if
11355 id->reset_location is true.
11356 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
11357 is set on the function to be inlined.
11359 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
11361 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
11363 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
11365 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
11366 registers for Pmode.
11367 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
11368 hard registers for the clobbered pseudo.
11370 2018-06-27 Paul Koning <ni1d@arrl.net>
11372 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
11373 mutually exclusive options.
11374 * config/pdp11/constraints.md (h): New constraint.
11375 (O): Update definition to match shift code generation.
11376 (D): New constraint.
11377 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
11379 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
11381 (output_jump): Change arguments.
11382 (pdp11_fixed_cc_regs): New function.
11383 (pdp11_cc_mode): Ditto.
11384 (pdp11_expand_shift): Ditto.
11385 (pdp11_assemble_shift): Ditto.
11386 (pdp11_small_shift): Ditto.
11387 (pdp11_branch_cost): Remove.
11388 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
11390 (pdp11_register_move_cost): Update for CC registers.
11391 (pdp11_rtx_costs): Add case for LSHIFTRT.
11392 (pdp11_output_jump): Add CCNZ mode conditional branches.
11393 (notice_update_cc_on_set): Remove.
11394 (pdp11_cc_mode): New function.
11395 (simple_memory_operand): Correct pre/post decrement case.
11396 (no_side_effect_operand): New function.
11397 (pdp11_regno_reg_class): Add CC_REGS class.
11398 (pdp11_fixed_cc_regs): New function.
11399 (pdp11_small_shift): New function.
11400 (pdp11_expand_shift): New function to expand shift insns.
11401 (pdp11_assemble_shift): New function to output shifts.
11402 (pdp11_branch_cost): Remove.
11403 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
11404 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
11405 (WCHAR_TYPE): Ditto.
11406 (PTRDIFF_TYPE): Ditto.
11407 (ADJUST_INSN_LENGTH): New macro.
11408 (FIXED_REGISTERS): Add CC registers.
11409 (CALL_USED_REGISTERS): Ditto.
11410 (reg_class): Ditto.
11411 (REG_CLASS_NAMES): Ditto.
11412 (REG_CLASS_CONTENTS): Ditto.
11413 (SELECT_CC_MODE): Use new function.
11414 (TARGET_FLAGS_REGNUM): New macro.
11415 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
11416 (cc0_reg_rtx): Remove.
11417 (CC_STATUS_MDEP): Remove.
11418 (CC_STATUS_MDEFP_INIT): Remove.
11419 (CC_IN_FPU): Remove.
11420 (NOTICE_UPDATE_CC): Remove.
11421 (REGISTER_NAMES): Add CC registers.
11422 (BRANCH_COST): Change to constant 1.
11423 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
11425 * config/pdp11/pdp11.opt (mbcopy): Remove.
11426 (mbcopy-builtin): Remove.
11427 (mbranch-cheap): Remove.
11428 (mbranch-expensive): Remove.
11429 * config/pdp11/predicates.md (expand_shift_operand): Update to
11430 match shift code generation.
11431 (ccnz_operator): New predicate.
11432 * doc/invoke.texi (PDP-11 Options): Remove deleted options
11433 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
11434 Remove non-existent option -mabshi, -mno-abshi. Document mutually
11436 * doc/md.texi (PDP-11): Document new D and h constraints. Update
11437 description of O constraint.
11439 2018-06-27 Jeff Law <law@redhat.com>
11440 Austin Law <austinklaw@gmail.com>
11442 * config/v850/v850.md (addsi3_set_flags): New pattern.
11443 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
11444 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
11445 (zero_extendhisi2_v850_set_flags): Likewise.
11446 (zero_extendqisi2_v850_set_flags): Likewise.
11447 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
11448 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
11449 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
11451 * config/v850/v850-protos.h (notice_update_cc): Remove.
11452 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
11453 (v850_print_operand): Handle 'D' and "d".
11454 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
11455 Add handling of arithmetic/logical operations compared against zero.
11456 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
11457 Do not look at v850_compare_op, instead get mode from last argument.
11458 (v850_gen_compare): Remove
11459 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
11460 after reload for prologue insns.
11461 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
11463 (construct_save_jarl): Likewise.
11464 (TARGET_FLAGS_REGNUM): Define.
11465 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
11466 (NOTICE_UPDATE_CC): Remove.
11467 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
11468 than cc0. Conditionalize on reload_completed.
11469 (cmpsi_insn, setfcc_insn): Likewise.
11470 (tst1 splitter): Turn into define_and_split which sets the flags
11472 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
11473 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
11474 (cstoresf4, cstoredf4): Clobber the flags.
11475 (cmpsi, cmpsf, cmpdf): Remove expanders.
11476 (setf_insn): Remove pattern.
11477 (addsi3): Turn into define_and_split which clobbers the flags after
11478 reload and a suitable pattern (addsi3_clobber_flags) for use after
11480 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
11481 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
11482 (ashrsi3, ashrsi3_v850e2): Likewise.
11483 (bins): Clobber the flags.
11484 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
11485 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
11486 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
11487 (call_value_internal_short, call_value_internal_long): Likewise.
11488 (callt_save_interrupt, callt_return_interrupt): Likewise.
11489 (save_interrupt, return_interrupt): Likewise.
11490 (callt_save_all_interrupt, save_all_interrupt): Likewise.
11491 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
11492 (restore_all_interrupt, _restore_all_interrupt): Likewise.
11493 (All FP comparisons): Only allow after reload has completed.
11495 (divh, divhu): Tweak output template.
11496 (branch_z_normal, branch_z_invert): Remove
11497 (branch_nz_normal, branch_nz_invert): Likewise.
11498 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
11500 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
11501 * config/v850/v850.c (notice_update_cc): Remove.
11502 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
11503 (CC_NO_CARRY): Likewise.
11504 (NOTICE_UPDATE_CC): Define to nothing.
11505 * config/v850/v850.md: Remove block comment on cc0 handling
11506 Remove "cc" attribute from all patterns. Remove cc_status handling
11507 from all patterns. Minor formatting fixes.
11509 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11511 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
11512 (cortex-a76.cortex-a55): Likewise.
11513 * config/aarch64/aarch64-tune.md: Regenerate.
11514 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
11515 cortex-a76.cortex-a55.
11517 2018-06-27 Jeff Law <law@redhat.com>
11519 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
11520 (MULTILIB_DIRNAMES): Similarly.
11522 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
11524 * gimple.h (gimple_return_retbnd): Delete.
11525 (gimple_return_set_retbnd): Likewise.
11526 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
11527 gimple_return_set_retbnd.
11528 * gimple-pretty-print.c (dump_gimple_return): Remove call to
11529 gimple_return_retbnd and adjust.
11530 * tree-inline.h (struct copy_body_data): Remove retbnd field.
11531 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
11532 Explicitly return NULL in a couple more cases. Move assertion
11533 on debug statements and remove unreachable code.
11534 (reset_debug_binding): Do not test id->retbnd.
11535 (expand_call_inline): Do not set it.
11537 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
11539 * configure.ac: Add --disable-gcov option.
11540 * configure: Regenerate.
11541 * Makefile.in: Honour @enable_gcov@.
11542 * doc/install.texi: Document --disable-gcov.
11544 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11546 * config/arm/arm-cpus.in (cortex-a76): New entry.
11547 (cortex-a76.cortex-a55): Likewise.
11548 * config/arm/arm-tables.opt: Regenerate.
11549 * config/arm/arm-tune.md: Likewise.
11550 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
11551 * doc/invoke.texi (ARM Options): Document cortex-a76 and
11552 cortex-a76.cortex-a55.
11554 2018-06-27 Tamar Christina <tamar.christina@arm.com>
11557 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
11559 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
11561 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
11563 (EPILOGUE_USES): Likewise.
11565 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
11567 * tree-inline.c (remap_location): New function extracted from...
11568 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
11569 (copy_phis_for_bb): ...here. Call remap_location.
11570 (copy_cfg_body): Adjust call to copy_edges_for_bb.
11572 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
11574 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
11575 unaligned vsx for 16B memset.
11577 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
11580 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
11581 ieee128_float_type_node to long_double_type_node unless
11582 TARGET_LONG_DOUBLE_128 is set.
11584 2018-06-26 David Malcolm <dmalcolm@redhat.com>
11586 * cfgloop.c (get_loop_location): Convert return type from
11587 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
11588 by implicit construction from rtx_insn *, and using
11589 dump_user_location_t::from_function_decl for the fallback case.
11590 * cfgloop.h (get_loop_location): Convert return type from
11591 location_t to dump_user_location_t.
11592 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
11593 dump_printf_loc to pass in a dump_location_t rather than a
11594 location_t, via the gimple stmt.
11595 * coverage.c (get_coverage_counts): Update calls to
11596 dump_printf_loc to pass in dump_location_t rather than a
11598 * doc/optinfo.texi (Dump types): Convert example of
11599 dump_printf_loc from taking "locus" to taking "insn". Update
11600 description of the "_loc" calls to cover dump_location_t.
11601 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
11603 (dump_user_location_t::dump_user_location_t): New constructors,
11604 from gimple *stmt and rtx_insn *.
11605 (dump_user_location_t::from_function_decl): New function.
11606 (dump_loc): Make static.
11607 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
11608 const dump_location_t &.
11609 (dump_generic_expr_loc): Delete.
11610 (dump_printf_loc): Convert param "loc" from location_t to
11611 const dump_location_t &.
11612 (selftest::test_impl_location): New function.
11613 (selftest::dumpfile_c_tests): New function.
11614 * dumpfile.h: Include "profile-count.h".
11615 (class dump_user_location_t): New class.
11616 (struct dump_impl_location_t): New struct.
11617 (class dump_location_t): New class.
11618 (dump_printf_loc): Convert 2nd param from source_location to
11619 const dump_location_t &.
11620 (dump_generic_expr_loc): Delete.
11621 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
11622 const dump_location_t &.
11623 * gimple-fold.c (fold_gimple_assign): Update call to
11624 dump_printf_loc to pass in a dump_location_t rather than a
11625 location_t, via the gimple stmt.
11626 (gimple_fold_call): Likewise.
11627 * gimple-loop-interchange.cc
11628 (loop_cand::analyze_iloop_reduction_var): Update for change to
11629 check_reduction_path.
11630 (tree_loop_interchange::interchange): Update for change to
11631 find_loop_location.
11632 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
11633 change in return-type of find_loop_location.
11634 (graphite_regenerate_ast_isl): Likewise.
11635 * graphite-optimize-isl.c (optimize_isl): Likewise.
11636 * graphite.c (graphite_transform_loops): Likewise.
11637 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
11638 pass in a dump_location_t rather than a location_t, via the
11640 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
11641 * ipa.c (walk_polymorphic_call_targets): Likewise.
11642 * loop-unroll.c (report_unroll): Convert "locus" param from
11643 location_t to dump_location_t.
11644 (decide_unrolling): Update for change to get_loop_location's
11646 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
11647 location_t to dump_user_location_t.
11648 (grid_find_single_omp_among_assignments_1): Updates calls to
11649 dump_printf_loc to pass in a dump_location_t rather than a
11650 location_t, via the gimple stmt.
11651 (grid_parallel_clauses_gridifiable): Convert "tloc" from
11652 location_t to dump_location_t. Updates calls to dump_printf_loc
11653 to pass in a dump_location_t rather than a location_t, via the
11655 (grid_inner_loop_gridifiable_p): Likewise.
11656 (grid_dist_follows_simple_pattern): Likewise.
11657 (grid_gfor_follows_tiling_pattern): Likewise.
11658 (grid_target_follows_gridifiable_pattern): Likewise.
11659 (grid_attempt_target_gridification): Convert initialization
11660 of local "grid" from memset to zero-initialization; FIXME: does
11661 this require C++11? Update call to dump_printf_loc to pass in a
11662 optinfo_location rather than a location_t, via the gimple stmt.
11663 * profile.c (read_profile_edge_counts): Updates call to
11664 dump_printf_loc to pass in a dump_location_t rather than a
11666 (compute_branch_probabilities): Likewise.
11667 * selftest-run-tests.c (selftest::run_tests): Call
11669 * selftest.h (dumpfile_c_tests): New decl.
11670 * tree-loop-distribution.c (pass_loop_distribution::execute):
11671 Update for change in return type of find_loop_location.
11672 * tree-parloops.c (parallelize_loops): Likewise.
11673 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
11674 "locus" from location_t to dump_user_location_t.
11675 (canonicalize_loop_induction_variables): Likewise.
11676 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
11677 for change in return type of find_loop_location.
11678 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
11679 to dump_printf_loc to pass in a dump_location_t rather than a
11680 location_t, via the stmt.
11681 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
11683 * tree-vect-loop-manip.c (find_loop_location): Convert return
11684 type from source_location to dump_user_location_t.
11685 (vect_do_peeling): Update for above change.
11686 (vect_loop_versioning): Update for change in type of
11688 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
11689 from location_t to dump_user_location_t.
11690 (vect_estimate_min_profitable_iters): Update for change in type
11692 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
11693 location_t to dump_location_t.
11694 (vect_slp_bb): Update for change in type of vect_location.
11695 * tree-vectorizer.c (vect_location): Convert from source_location
11696 to dump_user_location_t.
11697 (try_vectorize_loop_1): Update for change in vect_location's type.
11698 (vectorize_loops): Likewise.
11699 (increase_alignment): Likewise.
11700 * tree-vectorizer.h (vect_location): Convert from source_location
11701 to dump_user_location_t.
11702 (find_loop_location): Convert return type from source_location to
11703 dump_user_location_t.
11704 (check_reduction_path): Convert 1st param from location_t to
11705 dump_user_location_t.
11706 * value-prof.c (check_counter): Update call to dump_printf_loc to
11707 pass in a dump_user_location_t rather than a location_t; update
11708 call to error_at for change in type of "locus".
11709 (check_ic_target): Update call to dump_printf_loc to
11710 pass in a dump_user_location_t rather than a location_t, via the
11713 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
11715 * config/s390/s390.h (enum processor_flags): Do not use
11716 default tune parameter when -march was specified.
11718 2018-06-26 Jakub Jelinek <jakub@redhat.com>
11721 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
11722 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
11725 2018-06-26 Richard Biener <rguenther@suse.de>
11727 PR tree-optimization/86287
11729 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
11730 (vect_analyze_loop): Initialize n_stmts.
11732 2018-06-26 Richard Biener <rguenther@suse.de>
11734 PR middle-end/86271
11735 * fold-const.c (fold_convertible_p): Pointer extension
11738 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
11741 * dwarf2out.c (loc_list_has_views): Adjust comments.
11742 (dw_loc_list): Split single cross-partition range with
11745 2018-06-25 Jeff Law <law@redhat.com>
11747 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
11748 on -mbig-switch by default.
11750 * config/v850/predicates.md (const_float_1_operand): Fix match_code
11752 (const_float_0_operand): Remove unused predicate.
11753 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
11754 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
11755 (recipsf2): New expander. Original pattern now called
11757 (recipdf2, recipdf2_insn): Similarly.
11758 (rsqrtsf2, rsqrtsf2_insn): Similarly
11759 (rsqrtdf2, rsqrtdf2_insn): Similarly
11761 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
11763 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
11764 Simplify logic for FreeBSD (twice).
11766 2018-06-25 Martin Sebor <msebor@redhat.com>
11768 PR tree-optimization/86204
11769 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
11770 a strnlen result if it's less than the length of the string.
11772 2018-06-25 Martin Sebor <msebor@redhat.com>
11774 PR tree-optimization/85700
11775 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
11776 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
11777 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
11779 2018-06-25 Martin Sebor <msebor@redhat.com>
11781 * doc/extend.texi (Zero-length arrays): Update and clarify.
11783 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
11785 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
11786 added IEEE/IBM long double multilib support on PowerPC little
11787 endian Linux systems.
11788 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
11789 (MULTILIB_DEFAULTS): Likewise.
11790 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11792 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
11793 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
11794 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
11796 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
11798 PR middle-end/86311
11799 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
11800 (REORDER_45): Likewise.
11802 2018-06-25 Jeff Law <law@redhat.com>
11804 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
11805 dividend to 32 bits. Adjust length.
11806 (udivmodhi4): Cleanup output template. Fix length.
11808 2018-06-25 Carl Love <cel@us.ibm.com>
11810 * config/rs6000/vsx.md: Change word selector to prefered location.
11812 2018-06-25 Richard Biener <rguenther@suse.de>
11814 PR tree-optimization/86304
11815 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
11816 epilogue-if-converted loops as well.
11818 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
11820 * lto-section-out.c (lto_begin_section): Do not print section
11821 name for noaddr and unnumbered dumps.
11823 2018-06-25 Richard Biener <rguenther@suse.de>
11825 * tree-vectorizer.h (struct vec_info_shared): New structure
11826 with parts split out from struct vec_info and loop_nest from
11827 struct _loop_vec_info.
11828 (struct vec_info): Adjust accordingly.
11829 (struct _loop_vec_info): Likewise.
11830 (LOOP_VINFO_LOOP_NEST): Adjust.
11831 (LOOP_VINFO_DATAREFS): Likewise.
11832 (LOOP_VINFO_DDRS): Likewise.
11833 (struct _bb_vec_info): Likewise.
11834 (BB_VINFO_DATAREFS): Likewise.
11835 (BB_VINFO_DDRS): Likewise.
11836 (struct _stmt_vec_info): Add dr_aux member.
11837 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
11838 (DR_MISALIGNMENT_UNINITIALIZED): New.
11839 (set_dr_misalignment): Adjust.
11840 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
11841 (vect_analyze_loop): Adjust prototype.
11842 (vect_analyze_loop_form): Likewise.
11843 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
11844 Compute dependences lazily.
11845 (vect_record_base_alignments): Use shared datarefs/ddrs.
11846 (vect_verify_datarefs_alignment): Likewise.
11847 (vect_analyze_data_refs_alignment): Likewise.
11848 (vect_analyze_data_ref_accesses): Likewise.
11849 (vect_analyze_data_refs): Likewise.
11850 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
11851 constructor parameter for shared part.
11852 (vect_analyze_loop_form): Pass in shared part and adjust.
11853 (vect_analyze_loop_2): Pass in storage for the number of
11854 stmts. Move loop nest finding to the caller. Compute
11856 (vect_analyze_loop): Pass in shared part.
11857 (vect_transform_loop): Verify shared datarefs are unchanged.
11858 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
11859 constructor parameter for shared part.
11860 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
11861 (vect_slp_bb): Verify shared datarefs are unchanged before
11863 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
11865 (new_stmt_vec_info): Initialize DR_AUX misalignment to
11866 DR_MISALIGNMENT_UNINITIALIZED.
11867 * tree-vectorizer.c (vec_info::vec_info): Add constructor
11868 parameter for shared part.
11869 (vec_info::~vec_info): Adjust.
11870 (vec_info_shared::vec_info_shared): New.
11871 (vec_info_shared::~vec_info_shared): Likewise.
11872 (vec_info_shared::save_datarefs): Likewise.
11873 (vec_info_shared::check_datarefs): Likewise.
11874 (try_vectorize_loop_1): Construct shared part live for analyses
11875 of a single loop for multiple vector sizes.
11876 * tree-parloops.c (gather_scalar_reductions): Adjust.
11878 2018-06-25 Richard Biener <rguenther@suse.de>
11880 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
11881 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
11882 (vect_analyze_data_refs): Remove similar code from here and
11883 simplify accordingly.
11885 2018-06-25 Richard Biener <rguenther@suse.de>
11887 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
11888 for reverse storage order accesses rather than asserting
11889 they cannot happen here.
11891 2018-06-25 Tom de Vries <tdevries@suse.de>
11894 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
11895 Use data16 instead of .byte for insn prefix.
11897 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
11900 * parser.c (make_char_string_pack): Pass this literal chars
11901 through cpp_interpret_string.
11902 (cp_parser_userdef_numeric_literal): Check the result of
11903 make_char_string_pack.
11905 2018-06-24 Maya Rashish <coypu@sdf.org>
11907 * ginclude/stddef.h: Simplify conditions around avoiding
11908 re-definition of __size_t.
11910 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
11912 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
11913 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
11915 2018-06-22 Maya Rashish <coypu@sdf.org>
11917 * doc/invoke.texi (mno-fancy-math-387): Update for changes
11918 made to OpenBSD and NetBSD through the years.
11920 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
11922 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
11923 behavior of vec_pack (vector double, vector double) to match
11924 behavior of vec_float2 (vector double, vector double).
11926 2018-06-22 Olivier Hainque <hainque@adacore.com>
11928 * gimplify.c (gimplify_function_tree): Prevent creation
11929 of a trampoline for the address of the current function
11930 passed to entry/exit instrumentation hooks.
11932 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
11935 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
11938 2018-06-22 Martin Liska <mliska@suse.cz>
11940 PR tree-optimization/86263
11941 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
11942 Bail out if is_enabled is false.
11943 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
11945 (jump_table_cluster::is_enabled): New function.
11947 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
11949 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
11950 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
11951 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
11952 (lto_input_ts_binfo_tree_pointers): Likewise.
11953 * tree-streamer-out.c (streamer_write_tree_bitfields,
11954 write_ts_binfo_tree_pointers): Likewise.
11955 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
11957 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
11959 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
11961 2018-06-22 Martin Liska <mliska@suse.cz>
11963 * symbol-summary.h (get): Make it pure and inline move
11964 functionality from ::get function.
11965 (get): Remove and inline into ::get and ::get_create.
11966 (get_create): Move code from ::get function.
11968 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11971 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
11972 -x assembler-with-cpp.
11974 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11976 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
11977 _FILE_OFFSET_BITS=64 for C++.
11979 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
11981 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
11982 conversion insn that shows up when pr85657-3.c is compiled using
11983 IEEE 128-bit long double.
11984 (neg<mode>2_internal): Use the correct mode to check whether the
11985 mode is IBM extended.
11986 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
11987 multiply and divide external functions from being created more
11990 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
11992 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
11994 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
11995 the edge can be forwarded.
11996 (cfg_layout_merge_blocks): Likewise.
11998 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12000 * except.c (finish_eh_generation): Commit edge insertions only after
12001 the EH edges have been redirected from post-landing to landing pads.
12003 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12005 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
12006 create_tmp_var_for to create the FRAME decl.
12007 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
12009 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12011 * tree-inline.c (copy_edges_for_bb): Minor tweak.
12012 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
12013 debug statement when resetting its value.
12014 (expand_call_inline): Copy the locus of the call onto the assignment
12015 of the return value, if any. Use local variable in more cases.
12017 2018-06-21 Martin Liska <mliska@suse.cz>
12019 * ipa-pure-const.c (propagate_nothrow): Use
12020 funct_state_summaries->get.
12021 (dump_malloc_lattice): Likewise.
12022 (propagate_malloc): Likewise.
12024 2018-06-21 Richard Biener <rguenther@suse.de>
12026 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
12027 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
12028 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
12030 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
12031 BLOCK_ABSTRACT_ORIGIN unconditionally.
12033 2018-06-21 David Malcolm <dmalcolm@redhat.com>
12035 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
12037 * ipa-reference.c (ipa_reference_c_finalize): Delete
12038 ipa_ref_opt_sum_summaries and set it to NULL.
12040 2018-06-21 Tom de Vries <tdevries@suse.de>
12042 PR tree-optimization/85859
12043 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
12044 test with comment from bb_no_side_effects_p.
12046 2018-06-21 Richard Biener <rguenther@suse.de>
12048 PR tree-optimization/86232
12049 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
12050 max for constant niter.
12052 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
12054 * config/aarch64/aarch64-simd.md
12055 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
12057 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
12059 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
12060 Make opernads of the unspec commutative.
12062 2018-06-21 Richard Biener <rguenther@suse.de>
12064 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
12065 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
12066 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
12067 (vect_analyze_data_ref_dependence): Re-order checks to deal with
12069 (vect_record_base_alignments): Do not record base alignment
12070 for gathers or scatters.
12071 (vect_compute_data_ref_alignment): Drop return value that is always
12072 true. Bail out early for gathers or scatters.
12073 (vect_enhance_data_refs_alignment): Bail out early for gathers
12075 (vect_find_same_alignment_drs): Likewise.
12076 (vect_analyze_data_refs_alignment): Remove dead code.
12077 (vect_slp_analyze_and_verify_node_alignment): Likewise.
12078 (vect_analyze_data_refs): For possible gathers or scatters do
12079 not create an alternate DR, just check their possible validity
12080 and mark them. Adjust DECL_NONALIASED handling to not rely
12081 on DR_BASE_ADDRESS.
12082 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
12083 update inits of gathers or scatters.
12084 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
12085 Also copy gather/scatter flag to pattern vinfo.
12087 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
12089 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
12090 behavior of vec_packsu (vector unsigned long long, vector unsigned
12091 long long) to match behavior of vec_packs with same signature.
12093 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
12094 Thomas Schwinge <thomas@codesourcery.com>
12095 Cesar Philippidis <cesar@codesourcery.com>
12097 * gimplify.c (gimplify_scan_omp_clauses): Add support for
12098 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
12099 (gimplify_adjust_omp_clauses): Likewise.
12100 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
12101 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
12102 (gimplify_omp_target_update): Update handling of acc update and
12104 * omp-low.c (install_var_field): Remove unused parameter
12105 base_pointers_restrict.
12106 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
12107 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
12109 (omp_target_base_pointers_restrict_p): Delete.
12110 (scan_omp_target): Update call to scan_sharing_clauses.
12111 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
12113 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
12114 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
12115 (convert_local_omp_clauses): Likewise.
12116 * tree-pretty-print.c (dump_omp_clause): Likewise.
12117 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
12119 (omp_clause_code_name): Likewise.
12121 2018-06-20 Jakub Jelinek <jakub@redhat.com>
12124 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
12127 PR tree-optimization/86231
12128 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
12129 anti-range don't overwrite *vr0min before using it to compute *vr0max.
12131 2018-06-20 Tom de Vries <tdevries@suse.de>
12133 PR tree-optimization/86097
12134 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
12135 iv type if signedness of iv type is not the same as that of *nit.
12137 2018-06-20 Jakub Jelinek <jakub@redhat.com>
12139 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
12140 EDGE_EH edges, verify they are all EDGE_EH.
12142 2018-06-20 Maya Rashish <coypu@sdf.org>
12144 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
12146 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12148 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
12149 * config/aarch64/aarch64.c (xgene1_tunings): Add
12150 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
12151 (aarch64_mode_valid_for_sched_fusion_p):
12152 Allow 16-byte modes.
12153 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
12154 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
12156 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
12158 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
12159 * config/aarch64/iterators.md (VQ2): New mode iterator.
12161 2018-06-20 Martin Liska <mliska@suse.cz>
12163 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
12164 Change default ratio from 10 to 8.
12166 2018-06-20 Martin Liska <mliska@suse.cz>
12168 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
12170 (bit_test_cluster::find_bit_tests): Likewise.
12171 (switch_decision_tree::analyze_switch_statement): Find clusters.
12172 * tree-switch-conversion.h (struct jump_table_cluster): Document
12175 2018-06-20 Martin Liska <mliska@suse.cz>
12177 * tree-switch-conversion.c (switch_conversion::collect):
12178 Record m_uniq property.
12179 (switch_conversion::expand): Bail out for special conditions.
12180 (group_cluster::~group_cluster): New.
12181 (group_cluster::group_cluster): Likewise.
12182 (group_cluster::dump): Likewise.
12183 (jump_table_cluster::emit): New.
12184 (switch_decision_tree::fix_phi_operands_for_edges): New.
12185 (struct case_node): Remove struct.
12186 (jump_table_cluster::can_be_handled): New.
12187 (case_values_threshold): Moved to header.
12188 (reset_out_edges_aux): Likewise.
12189 (jump_table_cluster::is_beneficial): New.
12190 (bit_test_cluster::can_be_handled): Likewise.
12191 (add_case_node): Remove.
12192 (bit_test_cluster::is_beneficial): New.
12193 (case_bit_test::cmp): New.
12194 (bit_test_cluster::emit): New.
12195 (expand_switch_as_decision_tree_p): Remove.
12196 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
12197 (fix_phi_operands_for_edge): Likewise.
12198 (switch_decision_tree::analyze_switch_statement): New.
12199 (compute_cases_per_edge): Move ...
12200 (switch_decision_tree::compute_cases_per_edge): ... here.
12201 (try_switch_expansion): Likewise.
12202 (switch_decision_tree::try_switch_expansion): Likewise.
12203 (record_phi_operand_mapping): Likewise.
12204 (switch_decision_tree::record_phi_operand_mapping): Likewise.
12205 (emit_case_decision_tree): Likewise.
12206 (switch_decision_tree::emit): Likewise.
12207 (balance_case_nodes): Likewise.
12208 (switch_decision_tree::balance_case_nodes): Likewise.
12209 (dump_case_nodes): Likewise.
12210 (switch_decision_tree::dump_case_nodes): Likewise.
12211 (emit_jump): Likewise.
12212 (switch_decision_tree::emit_jump): Likewise.
12213 (emit_cmp_and_jump_insns): Likewise.
12214 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
12215 (emit_case_nodes): Likewise.
12216 (switch_decision_tree::emit_case_nodes): Likewise.
12217 (conditional_probability): Remove.
12218 * tree-switch-conversion.h (enum cluster_type): New.
12220 (struct cluster): Likewise.
12221 (cluster::cluster): Likewise.
12222 (struct simple_cluster): Likewise.
12223 (simple_cluster::simple_cluster): Likewise.
12224 (struct group_cluster): Likewise.
12225 (struct jump_table_cluster): Likewise.
12226 (struct bit_test_cluster): Likewise.
12227 (struct min_cluster_item): Likewise.
12228 (struct case_tree_node): Likewise.
12229 (case_tree_node::case_tree_node): Likewise.
12230 (jump_table_cluster::case_values_threshold): Likewise.
12231 (struct case_bit_test): Likewise.
12232 (struct switch_decision_tree): Likewise.
12233 (struct switch_conversion): Likewise.
12234 (switch_decision_tree::reset_out_edges_aux): Likewise.
12236 2018-06-20 Martin Liska <mliska@suse.cz>
12238 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
12239 (hoist_edge_and_branch_if_true): Likewise.
12240 (expand_switch_using_bit_tests_p): Likewise.
12241 (struct case_bit_test): Likewise.
12242 (case_bit_test_cmp): Likewise.
12243 (emit_case_bit_tests): Likewise.
12244 (switch_conversion::switch_conversion): New class.
12245 (struct switch_conv_info): Remove old struct.
12246 (collect_switch_conv_info): More to ...
12247 (switch_conversion::collect): ... this.
12248 (check_range): Likewise.
12249 (switch_conversion::check_range): Likewise.
12250 (check_all_empty_except_final): Likewise.
12251 (switch_conversion::check_all_empty_except_final): Likewise.
12252 (check_final_bb): Likewise.
12253 (switch_conversion::check_final_bb): Likewise.
12254 (create_temp_arrays): Likewise.
12255 (switch_conversion::create_temp_arrays): Likewise.
12256 (free_temp_arrays): Likewise.
12257 (gather_default_values): Likewise.
12258 (switch_conversion::gather_default_values): Likewise.
12259 (build_constructors): Likewise.
12260 (switch_conversion::build_constructors): Likewise.
12261 (constructor_contains_same_values_p): Likewise.
12262 (switch_conversion::contains_same_values_p): Likewise.
12263 (array_value_type): Likewise.
12264 (switch_conversion::array_value_type): Likewise.
12265 (build_one_array): Likewise.
12266 (switch_conversion::build_one_array): Likewise.
12267 (build_arrays): Likewise.
12268 (switch_conversion::build_arrays): Likewise.
12269 (gen_def_assigns): Likewise.
12270 (switch_conversion::gen_def_assigns): Likewise.
12271 (prune_bbs): Likewise.
12272 (switch_conversion::prune_bbs): Likewise.
12273 (fix_phi_nodes): Likewise.
12274 (switch_conversion::fix_phi_nodes): Likewise.
12275 (gen_inbound_check): Likewise.
12276 (switch_conversion::gen_inbound_check): Likewise.
12277 (process_switch): Use the newly created class.
12278 (switch_conversion::expand): New.
12279 (switch_conversion::~switch_conversion): New.
12280 * tree-switch-conversion.h: New file.
12282 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12284 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
12285 tree-vect-patterns.c.
12286 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
12287 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
12288 (vect_recog_sad_pattern): Likewise.
12289 (vect_recog_widen_sum_pattern): Likewise.
12290 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
12291 (vect_recog_widen_shift_pattern): Remove the type_in argument.
12292 (vect_recog_rotate_pattern): Likewise.
12293 (vect_recog_mult_pattern): Likewise.
12294 (vect_recog_vector_vector_shift_pattern): Likewise.
12295 (vect_recog_divmod_pattern): Likewise.
12296 (vect_recog_mixed_size_cond_pattern): Likewise.
12297 (vect_recog_bool_pattern): Likewise.
12298 (vect_recog_mask_conversion_pattern): Likewise.
12299 (vect_try_gather_scatter_pattern): Likewise.
12300 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
12301 (vect_recog_over_widening_pattern): Likewise.
12302 (vect_recog_gather_scatter_pattern): Likewise.
12303 (vect_recog_func_ptr): Move from tree-vectorizer.h
12304 (vect_vect_recog_func_ptrs): Move further down the file.
12305 (vect_recog_func): Likewise. Remove the third argument.
12306 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
12307 (vect_pattern_recog_1): Expect the pattern function to do any
12308 necessary target tests. Also expect it to provide a vector type.
12309 Remove the type_in handling.
12311 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12313 * tree-vect-patterns.c (vect_pattern_detected): New function.
12314 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
12315 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
12316 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
12317 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
12318 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
12319 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
12320 (vect_recog_mask_conversion_pattern)
12321 (vect_try_gather_scatter_pattern): Likewise.
12323 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12325 * tree-vect-patterns.c (vect_get_internal_def): New function.
12326 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
12327 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
12328 (search_type_for_mask_1): Use it.
12330 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12332 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
12333 redundant WIDEN_SUM_EXPR handling.
12334 (vect_recog_sad_pattern): Likewise.
12336 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12338 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
12339 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
12340 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
12341 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
12342 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
12344 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12346 * tree-vect-stmts.c (vectorizable_call): Make sure that we
12347 use the stmt_vec_info of the original bb statement for the
12348 new zero assignment, even if the call is part of a pattern.
12350 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12352 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
12353 that the sequence is attached to the original statement rather
12354 than the pattern statement.
12355 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
12356 PATTERN_DEF_SEQ from the original statement rather than
12357 the main pattern statement.
12358 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
12359 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
12360 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
12362 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12364 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
12365 definition statements before the early exit for statements that aren't
12367 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
12369 (vect_transform_loop): ...here. Process pattern definition
12370 statements without first checking whether the main pattern
12371 statement is live or relevant.
12373 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
12375 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
12376 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
12378 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
12380 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
12381 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
12382 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
12383 (expand_block_compare): Change select_block_compare_mode call.
12384 (expand_strncmp_align_check): Use new functions, fix comment.
12385 (emit_final_str_compare_gpr): New function.
12386 (expand_strn_compare): Refactor and clean up code.
12387 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
12389 2018-06-19 Tony Reix <tony.reix@atos.com>
12390 Damien Bergamini <damien.bergamini@atos.com>
12391 David Edelsohn <dje.gcc@gmail.com>
12393 * collect2.c (static_obj): New variable.
12394 (static_libs): New variable.
12395 (is_in_list): Uncomment declaration.
12396 (main): Track AIX libraries linked statically.
12397 (is_in_list): Uncomment definition.
12398 (scan_prog_file): Don't add AIX shared libraries initializer
12399 to constructor list if linking statically.
12401 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
12403 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
12405 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
12407 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
12409 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
12412 2018-06-19 Martin Liska <mliska@suse.cz>
12414 * config/i386/i386.c (ix86_can_inline_p): Do not use
12415 ipa_fn_summaries::get_create.
12416 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
12418 (devirtualization_time_bonus): Likewise.
12419 (ipcp_propagate_stage): Likewise.
12420 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
12421 (edge_set_predicate): Likewise.
12422 (evaluate_conditions_for_known_args): Likewise.
12423 (evaluate_properties_for_edge): Likewise.
12424 (ipa_call_summary::reset): Tranform to ...
12425 (ipa_call_summary::~ipa_call_summary): ... this.
12426 (ipa_fn_summary::reset): Transform to ...
12427 (ipa_fn_summary::~ipa_fn_summary): ... this.
12428 (ipa_fn_summary_t::remove): Rename to ...
12429 (ipa_fn_summary_t::remove_callees): ... this.
12430 (ipa_fn_summary_t::duplicate): Use placement new
12431 instead of memory copy.
12432 (ipa_call_summary_t::duplicate): Likewise.
12433 (ipa_call_summary_t::remove): Remove.
12434 (dump_ipa_call_summary): Change get_create to get.
12435 (ipa_dump_fn_summary): Dump only when summary exists.
12436 (analyze_function_body): Use symbol_summary::get instead
12438 (compute_fn_summary): Likewise.
12439 (estimate_edge_devirt_benefit): Likewise.
12440 (estimate_edge_size_and_time): Likewise.
12441 (inline_update_callee_summaries): Likewise.
12442 (remap_edge_change_prob): Likewise.
12443 (remap_edge_summaries): Likewise.
12444 (ipa_merge_fn_summary_after_inlining): Likewise.
12445 (write_ipa_call_summary): Likewise.
12446 (ipa_fn_summary_write): Likewise.
12447 (ipa_free_fn_summary): Likewise.
12448 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
12449 (struct ipa_call_summary): Likewise.
12450 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
12452 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
12453 (estimate_size_after_inlining): Likewise.
12454 (estimate_growth): Likewise.
12455 (growth_likely_positive): Likewise.
12456 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
12457 (inline_call): Likewise.
12458 * ipa-inline.c (caller_growth_limits): Likewise.
12459 (can_inline_edge_p): Likewise.
12460 (can_inline_edge_by_limits_p): Likewise.
12461 (compute_uninlined_call_time): Likewise.
12462 (compute_inlined_call_time): Likewise.
12463 (want_inline_small_function_p): Likewise.
12464 (edge_badness): Likewise.
12465 (update_caller_keys): Likewise.
12466 (update_callee_keys): Likewise.
12467 (inline_small_functions): Likewise.
12468 (inline_to_all_callers_1): Likewise.
12469 (dump_overall_stats): Likewise.
12470 (early_inline_small_functions): Likewise.
12471 (early_inliner): Likewise.
12472 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12473 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
12474 * ipa-pure-const.c (malloc_candidate_p): Likewise.
12475 * ipa-split.c (execute_split_functions): Likewise.
12476 * symbol-summary.h: Likewise.
12477 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
12479 2018-06-19 Richard Biener <rguenther@suse.de>
12481 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
12482 (vectorize_loops): ... here. Fix dbgcnt handling.
12483 (try_vectorize_loop): Wrap try_vectorize_loop_1.
12485 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
12488 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
12489 ieee128 argument takes up only one (vector) register, not two (floating
12492 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
12494 * gimplify.c (gimplify_init_constructor): Really never clear for an
12495 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
12497 2018-06-19 Richard Biener <rguenther@suse.de>
12499 PR tree-optimization/86179
12500 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
12501 after failed recognition.
12503 2018-06-18 Martin Sebor <msebor@redhat.com>
12505 PR middle-end/85602
12506 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
12507 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
12508 Handle integer subtraction.
12509 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
12510 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
12512 2018-06-18 David Malcolm <dmalcolm@redhat.com>
12514 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
12515 param from rtx to rtx_insn *.
12516 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
12518 (frv_ifcvt_modify_insn): Likwise.
12519 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
12520 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
12521 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
12522 as_a <rtx_insn *> cast to local "unprotected_region" once
12523 it's been established that it's not NULL or pc_rtx.
12524 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
12525 param "sethi" from rtx to rtx_insn *.
12526 (nds32_group_float_insns): Likewise for param "insn".
12527 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
12529 (vax_output_int_subtract): Likewise.
12530 * config/vax/vax.c (vax_output_int_add): Likewise for param
12532 (vax_output_int_subtract): Likewise.
12533 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
12534 (emit_pattern_after): Likewise for param "after".
12535 (emit_insn_after): Likewise.
12536 (emit_jump_insn_after): Likewise.
12537 (emit_call_insn_after): Likewise.
12538 (emit_debug_insn_after): Likewise.
12539 (emit_pattern_before): Likewise for param "before".
12540 (emit_insn_before): Likewise.
12541 (emit_jump_insn_before): Likewise.
12542 * final.c (get_insn_template): Likewise for param "insn", removing
12544 * output.h (get_insn_template): Likewise for 2nd param.
12545 * rtl.h (emit_insn_before): Likewise.
12546 (emit_jump_insn_before): Likewise.
12547 (emit_debug_insn_before_noloc): Likewise.
12548 (emit_insn_after): Likewise.
12549 (emit_jump_insn_after): Likewise.
12550 (emit_call_insn_after): Likewise.
12551 (emit_debug_insn_after): Likewise.
12552 (set_insn_deleted): Likewise for param.
12554 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
12557 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
12558 floating point modes, so that IFmode is numerically greater than
12559 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
12560 to declare the ordering. This prevents IFmode from being
12561 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
12562 machine. Include rs6000-modes.h to share the fractional values
12563 between genmodes* and the rest of the compiler.
12564 (IFmode): Likewise.
12565 (KFmode): Likewise.
12566 (TFmode): Likewise.
12567 * config/rs6000/rs6000-modes.h: New file.
12568 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
12569 meaning of rs6000_long_double_size so that 126..128 selects an
12570 appropriate 128-bit floating point type.
12571 (rs6000_option_override_internal): Likewise.
12572 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
12573 (TARGET_LONG_DOUBLE_128): Change the meaning of
12574 rs6000_long_double_size so that 126..128 selects an appropriate
12575 128-bit floating point type.
12576 (LONG_DOUBLE_TYPE_SIZE): Update comment.
12577 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
12578 source and destination to match the standard usage.
12579 (truncifkf2): Likewise.
12580 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
12581 ISA 2.07 to use an explicit clobber, instead of passing in a
12583 (copysign<mode>3_soft): Likewise.
12585 2018-06-18 David Malcolm <dmalcolm@redhat.com>
12587 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
12588 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
12589 (vect_slp_analyze_instance_dependence): Likewise.
12590 (vect_enhance_data_refs_alignment): Likewise.
12591 (vect_analyze_data_refs_alignment): Likewise.
12592 (vect_slp_analyze_and_verify_instance_alignment
12593 (vect_analyze_data_ref_accesses): Likewise.
12594 (vect_prune_runtime_alias_test_list): Likewise.
12595 (vect_analyze_data_refs): Likewise.
12596 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
12597 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
12598 (vect_analyze_scalar_cycles_1): Likewise.
12599 (vect_get_loop_niters): Likewise.
12600 (vect_analyze_loop_form_1): Likewise.
12601 (vect_update_vf_for_slp): Likewise.
12602 (vect_analyze_loop_operations): Likewise.
12603 (vect_analyze_loop): Likewise.
12604 (vectorizable_induction): Likewise.
12605 (vect_transform_loop): Likewise.
12606 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
12607 * tree-vect-slp.c (vect_analyze_slp): Likewise.
12608 (vect_make_slp_decision): Likewise.
12609 (vect_detect_hybrid_slp): Likewise.
12610 (vect_slp_analyze_operations): Likewise.
12611 (vect_slp_bb): Likewise.
12612 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
12613 (vectorizable_bswap): Likewise.
12614 (vectorizable_call): Likewise.
12615 (vectorizable_simd_clone_call): Likewise.
12616 (vectorizable_conversion): Likewise.
12617 (vectorizable_assignment): Likewise.
12618 (vectorizable_shift): Likewise.
12619 (vectorizable_operation): Likewise.
12620 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
12622 2018-06-18 Martin Sebor <msebor@redhat.com>
12624 PR tree-optimization/81384
12625 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
12626 * builtins.c (expand_builtin_strnlen): New function.
12627 (expand_builtin): Call it.
12628 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
12629 * builtins.def (BUILT_IN_STRNLEN): New.
12630 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
12631 Warn for bounds in excess of maximum object size.
12632 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
12633 single-value ranges. Handle strnlen.
12634 (handle_builtin_strlen): Handle strnlen.
12635 (strlen_check_and_optimize_stmt): Same.
12636 * doc/extend.texi (Other Builtins): Document strnlen.
12638 2018-06-18 Maya Rashish <coypu@sdf.org>
12640 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
12641 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
12642 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
12644 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
12646 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
12648 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12650 * tree.c (escaped_string::escape): Replace cast to char * by
12651 const_cast<char *> (unescaped).
12653 2018-06-18 Nick Clifton <nickc@redhat.com>
12656 * tree.c (escaped_string): New class. Converts an unescaped
12657 string into its escaped equivalent.
12658 (warn_deprecated_use): Use the new class to convert the
12659 deprecation message, if present.
12660 (test_escaped_strings): New self test.
12661 (test_c_tests): Add test_escaped_strings.
12662 * doc/extend.texi (deprecated): Add a note that the
12663 deprecation message is affected by the -fmessage-length
12664 option, and that control characters will be escaped.
12665 (#pragma GCC error): Document this pragma.
12666 (#pragma GCC warning): Likewise.
12667 * doc/invoke.texi (-fmessage-length): Document this option's
12668 effect on the #warning and #error preprocessor directives and
12669 the deprecated attribute.
12671 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
12673 * tree.c (decl_value_expr_lookup): Revert latest change.
12674 (decl_value_expr_insert): Likewise.
12676 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
12678 * gimplify.c (nonlocal_vlas): Delete.
12679 (nonlocal_vla_vars): Likewise.
12680 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
12682 (gimplify_body): Do not create and destroy nonlocal_vlas.
12683 * tree-nested.c: Include diagnostic.h.
12684 (use_pointer_in_frame): Tweak.
12685 (lookup_field_for_decl): Add assertion and declare the transformation.
12686 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
12687 internal error when the reference is in a wrong context. Do not
12688 create a debug decl by default.
12689 (note_nonlocal_block_vlas): Delete.
12690 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
12691 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
12692 create a debug decl by default.
12693 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
12694 call is in a wrong context.
12695 (fixup_vla_decls): New function.
12696 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
12697 debug variables were created.
12698 * tree.c (decl_value_expr_lookup): Add checking assertion.
12699 (decl_value_expr_insert): Likewise.
12701 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
12703 PR middle-end/82479
12704 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
12705 * tree-scalar-evolution.c (interpret_expr): Likewise.
12706 (expression_expensive_p): Likewise.
12707 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
12708 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
12709 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
12710 (ssa_defined_by_minus_one_stmt_p): New.
12712 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
12714 PR middle-end/64946
12715 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
12716 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
12717 * dojump.c (do_jump): Likewise.
12718 * expr.c (expand_expr_real_2): Check operand type's sign.
12719 * fold-const.c (const_unop): Handle ABSU_EXPR.
12720 (fold_abs_const): Likewise.
12721 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
12722 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
12723 (strip_sign_op_1): Likesise.
12724 * match.pd: Add new pattern to generate ABSU_EXPR.
12725 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
12726 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
12727 * tree-eh.c (operation_could_trap_helper_p): Likewise.
12728 * tree-inline.c (estimate_operator_cost): Likewise.
12729 * tree-pretty-print.c (dump_generic_node): Likewise.
12730 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
12731 * tree.def (ABSU_EXPR): New.
12733 2018-06-16 Jakub Jelinek <jakub@redhat.com>
12735 PR middle-end/86095
12736 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
12737 documented as preserved for backward compatibility only.
12738 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
12740 PR rtl-optimization/86108
12741 * bb-reorder.c (create_forwarder_block): Renamed to ...
12742 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
12743 jump from new landing pad to the second part.
12744 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
12747 2018-06-15 Jakub Jelinek <jakub@redhat.com>
12749 PR middle-end/85878
12750 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
12751 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
12752 Only call store_expr for halves if the mode is the same.
12754 PR middle-end/86123
12755 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
12756 Fix up comment formatting.
12758 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
12760 * typed-splay-tree.h (typed_splay_tree::remove): New function.
12761 (typed_splay_tree::closure,
12762 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
12763 (typed_splay_tree::typed_splay_tree,
12764 typed_splay_tree::operator =): Declared private.
12765 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
12766 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
12767 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
12768 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
12769 typed_splay_tree::splay_tree_splay,
12770 typed_splay_tree::splay_tree_foreach_helper,
12771 typed_splay_tree::splay_tree_insert,
12772 typed_splay_tree::splay_tree_remove,
12773 typed_splay_tree::splay_tree_lookup,
12774 typed_splay_tree::splay_tree_predecessor,
12775 typed_splay_tree::splay_tree_successor,
12776 typed_splay_tree::splay_tree_min,
12777 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
12778 (typed_splay_tree::root, typed_splay_tree::comp,
12779 typed_splay_tree::delete_key,
12780 typed_splay_tree::delete_value): New data members.
12781 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
12782 typed_splay_tree::remove.
12784 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
12786 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
12787 -mginv and -mno-ginv to the assembler.
12788 * config/mips/mips.opt (-mcrc): New option.
12789 (-mginv): Likewise.
12790 * doc/invoke.text (-mcrc): Document.
12791 (-mginv): Likewise.
12793 2018-06-15 Nick Clifton <nickc@redhat.com>
12796 * tree.c (escaped_string): New class. Converts an unescaped
12797 string into its escaped equivalent.
12798 (warn_deprecated_use): Use the new class to convert the
12799 deprecation message, if present.
12800 (test_escaped_strings): New self test.
12801 (test_c_tests): Add test_escaped_strings.
12802 * doc/extend.texi (deprecated): Add a note that the
12803 deprecation message is affected by the -fmessage-length
12804 option, and that control characters will be escaped.
12805 (#pragma GCC error): Document this pragma.
12806 (#pragma GCC warning): Likewise.
12807 * doc/invoke.texi (-fmessage-length): Document this option's
12808 effect on the #warning and #error preprocessor directives and
12809 the deprecated attribute.
12811 2018-06-15 Richard Biener <rguenther@suse.de>
12813 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
12814 here, also noting vector size used.
12815 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
12816 size used in MSG_OPTIMIZED_LOCATIONS dump.
12817 (pass_slp_vectorize::execute): Adjust.
12819 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
12822 * config/arc/arc.c (arc_return_address_register): Fix
12825 2018-06-15 Richard Biener <rguenther@suse.de>
12827 PR middle-end/86159
12828 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
12829 leave useless conversion stripping to force_gimple_operand_gsi.
12830 (gimplify_build2): Likewise.
12831 (gimplify_build1): Likewise.
12833 2018-06-15 Richard Biener <rguenther@suse.de>
12835 PR middle-end/86076
12836 * tree-cfg.c (move_stmt_op): unshare invariant addresses
12837 before adjusting their block.
12839 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
12841 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
12842 multilibs for *-*-rtems*.
12843 * config/riscv/t-rtems: New file.
12845 2018-06-14 Jakub Jelinek <jakub@redhat.com>
12847 PR middle-end/86122
12848 * match.pd ((A +- CST1) +- CST2): Punt if last resort
12849 unsigned_type_for returns NULL.
12852 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
12853 subregs of multi-word pseudos unless the float mode has word size.
12855 2018-06-14 Richard Biener <rguenther@suse.de>
12857 PR middle-end/86139
12858 * tree-vect-generic.c (build_word_mode_vector_type): Remove
12859 duplicate and harmful type_hash_canon.
12860 * tree.c (type_hash_canon): Assert we didn't find ourselves.
12862 2018-06-14 Richard Biener <rguenther@suse.de>
12865 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
12868 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
12870 * config/rtems.h (STDINT_LONG32): Define.
12872 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
12873 Prachi Godbole <prachi.godbole@imgtec.com>
12875 * config/mips/mips-cpus.def: Define P6600.
12876 * config/mips/mips-tables.opt: Regenerate.
12877 * config/mips/mips.c (mips_ucbranch_type): New enum.
12878 (mips_rtx_cost_data): Add support for P6600.
12879 (mips_issue_rate): Likewise.
12880 (mips_multipass_dfa_lookahead): Likewise.
12881 (mips_avoid_hazard): Likewise.
12882 (mips_reorg_process_insns): Likewise.
12883 (mips_classify_branch_p6600): New function.
12884 * config/mips/mips.h (TUNE_P6600): New define.
12885 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
12886 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
12887 * config/mips/mips.md: Include p6600.md.
12888 (processor): Add p6600.
12889 * config/mips/p6600.md: New file.
12890 * doc/invoke.texi: Add p6600 to supported architectures.
12892 2018-06-13 Martin Sebor <msebor@redhat.com>
12894 PR tree-optimization/86114
12895 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
12897 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
12899 2018-06-13 Richard Biener <rguenther@suse.de>
12901 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
12902 Properly set vector type of the intermediate stmt.
12903 * tree-vect-stmts.c (vectorizable_operation): The destination
12904 var always has vectype_out type.
12906 2018-06-13 Jeff Law <law@redhat.com>
12908 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
12909 integer 0 for argument to print_rtl_with_bb.
12910 (rl78_reorg): Likewise.
12912 2018-06-13 David Malcolm <dmalcolm@redhat.com>
12914 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
12915 from rtx to rtx_insn *.
12916 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
12918 (add_sched_insns_for_speculation): Likewise for local "target",
12919 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
12920 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
12921 from rtx_insn ** to rtx_code_label **.
12922 (reorg_emit_nops): Likewise.
12923 (c6x_reorg): Likewise for local "call_labels".
12924 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
12926 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
12927 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
12928 the loops over LABEL_REFS.
12929 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
12931 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
12932 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
12933 (split_branches): Strengthen local "olabel" from rtx to
12934 rtx_insn *, adding a safe_as_a cast.
12935 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
12937 (add_insn_after): Likewise for first two params.
12938 (add_insn_before): Likewise.
12939 (remove_insn): Likewise for param.
12940 (emit_pattern_before_noloc): Likewise for second and third params.
12941 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
12942 (emit_call_insn_before_noloc): Likewise.
12943 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
12945 (emit_barrier_before): Likewise.
12946 (emit_label_before): Strengthen "label" param from "rtx" to
12947 "rtx_code_label *". Strengthen "before" param from "rtx" to
12949 (emit_insn_after_1): Strengthen "after" param from "rtx" to
12951 (emit_pattern_after_noloc): Likewise.
12952 (emit_insn_after_noloc): Likewise.
12953 (emit_jump_insn_after_noloc): Likewise.
12954 (emit_call_insn_after_noloc): Likewise.
12955 (emit_debug_insn_after_noloc): Likewise.
12956 (emit_barrier_after): Likewise.
12957 (emit_label_after): Likewise for both params.
12958 (emit_pattern_after_setloc): Likewise for "after" param. Convert
12959 "loc" param from "int" to "location_t".
12960 (emit_insn_after_setloc): Likewise.
12961 (emit_jump_insn_after_setloc): Likewise.
12962 (emit_call_insn_after_setloc): Likewise.
12963 (emit_debug_insn_after_setloc): Likewise.
12964 (emit_pattern_before_setloc): Likewise for "before" param. Convert
12965 "loc" param from "int" to "location_t".
12966 (emit_pattern_before): Convert NULL_RTX to NULL.
12967 (emit_insn_before_setloc): Convert "loc" param from "int" to
12969 (emit_jump_insn_before_setloc): Likewise.
12970 (emit_call_insn_before_setloc): Likewise.
12971 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
12972 rtx_insn *. Convert "loc" param from "int" to "location_t".
12973 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
12974 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
12975 Convert 3rd param from "int" to "location_t".
12976 (emit_barrier_before, emit_barrier_after, next_real_insn):
12977 Strengthen param from rtx to rtx_insn *.
12978 (emit_label_before): Strengthen 1st param from "rtx" to
12979 "rtx_code_label *". Strengthen 2nd param from "rtx" to
12981 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
12982 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
12983 Strengthen 2nd param from "rtx" to "rtx_insn *".
12984 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
12985 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
12986 Likewise. Convert 3rd param from "int" to "location_t".
12987 (emit_label_after): Strengthen 1st param from "rtx" to
12988 "rtx_code_label *".
12989 (next_real_insn, remove_insn): Strengthen param from "rtx" to
12991 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
12992 from "rtx" to "rtx_insn *".
12994 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
12996 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
12997 bodies streamed in with -Q.
12998 * dumpfile.c (dump_files): Add lto-stream-out dump file.
12999 * dumpfile.h (tree_dump_index): Add lto_stream_out.
13000 * gimple-streamer-out.c: Include gimple-pretty-print.h
13001 (output_bb): Dump stmts streamed.
13002 * lto-section-out.c: Include print-tree.h
13003 (lto_begin_section): Dump sections created.
13004 (lto_output_decl_index): Dump decl encoded.
13005 * lto-streamer-out.c: Include print-tree.h
13006 (create_output_block): Dump output block created.
13007 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
13008 (output_function): Dump function output.
13009 (output_constructor): Dump constructor streamed.
13010 (write_global_stream): Output indexes encoded.
13011 (produce_asm_for_decls): Dump streams encoded.
13012 * lto-streamer.c (streamer_dump_file): New global var.
13013 * lto-streamer.h (streamer_dump_file): Declare.
13014 * passes.c (ipa_write_summaries): Initialize streamer dump.
13015 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
13018 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
13021 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
13022 offsets for register save directives. Emit a second batch of save
13023 directives, if need be, when the function accesses prior frames.
13025 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13027 * config/arc/fpu.md (fmasf4): Force operand to register.
13028 (fnmasf4): Likewise.
13030 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13032 * config/arc/arc-protos.h (arc_pad_return): Remove.
13033 * config/arc/arc.c (machine_function): Remove force_short_suffix
13035 (arc_print_operand): Adjust printing of '&'.
13036 (arc_verify_short): Remove conditional printing of short suffix.
13037 (arc_final_prescan_insn): Remove reference to size_reason.
13038 (pad_return): New function.
13039 (arc_reorg): Call pad_return.
13040 (arc_pad_return): Remove.
13041 (arc_init_machine_status): Remove reference to force_short_suffix.
13042 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
13043 (attr length): When attribute iscompact is true force to 2
13044 regardless; in the case of maybe check if we want to force the
13045 instruction to have 4 bytes length.
13046 (nopv): Change it to generate 4 byte long nop as well.
13047 (blockage): New pattern.
13048 (simple_return): Remove call to arc_pad_return.
13049 (p_return_i): Likewise.
13051 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13053 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
13055 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13057 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
13060 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13062 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
13063 for ARC700 and ARCv2.
13065 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
13068 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
13069 operands[2] instead of operands[1].
13072 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13074 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
13075 case, check whether the outer register overlaps an unallocatable
13076 register, not just whether it fits the required class.
13078 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13080 * poly-int.h (can_div_trunc_p): Add new overload in which all values
13082 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
13083 (memrefs_conflict_p): Likewise.
13084 (init_alias_analysis): Likewise.
13085 * cfgexpand.c (expand_debug_expr): Likewise.
13086 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
13087 * cse.c (fold_rtx): Likewise.
13088 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
13089 * expr.c (emit_block_move_hints): Likewise.
13090 (clear_storage_hints, push_block, emit_push_insn): Likewise.
13091 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
13092 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
13093 (emit_group_store): Likewise.
13094 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
13095 to read the PRE/POST_MODIFY increment.
13096 * calls.c (store_one_arg): Use strip_offset.
13097 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
13099 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
13101 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
13102 (simplify_binary_operation_1): Extend CONST_INT handling to
13104 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
13105 than a HOST_WIDE_INT.
13106 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
13108 (adjust_mems, add_stores): Update accodingly.
13109 (vt_canonicalize_addr): Track polynomial offsets.
13110 (emit_note_insn_var_location): Likewise.
13111 (vt_add_function_parameter): Likewise.
13112 (vt_initialize): Likewise.
13114 2018-06-12 Jeff Law <law@redhat.com>
13116 * config.gcc (alpha*-*-freebsd*): Remove.
13117 * config/alpha/freebsd.h: Remove.
13119 2018-06-12 David Malcolm <dmalcolm@redhat.com>
13122 * spellcheck-tree.c (levenshtein_distance): Rename to...
13123 (get_edit_distance): ...this, and update for underlying renaming.
13124 * spellcheck-tree.h (levenshtein_distance): Rename to...
13125 (get_edit_distance): ...this.
13126 * spellcheck.c (levenshtein_distance): Rename to...
13127 (get_edit_distance): ...this. Convert from Levenshtein distance
13128 to Damerau-Levenshtein distance by supporting transpositions of
13129 adjacent characters. Rename "v1" to "v_next" and "v0" to
13131 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
13132 (selftest::test_edit_distance_unit_test_oneway): ...this, and
13133 update for underlying renaming.
13134 (selftest::levenshtein_distance_unit_test): Rename to...
13135 (selftest::test_get_edit_distance_unit): ...this, and update for
13136 underlying renaming.
13137 (selftest::test_find_closest_string): Add example from PR 69968
13138 where transposition helps
13139 (selftest::test_metric_conditions): Update for renaming.
13140 (selftest::test_metric_conditions): Likewise.
13141 (selftest::spellcheck_c_tests): Likewise.
13142 * spellcheck.h (levenshtein_distance): Rename both overloads to...
13143 (get_edit_distance): ...this.
13144 (best_match::consider): Update for renaming.
13146 2018-06-12 Martin Sebor <msebor@redhat.com>
13148 PR tree-optimization/85259
13149 * builtins.c (compute_objsize): Handle constant offsets.
13150 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
13151 true iff a warning has been issued.
13152 * gimple.h (gimple_nonartificial_location): New function.
13153 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
13154 gimple_nonartificial_location and handle -Wno-system-headers.
13155 (handle_builtin_stxncpy): Same.
13157 2018-06-12 Martin Sebor <msebor@redhat.com>
13160 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
13162 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13164 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13165 BUILTIN_VEC_XST entries for pointer to double and long long.
13167 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
13170 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
13172 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
13175 2018-06-12 Martin Liska <mliska@suse.cz>
13177 * doc/options.texi: Document IntegerRange.
13179 2018-06-12 Martin Liska <mliska@suse.cz>
13181 * config/i386/i386.opt: Make MPX-related options as Deprecated.
13182 * opt-functions.awk: Handle Deprecated flag.
13183 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
13185 (read_cmdline_option): Report warning for a deprecated option.
13186 * opts.h (struct cl_option): Add new field cl_deprecated.
13187 (CL_ERR_DEPRECATED): New.
13189 2018-06-12 Martin Liska <mliska@suse.cz>
13191 * doc/options.texi: Document Deprecated option flag.
13193 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13195 * config/arc/arc-arch.h (arc_extras): New enum.
13196 (arc_cpu_t):Add field extra.
13197 (arc_cpu_types): Consider the extras.
13198 * config/arc/arc-cpus.def: Add extras info.
13199 * config/arc/arc-opts.h (processor_type): Consider extra field.
13200 * config/arc/arc.c (arc_override_options): Handle extra field.
13202 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13204 * config/arc/arc-arch.h: Update ARC_OPTX macro.
13205 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
13207 * config/arc/arc.c (arc_init): Update pic warning.
13208 (irq_range): Update irq range parsing warnings.
13209 (arc_override_options): Update various warning messages.
13210 (arc_handle_aux_attribute): Likewise.
13212 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
13214 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
13216 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
13218 * doc/sourcebuild.texi: Document usage of line number 0 in verify
13219 compiler messages directives.
13221 2018-06-12 Matthew Fortune <mfortune@gmail.com>
13223 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
13224 * config/mips/mips-tables.opt: Regenerate.
13225 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
13227 * doc/invoke.texi: Document -march=i6500.
13229 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
13231 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
13232 (i6400_gpmul): Add cpu_unit.
13233 (i6400_gpdiv): Likewise.
13234 (i6400_msa_add_d): Update reservations.
13235 (i6400_msa_int_add) Likewise.
13236 (i6400_msa_short_logic3) Likewise.
13237 (i6400_msa_short_logic2) Likewise.
13238 (i6400_msa_short_logic) Likewise.
13239 (i6400_msa_move) Likewise.
13240 (i6400_msa_cmp) Likewise.
13241 (i6400_msa_short_float2) Likewise.
13242 (i6400_msa_div_d) Likewise.
13243 (i6400_msa_long_logic1) Likewise.
13244 (i6400_msa_long_logic2) Likewise.
13245 (i6400_msa_mult) Likewise.
13246 (i6400_msa_long_float2) Likewise.
13247 (i6400_msa_long_float4) Likewise.
13248 (i6400_msa_long_float5) Likewise.
13249 (i6400_msa_long_float8) Likewise.
13250 (i6400_fpu_fadd): Include frint type.
13251 (i6400_fpu_store): New define_insn_reservation.
13252 (i6400_fpu_load): Likewise.
13253 (i6400_fpu_move): Likewise.
13254 (i6400_fpu_fcmp): Likewise.
13255 (i6400_fpu_fmadd): Likewise.
13256 (i6400_int_mult): Include imul3nc type and update reservation.
13257 (i6400_int_div): Include idiv3 type and update reservation.
13258 (i6400_int_load): Update to check type not move_type.
13259 (i6400_int_store): Likewise.
13260 (i6400_int_prefetch): Set zero latency.
13262 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
13264 * gcc.c: Document new %@{...} sequence.
13265 (LINK_COMMAND_SPEC): Use it for the -L switches.
13266 (cpp_unique_options): Use it for the -I switches.
13267 (at_file_argbuf): New global variable.
13268 (in_at_file): Likewise.
13269 (alloc_args): Create at_file_argbuf.
13270 (clear_args): Truncate at_file_argbuf.
13271 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
13272 (open_at_file): New function.
13273 (close_at_file): Likewise.
13274 (create_at_file): Delete.
13275 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
13278 (validate_switches_from_spec): Deal with %@{...} sequence.
13279 (validate_switches): Likewise.
13280 (driver::finalize): Call clear_args.
13282 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
13284 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
13286 2018-06-11 Martin Sebor <msebor@redhat.com>
13288 * doc/invoke.texi (-Wall): List -Wc++17-compat.
13289 (Wno-class-memaccess): Add @opindex.
13290 (Wno-templates, Wno-multiple-inheritance): Same.
13291 (Wno-virtual-inheritance, Wno-namespaces): Same.
13292 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
13293 (Wno-format-overflow, Wno-format-truncation): Same.
13294 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
13295 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
13296 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
13297 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
13298 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
13299 (Wno-misspelled-isr): Same.
13301 2018-06-11 Martin Sebor <msebor@redhat.com>
13303 * PR tree-optimization/86083
13304 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
13306 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
13308 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
13310 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
13313 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
13314 on the correct operand.
13315 (*movdi_internal64): Ditto.
13317 2018-06-11 Martin Liska <mliska@suse.cz>
13319 PR tree-optimization/86089
13320 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
13322 2018-06-11 Julia Koval <julia.koval@intel.com>
13324 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
13325 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
13326 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
13328 2018-06-11 Olivier Hainque <hainque@adacore.com>
13330 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
13331 for Ada with strict dwarf2.
13333 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
13336 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
13339 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
13341 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
13343 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
13345 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
13348 2018-06-08 Martin Liska <mliska@suse.cz>
13350 * tree-cfg.h (debug_function): Fix argument type to match
13353 2018-06-08 Martin Liska <mliska@suse.cz>
13355 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
13356 Remove usage of MPX-related (and removed) fields.
13357 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
13359 2018-06-08 David Malcolm <dmalcolm@redhat.com>
13361 * cfg.c (debug): Use TDF_NONE rather than 0.
13362 * cfghooks.c (debug): Likewise.
13363 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
13364 (struct dump_option_value_info): Convert to...
13365 (struct kv_pair): ...this template type.
13366 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
13368 (optinfo_verbosity_options): Likewise.
13369 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
13371 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
13372 than int for "optgroup_flags" param.
13373 (dump_generic_expr_loc): Use dump_flags_t rather than int for
13375 (dump_dec): Likewise.
13376 (dump_finish): Use TDF_NONE rather than 0.
13377 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
13378 rather than int for "optgroup_flags" param. Use TDF_NONE rather
13379 than 0. Update for change to option_ptr.
13380 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
13381 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
13382 0. Update for changes to optinfo_verbosity_options and
13384 (opt_info_switch_p): Convert optgroup_flags from int to
13386 (dump_basic_block): Use dump_flags_t rather than int
13387 for "dump_kind" param.
13388 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
13389 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
13390 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
13391 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
13392 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
13393 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
13394 TDF_NONE): Convert from macros to...
13395 (enum dump_flag): ...this new enum.
13396 (dump_flags_t): Update to use enum.
13397 (operator|, operator&, operator~, operator|=, operator&=):
13398 Implement for dump_flags_t.
13399 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
13400 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
13401 Convert from macros to...
13402 (enum optgroup_flag): ...this new enum.
13403 (optgroup_flags_t): New typedef.
13404 (operator|, operator|=): Implement for optgroup_flags_t.
13405 (struct dump_file_info): Convert field "alt_flags" to
13406 dump_flags_t. Convert field "optgroup_flags" to
13408 (dump_basic_block): Use dump_flags_t rather than int for param.
13409 (dump_generic_expr_loc): Likewise.
13410 (dump_dec): Likewise.
13411 (dump_register): Convert param "optgroup_flags" to
13413 (opt_info_enable_passes): Likewise.
13414 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
13416 * gimple-pretty-print.c (debug): Likewise.
13417 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
13418 (merged_store_group::apply_stores): Likewise.
13419 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
13420 * gimple.c (verify_gimple_pp): Likewise.
13421 * graphite-poly.c (print_pbb_body): Likewise.
13422 * passes.c (pass_manager::register_one_dump_file): Convert
13423 local "optgroup_flags" to optgroup_flags_t.
13424 * print-tree.c (print_node): Use TDF_NONE rather than 0.
13426 (debug_body): Likewise.
13427 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
13428 to optgroup_flags_t.
13429 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
13431 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
13432 (convert_mult_to_fma): Likewise.
13433 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
13434 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
13435 * tree-vect-data-refs.c (dump_lower_bound): Convert param
13436 "dump_kind" to dump_flags_t.
13438 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
13440 * config/rs6000/rs6000.c (min, max): Delete.
13442 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
13444 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
13445 -mabi=spe and -mabi=no-spe.
13447 2018-06-08 Martin Liska <mliska@suse.cz>
13449 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
13450 where we expect an existing summary.
13452 2018-06-08 Martin Liska <mliska@suse.cz>
13454 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
13455 * ipa-inline.h (estimate_edge_growth): Likewise.
13457 2018-06-08 Martin Liska <mliska@suse.cz>
13459 * cgraph.c (function_version_hasher::hash): Use
13460 cgraph_node::get_uid ().
13461 (function_version_hasher::equal):
13462 * cgraph.h (cgraph_node::get_uid): New method.
13463 * ipa-inline.c (update_caller_keys): Use
13464 cgraph_node::get_uid ().
13465 (update_callee_keys): Likewise.
13466 * ipa-utils.c (searchc): Likewise.
13467 (ipa_reduced_postorder): Likewise.
13468 * lto-cgraph.c (input_node): Likewise.
13469 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
13470 * symbol-summary.h (symtab_insertion): Likewise.
13471 (symtab_removal): Likewise.
13472 (symtab_duplication): Likewise.
13473 * tree-pretty-print.c (dump_function_header): Likewise.
13474 * tree-sra.c (convert_callers_for_node): Likewise.
13476 2018-06-08 Martin Liska <mliska@suse.cz>
13478 * cgraph.c (symbol_table::create_edge): Always assign a new
13480 (symbol_table::free_edge): Do not recycle numbers.
13481 * cgraph.h (cgraph_edge::get): New method.
13482 * symbol-summary.h (symtab_removal): Use it.
13483 (symtab_duplication): Likewise.
13484 (call_summary::hashable_uid): Remove.
13486 2018-06-08 Martin Liska <mliska@suse.cz>
13488 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
13489 (initialize_growth_caches): Remove.
13490 (free_growth_caches): Likewise.
13491 (do_estimate_edge_time): Use edge_growth_cache.
13492 (do_estimate_edge_size): Likewise.
13493 (do_estimate_edge_hints): Likewise.
13494 * ipa-inline.c (reset_edge_caches): Likewise.
13495 (recursive_inlining): Likewise.
13496 (inline_small_functions): Likewise.
13497 * ipa-inline.h (initialize_growth_caches): Remove.
13498 (estimate_edge_size): Likewise.
13499 (estimate_edge_time): Likewise.
13500 (estimate_edge_hints): Likewise.
13501 (reset_edge_growth_cache): Likewise.
13502 * symbol-summary.h (call_summary::remove): New method.
13504 2018-06-08 Martin Liska <mliska@suse.cz>
13506 * ipa-cp.c (class edge_clone_summary): New summary.
13507 (grow_edge_clone_vectors): Remove.
13508 (ipcp_edge_duplication_hook): Remove.
13509 (class edge_clone_summary_t): New call_summary class.
13510 (ipcp_edge_removal_hook): Remove.
13511 (edge_clone_summary_t::duplicate): New function.
13512 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
13513 (create_specialized_node): Likewise.
13514 (ipcp_driver): Initialize edge_clone_summaries and do not
13517 2018-06-08 Martin Liska <mliska@suse.cz>
13519 * symbol-summary.h (get): New function.
13520 (call_summary::m_initialize_when_cloning): New class member.
13522 2018-06-08 Martin Liska <mliska@suse.cz>
13524 * cgraph.c (cgraph_node::remove): Do not recycle uid.
13525 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
13526 (symbol_table::allocate_cgraph_symbol): Do not set uid.
13527 * passes.c (uid_hash_t): Record removed_nodes by their uids.
13528 (remove_cgraph_node_from_order): Use the removed_nodes set.
13529 (do_per_function_toporder): Likwise.
13530 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
13531 instead of summary_uid.
13532 (symtab_removal): Likewise.
13533 (symtab_duplication): Likewise.
13535 2018-06-08 Martin Liska <mliska@suse.cz>
13537 * ipa-cp.c (ipcp_store_bits_results): Use
13538 ipcp_transformation_sum.
13539 (ipcp_store_vr_results): Likewise.
13540 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
13542 (ipcp_transformation_initialize): ... this.
13543 (ipa_set_node_agg_value_chain):
13544 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
13545 (write_ipcp_transformation_info): Likewise.
13546 (read_ipcp_transformation_info): Likewise.
13547 (ipcp_update_bits): Likewise.
13548 (ipcp_update_vr): Likewise.
13549 (ipcp_transform_function): Likewise.
13550 * ipa-prop.h: Rename ipcp_transformation_summary to
13551 ipcp_transformation.
13552 (class ipcp_transformation_t): New function summary.
13553 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
13554 (ipa_get_agg_replacements_for_node): Likewise.
13556 2018-06-08 Martin Liska <mliska@suse.cz>
13558 * ipa-pure-const.c (struct funct_state_d): Do it class instead
13560 (class funct_state_summary_t): New function_summary class.
13561 (has_function_state): Remove.
13562 (get_function_state): Likewise.
13563 (set_function_state): Likewise.
13564 (add_new_function): Likewise.
13565 (funct_state_summary_t::insert): New function.
13566 (duplicate_node_data): Remove.
13567 (remove_node_data): Remove.
13568 (funct_state_summary_t::duplicate): New function.
13569 (register_hooks): Create new funct_state_summaries.
13570 (pure_const_generate_summary): Use it.
13571 (pure_const_write_summary): Likewise.
13572 (pure_const_read_summary): Likewise.
13573 (propagate_pure_const): Likewise.
13574 (propagate_nothrow): Likewise.
13575 (dump_malloc_lattice): Likewise.
13576 (propagate_malloc): Likewise.
13577 (execute): Do not register hooks, just remove summary
13579 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
13582 2018-06-08 Martin Liska <mliska@suse.cz>
13584 * ipa-reference.c (remove_node_data): Remove.
13585 (duplicate_node_data): Likewise.
13586 (class ipa_ref_var_info_summary_t): New class.
13587 (class ipa_ref_opt_summary_t): Likewise.
13588 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
13589 (get_reference_optimization_summary): Use
13590 ipa_ref_opt_sum_summaries.
13591 (set_reference_vars_info): Remove.
13592 (set_reference_optimization_summary): Likewise.
13593 (ipa_init): Create summaries.
13594 (init_function_info): Use function summary.
13595 (ipa_ref_opt_summary_t::duplicate): New function.
13596 (ipa_ref_opt_summary_t::remove): New function.
13597 (get_read_write_all_from_node): Fix GNU coding style.
13598 (propagate): Use function summary.
13599 (write_node_summary_p): Fix GNU coding style.
13600 (stream_out_bitmap): Likewise.
13601 (ipa_reference_read_optimization_summary): Use function summary.
13602 (ipa_reference_c_finalize): Do not release hooks.
13604 2018-06-08 Martin Liska <mliska@suse.cz>
13606 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
13607 (analyze_function_body): Extract multiple calls of get_create.
13608 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
13609 * ipa-inline.c (recursive_inlining): Use ::get method.
13610 * ipa-inline.h (estimate_edge_growth): Likewise.
13612 2018-06-08 Martin Liska <mliska@suse.cz>
13614 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
13616 (hsa_function_summary::hsa_function_summary): Use the new enum
13618 (hsa_gpu_implementation_p): Use hsa_summaries::get.
13619 * hsa-gen.c (hsa_get_host_function): Likewise.
13620 (get_brig_function_name): Likewise.
13621 * ipa-hsa.c (process_hsa_functions): Likewise.
13622 (ipa_hsa_write_summary): Likewise.
13623 * symbol-summary.h (symtab_duplication): Use ::get function/
13624 (get): New function.
13626 2018-06-08 Martin Liska <mliska@suse.cz>
13628 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
13630 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
13631 (hsa_register_kernel): Likewise.
13632 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
13633 * hsa-gen.c (hsa_get_host_function): Likewise.
13634 (get_brig_function_name): Likewise.
13635 (generate_hsa): Likewise.
13636 (pass_gen_hsail::execute): Likewise.
13637 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
13638 (devirtualization_time_bonus): Likewise.
13639 (ipcp_propagate_stage): Likewise.
13640 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
13641 (edge_set_predicate): Likewise.
13642 (evaluate_conditions_for_known_args): Likewise.
13643 (evaluate_properties_for_edge): Likewise.
13644 (ipa_fn_summary::reset): Likewise.
13645 (ipa_fn_summary_t::duplicate): Likewise.
13646 (dump_ipa_call_summary): Likewise.
13647 (ipa_dump_fn_summary): Likewise.
13648 (analyze_function_body): Likewise.
13649 (compute_fn_summary): Likewise.
13650 (estimate_edge_devirt_benefit): Likewise.
13651 (estimate_edge_size_and_time): Likewise.
13652 (estimate_calls_size_and_time): Likewise.
13653 (estimate_node_size_and_time): Likewise.
13654 (inline_update_callee_summaries): Likewise.
13655 (remap_edge_change_prob): Likewise.
13656 (remap_edge_summaries): Likewise.
13657 (ipa_merge_fn_summary_after_inlining): Likewise.
13658 (ipa_update_overall_fn_summary): Likewise.
13659 (read_ipa_call_summary): Likewise.
13660 (inline_read_section): Likewise.
13661 (write_ipa_call_summary): Likewise.
13662 (ipa_fn_summary_write): Likewise.
13663 (ipa_free_fn_summary): Likewise.
13664 * ipa-hsa.c (process_hsa_functions): Likewise.
13665 (ipa_hsa_write_summary): Likewise.
13666 (ipa_hsa_read_section): Likewise.
13667 * ipa-icf.c (sem_function::merge): Likewise.
13668 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
13669 (do_estimate_edge_time): Likewise.
13670 (estimate_size_after_inlining): Likewise.
13671 (estimate_growth): Likewise.
13672 (growth_likely_positive): Likewise.
13673 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
13674 (inline_call): Likewise.
13675 * ipa-inline.c (caller_growth_limits): Likewise.
13676 (can_inline_edge_p): Likewise.
13677 (can_inline_edge_by_limits_p): Likewise.
13678 (compute_uninlined_call_time): Likewise.
13679 (compute_inlined_call_time): Likewise.
13680 (want_inline_small_function_p): Likewise.
13681 (edge_badness): Likewise.
13682 (update_caller_keys): Likewise.
13683 (update_callee_keys): Likewise.
13684 (recursive_inlining): Likewise.
13685 (inline_small_functions): Likewise.
13686 (inline_to_all_callers_1): Likewise.
13687 (dump_overall_stats): Likewise.
13688 (early_inline_small_functions): Likewise.
13689 (early_inliner): Likewise.
13690 * ipa-inline.h (estimate_edge_growth): Likewise.
13691 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13692 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
13693 * ipa-prop.h (IPA_NODE_REF): Likewise.
13694 (IPA_EDGE_REF): Likewise.
13695 * ipa-pure-const.c (malloc_candidate_p): Likewise.
13696 (propagate_malloc): Likewise.
13697 * ipa-split.c (execute_split_functions): Likewise.
13698 * symbol-summary.h: Rename get to get_create.
13700 (get_create): Likewise.
13701 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
13703 2018-06-08 Martin Liska <mliska@suse.cz>
13705 * symbol-summary.h (release): Move definition out of class
13707 (symtab_removal): Likewise.
13708 (symtab_duplication): Likewise.
13710 2018-06-08 Martin Liska <mliska@suse.cz>
13712 * symbol-summary.h (function_summary): Move constructor
13713 implementation out of class declaration.
13714 (release): Likewise.
13715 (symtab_insertion): Likewise.
13716 (symtab_removal): Likewise.
13717 (symtab_duplication): Likewise.
13720 2018-06-08 Martin Liska <mliska@suse.cz>
13722 * Makefile.in: Remove support for MPX (macros, related functions,
13723 fields in cgraph_node, ...).
13724 * builtin-types.def (BT_BND): Likewise.
13725 (BT_FN_BND_CONST_PTR): Likewise.
13726 (BT_FN_CONST_PTR_BND): Likewise.
13727 (BT_FN_VOID_PTR_BND): Likewise.
13728 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
13729 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
13730 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
13731 (expand_builtin_mempcpy_with_bounds): Likewise.
13732 (expand_builtin_memset_with_bounds): Likewise.
13733 (expand_builtin_memset_args): Likewise.
13734 (std_expand_builtin_va_start): Likewise.
13735 (expand_builtin): Likewise.
13736 (expand_builtin_with_bounds): Likewise.
13737 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
13738 (DEF_LIB_BUILTIN_CHKP): Likewise.
13739 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
13740 (DEF_CHKP_BUILTIN): Likewise.
13741 (BUILT_IN_MEMCPY): Likewise.
13742 (BUILT_IN_MEMMOVE): Likewise.
13743 (BUILT_IN_MEMPCPY): Likewise.
13744 (BUILT_IN_MEMSET): Likewise.
13745 (BUILT_IN_STPCPY): Likewise.
13746 (BUILT_IN_STRCAT): Likewise.
13747 (BUILT_IN_STRCHR): Likewise.
13748 (BUILT_IN_STRCPY): Likewise.
13749 (BUILT_IN_STRLEN): Likewise.
13750 (BUILT_IN_MEMCPY_CHK): Likewise.
13751 (BUILT_IN_MEMMOVE_CHK): Likewise.
13752 (BUILT_IN_MEMPCPY_CHK): Likewise.
13753 (BUILT_IN_MEMSET_CHK): Likewise.
13754 (BUILT_IN_STPCPY_CHK): Likewise.
13755 (BUILT_IN_STRCAT_CHK): Likewise.
13756 (BUILT_IN_STRCPY_CHK): Likewise.
13757 * calls.c (store_bounds): Likewise.
13758 (emit_call_1): Likewise.
13759 (special_function_p): Likewise.
13760 (maybe_warn_nonstring_arg): Likewise.
13761 (initialize_argument_information): Likewise.
13762 (finalize_must_preallocate): Likewise.
13763 (compute_argument_addresses): Likewise.
13764 (expand_call): Likewise.
13765 * cfgexpand.c (expand_call_stmt): Likewise.
13766 (expand_return): Likewise.
13767 (expand_gimple_stmt_1): Likewise.
13768 (pass_expand::execute): Likewise.
13769 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
13770 (cgraph_node::remove): Likewise.
13771 (cgraph_node::dump): Likewise.
13772 (cgraph_node::verify_node): Likewise.
13773 * cgraph.h (chkp_function_instrumented_p): Likewise.
13774 (symtab_node::get_alias_target): Likewise.
13775 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
13776 (cgraph_local_p): Likewise.
13777 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
13778 (cgraph_edge::rebuild_references): Likewise.
13779 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
13780 (walk_polymorphic_call_targets): Likewise.
13781 (cgraph_node::expand_thunk): Likewise.
13782 (symbol_table::output_weakrefs): Likewise.
13783 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
13784 (ix86_handle_option): Likewise.
13785 * config/i386/constraints.md: Likewise.
13786 * config/i386/i386-builtin-types.def (BND): Likewise.
13790 * config/i386/i386-builtin.def (BDESC_END): Likewise.
13791 (BDESC_FIRST): Likewise.
13793 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
13794 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
13795 * config/i386/i386.c (enum reg_class): Likewise.
13796 (ix86_target_string): Likewise.
13797 (ix86_option_override_internal): Likewise.
13798 (ix86_conditional_register_usage): Likewise.
13799 (ix86_valid_target_attribute_inner_p): Likewise.
13800 (ix86_set_indirect_branch_type): Likewise.
13801 (ix86_set_current_function): Likewise.
13802 (ix86_function_arg_regno_p): Likewise.
13803 (init_cumulative_args): Likewise.
13804 (ix86_function_arg_advance): Likewise.
13805 (ix86_function_arg): Likewise.
13806 (ix86_pass_by_reference): Likewise.
13807 (ix86_function_value_regno_p): Likewise.
13808 (ix86_function_value_1): Likewise.
13809 (ix86_function_value_bounds): Likewise.
13810 (ix86_return_in_memory): Likewise.
13811 (ix86_setup_incoming_vararg_bounds): Likewise.
13812 (ix86_va_start): Likewise.
13813 (indirect_thunk_need_prefix): Likewise.
13814 (print_reg): Likewise.
13815 (ix86_print_operand): Likewise.
13816 (ix86_expand_call): Likewise.
13817 (ix86_output_function_return): Likewise.
13818 (reg_encoded_number): Likewise.
13819 (BDESC_VERIFYS): Likewise.
13820 (ix86_init_mpx_builtins): Likewise.
13821 (ix86_init_builtins): Likewise.
13822 (ix86_emit_cmove): Likewise.
13823 (ix86_emit_move_max): Likewise.
13824 (ix86_expand_builtin): Likewise.
13825 (ix86_builtin_mpx_function): Likewise.
13826 (ix86_get_arg_address_for_bt): Likewise.
13827 (ix86_load_bounds): Likewise.
13828 (ix86_store_bounds): Likewise.
13829 (ix86_load_returned_bounds): Likewise.
13830 (ix86_store_returned_bounds): Likewise.
13831 (ix86_class_likely_spilled_p): Likewise.
13832 (ix86_hard_regno_mode_ok): Likewise.
13833 (x86_order_regs_for_local_alloc): Likewise.
13834 (ix86_mitigate_rop): Likewise.
13835 (ix86_bnd_prefixed_insn_p): Likewise.
13836 (ix86_mpx_bound_mode): Likewise.
13837 (ix86_make_bounds_constant): Likewise.
13838 (ix86_initialize_bounds): Likewise.
13839 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
13840 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
13841 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
13842 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
13843 (TARGET_CHKP_BOUND_MODE): Likewise.
13844 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
13845 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
13846 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
13847 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
13848 * config/i386/i386.h (TARGET_MPX): Likewise.
13849 (TARGET_MPX_P): Likewise.
13850 (VALID_BND_REG_MODE): Likewise.
13851 (FIRST_BND_REG): Likewise.
13852 (LAST_BND_REG): Likewise.
13853 (enum reg_class): Likewise.
13854 (BND_REG_P): Likewise.
13855 (BND_REGNO_P): Likewise.
13856 (BNDmode): Likewise.
13857 (ADJUST_INSN_LENGTH): Likewise.
13858 * config/i386/i386.md: Likewise.
13859 * config/i386/i386.opt: Likewise.
13860 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
13861 (defined): Likewise.
13862 (LINK_MPX): Likewise.
13863 (MPX_SPEC): Likewise.
13864 (LIBMPX_SPEC): Likewise.
13865 (LIBMPXWRAPPERS_SPEC): Likewise.
13866 (CHKP_SPEC): Likewise.
13867 * config/i386/predicates.md: Likewise.
13868 * dbxout.c (dbxout_type): Likewise.
13869 * doc/extend.texi: Likewise.
13870 * doc/invoke.texi: Likewise.
13871 * doc/md.texi: Likewise.
13872 * doc/tm.texi: Likewise.
13873 * doc/tm.texi.in: Likewise.
13874 * dwarf2out.c (is_base_type): Likewise.
13875 (gen_formal_types_die): Likewise.
13876 (gen_subprogram_die): Likewise.
13877 (gen_type_die_with_usage): Likewise.
13878 (gen_decl_die): Likewise.
13879 (dwarf2out_late_global_decl): Likewise.
13880 * expr.c (expand_assignment): Likewise.
13881 (emit_storent_insn): Likewise.
13882 (store_expr_with_bounds): Likewise.
13883 (store_expr): Likewise.
13884 (expand_expr_real_1): Likewise.
13885 * expr.h (store_expr_with_bounds): Likewise.
13886 * function.c (use_register_for_decl): Likewise.
13887 (struct bounds_parm_data): Likewise.
13888 (assign_parms_augmented_arg_list): Likewise.
13889 (assign_parm_find_entry_rtl): Likewise.
13890 (assign_parm_is_stack_parm): Likewise.
13891 (assign_parm_load_bounds): Likewise.
13892 (assign_bounds): Likewise.
13893 (assign_parms): Likewise.
13894 (expand_function_start): Likewise.
13895 * gcc.c (CHKP_SPEC): Likewise.
13896 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
13897 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
13898 (wrestrict_dom_walker::check_call): Likewise.
13899 * gimple.c (gimple_build_call_from_tree): Likewise.
13900 * gimple.h (enum gf_mask): Likewise.
13901 (gimple_call_with_bounds_p): Likewise.
13902 (gimple_call_set_with_bounds): Likewise.
13903 * gimplify.c (gimplify_init_constructor): Likewise.
13904 * ipa-cp.c (initialize_node_lattices): Likewise.
13905 (propagate_constants_across_call): Likewise.
13906 (find_more_scalar_values_for_callers_subset): Likewise.
13907 * ipa-hsa.c (process_hsa_functions): Likewise.
13908 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
13909 * ipa-icf.c (sem_function::merge): Likewise.
13910 * ipa-inline.c (early_inliner): Likewise.
13911 * ipa-pure-const.c (warn_function_noreturn): Likewise.
13912 (warn_function_cold): Likewise.
13913 (propagate_pure_const): Likewise.
13914 * ipa-ref.h (enum GTY): Likewise.
13915 * ipa-split.c (find_retbnd): Likewise.
13916 (consider_split): Likewise.
13917 (split_function): Likewise.
13918 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
13919 * ipa.c (walk_polymorphic_call_targets): Likewise.
13920 (symbol_table::remove_unreachable_nodes): Likewise.
13921 (process_references): Likewise.
13922 (cgraph_build_static_cdtor_1): Likewise.
13923 * lto-cgraph.c (lto_output_node): Likewise.
13924 (output_refs): Likewise.
13925 (compute_ltrans_boundary): Likewise.
13926 (input_overwrite_node): Likewise.
13927 (input_node): Likewise.
13928 (input_cgraph_1): Likewise.
13929 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
13930 * passes.c (pass_manager::execute_early_local_passes): Likewise.
13931 (class pass_chkp_instrumentation_passes): Likewise.
13932 (make_pass_chkp_instrumentation_passes): Likewise.
13933 * passes.def: Likewise.
13934 * rtl.h (struct GTY): Likewise.
13935 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
13936 * stor-layout.c (layout_type): Likewise.
13937 * symtab.c: Likewise.
13938 * target.def: Likewise.
13939 * targhooks.c (default_chkp_bound_type): Likewise.
13940 (default_chkp_bound_mode): Likewise.
13941 (default_builtin_chkp_function): Likewise.
13942 (default_chkp_function_value_bounds): Likewise.
13943 (default_chkp_make_bounds_constant): Likewise.
13944 (default_chkp_initialize_bounds): Likewise.
13945 * targhooks.h (default_chkp_bound_type): Likewise.
13946 (default_chkp_bound_mode): Likewise.
13947 (default_builtin_chkp_function): Likewise.
13948 (default_chkp_function_value_bounds): Likewise.
13949 (default_chkp_make_bounds_constant): Likewise.
13950 (default_chkp_initialize_bounds): Likewise.
13951 * toplev.c (compile_file): Likewise.
13952 (process_options): Likewise.
13953 * tree-core.h (DEF_BUILTIN): Likewise.
13954 (DEF_BUILTIN_CHKP): Likewise.
13955 * tree-inline.c (declare_return_variable): Likewise.
13956 (remap_gimple_stmt): Likewise.
13957 (copy_bb): Likewise.
13958 (initialize_inlined_parameters): Likewise.
13959 (expand_call_inline): Likewise.
13960 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
13961 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
13962 (make_pass_ipa_chkp_produce_thunks): Likewise.
13963 (make_pass_chkp): Likewise.
13964 (make_pass_chkp_opt): Likewise.
13965 (make_pass_chkp_instrumentation_passes): Likewise.
13966 * tree-pretty-print.c (dump_generic_node): Likewise.
13967 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
13968 * tree-ssa-dce.c (propagate_necessity): Likewise.
13969 (eliminate_unnecessary_stmts): Likewise.
13970 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
13971 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
13972 * tree-ssa-sccvn.h: Likewise.
13973 * tree-ssa-strlen.c (get_string_length): Likewise.
13974 (valid_builtin_call): Likewise.
13975 (adjust_last_stmt): Likewise.
13976 (handle_builtin_strchr): Likewise.
13977 (handle_builtin_strcpy): Likewise.
13978 (handle_builtin_stxncpy): Likewise.
13979 (handle_builtin_memcpy): Likewise.
13980 (handle_builtin_strcat): Likewise.
13981 (strlen_check_and_optimize_stmt): Likewise.
13982 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
13983 * tree-streamer-in.c: Likewise.
13984 * tree-streamer.c (record_common_node): Likewise.
13985 * tree.c (tree_code_size): Likewise.
13986 (wide_int_to_tree_1): Likewise.
13987 (type_contains_placeholder_1): Likewise.
13988 (build_common_tree_nodes): Likewise.
13989 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
13990 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
13991 (POINTER_BOUNDS_P): Likewise.
13992 (BOUNDED_TYPE_P): Likewise.
13993 (BOUNDED_P): Likewise.
13994 (CALL_WITH_BOUNDS_P): Likewise.
13995 (pointer_bounds_type_node): Likewise.
13996 * value-prof.c (gimple_ic): Likewise.
13997 * var-tracking.c (vt_add_function_parameters): Likewise.
13998 * varasm.c (make_decl_rtl): Likewise.
13999 (assemble_start_function): Likewise.
14000 (output_constant): Likewise.
14001 (maybe_assemble_visibility): Likewise.
14002 * varpool.c (ctor_for_folding): Likewise.
14003 * chkp-builtins.def: Remove.
14004 * ipa-chkp.c: Remove.
14005 * ipa-chkp.h: Remove.
14006 * rtl-chkp.c: Remove.
14007 * rtl-chkp.h: Remove.
14008 * tree-chkp-opt.c: Remove.
14009 * tree-chkp.c: Remove.
14010 * tree-chkp.h: Remove.
14012 2018-06-07 Carl Love <cel@us.ibm.com>
14014 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14015 vextract_fp_from_shortl): Add BE support.
14017 2018-06-07 Paul Koning <ni1d@arrl.net>
14019 * compare-elim.c (try_merge_compare): Don't merge compare if
14020 address contains a side effect.
14021 (try_eliminate_compare): Likewise.
14023 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
14025 * config.gcc: Support "tremont".
14026 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
14027 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14029 * config/i386/i386.c (m_TREMONT): Define.
14030 (processor_target_table): Add "tremont".
14031 (PTA_TREMONT): Define.
14032 (ix86_lea_outperforms): Add TARGET_TREMONT.
14033 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
14034 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
14035 and M_INTEL_GOLDMONT_PLUS.
14036 (fold_builtin_cpu): Add "tremont".
14037 (ix86_add_stmt_cost): Add TARGET_TREMONT.
14038 (ix86_option_override_internal): Add "tremont".
14039 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
14040 (processor_type): Add PROCESSOR_TREMONT.
14041 * config/i386/x86-tune.def: Add m_TREMONT.
14042 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
14044 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14046 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
14047 symbol defined for msp430i* devices to be lower case.
14049 2018-06-07 Richard Biener <rguenther@suse.de>
14051 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
14052 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
14053 Properly wrap signed arithmetic if overflow wraps.
14055 2018-06-07 Jakub Jelinek <jakub@redhat.com>
14057 PR tree-optimization/69615
14058 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
14059 of a cast from a same precision integral SSA_NAME in a bb dominated
14060 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
14061 cast to utype if rhs2 has already a compatible type.
14063 2018-06-07 Richard Biener <rguenther@suse.de>
14065 PR tree-optimization/85935
14066 * graphite-scop-detection.c (find_params_in_bb): Analyze
14067 condition operands with respect to the correct loop. Assert
14068 the analysis doesn't fail.
14070 2018-06-04 Carl Love <cel@us.ibm.com>
14072 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
14073 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
14074 as it is slightly cheaper.
14075 (first_match_or_eos_index_<mode>):
14076 Calculate index using natural element order.
14077 (first_match_index_<mode>):
14078 Calculate index using natural element order.
14079 (first_match_or_eos_index_<mode>):
14080 Calculate index using natural order.
14081 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
14082 for BE and LE modes.
14083 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
14084 P9V_BUILTIN_VCLZLSBB_V16QI.
14085 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
14088 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
14090 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
14091 indentation and line wrap for many prototypes. Add missing
14092 @smallexample directives around block of prototypes for vec_xl and
14095 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
14097 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
14098 track if we pass or return IEEE 128-bit floating point.
14099 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
14100 C++ mangling that is compatible with GCC 8.1.
14101 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
14102 (init_cumulative_args): Note if we pass or return IEEE 128-bit
14103 floating point types.
14104 (rs6000_function_arg_advance_1): Likewise.
14105 (rs6000_mangle_type): Optionally generate mangled names that match
14106 what GCC 8.1 generated for IEEE 128-bit floating point types.
14107 (rs6000_globalize_decl_name): If we have an external function that
14108 passes or returns IEEE 128-bit types, generate a weak reference
14109 from the mangled name used in GCC 8.1 to the current mangled
14111 (rs6000_init_builtins): Make __ibm128 use the long double type if
14112 long double is IBM extended double. Make __float128 use the long
14113 double type if long double is IEEE 128-bit.
14116 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
14117 macro for __ibm128 built-in functions.
14118 (PACK_IF): Add __ibm128 pack/unpack functions.
14119 (UNPACK_IF): Likewise.
14120 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
14121 enable long double built-in functions if long double is IEEE
14122 128-bit floating point.
14123 (rs6000_invalid_builtin): Update long double built-in function
14125 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
14126 functions, adjust the built-in function to use the long double
14127 built-in function if __ibm128 and long double are the same type.
14128 * doc/extend.texi (PowerPC builtins): Update documention for
14129 __builtin_{,un}pack_longdouble. Add documentation for
14130 __builtin_{,un}pack_ibm128.
14132 2018-06-06 Jim Wilson <jimw@sifive.com>
14134 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
14135 (struct machine_function): New field interrupt_mode.
14136 (riscv_handle_type_attribute): New function. Add forward declaration.
14137 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
14138 (riscv_expand_epilogue): Check interrupt_mode field.
14139 (riscv_set_current_function): Check interrupt attribute args and
14140 set interrupt_mode field.
14141 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
14142 (riscv_sret, riscv_uret): New.
14143 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
14144 new arguments to interrupt attribute.
14146 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
14149 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
14150 Don't handle -mcpu=power8 if -mpower9-vector is also used.
14152 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
14154 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
14155 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
14156 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
14157 several redundant entries.
14159 2018-06-06 David Malcolm <dmalcolm@redhat.com>
14161 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
14162 type from "rtx" to "rtx_insn *".
14163 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
14164 for local "call_insn", removing cast.
14165 (ix86_expand_call): Likewise, introducing a "call_insn" local.
14167 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
14169 PR tree-optimization/86066
14170 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
14171 for BIT_INSERT_EXPR stores.
14173 2018-06-06 Richard Biener <rguenther@suse.de>
14175 PR tree-optimization/86062
14176 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
14177 component refs ontop
14178 of to be offsetted base.
14180 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14182 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
14183 to be static and remove check on interrupt attribute name.
14185 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
14187 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
14188 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
14190 2018-06-05 Steve Ellcey <sellcey@cavium.com>
14193 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
14195 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
14196 Remove second argument, change how error is called.
14197 (aarch64_layout_arg): Remove second argument from
14198 aarch64_err_no_fpadvsimd call.
14199 (aarch64_init_cumulative_args): Ditto.
14200 (aarch64_gimplify_va_arg_expr): Ditto.
14201 * config/aarch64/aarch64.md (mov<mode>): Ditto.
14203 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
14205 * config/i386/i386.md (simple_return_indirect_internal): New expander.
14206 (*simple_return_indirect_internal<mode>): Rename from
14207 simple_return_indirect_internal. Use W mode iterator.
14208 (rstorssp): New expander.
14209 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
14210 (clrssbsy): New expander.
14211 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
14213 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
14215 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
14217 (cmse_check_pointed_object): Likewise.
14219 2018-06-05 Martin Liska <mliska@suse.cz>
14221 PR gcov-profile/47618
14222 * doc/invoke.texi: Document how -fprofile-dir format
14225 2018-06-05 Richard Biener <rguenther@suse.de>
14227 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
14228 removal pretend DOM info isn't available so we do not update
14229 it and only remove edges, not dominated blocks. Actually free
14230 DOM info in case we removed something. Remove unreachable blocks.
14231 (mfb_keep_latches): Work with either DOM info or marked backedges.
14232 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
14233 first. Mark backedges if DOM info isn't available.
14234 (Re-)compute DOM info after cleanup_control_flow_pre.
14236 2018-06-05 Richard Biener <rguenther@suse.de>
14238 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
14239 (locus_discrim_hasher::hash): Adjust.
14240 (locus_discrim_hasher::equal): Likewise.
14241 (next_discriminator_for_locus): Work on line directly.
14242 (same_line_p): Pass in expanded locus1 as well.
14243 (assign_discriminators): Avoid redundant location expansions.
14245 2018-06-05 Richard Biener <rguenther@suse.de>
14247 PR tree-optimization/86046
14248 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
14249 if required after clearing TREE_ADDRESSABLE.
14251 2018-06-05 Richard Biener <rguenther@suse.de>
14253 PR tree-optimization/86047
14254 * tree-ssa-loop.c (for_each_index): Glob handling of all
14255 decls and constants and really handle all of them.
14257 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14260 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
14261 qualifier_void_pointer and qualifier_const_void_pointer.
14262 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
14263 (arm_init_builtins): Handle the above.
14264 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
14265 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
14266 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
14269 2018-06-05 Martin Liska <mliska@suse.cz>
14271 * auto-profile.c (read_autofdo_file): Do not use
14272 gcov_ctr_summary struct.
14273 (afdo_callsite_hot_enough_for_early_inline): Likewise.
14274 * coverage.c (struct counts_entry): Likewise.
14275 (read_counts_file): Read just single summary entry.
14276 (get_coverage_counts): Use gcov_summary struct.
14277 * coverage.h (get_coverage_counts): Likewise.
14278 * gcov-dump.c (dump_working_sets): Likewise.
14279 (tag_summary): Dump just single summary.
14280 * gcov-io.c (gcov_write_summary): Write just histogram
14282 (gcov_read_summary): Read just single summary.
14283 (compute_working_sets): Use gcov_summary struct.
14284 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
14285 of GCOV_COUNTERS_SUMMABLE.
14286 (GCOV_COUNTERS_SUMMABLE): Remove.
14287 (GCOV_FIRST_VALUE_COUNTER): Replace with
14288 GCOV_COUNTER_V_INTERVAL.
14289 (struct gcov_ctr_summary): Remove.
14290 (struct gcov_summary): Directly use fields of former
14292 (compute_working_sets): Use gcov_summary struct.
14293 * gcov.c (read_count_file): Do not use ctrs fields.
14294 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
14296 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
14298 * profile.c: Likewise.
14299 * profile.h: Likewise.
14301 2018-06-05 Martin Liska <mliska@suse.cz>
14303 PR gcov-profile/84846
14304 * gcov.c (output_lines): Print working directory only
14305 in intermediate format.
14307 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
14309 * config/s390/s390-builtin-types.def: Add void function type.
14310 * config/s390/s390-builtins.def: Use the function type for the
14313 2018-06-04 Jim Wilson <jimw@sifive.com>
14315 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
14317 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
14318 and maybe_eh_return. Change regno to unsigned int. Use new args to
14319 handle EH_RETURN_DATA_REGNO registers properly.
14320 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
14321 (riscv_expand_epilogue): Update comment. Change argument name and
14322 type. Update code to use new name and type. Pass new args to
14323 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
14325 * config/riscv/riscv.md (NORMAL_RETURN): New.
14326 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
14327 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
14328 (eh_return): Call gen_eh_return_internal and emit barrier.
14329 (eh_return_internal): Call riscv_expand_epilogue.
14331 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
14333 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
14334 bit_insertion field and declare can_be_merged_into method.
14335 (merged_store_group::can_be_merged_into): New method.
14336 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
14337 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
14338 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
14340 2018-06-04 Richard Biener <rguenther@suse.de>
14342 PR tree-optimization/85955
14343 * builtins.c (fold_builtin_sincos): Convert pointers to
14344 destination to appropriate type before dereferencing.
14346 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
14348 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
14350 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
14352 * expr.c (expand_expr_real_1): Force the operand into memory if
14353 its TYPE_MODE is BLKmode and if there is no integer mode for
14354 the number of bits being extracted.
14356 2018-06-04 Jakub Jelinek <jakub@redhat.com>
14360 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
14361 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
14363 2018-06-04 Richard Biener <rguenther@suse.de>
14365 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
14366 (cleanup_tree_cfg_noloop): ... single caller. Do
14367 start_recording_case_labels later.
14369 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
14371 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
14372 to _IMMINTRIN_H_INCLUDED.
14373 * config/i386/pconfigintrin.h: Ditto.
14374 * config/i386/waitpkgintrin.h: Ditto.
14375 * config/i386/immintrin.h: Add includes for sgxintrin.h,
14376 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
14377 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
14378 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
14379 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
14380 waitpkgintrin.h and cldemoteintrin.h.
14382 2018-06-04 Richard Biener <rguenther@suse.de>
14384 PR tree-optimization/86038
14385 * tracer.c (find_best_successor): Check probability for
14386 being initialized, bail out if not.
14388 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
14391 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
14392 of bits to ignore when comparing architectures.
14394 2018-06-04 Jakub Jelinek <jakub@redhat.com>
14396 PR tree-optimization/69615
14397 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
14398 maximum or minimum of the type, try to merge it also as if
14399 range1 is + [-, x - 1] or + [x + 1, -].
14402 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
14404 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
14406 PR tree-optimization/86034
14407 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
14408 the unsigned bitfield type in a bit insertion sequence if it does not
14409 have a larger precision than the bitfield size.
14410 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
14412 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
14414 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
14416 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
14418 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
14419 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
14420 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
14421 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
14423 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
14425 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
14426 Disable -fdelete-null-pointer-checks for ELF toolchain.
14428 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
14429 Kito Cheng <kito.cheng@gmail.com>
14431 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
14432 (nds32le-*-*, nds32be-*-*): Integrate checking process.
14433 (nds32*-*-*): Add glibc and uclibc conditions.
14434 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
14435 (TARGET_EXCEPT_UNWIND_INFO): Define.
14436 * config/nds32/elf.h: New file.
14437 * config/nds32/linux.h: New file.
14438 * config/nds32/nds32-elf.opt: New file.
14439 * config/nds32/nds32-linux.opt: New file.
14440 * config/nds32/nds32-fp-as-gp.c
14441 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
14442 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
14444 (nds32_asm_file_end): Ditto.
14445 (nds32_print_operand): Ditto.
14446 (nds32_insert_attributes): Ditto.
14447 (nds32_init_libfuncs): New function.
14448 (TARGET_HAVE_TLS): Define.
14449 (TARGET_INIT_LIBFUNCS): Define.
14450 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
14452 (TARGET_ELF): Apply different mcmodel setting.
14453 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
14454 been migrated into elf.h and linux.h files.
14455 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
14456 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
14457 (mcmodel): The content has been migrated into nds32-elf.opt and
14458 nds32-linux.opt files.
14459 * config/nds32/t-elf: New file.
14460 * config/nds32/t-linux: New file.
14462 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
14463 Shiva Chen <shiva0217@gmail.com>
14465 * config/nds32/constants.md (unspec_volatile_element): Add
14466 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
14467 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
14469 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
14470 (make_pass_nds32_fp_as_gp): Declare.
14471 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
14473 (nds32_asm_function_end_prologue): Remove unused asm output.
14474 (nds32_asm_function_begin_epilogue): Remove unused asm output.
14475 (nds32_asm_file_start): Output necessary fp_as_gp information.
14476 (nds32_option_override): Adjust register usage.
14477 (nds32_expand_prologue): Consider fp_as_gp situation.
14478 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
14479 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
14482 (simple_return): Ditto.
14483 (omit_fp_begin): Output special directive for fp_as_gp.
14484 (omit_fp_end): Output special directive for fp_as_gp.
14485 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
14486 mforbid-fp-as-gp): New options.
14488 2018-06-01 Mark Wielaard <mark@klomp.org>
14490 * dwarf2out.c (dwarf2out_finish): Remove generation of
14491 DW_AT_loclists_base.
14493 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
14495 * gimple-ssa-store-merging.c: Include gimple-fold.h.
14496 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
14497 (struct merged_store_group): Add bit_insertion field.
14498 (dump_char_array): Use standard hexadecimal format.
14499 (merged_store_group::merged_store_group): Set bit_insertion to false.
14500 (merged_store_group::apply_stores): Use optimal buffer size. Deal
14501 with BIT_INSERT_EXPR stores. Move up code updating the mask and
14502 also print the mask in the dump file.
14503 (pass_store_merging::gate): Minor tweak.
14504 (imm_store_chain_info::coalesce_immediate): Fix wrong association
14505 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
14506 stores with INTEGER_CST stores.
14507 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
14508 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
14509 and use it throughout. Generate bit insertion sequences if need be.
14510 (pass_store_merging::process_store): Remove redundant condition.
14511 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
14513 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
14515 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
14516 the 128-bit floating point types. Fix function comment.
14518 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14520 * config/aarch64/aarch64-simd.md
14521 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
14523 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
14526 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
14528 PR tree-optimization/85989
14529 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
14531 (backprop::intersect_uses): Check it when deciding whether this
14532 is a backedge reference.
14533 (backprop::process_block): Add each phi to m_visited_phis
14534 after visiting it, then clear it at the end.
14536 2018-06-01 Richard Biener <rguenther@suse.de>
14538 * tree-vectorizer.h (vect_dr_stmt): New function.
14539 (vect_get_load_cost): Adjust.
14540 (vect_get_store_cost): Likewise.
14541 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
14542 Use vect_dr_stmt instead of DR_SMTT.
14543 (vect_record_base_alignments): Likewise.
14544 (vect_calculate_target_alignment): Likewise.
14545 (vect_compute_data_ref_alignment): Likewise and make static.
14546 (vect_update_misalignment_for_peel): Likewise.
14547 (vect_verify_datarefs_alignment): Likewise.
14548 (vector_alignment_reachable_p): Likewise.
14549 (vect_get_data_access_cost): Likewise. Pass down
14550 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
14551 (vect_get_peeling_costs_all_drs): Likewise.
14552 (vect_peeling_hash_get_lowest_cost): Likewise.
14553 (vect_enhance_data_refs_alignment): Likewise.
14554 (vect_find_same_alignment_drs): Likewise.
14555 (vect_analyze_data_refs_alignment): Likewise.
14556 (vect_analyze_group_access_1): Likewise.
14557 (vect_analyze_group_access): Likewise.
14558 (vect_analyze_data_ref_access): Likewise.
14559 (vect_analyze_data_ref_accesses): Likewise.
14560 (vect_vfa_segment_size): Likewise.
14561 (vect_small_gap_p): Likewise.
14562 (vectorizable_with_step_bound_p): Likewise.
14563 (vect_prune_runtime_alias_test_list): Likewise.
14564 (vect_analyze_data_refs): Likewise.
14565 (vect_supportable_dr_alignment): Likewise.
14566 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
14567 (vect_gen_prolog_loop_niters): Likewise.
14568 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
14569 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
14571 (vect_recog_mask_conversion_pattern): Likewise.
14572 (vect_try_gather_scatter_pattern): Likewise.
14573 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
14574 to vect_get_store_cost.
14575 (vect_get_store_cost): Get stmt_info instead of DR.
14576 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
14577 (vect_get_load_cost): Get stmt_info instead of DR.
14579 2018-06-01 Richard Biener <rguenther@suse.de>
14581 PR middle-end/86017
14582 * gimple-fold.c (var_decl_component_p): Also allow offsetted
14583 vars wrapped in MEM_REFs.
14585 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
14587 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
14588 Fix subreg tests so that we only return a choice between
14589 GENERAL_REGS and FP_REGS if the original classes included both.
14591 2018-06-01 Richard Biener <rguenther@suse.de>
14594 * tree-ssa-structalias.c (get_function_part_constraint):
14595 Handle NULL fi->decl.
14596 (find_func_aliases_for_call): Properly handle indirect
14597 fi from direct call.
14598 (find_func_clobbers): Likewise.
14599 (ipa_pta_execute): Likewise.
14600 (create_variable_info_for): For functions that are ifunc_resolver
14601 resolve to a varinfo that contains the result of the resolver call.
14602 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
14605 2018-05-31 Michael Collison <michael.collison@arm.com>
14607 * config/aarch64/aarch64.md:
14608 (*fix_to_zero_extenddfdi2): New pattern.
14609 * gcc.target/aarch64/fix_extend1.c: New testcase.
14611 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
14613 PR middle-end/78809
14614 PR middle-end/83026
14615 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
14616 and BUILT_IN_STRNCMP_EQ.
14617 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
14618 BUILT_IN_STRNCMP_EQ.
14619 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
14620 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
14621 (gimple_fold_builtin): Likewise.
14622 * tree-ssa-strlen.c (compute_string_length): New function.
14623 (determine_min_obsize): New function.
14624 (handle_builtin_string_cmp): New function to handle calls to
14625 string compare functions.
14626 (strlen_optimize_stmt): Add handling to builtin string compare
14628 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
14629 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
14630 * tree.c (build_common_builtin_nodes): Add new defines of
14631 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
14633 2018-05-31 Jakub Jelinek <jakub@redhat.com>
14636 * bb-reorder.c (pass_partition_blocks::gate): Return false for
14637 functions with naked attribute.
14639 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
14641 * config/i386/sse.md (avx_vec_concat<mode>):
14642 Substitute concat_tg_mode mode attribute with xtg_mode.
14643 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
14644 (concat_tg_mode): Remove mode attribute.
14646 2018-05-31 Martin Sebor <msebor@redhat.com>
14649 * calls.c (alloc_max_size): Correct a logic error/typo.
14650 Treat excessive arguments as infinite. Warn for invalid arguments.
14651 * doc/invoke.texi (-Walloc-size-larger-than): Update.
14653 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
14656 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
14657 and movx for Haswell.
14659 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
14660 Cesar Philippidis <cesar@codesourcery.com>
14662 PR middle-end/85879
14663 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
14664 when emitting error on private/firstprivate reductions.
14665 * omp-low.c (lower_omp_target): Avoid reference-type processing
14666 on pointers for firstprivate clause.
14668 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
14670 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
14673 * config/aarch64/aarch64-simd.md
14674 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
14675 (aarch64_ld1_x3_<mode>): Likewise
14676 (aarch64_st1x2<VALLDIF:mode>): Likewise
14677 (aarch64_st1_x2_<mode>): Likewise
14678 (aarch64_st1x3<VALLDIF:mode>): Likewise
14679 (aarch64_st1_x3_<mode>): Likewise
14680 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
14681 (vld1_s8_x3): Likewise.
14682 (vld1_u16_x3): Likewise.
14683 (vld1_s16_x3): Likewise.
14684 (vld1_u32_x3): Likewise.
14685 (vld1_s32_x3): Likewise.
14686 (vld1_u64_x3): Likewise.
14687 (vld1_s64_x3): Likewise.
14688 (vld1_f16_x3): Likewise.
14689 (vld1_f32_x3): Likewise.
14690 (vld1_f64_x3): Likewise.
14691 (vld1_p8_x3): Likewise.
14692 (vld1_p16_x3): Likewise.
14693 (vld1_p64_x3): Likewise.
14694 (vld1q_u8_x3): Likewise.
14695 (vld1q_s8_x3): Likewise.
14696 (vld1q_u16_x3): Likewise.
14697 (vld1q_s16_x3): Likewise.
14698 (vld1q_u32_x3): Likewise.
14699 (vld1q_s32_x3): Likewise.
14700 (vld1q_u64_x3): Likewise.
14701 (vld1q_s64_x3): Likewise.
14702 (vld1q_f16_x3): Likewise.
14703 (vld1q_f32_x3): Likewise.
14704 (vld1q_f64_x3): Likewise.
14705 (vld1q_p8_x3): Likewise.
14706 (vld1q_p16_x3): Likewise.
14707 (vld1q_p64_x3): Likewise.
14708 (vst1_s64_x2): Likewise.
14709 (vst1_u64_x2): Likewise.
14710 (vst1_f64_x2): Likewise.
14711 (vst1_s8_x2): Likewise.
14712 (vst1_p8_x2): Likewise.
14713 (vst1_s16_x2): Likewise.
14714 (vst1_p16_x2): Likewise.
14715 (vst1_s32_x2): Likewise.
14716 (vst1_u8_x2): Likewise.
14717 (vst1_u16_x2): Likewise.
14718 (vst1_u32_x2): Likewise.
14719 (vst1_f16_x2): Likewise.
14720 (vst1_f32_x2): Likewise.
14721 (vst1_p64_x2): Likewise.
14722 (vst1q_s8_x2): Likewise.
14723 (vst1q_p8_x2): Likewise.
14724 (vst1q_s16_x2): Likewise.
14725 (vst1q_p16_x2): Likewise.
14726 (vst1q_s32_x2): Likewise.
14727 (vst1q_s64_x2): Likewise.
14728 (vst1q_u8_x2): Likewise.
14729 (vst1q_u16_x2): Likewise.
14730 (vst1q_u32_x2): Likewise.
14731 (vst1q_u64_x2): Likewise.
14732 (vst1q_f16_x2): Likewise.
14733 (vst1q_f32_x2): Likewise.
14734 (vst1q_f64_x2): Likewise.
14735 (vst1q_p64_x2): Likewise.
14736 (vst1_s64_x3): Likewise.
14737 (vst1_u64_x3): Likewise.
14738 (vst1_f64_x3): Likewise.
14739 (vst1_s8_x3): Likewise.
14740 (vst1_p8_x3): Likewise.
14741 (vst1_s16_x3): Likewise.
14742 (vst1_p16_x3): Likewise.
14743 (vst1_s32_x3): Likewise.
14744 (vst1_u8_x3): Likewise.
14745 (vst1_u16_x3): Likewise.
14746 (vst1_u32_x3): Likewise.
14747 (vst1_f16_x3): Likewise.
14748 (vst1_f32_x3): Likewise.
14749 (vst1_p64_x3): Likewise.
14750 (vst1q_s8_x3): Likewise.
14751 (vst1q_p8_x3): Likewise.
14752 (vst1q_s16_x3): Likewise.
14753 (vst1q_p16_x3): Likewise.
14754 (vst1q_s32_x3): Likewise.
14755 (vst1q_s64_x3): Likewise.
14756 (vst1q_u8_x3): Likewise.
14757 (vst1q_u16_x3): Likewise.
14758 (vst1q_u32_x3): Likewise.
14759 (vst1q_u64_x3): Likewise.
14760 (vst1q_f16_x3): Likewise.
14761 (vst1q_f32_x3): Likewise.
14762 (vst1q_f64_x3): Likewise.
14763 (vst1q_p64_x3): Likewise.
14765 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14767 * config/msp430/msp430.c (msp430_output_labelref): Prepend
14768 user_label_prefix to name.
14770 * tree-core.h: Update comment about the format of NAME string
14771 passed to handler in attribute_spec.
14773 * config/msp430/msp430.md: Remove erroneous subreg expression from
14774 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
14775 zero_extend{q,h}isi2.
14777 2018-05-30 Borislav Petkov <bp@suse.de>
14779 * doc/extend.texi: Document some architecture specific
14780 constraints and sort entries.
14782 2018-05-30 Martin Sebor <msebor@redhat.com>
14784 PR middle-end/85369
14785 * builtins.c (expand_builtin_stpcpy_1): New function.
14786 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
14787 only if the former succeeds.
14789 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
14791 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
14794 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
14796 * doc/invoke.texi (-flinker-output): Document
14798 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
14800 * passes.c (ipa_write_summaries): Only modify statements if body
14802 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
14803 incrementally linking.
14804 (ipa_passes): Likewise.
14805 * lto-cgraph.c (lto_output_node): When incrementally linking do not
14806 pass down resolution info.
14807 * common.opt (flag_incremental_link): Update info.
14808 * gcc.c (plugin specs): Turn flinker-output=* to
14809 -plugin-opt=-linker-output-known
14810 * toplev.c (compile_file): Also cut compilation when doing incremental
14812 * flag-types. (enum lto_partition_model): Add
14813 LTO_LINKER_OUTPUT_NOLTOREL.
14814 (invoke.texi): Add -flinker-output docs.
14815 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
14816 link same way as WPA; do not stream in dead initializers.
14818 * dwarf2out.c (dwarf2out_die_ref_for_decl,
14819 darf2out_register_external_decl): Support incremental link.
14821 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
14823 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
14825 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
14827 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
14828 it down to simple_object_copy_lto_debug_sections.
14829 (run_gcc): Determine incremental LTO link time and configure
14830 lto1 into non-wpa mode, disable renaming of debug sections.
14832 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
14834 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
14835 descriptions of various incorrectly documented functions.
14837 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
14840 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
14841 address check not strict.
14843 2018-05-30 Richard Biener <rguenther@suse.de>
14845 PR tree-optimization/85964
14846 * tracer.c (better_p): Drop initialized count check, we only
14847 call the function with initialized counts now.
14848 (find_best_successor): Do find a best edge if one
14849 has uninitialized count.
14850 (find_best_predecessor): Likewise. Do BB frequency check only
14851 if count is initialized.
14853 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
14855 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
14856 (aarch64_ldrstr_offset_compare): New.
14857 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
14858 load/store orderings.
14859 (aarch64_gen_adjusted_ldpstp): Likewise.
14861 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
14863 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
14864 Check for subset of GENERAL_REGS and FP_REGS.
14865 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
14868 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
14870 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
14871 and wi::to_poly_offset. Add the current offset and then check
14872 whether the sum fits, rather than using an unchecked addition of
14873 a checked term. Check for a shwi rather than a uhwi.
14874 * expr.c (get_bit_range): Use tree_to_poly_uint64.
14875 (store_constructor): Use poly_int_tree_p.
14876 (expand_expr_real_1): Likewise.
14877 * function.c (assign_temp): Likewise.
14878 * fold-const.c (const_binop): Use poly_int_tree_p and
14879 wi::to_poly_offset.
14880 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
14882 * ipa-icf-gimple.c (func_checker::compare_operand): Use
14883 to_poly_offset for MEM offsets.
14884 * ipa-icf.c (sem_variable::equals): Likewise.
14885 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
14886 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
14887 wi::to_poly_offset for BIT_FIELD_REF offsets.
14888 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
14889 wi::to_poly_offset.
14890 * var-tracking.c (emit_note_insn_var_location): Use
14891 tree_to_poly_uint64.
14893 2018-05-29 Jim Wilson <jimw@sifive.com>
14895 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
14897 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
14900 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
14901 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
14903 (sse4_1_round<mode>2): Use nonimmediate_operand
14904 for operand 1 predicate.
14906 2018-05-29 Martin Sebor <msebor@redhat.com>
14907 Richard Biener <rguenther@suse.de>
14910 * calls.c (get_size_range): Call determine_value_range instead
14911 of get_value_range..
14912 * tree-vrp.h (determine_value_range): Declared new function.
14913 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
14915 2018-05-29 Richard Biener <rguenther@suse.de>
14917 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
14918 sure to use non-pattern stmts for get_earlier_stmt arguments.
14919 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
14920 called on pattern stmts.
14921 (get_later_stmt): Likewise.
14923 2018-05-29 Martin Liska <mliska@suse.cz>
14925 PR gcov-profile/85759
14926 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
14929 2018-05-29 Jakub Jelinek <jakub@redhat.com>
14931 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
14933 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
14937 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
14938 VEC_PACK_FLOAT_EXPR): New tree codes.
14939 * tree-pretty-print.c (op_code_prio): Handle
14940 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
14941 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
14942 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
14943 * tree-inline.c (estimate_operator_cost): Likewise.
14944 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
14945 * fold-const.c (const_binop): Likewise.
14946 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
14947 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
14948 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
14949 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
14950 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
14951 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
14952 * expr.c (expand_expr_real_2): Likewise.
14953 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
14954 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
14955 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
14957 * optabs.c (expand_widen_pattern_expr): For
14958 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
14959 sign from result type rather than operand's type.
14960 (expand_binop_directly): For vec_packu_float_optab and
14961 vec_packs_float_optab allow result type to be different from operand's
14963 * optabs-tree.c (optab_for_tree_code): Handle
14964 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
14965 VEC_PACK_FLOAT_EXPR. Formatting fixes.
14966 * tree-vect-generic.c (expand_vector_operations_1): Handle
14967 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
14968 VEC_PACK_FLOAT_EXPR.
14969 * tree-vect-stmts.c (supportable_widening_operation): Handle
14971 (supportable_narrowing_operation): Handle FLOAT_EXPR.
14972 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
14973 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
14974 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
14975 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
14977 (vec_pack<floatprefix>_float_<mode>): New expander.
14978 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
14980 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
14981 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
14982 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
14983 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
14984 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
14986 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
14987 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
14988 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
14989 VEC_PACK_FLOAT_EXPR): Document.
14991 2018-05-29 Richard Biener <rguenther@suse.de>
14993 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
14995 (stmt_vec_info_vec): Make pointer.
14996 (init_stmt_vec_info_vec): Remove.
14997 (free_stmt_vec_info_vec): Likewise.
14998 (set_stmt_vec_info_vec): New function.
14999 (free_stmt_vec_infos): Likewise.
15000 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
15001 (set_vinfo_for_stmt): Likewise.
15002 (get_earlier_stmt): Likewise.
15003 (get_later_stmt): Likewise.
15004 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
15005 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
15006 (vec_info::~vec_info): Free stmt_vec_infos.
15007 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
15008 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
15009 (pass_slp_vectorize::execute): Likewise.
15010 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
15011 (free_stmt_vec_info_vec): Likewise.
15012 (set_stmt_vec_info_vec): New function.
15013 (free_stmt_vec_infos): Likewise.
15014 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
15015 the global stmt_vec_info_vec.
15016 * tree-parloops.c (gather_scalar_reductions): Use
15017 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
15020 2018-05-29 Richard Biener <rguenther@suse.de>
15022 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
15024 2018-05-29 Martin Liska <mliska@suse.cz>
15025 David Malcolm <dmalcolm@redhat.com>
15027 * vec.c (test_reverse): New.
15028 (vec_c_tests): Add new test.
15029 * vec.h (vl_ptr>::reverse): New function.
15031 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
15033 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
15035 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
15038 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
15040 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
15042 2018-05-28 Richard Biener <rguenther@suse.de>
15044 PR tree-optimization/85933
15045 * tree-vect-data-refs.c (vect_record_base_alignments): Only
15046 look at stmts marked as vectorizable.
15048 2018-05-28 Richard Biener <rguenther@suse.de>
15050 PR tree-optimization/85934
15051 * tree-vect-generic.c (expand_vector_operations_1): Hoist
15052 vector boolean check before scalar optimization.
15054 2018-05-28 Jakub Jelinek <jakub@redhat.com>
15056 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
15059 2018-05-28 Mark Wielaard <mark@klomp.org>
15061 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
15062 if it is an expression containing a minus sign.
15064 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
15066 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
15068 2018-05-27 Paul Koning <ni1d@arrl.net>
15070 * config/pdp11/pdp11.md (truncsihi2): Remove.
15072 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
15073 Chung-Ju Wu <jasonwucj@gmail.com>
15075 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
15077 (unaligned_store_dw): Ditto.
15078 * config/nds32/nds32-memory-manipulation.c
15079 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
15080 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
15081 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
15082 (emit_setmem_word_loop): Rename to ...
15083 (emit_setmem_doubleword_loop): ... this.
15084 (nds32_gen_dup_4_byte_to_word_value): New function.
15085 (nds32_gen_dup_8_byte_to_double_word_value): New function.
15086 (nds32_expand_setmem_loop): Refine implementation.
15087 (nds32_expand_setmem_loop_v3m): Ditto.
15088 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
15091 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
15093 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
15095 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
15097 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
15098 (nds32_init_machine_status): Initialize machine->attr_naked_p and
15099 machine->attr_no_prologue_p.
15100 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
15101 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
15102 (nds32_expand_epilogue): Consider attr_naked_p.
15103 (nds32_expand_epilogue_v3pop): Likewise.
15104 (nds32_can_use_return_insn): Likewise.
15105 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
15106 attr_no_prologue_p fields.
15107 * config/nds32/nds32.opt (mret-in-naked-func): New option.
15109 2018-05-27 Jakub Jelinek <jakub@redhat.com>
15112 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
15114 * config/i386/sse.md
15115 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
15117 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
15119 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
15121 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
15123 (*<floatsuffix>floatv2div2sf2): Rename to ...
15124 (*float<floatunssuffix>v2div2sf2): ... this.
15125 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
15126 (float<floatunssuffix>v2div2sf2_mask): ... this.
15127 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
15128 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
15129 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
15131 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
15133 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
15135 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
15137 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
15139 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
15141 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
15142 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
15143 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
15144 gen_ufix_truncv8dfv8si2.
15145 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
15146 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
15147 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
15148 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
15149 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
15150 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
15151 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
15152 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
15154 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
15158 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
15160 2018-05-25 Jim Wilson <jimw@sifive.com>
15162 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
15163 * config/riscv/riscv.c (struct machine_function): Add
15164 interrupt_handler_p and attribute_checked_p fields.
15165 (riscv_attribute_table): Add interrupt.
15166 (riscv_interrupt_type_p): New.
15167 (riscv_save_reg_p): Save extra regs for interrupt handler.
15168 (riscv_use_save_libcall): Return false for interrupt handler.
15169 (riscv_first_stack_step): Add forward declaration.
15170 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
15171 for interrupt handler with large frame. Use it for saved reg list.
15172 (riscv_expand_prologue): Move flag_stack_usage_info support to
15173 eliminate duplication.
15174 (riscv_expand_epilogue): Generate mret for interrupt handler.
15175 (riscv_epilogue_uses): New.
15176 (riscv_can_use_return_insn): Return false for interrupt handler.
15177 (riscv_function_ok_for_sibcall): Likewise.
15178 (riscv_set_current_function): Add interrupt handler support.
15179 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
15180 * config/riscv/riscv.md (UNSPECV_MRET): New.
15182 (riscv_frflags, riscv_fsflags): Use tab after opcode.
15184 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
15186 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
15188 PR tree-optimization/85712
15189 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
15190 this candidate has already been replaced in-situ by a copy.
15192 2018-05-25 Jason Merrill <jason@redhat.com>
15194 PR c++/80485 - inline function non-zero address.
15195 * symtab.c (nonzero_address): Check DECL_COMDAT.
15197 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
15200 * config/alpha/alpha.md (ashlsi3): New insn pattern.
15201 (*ashlsi_se): Rename from *ashldi_se. Define as sign
15202 extension of SImode operation. Use const123_operand predicate.
15203 (*saddsi_1): Remove.
15204 (*saddl_se_1): Ditto.
15205 (*ssubsi_1): Ditto.
15206 (*ssubl_se_1): Ditto.
15207 * config/alpha/predicates.md (const123_operand): New predicate.
15208 * config/alpha/constraints.md (P): Use IN_RANGE.
15210 2018-05-25 Richard Biener <rguenther@suse.de>
15212 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
15214 (ref_maybe_used_by_stmt_p): Likewise.
15215 (stmt_may_clobber_ref_p): Likewise.
15216 (stmt_may_clobber_ref_p_1): Likewise.
15217 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
15219 (ref_maybe_used_by_stmt_p): Likewise.
15220 (stmt_may_clobber_ref_p): Likewise.
15221 (stmt_may_clobber_ref_p_1): Likewise.
15222 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
15223 the alias oracle to disambiguate DRs with stmts DR analysis
15225 (vect_analyze_data_refs): Do not give up on not analyzable
15226 DRs for BB vectorization. Remove code truncating the dataref
15229 2018-05-25 Jakub Jelinek <jakub@redhat.com>
15232 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
15233 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
15234 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
15236 2018-05-25 Richard Biener <rguenther@suse.de>
15238 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
15239 function, combining stmt data ref gathering and fatal analysis
15241 (vect_analyze_data_refs): Remove now redudnant code and simplify.
15242 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
15243 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
15244 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
15245 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
15247 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15249 PR tree-optimization/85720
15250 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
15251 SCC if all partitions are builtins.
15252 (version_loop_by_alias_check): New parameter. Generate cancelable
15253 runtime alias check if all partitions are builtins.
15254 (distribute_loop): Update call to above function.
15256 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15258 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
15259 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
15260 (parm_default_def_partition_arg): Ditto.
15261 (set_parm_default_def_partition): Ditto.
15262 (get_parm_default_def_partitions): Ditto and make it static.
15263 (get_undefined_value_partitions): Ditto and make it static.
15264 (remove_ssa_form): Refactor call to init_var_map here.
15265 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
15266 computation for loop region.
15267 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
15268 (register_default_def): Delete.
15269 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
15270 (parm_default_def_partition_arg): Ditto.
15271 (set_parm_default_def_partition): Ditto.
15272 (get_parm_default_def_partitions): Ditto and make it static.
15273 (get_undefined_value_partitions): Ditto and make it static.
15274 (coalesce_with_default, coalesce_with_default): Update comment.
15275 (create_coalesce_list_for_region): New func factored out from
15276 create_outofssa_var_map.
15277 (populate_coalesce_list_for_outofssa): New func factored out from
15278 create_outofssa_var_map and coalesce_ssa_name.
15279 (create_outofssa_var_map): Delete.
15280 (coalesce_ssa_name): Refactor to support live range computation.
15281 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
15282 (get_parm_default_def_partitions): Delete.
15283 (get_undefined_value_partitions): Ditto.
15284 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
15285 computation for loop region.
15286 (new_tree_live_info, loe_visit_block): Ditto.
15287 (live_worklist, set_var_live_on_entry): Ditto.
15288 (calculate_live_on_exit, verify_live_on_entry): Ditto.
15289 * tree-ssa-live.h (struct _var_map): New fields.
15290 (init_var_map): Change decl.
15291 (region_contains_p): New.
15293 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15295 * tree-ssa-live.h (live_merge_and_clear): Delete.
15297 2018-05-25 Richard Biener <rguenther@suse.de>
15300 * tree-dump.c (dequeue_and_dump): Remove access to removed
15301 operand 2 of a SWITCH_EXPR.
15303 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15305 * doc/sourcebuild.texi (vect_double_cond_arith): Include
15306 multiplication and division.
15307 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
15308 (cond_udiv@var{m}, cond_umod@var{m}): Document.
15309 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
15310 (cond_udiv_optab, cond_umod_optab): New optabs.
15311 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
15312 (IFN_COND_RDIV): New internal functions.
15313 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
15314 TRUNC_MOD_EXPR and RDIV_EXPR.
15315 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
15316 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
15318 (SVE_INT_BINARY): Include mult.
15319 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
15320 (optab, sve_int_op): Handle mult.
15321 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
15323 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
15324 for SVE_INT_BINARY_SD.
15326 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15328 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
15329 (optab, sve_int_op): Handle div and udiv.
15330 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
15331 for SVE_INT_BINARY_SD.
15332 (*<optab><mode>3): New insn for the same.
15334 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15336 * tree-vect-patterns.c: Include predict.h.
15337 (vect_recog_divmod_pattern): Restrict check for division support
15338 to when optimizing for size.
15340 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15342 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
15343 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
15344 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
15345 (gimple_match_op::set_op): Likewise.
15346 (gimple_resimplify4): Declare.
15347 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
15348 (expr::gen_transform): Likewise.
15349 (decision_tree::gen): Generate a simplification routine for 4 operands.
15350 * gimple-match-head.c (gimple_simplify): Add an overload for
15351 4 operands. In the top-level function, handle up to 4 call
15352 arguments and call gimple_resimplify4.
15353 (gimple_resimplify4): New function.
15354 (build_call_internal): Pass a fourth operand.
15355 (maybe_push_to_seq): Likewise.
15356 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
15357 Fold VEC_COND_EXPRs of an operation and a default value into
15358 an IFN_COND_* function if possible.
15359 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
15361 (SVE_COND_FP_BINARY): Include them.
15362 (optab, sve_fp_op): Handle them.
15363 (SVE_INT_BINARY_REV): New code iterator.
15364 (SVE_COND_FP_BINARY_REV): New int iterator.
15365 (commutative): New int attribute.
15366 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
15368 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
15370 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
15371 (*cond_<optab><mode>): New patterns for reversed operands.
15373 2018-05-25 Richard Biener <rguenther@suse.de>
15375 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
15376 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
15377 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
15378 (STMT_VINFO_GROUPED_ACCESS): Adjust.
15379 * tree-vect-data-refs.c (everywhere): Adjust users.
15380 * tree-vect-loop.c (everywhere): Likewise.
15381 * tree-vect-slp.c (everywhere): Likewise.
15382 * tree-vect-stmts.c (everywhere): Likewise.
15383 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
15385 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15387 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
15389 (gcc_cv_as_section_exclude): ... this.
15390 Try Solaris as #exclude syntax.
15391 * configure: Regenerate.
15392 * config.in: Regenerate.
15393 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
15395 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
15396 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
15398 * varasm.c (default_elf_asm_named_section): Don't check if
15399 HAVE_GAS_SECTION_EXCLUDE is defined.
15401 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15403 * doc/md.texi: Update the documentation of the cond_* optabs
15404 to mention the new final operand. Fix GET_MODE_NUNITS call.
15405 Describe the scalar case too.
15406 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
15407 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
15409 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
15410 (get_conditional_internal_fn): Update comment.
15411 * tree-vect-loop.c (vectorizable_reduction): Pass the original
15412 accumulator value as a final argument to conditional functions.
15413 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
15414 a define_expand and add an "else" operand. Assert for now that
15415 the else operand is equal to operand 2. Use SVE_INT_BINARY and
15416 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
15417 (*cond_<optab><mode>): New patterns.
15418 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
15419 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
15420 (UNSPEC_COND_EOR): Delete.
15421 (optab): Remove associated mappings.
15422 (SVE_INT_BINARY): New code iterator.
15423 (sve_int_op): Remove int attribute and add "minus" to the code
15425 (SVE_COND_INT_OP): Delete.
15426 (SVE_COND_FP_OP): Rename to...
15427 (SVE_COND_FP_BINARY): ...this.
15429 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15431 * optabs.c (can_reuse_operands_p): New function.
15432 (maybe_legitimize_operands): Try to reuse the results for
15435 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
15437 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
15438 Add {q} suffix to insn mnemonic.
15440 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
15442 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
15443 (msp430_warn_func_return): New.
15445 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
15447 * fold-const.c (tree_nonzero_bits): New function.
15448 * fold-const.h (tree_nonzero_bits): Likewise.
15449 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
15450 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
15452 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
15456 * varasm.c (assemble_alias): Check ifunc_resolver only on
15459 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
15462 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
15463 when memory input operand is handled.
15465 2018-05-24 Luis Machado <luis.machado@linaro.org>
15467 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
15469 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
15471 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
15473 * match.pd: Delay FMA folds until after vectorization.
15475 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
15478 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
15479 address check not strict.
15481 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
15483 * gimple-match.h (gimple_match_op): New class.
15484 (mprts_hook): Replace parameters with a gimple_match_op *.
15485 (maybe_build_generic_op): Likewise.
15486 (gimple_simplified_result_is_gimple_val): Replace parameters with
15487 a const gimple_match_op *.
15488 (gimple_simplify): Replace code_helper * and tree * parameters with
15489 a gimple_match_op * parameter.
15490 (gimple_resimplify1): Replace code_helper *, tree and tree *
15491 parameters with a gimple_match_op * parameter.
15492 (gimple_resimplify2): Likewise.
15493 (gimple_resimplify3): Likewise.
15494 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
15495 parameters with a gimple_match_op * parameter.
15496 * gimple-match-head.c (gimple_simplify): Change prototypes of
15497 auto-generated functions to take a gimple_match_op * instead of
15498 separate code_helper * and tree * parameters. Make the same
15499 change in the top-level overload and update calls to the
15500 gimple_resimplify routines. Update calls to the auto-generated
15501 functions and to maybe_push_res_to_seq in the publicly-facing
15502 operation-specific gimple_simplify overloads.
15503 (gimple_match_op::MAX_NUM_OPS): Define.
15504 (gimple_resimplify1): Replace rcode and ops with a single res_op
15505 parameter. Update call to gimple_simplify.
15506 (gimple_resimplify2): Likewise.
15507 (gimple_resimplify3): Likewise.
15508 (mprts_hook): Replace parameters with a gimple_match_op *.
15509 (maybe_build_generic_op): Likewise.
15510 (build_call_internal): Replace type, nargs and ops with
15511 a gimple_match_op *.
15512 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
15513 with a single gimple_match_op *. Update calls to mprts_hook,
15514 build_call_internal and gimple_simplified_result_is_gimple_val.
15515 Factor out code that is common to the tree_code and combined_fn cases.
15516 * genmatch.c (expr::gen_transform): Replace tem_code and
15517 tem_ops with a gimple_match_op called tem_op. Update calls
15518 to the gimple_resimplify functions and maybe_push_res_to_seq.
15519 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
15520 res_ops. Update call to the gimple_resimplify functions.
15521 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
15522 (decision_tree::gen): Make the functions take a gimple_match_op *
15523 called res_op instead of separate res_code and res_ops parameters.
15524 Update call accordingly.
15525 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
15526 and ops with a single res_op parameter. Update calls to
15527 maybe_build_generic_op and maybe_push_res_to_seq.
15528 (fold_stmt_1): Update calls to gimple_simplify and
15529 replace_stmt_with_simplification.
15530 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
15531 and gimple_simplified_result_is_gimple_val.
15532 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
15534 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
15535 with a gimple_match_op *.
15536 (vn_nary_build_or_lookup): Likewise. Update call to
15537 vn_nary_build_or_lookup_1.
15538 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
15539 gimple_match_op *. Update calls to the gimple_resimplify routines
15540 and to gimple_simplified_result_is_gimple_val.
15541 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
15542 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
15543 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
15544 (visit_nary_op): Likewise.
15545 (visit_reference_op_load): Likewise.
15547 2018-05-23 Luis Machado <luis.machado@linaro.org>
15549 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
15550 modifier for printing the step amount.
15552 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15555 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
15558 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
15560 * doc/sourcebuild.texi (Endianness): New subsubsection.
15562 2018-05-23 Luis Machado <luis.machado@linaro.org>
15564 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
15565 <prefetch_dynamic_strides>: New const bool field.
15566 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
15567 prefetch_dynamic_strides.
15568 (exynosm1_prefetch_tune): Likewise.
15569 (thunderxt88_prefetch_tune): Likewise.
15570 (thunderx_prefetch_tune): Likewise.
15571 (thunderx2t99_prefetch_tune): Likewise.
15572 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
15574 (aarch64_override_options_internal): Update to set
15575 PARAM_PREFETCH_DYNAMIC_STRIDES.
15576 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
15577 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
15578 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
15579 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
15580 prefetch-dynamic-strides setting.
15582 2018-05-23 Luis Machado <luis.machado@linaro.org>
15584 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
15585 <minimum_stride>: New const int field.
15586 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
15587 minimum_stride field defaulting to -1.
15588 (exynosm1_prefetch_tune): Likewise.
15589 (thunderxt88_prefetch_tune): Likewise.
15590 (thunderx_prefetch_tune): Likewise.
15591 (thunderx2t99_prefetch_tune): Likewise.
15592 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
15593 <default_opt_level>: Set to 3.
15594 (aarch64_override_options_internal): Update to set
15595 PARAM_PREFETCH_MINIMUM_STRIDE.
15596 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
15597 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
15598 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
15599 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
15600 stride is constant and is below the minimum stride threshold.
15602 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15604 * config/arm/arm-cpus.in (mode26): Delete.
15605 (armv4): Delete mode26 reference.
15606 * config/arm/arm.c (arm_configure_build_target): Delete use of
15609 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
15611 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
15613 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
15614 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
15616 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
15617 (floatunsdidf2): Ditto.
15619 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
15621 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
15622 (fixuns_trunc<mode>si2_avx512f): Ditto.
15623 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
15624 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
15625 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
15627 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
15629 PR rtl-optimization/79985
15630 * df-scan.c (df_insn_refs_collect): Remove special case for
15631 global registers and asm statements.
15633 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
15635 * extend.texi (Global Register Variables): Rewrite the bullet list.
15636 Note that the register is available for allocation. Note that access
15637 via inline asm must use constraints. Add note about async-signal
15638 handlers. Remove paragraph about automagic register selection.
15640 2018-05-23 Richard Biener <rguenther@suse.de>
15642 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
15643 of fixed offset from memset VN.
15645 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
15647 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
15648 first_interp field.
15649 (alloc_cand_and_find_basis): Initialize first_interp field.
15650 (slsr_process_mul): Modify first_interp field.
15651 (slsr_process_add): Likewise.
15652 (slsr_process_cast): Modify first_interp field for each new
15654 (slsr_process_copy): Likewise.
15655 (dump_candidate): Dump first_interp field.
15656 (replace_mult_candidate): Process all interpretations, not just
15658 (replace_rhs_if_not_dup): Likewise.
15659 (replace_one_candidate): Likewise.
15661 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
15663 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
15665 (aarch64_needs_frame_chain): New function.
15666 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
15668 2018-05-23 Sudakshina Das <sudi.das@arm.com>
15671 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
15672 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
15673 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
15674 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
15675 as true for strict-align.
15676 (aarch64_can_inline_p): Perform checks even when callee has no
15677 attributes to check for strict alignment.
15678 * doc/extend.texi (AArch64 Function Attributes): Document
15680 * doc/invoke.texi: (AArch64 Options): Likewise.
15682 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
15684 PR tree-optimization/85853
15685 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
15686 the handling of the root of the node to...
15687 (vect_slp_analyze_node_operations_1): ...this new function,
15688 and run the whole thing with the child nodes' def types
15689 set according to their SLP node's def type.
15691 2018-05-23 Richard Biener <rguenther@suse.de>
15693 PR middle-end/85874
15694 * tree-data-ref.c (create_runtime_alias_checks): Defer
15695 and ignore overflow warnings.
15697 2018-05-23 Yury Gribov <tetra2005@gmail.com>
15699 PR tree-optimization/85822
15700 * tree-vrp.c (is_masked_range_test): Fix handling of negative
15703 2018-05-23 Richard Biener <rguenther@suse.de>
15705 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
15706 memset constants via native_interpret_expr.
15708 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
15711 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
15713 (cgraph_node::create_alias): Likewise.
15714 (cgraph_node::get_availability): Check ifunc_resolver instead
15715 of looking up ifunc attribute.
15716 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
15717 * varasm.c (do_assemble_alias): Likewise.
15718 (assemble_alias): Likewise.
15719 (default_binds_local_p_3): Likewise.
15720 * cgraph.h (cgraph_node): Add ifunc_resolver.
15721 (cgraph_node::only_called_directly_or_aliased_p): Return false
15722 for IFUNC resolver.
15723 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
15725 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
15726 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
15727 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
15728 instead of looking up ifunc attribute.
15730 2018-05-22 Luis Machado <luis.machado@linaro.org>
15732 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
15734 2018-05-22 Martin Sebor <msebor@redhat.com>
15736 PR middle-end/85359
15737 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
15738 only when expasion succeeds.
15739 (expand_builtin_strcmp): Same.
15740 (expand_builtin_strncmp): Same.
15742 2018-05-22 Martin Sebor <msebor@redhat.com>
15744 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
15746 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
15747 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15749 * config/aarch64/aarch64-ldpstp.md: Replace uses of
15750 aarch64_mem_pair_operand with memory_operand and delete operand swapping
15752 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
15753 Add check for legitimate_address.
15754 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
15755 (aarch64_swap_ldrstr_operands): New.
15756 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
15759 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
15760 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15762 * config/aarch64/aarch64.md: New patterns to generate stp
15764 (store_pair_sw, store_pair_dw): New patterns to generate stp for
15765 single words and double words.
15766 (load_pair_sw, load_pair_dw): Likewise.
15767 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
15769 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
15771 * config/aarch64/aarch64-ldpstp.md: Modify peephole
15772 for different mode ldpstp and add peephole for merged zero stores.
15773 Likewise for loads.
15774 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
15776 (aarch64_gen_store_pair): Rename calls to match new patterns.
15777 (aarch64_gen_load_pair): Rename calls to match new patterns.
15778 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
15779 (load_pair<DREG:mode><DREG2:mode>): ... This.
15780 (store_pair<mode>): Rename to...
15781 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
15782 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
15783 New mode iterators.
15784 (V_INT_EQUIV): Handle SImode.
15785 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
15788 2018-05-22 Martin Sebor <msebor@redhat.com>
15791 * calls.c (maybe_warn_nonstring_arg): Use string length to set
15792 or ajust the presumed bound on an operation to avoid unnecessary
15795 2018-05-22 Martin Sebor <msebor@redhat.com>
15797 PR tree-optimization/85826
15798 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
15799 assuming that a DECL necesarily has a constant size.
15801 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
15803 PR middle-end/85862
15804 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
15806 2018-05-22 Richard Biener <rguenther@suse.de>
15808 PR tree-optimization/85834
15809 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
15810 non-constant and non-zero memset arguments.
15812 2018-05-22 Martin Liska <mliska@suse.cz>
15815 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
15817 2018-05-22 Richard Biener <rguenther@suse.de>
15819 PR tree-optimization/85863
15820 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
15821 comparisons when vectype is specified.
15822 (vectorizable_condition): Do not specify vectype for
15823 vect_is_simple_cond when SLP vectorizing.
15825 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
15828 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
15829 define __ibm128 as long double.
15830 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
15831 as a distinct type when IEEE 128-bit support is enabled.
15832 (init_float128_ieee): Fix up conversions between IFmode and IEEE
15833 128-bit types to use the correct functions.
15834 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
15835 convert between 128-bit floating point types that have different
15836 modes but the same representation, instead of using gen_lowpart to
15838 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
15840 (IFKF_reg): New attributes to give the register constraints for
15842 (extend<mode>tf2_internal): New insns to mark an explicit
15843 conversion between 128-bit floating point types that have a
15844 different mode but share the same representation.
15846 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
15848 PR tree-optimization/85814
15849 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
15850 a null return from get_strinfo when unsharing the next
15851 strinfo in the chain.
15853 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
15856 * varasm.c (weak_finish): Clean up weak_decls.
15858 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15860 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
15861 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
15862 UNSPEC_UADALP values.
15863 * config/aarch64/iterators.md (ABAL): New int iterator.
15866 (sur): Add mappings for the above.
15867 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
15869 (aarch64_<sur>abal<mode>_4): Likewise.
15870 (aarch64_<sur>adalp<mode>_3): Likewise.
15871 (<sur>sadv16qi): New define_expand.
15873 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
15875 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
15876 (*movdf_internal): Ditto.
15877 (*rcpsf2_sse): Ditto.
15878 (*rsqrtsf2_sse): Ditto.
15879 (*sqrt<mode>2_sse): Ditto.
15881 2018-05-21 Tamar Christina <tamar.christina@arm.com>
15883 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
15885 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
15886 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
15887 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
15888 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
15890 * config/aarch64/arm_neon.h: Likewise.
15891 * config/aarch64/iterators.md (VQ_I): New.
15893 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
15895 * config.gcc: Add arc/t-multilib-linux to tmake_file for
15897 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
15900 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
15902 * config/nds32/constraints.md (S): New constraint.
15903 * config/nds32/nds32.md (call_internal): Use constraint S.
15904 (call_value_internal): Likewise.
15905 (sibcall_internal): Likewise.
15906 (sibcall_value_internal): Likewise.
15908 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
15909 Chung-Ju Wu <jasonwucj@gmail.com>
15911 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
15912 into consideration.
15914 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
15915 Chung-Ju Wu <jasonwucj@gmail.com>
15917 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
15918 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
15919 (nds32_rtx_costs_impl): Simplify.
15920 (nds32_address_cost_impl): Simplify.
15921 (nds32_init_rtx_costs): New function.
15922 (nds32_rtx_costs_speed_prefer): Likewise.
15923 (nds32_rtx_costs_size_prefer): Likewise.
15924 (nds32_address_cost_speed_prefer): Likewise.
15925 (nds32_address_cost_speed_fwprop): Likewise.
15926 (nds32_address_cost_size_prefer): Likewise.
15927 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
15928 * config/nds32/nds32.c (nds32_option_override): Use
15929 nds32_init_rtx_costs function.
15931 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
15933 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
15934 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
15935 (TARGET_PIPELINE_N8): Likewise.
15936 (TARGET_PIPELINE_N10): Likewise.
15937 (TARGET_PIPELINE_N13): Likewise.
15938 (TARGET_PIPELINE_GRAYWOLF): Likewise.
15940 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
15942 * config/nds32/nds32-fpu.md: Update copyright year.
15944 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
15946 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
15948 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
15950 * config/nds32/nds32.c
15951 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
15952 * config/nds32/nds32.opt (minline-asm-r15): New option.
15954 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
15956 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
15958 * config/nds32/nds32.md (abssi2): New pattern.
15960 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
15962 * config/i386/i386.md (rex64namesuffix): New mode attribute.
15963 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
15964 Merge insn pattern from sse_cvtsi2ss<round_name> and
15965 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
15966 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
15967 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
15968 using SWI48 mode iterator.
15969 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
15970 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
15971 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
15972 pattern from sse_cvttss2si<round_saeonly_name>
15973 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
15974 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
15975 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
15976 using SWI48 mode iterator.
15977 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
15978 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
15979 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
15980 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
15981 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
15982 using SWI48 mode iterator.
15983 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
15984 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
15985 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
15986 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
15987 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
15988 SWI48 mode iterator.
15989 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
15990 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
15991 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
15992 pattern from sse_cvttsd2si<round_saeonly_name>
15993 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
15995 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
15997 * config/nds32/nds32-md-auxiliary.c
15998 (nds32_valid_smw_lwm_base_p): Refine.
15999 (nds32_output_smw_single_word): Refine.
16000 (nds32_output_smw_double_word): New.
16001 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
16003 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16005 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
16006 (nds32_output_stack_pop): Refine.
16007 (nds32_expand_unaligned_load): Refine.
16008 (nds32_expand_unaligned_store): Refine.
16010 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16011 Chung-Ju Wu <jasonwucj@gmail.com>
16013 * config/nds32/constants.md: Add TP_REGNUM constant.
16014 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
16015 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
16017 * config/nds32/nds32-doubleword.md: Consider flag_pic.
16018 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
16019 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
16020 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
16021 and PIC code generation.
16022 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
16024 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
16026 * config/nds32/nds32.md: Support TLS and PIC.
16027 * config/nds32/nds32.c: Support TLS and PIC.
16028 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
16029 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
16032 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16034 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
16035 mode with E_ prefix.
16037 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16038 Chung-Ju Wu <jasonwucj@gmail.com>
16040 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
16041 * config/nds32/nds32-md-auxiliary.c
16042 (symbolic_reference_mentioned_p): New.
16043 (nds32_legitimize_ict_address): New.
16044 (nds32_expand_ict_move): New.
16045 (nds32_indirect_call_referenced_p): New.
16046 (nds32_symbol_binds_local_p): Delete.
16047 (nds32_long_call_p): Modify.
16048 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
16049 * config/nds32/nds32-protos.h
16050 (symbolic_reference_mentioned_p): Declare.
16051 (nds32_legitimize_ict_address): Declare.
16052 (nds32_expand_ict_move): Declare.
16053 (nds32_indirect_call_referenced_p): Declare.
16054 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
16055 (nds32_relax_group): Use nds32_ict_const_p as condition.
16056 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
16057 (nds32_asm_file_start): Output ict_model directive in asm code.
16058 (nds32_legitimate_address_p): Consider indirect call.
16059 (nds32_print_operand): Consider indirect call.
16060 (nds32_print_operand_address): Consider indirect call.
16061 (nds32_insert_attributes): Handle "indirect_call" attribute.
16062 (TARGET_LEGITIMATE_ADDRESS_P): Define.
16063 (TARGET_LEGITIMATE_CONSTANT_P): Define.
16064 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
16065 (TARGET_DELEGITIMIZE_ADDRESS): Define.
16066 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
16067 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
16068 (TARGET_ICT_MODEL_SMALL): Define.
16069 (TARGET_ICT_MODEL_LARGE): Define.
16070 * config/nds32/nds32.md (movsi): Consider ict model.
16071 (call, call_value): Consider ict model.
16072 (sibcall, sibcall_value): Consider ict model.
16073 * config/nds32/nds32.opt (mict-model): New option.
16074 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
16077 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
16078 Monk Chiang <sh.chiang04@gmail.com>
16079 Jim Wilson <jimw@sifive.com>
16081 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
16082 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
16083 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
16084 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
16085 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
16086 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
16087 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
16088 compute save_libcall_adjustment properly.
16089 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
16090 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
16091 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
16092 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
16093 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
16094 (ABI_SPEC): Handle mabi=ilp32e.
16095 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
16096 (RVE): Add RVE mask.
16097 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
16098 <-march>: Add rv32e as an example.
16100 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
16103 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
16104 (intra_create_variable_infos): Handle C++ constructors.
16106 2018-05-18 Martin Liska <mliska@suse.cz>
16108 * passes.def: Remove a redundant pass.
16110 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
16113 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
16115 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16117 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
16119 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
16121 (armv2, armv2a, armv3, armv3m): Delete architectures.
16122 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
16123 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
16124 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
16126 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
16127 (*mulsidi3adddi): Likewise.
16128 (mulsidi3): Likewise.
16129 (*mulsidi3_nov6): Likewise.
16130 (umulsidi3): Likewise.
16131 (umulsidi3_nov6): Likewise.
16132 (umaddsidi4): Likewise.
16133 (*umulsidi3adddi): Likewise.
16134 (smulsi3_highpart): Likewise.
16135 (*smulsi3_highpart_nov6): Likewise.
16136 (umulsi3_highpart): Likewise.
16137 (*umulsi3_highpart_nov6): Likewise.
16138 * config/arm/arm.h (arm_arch3m): Delete.
16139 * config/arm/arm.c (arm_arch3m): Delete.
16140 (arm_option_override_internal): Update armv3-related comment.
16141 (arm_configure_build_target): Delete use of isa_bit_mode32.
16142 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
16143 (arm_rtx_costs_internal): Delete check of arm_arch3m.
16144 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
16145 (mulsa3): Likewise.
16146 (mulusa3): Likewise.
16147 * config/arm/arm-protos.h (arm_arch3m): Delete.
16148 * config/arm/arm-tables.opt: Regenerate.
16149 * config/arm/arm-tune.md: Likewise.
16150 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
16151 deleted architectures.
16153 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16155 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
16156 (armv5t, armv5te): New features.
16157 (ARMv5, ARMv5e): Delete fgroups.
16158 (ARMv5t, ARMv5te): Adjust for above changes.
16159 (ARMv6m): Likewise.
16160 (armv5, armv5e): Delete arches.
16161 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
16163 (*call_reg_arm): Likewise.
16164 (*call_value_reg_armv5): Likewise.
16165 (*call_value_reg_arm): Likewise.
16166 (*call_symbol): Likewise.
16167 (*call_value_symbol): Likewise.
16168 (*sibcall_insn): Likewise.
16169 (*sibcall_value_insn): Likewise.
16170 (clzsi2): Likewise.
16171 (prefetch): Likewise.
16172 (define_split and define_peephole2 dependent on arm_arch5):
16174 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
16176 (TARGET_ARM_QBIT): Likewise.
16177 (TARGET_DSP_MULTIPLY): Likewise.
16178 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
16179 (arm_arch5, arm_arch5e): Delete.
16180 (arm_arch5t, arm_arch5te): Declare.
16181 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
16182 (arm_arch5t): Declare.
16183 (arm_option_reconfigure_globals): Update for the above.
16184 (arm_options_perform_arch_sanity_checks): Update comment, replace
16185 use of arm_arch5 with arm_arch5t.
16186 (use_return_insn): Likewise.
16187 (arm_emit_call_insn): Likewise.
16188 (output_return_instruction): Likewise.
16189 (arm_final_prescan_insn): Likewise.
16190 (arm_coproc_builtin_available): Likewise.
16191 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
16192 arm_arch5e with arm_arch5t and arm_arch5te.
16193 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
16194 (arm_arch5t, arm_arch5te): Declare.
16195 * config/arm/arm-tables.opt: Regenerate.
16196 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
16197 * config/arm/t-multilib: Likewise.
16198 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
16199 instead of arm_arch5.
16200 (*call_reg_thumb1): Likewise.
16201 (*call_value_reg_thumb1_v5): Likewise.
16202 (*call_value_reg_thumb1): Likewise.
16203 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
16205 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
16207 2018-05-18 Martin Liska <mliska@suse.cz>
16209 PR gcov-profile/84846
16210 * doc/gcov.texi: Document -t option of gcov tool.
16212 2018-05-18 Martin Liska <mliska@suse.cz>
16214 PR gcov-profile/84846
16215 * gcov.c (print_usage): Add new -t option.
16216 (process_args): Handle the option.
16217 (generate_results): Use stdout as output when requested by
16220 2018-05-18 Martin Liska <mliska@suse.cz>
16222 PR gcov-profile/84846
16223 * coverage.c (coverage_init): Write PWD to .gcno file.
16224 * doc/gcov.texi: Document how working directory is printed.
16225 * gcov-dump.c (dump_gcov_file): Print PWD.
16226 * gcov.c (output_intermediate_file): Likewise.
16227 (read_graph_file): Read PWD string.
16228 (output_lines): Print PWD.
16230 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16232 PR middle-end/85817
16233 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
16234 for retval and return false if all args to phi are zero.
16236 2018-05-18 Richard Biener <rguenther@suse.de>
16238 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
16240 (evrp_dom_walker::before_dom_children): Call it.
16242 2018-05-18 Richard Biener <rguenther@suse.de>
16244 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
16245 results when processing array refs with variable index.
16247 2018-05-18 Toon Moene <toon@moene.org>
16249 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
16250 directly after that of -floop-interchange. Indicate that both
16251 options are enabled by default when specifying -O3.
16253 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16255 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
16256 iterator. Delete separate integer-mode vec_set<mode> expander.
16257 (aarch64_simd_vec_setv2di): Delete.
16258 (vec_setv2di): Delete.
16259 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
16260 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
16261 the "w, r" alternative.
16263 2018-05-18 Martin Liska <mliska@suse.cz>
16265 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
16266 * tree-pass.h (make_pass_lower_switch_O0): New function.
16267 * tree-switch-conversion.c (node_has_low_bound): Remove.
16268 (node_has_high_bound): Likewise.
16269 (node_is_bounded): Likewise.
16270 (class pass_lower_switch): Make it a template type and create
16272 (pass_lower_switch::execute): Add template argument.
16273 (make_pass_lower_switch): New function.
16274 (make_pass_lower_switch_O0): New function.
16275 (do_jump_if_equal): Remove.
16276 (emit_case_nodes): Simplify to just handle all 3 cases and leave
16277 all the hard work to tree optimization passes.
16279 2018-05-18 Martin Liska <mliska@suse.cz>
16281 * dbgcnt.c (limit_low): Renamed from limit.
16282 (limit_high): New variable.
16283 (dbg_cnt_is_enabled): Check for upper limit.
16284 (dbg_cnt): Adjust dumping.
16285 (dbg_cnt_set_limit_by_index): Add new argument for high
16287 (dbg_cnt_set_limit_by_name): Likewise.
16288 (dbg_cnt_process_single_pair): Parse new format.
16289 (dbg_cnt_process_opt): Use strtok.
16290 (dbg_cnt_list_all_counters): Remove 'value' and add
16292 * doc/invoke.texi: Document changes.
16294 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
16296 * doc/sourcebuild.texi (scalar_all_fma): Document.
16297 * tree.def (FMA_EXPR): Delete.
16298 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
16299 * internal-fn.c (ternary_direct): New macro.
16300 (expand_ternary_optab_fn): Likewise.
16301 (direct_ternary_optab_supported_p): Likewise.
16302 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
16303 * builtins.c (fold_builtin_fma): Delete.
16304 (fold_builtin_3): Don't call it.
16305 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
16306 * expr.c (expand_expr_real_2): Likewise.
16307 * fold-const.c (operand_equal_p): Likewise.
16308 (fold_ternary_loc): Likewise.
16309 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
16310 * gimple.c (DEFTREECODE): Likewise.
16311 * gimplify.c (gimplify_expr): Likewise.
16312 * optabs-tree.c (optab_for_tree_code): Likewise.
16313 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16314 * tree-eh.c (operation_could_trap_p): Likewise.
16315 (stmt_could_throw_1_p): Likewise.
16316 * tree-inline.c (estimate_operator_cost): Likewise.
16317 * tree-pretty-print.c (dump_generic_node): Likewise.
16318 (op_code_prio): Likewise.
16319 * tree-ssa-loop-im.c (stmt_cost): Likewise.
16320 * tree-ssa-operands.c (get_expr_operands): Likewise.
16321 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
16322 * fold-const-call.h (fold_fma): Delete.
16323 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
16324 CFN_FNMA and CFN_FNMS.
16325 (fold_fma): Delete.
16326 * genmatch.c (combined_fn): New enum.
16327 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
16328 (commutative_op): New function.
16329 (commutate): Use it. Handle more than 2 operands.
16330 (dt_operand::gen_gimple_expr): Use commutative_op.
16331 (parser::parse_expr): Allow :c to be used with non-binary
16332 operators if the commutative operand is known.
16333 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
16334 CFN_FMS, CFN_FNMA and CFN_FNMS.
16335 (backprop::process_assign_use): Remove FMA_EXPR handling.
16336 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
16337 (gen_hsa_fma): New function.
16338 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
16339 IFN_FNMA and IFN_FNMS.
16340 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
16341 * gimple-fold.h (follow_all_ssa_edges): Declare.
16342 * gimple-fold.c (follow_all_ssa_edges): New function.
16343 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
16344 gimple_build interface and use follow_all_ssa_edges to fold the result.
16345 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
16346 instead of checking for optabs directly.
16347 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
16348 rather than FMA_EXPRs.
16349 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
16350 call to IFN_FMA instead of an FMA_EXPR.
16352 2018-05-17 Jim Wilson <jimw@sifive.com>
16354 * expr.c (do_tablejump): When converting index to Pmode, if we have a
16355 sign extended promoted subreg, and the range does not have the sign bit
16356 set, then do a sign extend.
16358 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
16359 test, check for sign extended subreg and/or constant operands, and
16360 do a sign extend in that case.
16362 2018-05-17 Steve Ellcey <sellcey@cavium.com>
16364 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
16365 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
16367 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
16368 Change logics_shift_reg to logics_shift_imm.
16369 (thunderx2t99_fp_loadpair_basic): Delete.
16370 (thunderx2t99_fp_storepair_basic): Delete.
16371 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
16372 (thunderx2t99_asimd_polynomial): Delete.
16373 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
16374 and neon_fp_mul_d_scalar_q.
16375 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
16376 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
16377 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
16378 (thunderx2t99_asimd_lut): Add missing tbl types.
16379 (thunderx2t99_asimd_ext): Delete.
16380 (thunderx2t99_asimd_load1_1_mult): Delete.
16381 (thunderx2t99_asimd_load1_2_mult): Delete.
16382 (thunderx2t99_asimd_load1_ldp): New.
16383 (thunderx2t99_asimd_load1): New.
16384 (thunderx2t99_asimd_load2): Add missing *load2* types.
16385 (thunderx2t99_asimd_load3): New.
16386 (thunderx2t99_asimd_load4): New.
16387 (thunderx2t99_asimd_store1_1_mult): Delete.
16388 (thunderx2t99_asimd_store1_2_mult): Delete.
16389 (thunderx2t99_asimd_store2_mult): Delete.
16390 (thunderx2t99_asimd_store2_onelane): Delete.
16391 (thunderx2t99_asimd_store_stp): New.
16392 (thunderx2t99_asimd_store1): New.
16393 (thunderx2t99_asimd_store2): New.
16394 (thunderx2t99_asimd_store3): New.
16395 (thunderx2t99_asimd_store4): New.
16397 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
16399 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
16400 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
16402 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
16403 Segher Boessenkool <segher@kernel.crashing.org>
16406 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
16409 2018-05-17 Richard Biener <rguenther@suse.de>
16411 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
16412 for pruning loop and prune defs feeding only already visited PHIs.
16414 2018-05-17 Richard Biener <rguenther@suse.de>
16416 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
16418 2018-05-17 Bin Cheng <bin.cheng@arm.com>
16419 Richard Biener <rguenther@suse.de>
16421 PR tree-optimization/85793
16422 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
16423 for VMAT_ELEMENTWISE.
16425 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
16427 * internal-fn.h (lookup_internal_fn): Declare
16428 * internal-fn.c (lookup_internal_fn): New function.
16429 * gimple.c (gimple_build_call_from_tree): Handle calls to
16430 internal functions.
16431 * gimple-pretty-print.c (dump_gimple_call): Print "." before
16432 internal function names.
16433 * tree-pretty-print.c (dump_generic_node): Likewise.
16434 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
16436 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
16438 * gimple-fold.h (gimple_build): Make the function forms take
16439 combined_fn rather than built_in_function.
16440 (gimple_simplify): Likewise.
16441 * gimple-match-head.c (gimple_simplify): Likewise.
16442 * gimple-fold.c (gimple_build): Likewise.
16443 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
16444 rather than gimple_build_call_internal.
16445 (get_initial_defs_for_reduction): Likewise.
16446 (vect_create_epilog_for_reduction): Likewise.
16447 (vectorizable_live_operation): Likewise.
16449 2018-05-17 Martin Liska <mliska@suse.cz>
16451 * gimple-ssa-sprintf.c (format_directive): Do not use
16452 space in between 'G_' and '('.
16454 2018-05-17 Jakub Jelinek <jakub@redhat.com>
16457 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
16458 even if the mask is not all ones.
16461 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
16463 (ix86_gimple_fold_builtin): Likewise.
16466 * config/i386/i386.c: Include tree-vector-builder.h.
16467 (ix86_vector_shift_count): New function.
16468 (ix86_fold_builtin): Fold shift builtins by scalar count.
16469 (ix86_gimple_fold_builtin): Likewise.
16471 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
16472 _mm512_setzero): New intrinsics.
16474 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
16475 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16477 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
16478 code generation for cases where splatting a value is not useful.
16479 * simplify-rtx.c (simplify_ternary_operation): Simplify
16480 vec_merge across a vec_duplicate and a paradoxical subreg forming
16481 a vector mode to a vec_concat.
16483 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
16485 * config.gcc: Support "goldmont-plus".
16486 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
16488 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
16489 PROCESSOR_GOLDMONT_PLUS.
16490 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
16491 (processor_target_table): Add "goldmont-plus".
16492 (PTA_GOLDMONT_PLUS): Define.
16493 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
16494 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
16495 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
16496 (fold_builtin_cpu): Add "goldmont-plus".
16497 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
16498 (ix86_option_override_internal): Add "goldmont-plus".
16499 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
16500 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
16501 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
16502 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
16504 2018-05-17 Richard Biener <rguenther@suse.de>
16506 PR tree-optimization/85757
16507 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
16508 remove defs that only feed that PHI from further processing.
16510 2018-05-16 Jim Wilson <jimw@sifive.com>
16512 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
16514 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
16516 2018-05-16 Mark Wielaard <mark@klomp.org>
16518 * dwarf2out.c (count_index_strings): New function.
16519 (output_indirect_strings): Call count_index_strings and generate
16520 header for dwarf_version >= 5.
16522 2018-05-16 Mark Wielaard <mark@klomp.org>
16524 * dwarf2out.c (dwarf_FORM): New function.
16525 (set_indirect_string): Use dwarf_FORM.
16526 (reset_indirect_string): Likewise.
16527 (size_of_die): Likewise.
16528 (value_format): Likewise.
16529 (output_die): Likewise.
16530 (add_skeleton_AT_string): Likewise.
16531 (output_macinfo_op): Likewise.
16532 (index_string): Likewise.
16533 (output_index_string_offset): Likewise.
16534 (output_index_string): Likewise.
16535 (count_index_strings): Likewise.
16537 2018-05-16 Carl Love <cel@us.ibm.com>
16539 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
16540 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
16542 2018-05-16 Martin Jambor <mjambor@suse.cz>
16544 * ipa-prop.c (ipa_free_all_edge_args): Remove.
16545 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
16547 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
16549 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
16550 (fnma<mode>4): Likewise.
16551 (fms<mode>4): Likewise.
16552 (fnms<mode>4): Likewise.
16553 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
16554 (aarch64_fnma<mode>4): Likewise.
16555 (aarch64_fms<mode>4): Likewise.
16556 (aarch64_fnms<mode>4): Likewise.
16557 (aarch64_fnmadd<mode>4): Likewise.
16559 2018-05-16 Jason Merrill <jason@redhat.com>
16561 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
16563 2018-05-16 Richard Biener <rguenther@suse.de>
16565 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
16566 (dump_stmt_cost): Declare.
16567 (add_stmt_cost): Dump cost we add.
16568 (add_stmt_costs): New function.
16569 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
16570 No longer exported.
16571 (vect_analyze_stmt): Adjust prototype.
16572 (vectorizable_condition): Likewise.
16573 (vectorizable_live_operation): Likewise.
16574 (vectorizable_reduction): Likewise.
16575 (vectorizable_induction): Likewise.
16576 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
16577 cost vector to pass to vectorizable_ and record afterwards.
16578 (vect_model_reduction_cost): Take cost vector argument and adjust.
16579 (vect_model_induction_cost): Likewise.
16580 (vectorizable_reduction): Likewise.
16581 (vectorizable_induction): Likewise.
16582 (vectorizable_live_operation): Likewise.
16583 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
16584 SLP_TREE_NUMBER_OF_VEC_STMTS.
16585 (vect_analyze_slp_cost_1): Remove.
16586 (vect_analyze_slp_cost): Likewise.
16587 (vect_slp_analyze_node_operations): Take visited args and
16588 a target cost vector. Avoid processing already visited stmt sets.
16589 (vect_slp_analyze_operations): Use a local cost vector to gather
16590 costs and register those of non-discarded instances.
16591 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
16592 (vect_schedule_slp_instance): Remove copying of
16593 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
16595 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
16596 adding cost. Record cost entry location.
16597 (vect_prologue_cost_for_slp_op): Function to compute cost of
16598 a constant or invariant generated for SLP vect in the prologue,
16599 split out from vect_analyze_slp_cost_1.
16600 (vect_model_simple_cost): Make static. Adjust for SLP costing.
16601 (vect_model_promotion_demotion_cost): Likewise.
16602 (vect_model_store_cost): Likewise, make static.
16603 (vect_model_load_cost): Likewise.
16604 (vectorizable_bswap): Add cost vector arg and adjust.
16605 (vectorizable_call): Likewise.
16606 (vectorizable_simd_clone_call): Likewise.
16607 (vectorizable_conversion): Likewise.
16608 (vectorizable_assignment): Likewise.
16609 (vectorizable_shift): Likewise.
16610 (vectorizable_operation): Likewise.
16611 (vectorizable_store): Likewise.
16612 (vectorizable_load): Likewise.
16613 (vectorizable_condition): Likewise.
16614 (vectorizable_comparison): Likewise.
16615 (can_vectorize_live_stmts): Likewise.
16616 (vect_analyze_stmt): Likewise.
16617 (vect_transform_stmt): Adjust calls to vectorizable_*.
16618 * tree-vectorizer.c: Include gimple-pretty-print.h.
16619 (dump_stmt_cost): New function.
16621 2018-05-16 Richard Biener <rguenther@suse.de>
16623 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
16624 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
16625 * tree-ssa-dse.c: Include tree-ssa-loop.h.
16626 (check_name): New callback.
16627 (dse_classify_store): Track cycles via a visited bitmap of PHI
16628 defs and simplify handling of in-loop and across loop dead stores
16629 and properly fail for loop-variant refs. Handle byte-tracking with
16630 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
16633 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
16635 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
16636 (vect_get_mask_type_for_stmt): Likewise.
16637 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
16639 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
16640 to determine the statement's vector type and the vector type that
16641 should be used for calculating nunits. Deal with cases in which
16642 the type has to be deferred.
16643 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
16644 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
16645 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
16646 (vect_determine_vf_for_stmt): New functions, split out from...
16647 (vect_determine_vectorization_factor): ...here.
16648 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
16649 (vect_get_mask_type_for_stmt): New functions, split out from
16650 vect_determine_vectorization_factor.
16652 2018-05-16 Richard Biener <rguenther@suse.de>
16654 * tree-cfg.c (verify_gimple_assign_ternary): Properly
16655 verify the [VEC_]COND_EXPR embedded comparison.
16657 2018-05-15 Martin Sebor <msebor@redhat.com>
16659 PR tree-optimization/85753
16660 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
16661 RECORD_TYPE in addition to ARRAY_TYPE.
16663 2018-05-15 Martin Sebor <msebor@redhat.com>
16665 PR middle-end/85643
16666 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
16668 2018-05-15 Richard Biener <rguenther@suse.de>
16670 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
16671 add by_clobber_p one. Change algorithm to collect all defs
16672 representing uses we need to walk and try reducing them to
16673 a single one before failing.
16674 (dse_dom_walker::dse_optimize_stmt): Adjust.
16676 2018-05-13 Mark Wielaard <mark@klomp.org>
16678 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
16679 (size_of_loc_descr): Likewise.
16680 (output_loc_operands): Likewise.
16681 (output_loc_operands_raw): Likewise.
16682 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
16683 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
16684 (hash_loc_operands): Likewise.
16685 (compare_loc_operands): Likewise.
16687 2018-05-14 Mark Wielaard <mark@klomp.org>
16689 * dwarf2out.c (count_index_addrs): New function.
16690 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
16692 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16694 PR tree-optimization/83648
16695 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
16696 return value as malloc candidate.
16698 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16701 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
16702 param as true in call to suggest_attribute.
16704 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
16706 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
16707 -mreadonly-in-sdata.
16709 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16711 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
16713 (aarch64_crypto_aesd_fused): Likewise.
16715 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
16717 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
16718 (movsi_aarch64): Likewise.
16719 (load_pairsi): Likewise.
16720 (load_pairdi): Likewise.
16721 (store_pairsi): Likewise.
16722 (store_pairdi): Likewise.
16723 (load_pairsf): Likewise.
16724 (load_pairdf): Likewise.
16725 (store_pairsf): Likewise.
16726 (store_pairdf): Likewise.
16727 (zero_extend): Likewise.
16728 (trunc): Swap alternatives.
16729 (fcvt_target): Add '?' to prefer w over r.
16731 2018-05-14 Jakub Jelinek <jakub@redhat.com>
16734 * config/i386/i386.md: Disallow non-commutative arithmetics in
16735 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
16736 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
16737 in the peephole2 before it.
16739 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
16741 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
16742 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
16743 (ix86_handle_option): Handle -mcldemote.
16744 * config.gcc: New header.
16745 * config/i386/cldemoteintrin.h: New file.
16746 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
16747 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
16749 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
16750 OPTION_MASK_ISA_CLDEMOTE.
16751 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
16752 (ix86_valid_target_attribute_inner_p): Ditto.
16753 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
16754 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
16755 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
16756 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
16757 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
16759 * config/i386/i386.opt: Add -mcldemote.
16760 * config/i386/x86intrin.h: New header.
16761 * doc/invoke.texi: Add -mcldemote.
16763 2018-05-14 Richard Biener <rguenther@suse.de>
16765 * doc/match-and-simplify.texi: Adjust :s documentation.
16767 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
16769 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
16770 intended memcpy size.
16771 (REORDER_45): Likewise.
16773 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
16775 * sort.cc: New file.
16776 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
16777 * vec.c (qsort_chk): Use gcc_qsort.
16778 * Makefile.in (OBJS-libcommon): Add sort.o.
16779 (build/sort.o): New target. Use it...
16780 (BUILD_RTL): ... here, and...
16781 (build/gencfn-macros): ... here, and...
16782 (build/genmatch): ... here.
16784 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
16785 Chung-Ju Wu <jasonwucj@gmail.com>
16787 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
16788 * config/nds32/nds32-graywolf.md: New file.
16789 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
16790 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
16792 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
16793 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
16794 * config/nds32/nds32.md (pipeline_model): Add graywolf.
16795 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
16796 * config/nds32/pipelines.md: Include n15 settings.
16798 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
16799 Chung-Ju Wu <jasonwucj@gmail.com>
16801 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
16802 * config/nds32/nds32-n13.md: New file.
16803 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
16804 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
16806 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
16807 * config/nds32/nds32.md (pipeline_model): Add n13.
16808 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
16809 * config/nds32/pipelines.md: Include n13 settings.
16811 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
16812 Chung-Ju Wu <jasonwucj@gmail.com>
16814 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
16815 * config/nds32/nds32-n10.md: New file.
16816 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
16817 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
16819 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
16820 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
16821 * config/nds32/nds32.md (pipeline_model): Add n10.
16822 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
16823 * config/nds32/pipelines.md: Include n10 settings.
16825 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
16826 Kito Cheng <kito.cheng@gmail.com>
16827 Chung-Ju Wu <jasonwucj@gmail.com>
16829 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
16830 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
16831 Add enum values for DSP extension instructions.
16832 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
16834 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
16835 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
16836 New code iterators.
16837 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
16838 * config/nds32/nds32-dspext.md: New file for DSP implementation.
16839 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
16840 * config/nds32/nds32-intrinsic.md: Likewise.
16841 * config/nds32/nds32_intrinsic.h: Likewise.
16842 * config/nds32/nds32-md-auxiliary.c: Likewise.
16843 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
16844 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
16845 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
16846 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
16847 * config/nds32/nds32-protos.h: New declarations for DSP extension.
16848 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
16849 TYPE_DMAC in switch statement.
16850 * config/nds32/nds32.c: New checking and implementation for DSP
16851 extension instructions.
16852 * config/nds32/nds32.h: Likewise.
16853 * config/nds32/nds32.md: Likewise.
16854 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
16855 * config/nds32/predicates.md: Implement new predicates for DSP
16858 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
16860 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
16861 Reformat alternatives and attributes so it is easier to identify
16862 which constraints/attributes go with which instruction.
16863 (mov<mode>_hardfloat32, FMOVE64): Likewise.
16864 (mov<mode>_softfloat32, FMOVE64): Likewise.
16865 (mov<mode>_hardfloat64, FMOVE64): Likewise.
16866 (mov<mode>_softfloat64, FMOVE64): Likewise.
16868 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
16870 * doc/extend.texi (PowerPC Built-in Functions): Rename this
16872 (Basic PowerPC Built-in Functions): The new name of the
16873 subsection previously known as "PowerPC Built-in Functions".
16874 (Basic PowerPC Built-in Functions Available on all Configurations):
16876 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
16877 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
16878 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
16879 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
16881 2018-05-11 Martin Jambor <mjambor@suse.cz>
16884 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
16887 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
16890 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
16892 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
16894 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
16895 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
16896 (ix86_handle_option): Handle -mwaitpkg.
16897 * config.gcc: New header.
16898 * config/i386/cpuid.h (bit_WAITPKG): New bit.
16899 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
16900 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
16902 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
16903 OPTION_MASK_ISA_WAITPKG.
16904 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
16905 (ix86_option_override_internal): Add PTA_WAITPKG.
16906 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
16907 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
16908 IX86_BUILTIN_TPAUSE.
16909 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
16910 __builtin_ia32_umwait and __builtin_ia32_tpause.
16911 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
16912 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
16913 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
16914 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
16915 UNSPECV_TPAUSE): New.
16916 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
16917 * config/i386/i386.opt: Add -mwaitpkg.
16918 * config/i386/waitpkgintrin.h: New file.
16919 * config/i386/x86intrin.h: New header.
16920 * doc/invoke.texi: Add -mwaitpkg.
16922 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
16925 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
16927 (cortex-m0): Use armv6s-m isa.
16928 (cortex-m0plus): Likewise.
16929 (cortex-m1): Likewise.
16930 (cortex-m0.small-multiply): Likewise.
16931 (cortex-m0plus.small-multiply): Likewise.
16932 (cortex-m1.small-multiply): Likewise.
16934 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
16935 Jakub Jelinek <jakub@redhat.com>
16937 PR tree-optimization/85692
16938 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
16939 source permute as well.
16941 2018-05-11 Martin Liska <mliska@suse.cz>
16944 * doc/extend.texi: Document LLVM style format for no_sanitize
16947 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
16949 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
16950 mode_supports_vsx_dform_quad to mode_supports_dq_form.
16951 (mode_supports_vsx_dform_quad): Likewise.
16952 (mode_supports_vmx_dform): Move these functions to be next to the
16953 other mode_supports functions.
16954 (mode_supports_dq_form): Likewise.
16955 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
16956 mode_supports_dq_form.
16957 (reg_offset_addressing_ok_p): Likewise.
16958 (offsettable_ok_by_alignment): Likewise.
16959 (rs6000_legitimate_offset_address_p): Likewise.
16960 (legitimate_lo_sum_address_p): Likewise.
16961 (rs6000_legitimize_address): Likewise.
16962 (rs6000_legitimize_reload_address): Likewise.
16963 (rs6000_secondary_reload_inner): Likewise.
16964 (rs6000_preferred_reload_class): Likewise.
16965 (rs6000_output_move_128bit): Likewise.
16967 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
16969 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
16970 Generate SImode target register for null target.
16971 <case IX86_BUILTIN_XGETBV>: Ditto.
16972 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
16973 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
16975 2018-05-10 Carl Love <cel@us.ibm.com>
16977 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
16978 dcbtt and dcbtstt if operands[2] is 0.
16980 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
16983 * config/i386/sse.md (usadv64qi): New expander.
16985 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
16987 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
16988 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
16989 -maltivec=be support.
16990 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
16991 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
16992 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16993 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
16994 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
16995 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
16996 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
16997 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
16998 altivec_vsumsws): Adjust.
16999 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
17000 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
17001 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
17003 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
17004 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
17005 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
17006 (altivec_lve<VI_char>x): Delete expand.
17007 (*altivec_lve<VI_char>x_internal): Rename to...
17008 (altivec_lve<VI_char>x): ... this.
17009 (altivec_lvxl_<mode>): Delete expand.
17010 (*altivec_lvxl_<mode>_internal): Rename to ...
17011 (altivec_lvxl_<mode>): ... this.
17012 (altivec_stvxl_<mode>): Delete expand.
17013 (*altivec_stvxl_<mode>_internal): Rename to ...
17014 (altivec_stvxl_<mode>): ... this.
17015 (altivec_stve<VI_char>x): Delete expand.
17016 (*altivec_stve<VI_char>x_internal): Rename to ...
17017 (altivec_stve<VI_char>x): ... this.
17018 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
17019 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
17020 reduc_plus_scal_<mode>): Adjust.
17021 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
17023 (rs6000_cpu_cpp_builtins): Adjust.
17024 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
17025 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
17026 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
17027 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
17028 -maltivec=be support.
17029 (rs6000_split_vec_extract_var): Adjust.
17030 (rs6000_split_v4si_init): Adjust.
17031 (swap_selector_for_mode): Delete.
17032 (altivec_expand_lvx_be, altivec_expand_stvx_be,
17033 altivec_expand_stvex_be): Delete.
17034 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
17035 -maltivec=be support.
17036 (rs6000_gimple_fold_builtin): Ditto.
17037 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
17039 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
17040 (TARGET_DIRECT_MOVE_64BIT): Adjust.
17041 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
17042 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
17043 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
17044 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
17045 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
17046 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
17047 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
17048 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
17049 anonymous split): Adjust.
17050 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
17051 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
17053 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
17055 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
17056 when --with-gxx-include-dir is also specified.
17057 * configure: Regenerate.
17059 2018-05-09 Jim Wilson <jimw@sifive.com>
17062 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
17063 * config/riscv/t-withmultilib: New.
17064 * config/riscv/withmultilib.h: New.
17065 * doc/install.texi: Document RISC-V --with-multilib-list support.
17067 2018-05-09 Richard Biener <rguenther@suse.de>
17069 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
17071 (vect_bb_vectorization_profitable_p): Adjust. Compute
17072 actual scalar cost using the cost vector and the add_stmt_cost
17075 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17077 PR rtl-optimization/85645
17078 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
17079 in the REG_CFA_REGISTER note for LR, don't leave it empty.
17081 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17083 PR rtl-optimization/85645
17084 * shrink-wrap.c (spread_components): Return a boolean saying if
17085 anything was changed.
17086 (try_shrink_wrapping_separate): Iterate spread_components until
17087 nothing changes anymore.
17089 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17091 PR rtl-optimization/85645
17092 * regrename.c (build_def_use): Also kill the chains that include the
17093 destination of a REG_CFA_REGISTER note.
17095 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17097 PR rtl-optimization/85645
17098 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
17099 insn that has a REG_CFA_REGISTER note.
17101 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
17103 * cfgexpand.c (expand_clobber): New function.
17104 (expand_gimple_stmt_1): Use it.
17105 * tree-vect-stmts.c (vect_clobber_variable): New function,
17107 (vectorizable_simd_clone_call): ...here.
17108 (vectorizable_store): Emit a clobber either side of an
17109 IFN_STORE_LANES sequence.
17110 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
17112 2018-05-09 Tom de Vries <tom@codesourcery.com>
17115 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
17116 (define_insn "trap_if_false"): Add exit after trap.
17118 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
17120 PR rtl-optimization/85638
17121 * bb-reorder.c: Include common/common-target.h.
17122 (create_forwarder_block): New function extracted from...
17123 (fix_up_crossing_landing_pad): ...here. Rename into...
17124 (dw2_fix_up_crossing_landing_pad): ...this.
17125 (sjlj_fix_up_crossing_landing_pad): New function.
17126 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
17127 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
17128 from both partitions and exit the loop after one iteration.
17130 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17133 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17135 (Basic PowerPC Built-in Functions): The new name of the
17136 subsection previously known as "PowerPC Built-in Functions".
17137 (Basic PowerPC Built-in Functions Available on all Configurations):
17139 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
17141 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
17143 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
17145 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
17148 2018-05-08 Jim Wilson <jimw@sifive.com>
17150 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
17151 (LD_EMUL_SUFFIX): New.
17152 (LINK_SPEC): Use it.
17154 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17156 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17158 (Basic PowerPC Built-in Functions): The new name of the
17159 subsection previously known as "PowerPC Built-in Functions".
17160 (Basic PowerPC Built-in Functions Available on all Configurations):
17162 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
17164 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
17166 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
17168 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
17171 2018-05-08 Jakub Jelinek <jakub@redhat.com>
17174 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
17175 after cmpelim optimization.
17177 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
17179 * config.gcc: Support "goldmont".
17180 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
17181 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17182 PROCESSOR_GOLDMONT.
17183 * config/i386/i386.c (m_GOLDMONT): Define.
17184 (processor_target_table): Add "goldmont".
17185 (PTA_GOLDMONT): Define.
17186 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
17187 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
17188 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
17189 (fold_builtin_cpu): Add "goldmont".
17190 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
17191 (ix86_option_override_internal): Add "goldmont".
17192 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
17193 (processor_type): Add PROCESSOR_GOLDMONT.
17194 * config/i386/i386.md: Add CPU "glm".
17195 * config/i386/glm.md: New file.
17196 * config/i386/x86-tune.def: Add m_GOLDMONT.
17197 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
17199 2018-05-08 Jakub Jelinek <jakub@redhat.com>
17202 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
17204 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
17205 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
17206 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
17209 * config/i386/i386.c (ix86_fold_builtin): Handle
17210 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
17213 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
17214 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
17216 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
17219 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
17220 (check_arch): Likewise.
17221 (check_fpu): Return the result rather than printing it.
17222 (end arch): Fix operator precedence.
17223 (end cpu): Likewise.
17224 (END): Print the result from check_fpu.
17226 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
17227 Alan Hayward <alan.hayward@arm.com>
17228 David Sherwood <david.sherwood@arm.com>
17230 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
17231 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
17232 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
17233 (*fcmuo<mode>_and): New patterns.
17235 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
17237 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
17238 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
17239 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
17240 (cmp_op, sve_imm_con): New code attributes.
17241 (SVE_COND_INT_CMP, imm_con): Delete.
17242 (cmp_op): Remove above unspecs from int attribute.
17243 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
17245 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
17246 comparison-specific unspecs.
17247 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
17248 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
17249 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
17250 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
17251 (*vec_fcm<cmp_op><mode>): Rename to...
17252 (*fcm<cmp_op><mode>): ...this and adjust likewise.
17253 (*vec_fcmuo<mode>): Rename to...
17254 (*fcmuo<mode>): ...this and adjust likewise.
17255 (*pred_fcm<cmp_op><mode>): New pattern.
17256 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
17257 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
17259 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
17261 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
17262 (aarch64_emit_sve_predicated_cond): New function.
17263 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
17264 (aarch64_emit_unspec_cond_or): Replace with...
17265 (aarch64_emit_sve_or_conds): ...this new function. Use
17266 aarch64_emit_sve_ptrue_op for the individual comparisons and
17267 aarch64_emit_binop to OR them together.
17268 (aarch64_emit_inverted_unspec_cond): Replace with...
17269 (aarch64_emit_sve_inverted_cond): ...this new function. Use
17270 aarch64_emit_sve_ptrue_op for the comparison and
17271 aarch64_emit_unop to invert the result.
17272 (aarch64_expand_sve_vec_cmp_float): Update after the above
17273 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
17275 2018-05-07 Nathan Sidwell <nathan@acm.org>
17277 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
17278 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
17279 (Backwards Compatibility): Likewise.
17281 2018-05-07 Luis Machado <luis.machado@linaro.org>
17285 2018-05-07 Luis Machado <luis.machado@linaro.org>
17287 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17288 <prefetch_dynamic_strides>: New const bool field.
17289 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17290 prefetch_dynamic_strides.
17291 (exynosm1_prefetch_tune): Likewise.
17292 (thunderxt88_prefetch_tune): Likewise.
17293 (thunderx_prefetch_tune): Likewise.
17294 (thunderx2t99_prefetch_tune): Likewise.
17295 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
17297 (aarch64_override_options_internal): Update to set
17298 PARAM_PREFETCH_DYNAMIC_STRIDES.
17299 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
17300 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
17301 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
17302 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
17303 prefetch-dynamic-strides setting.
17305 2018-05-07 Luis Machado <luis.machado@linaro.org>
17307 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17308 <minimum_stride>: New const int field.
17309 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17310 minimum_stride field.
17311 (exynosm1_prefetch_tune): Likewise.
17312 (thunderxt88_prefetch_tune): Likewise.
17313 (thunderx_prefetch_tune): Likewise.
17314 (thunderx2t99_prefetch_tune): Likewise.
17315 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
17316 (aarch64_override_options_internal): Update to set
17317 PARAM_PREFETCH_MINIMUM_STRIDE.
17318 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
17319 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
17320 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
17321 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
17322 stride is constant and is below the minimum stride threshold.
17324 2018-05-07 Luis Machado <luis.machado@linaro.org>
17326 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
17329 2018-05-07 Luis Machado <luis.machado@linaro.org>
17331 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17332 <prefetch_dynamic_strides>: New const bool field.
17333 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17334 prefetch_dynamic_strides.
17335 (exynosm1_prefetch_tune): Likewise.
17336 (thunderxt88_prefetch_tune): Likewise.
17337 (thunderx_prefetch_tune): Likewise.
17338 (thunderx2t99_prefetch_tune): Likewise.
17339 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
17341 (aarch64_override_options_internal): Update to set
17342 PARAM_PREFETCH_DYNAMIC_STRIDES.
17343 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
17344 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
17345 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
17346 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
17347 prefetch-dynamic-strides setting.
17349 2018-05-07 Luis Machado <luis.machado@linaro.org>
17351 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17352 <minimum_stride>: New const int field.
17353 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17354 minimum_stride field.
17355 (exynosm1_prefetch_tune): Likewise.
17356 (thunderxt88_prefetch_tune): Likewise.
17357 (thunderx_prefetch_tune): Likewise.
17358 (thunderx2t99_prefetch_tune): Likewise.
17359 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
17360 (aarch64_override_options_internal): Update to set
17361 PARAM_PREFETCH_MINIMUM_STRIDE.
17362 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
17363 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
17364 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
17365 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
17366 stride is constant and is below the minimum stride threshold.
17368 2018-05-06 Jakub Jelinek <jakub@redhat.com>
17371 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
17372 the type is addressable. Don't force op into register if it has
17375 2018-05-05 Roland McGrath <mcgrathr@google.com>
17378 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
17379 any section for which we don't know a specific type it should have,
17380 regardless of name. Previously this was done only for the exact
17381 names ".init_array", ".fini_array", and ".preinit_array".
17382 (default_elf_asm_named_section): Add comment about
17383 relationship with default_section_type_flags and SECTION_NOTYPE.
17384 (get_section): Don't consider it a type conflict if one side has
17385 SECTION_NOTYPE and the other doesn't, as long as neither has the
17386 SECTION_BSS et al used in the default_section_type_flags logic.
17388 2018-05-05 Tom de Vries <tom@codesourcery.com>
17391 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
17392 (workaround_barsyncs): New function.
17393 (nvptx_reorg): Use workaround_barsyncs.
17394 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
17395 (define_expand "nvptx_membar_cta"): New define_expand.
17396 (define_insn "*nvptx_membar_cta"): New insn.
17398 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17400 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
17401 To improve optimization opportunities.
17402 * builtin-types.def: The new needed builtin types for the above.
17404 2018-05-04 Richard Biener <rguenther@suse.de>
17406 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
17407 * gimple-ssa-store-merging.c
17408 (imm_store_chain_info::output_merged_store): Remove redundant create,
17409 release split_store vector contents on failure.
17410 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
17411 scalar stmt vector on cache hit.
17413 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
17415 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
17417 * config.gcc (powerpc-xilinx-eabi*): Remove.
17418 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
17420 (fusion_addis_mem_combo_load): Ditto.
17421 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
17423 (rs6000_cpu_cpp_builtins): Ditto.
17424 * config/rs6000/rs6000-linux.c
17425 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
17426 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
17427 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
17429 (rs6000_setup_reg_addr_masks): Ditto.
17430 (rs6000_init_hard_regno_mode_ok): Ditto.
17431 (rs6000_option_override_internal): Ditto.
17432 (legitimate_lo_sum_address_p): Ditto.
17433 (rs6000_legitimize_address): Ditto.
17434 (rs6000_legitimize_reload_address): Ditto.
17435 (rs6000_legitimate_address_p): Ditto.
17436 (abi_v4_pass_in_fpr): Ditto.
17437 (setup_incoming_varargs): Ditto.
17438 (rs6000_gimplify_va_arg): Ditto.
17439 (rs6000_split_multireg_move): Ditto.
17440 (rs6000_savres_strategy): Ditto.
17441 (rs6000_emit_prologue_components): Ditto.
17442 (rs6000_emit_epilogue_components): Ditto.
17443 (rs6000_emit_prologue): Ditto.
17444 (rs6000_emit_epilogue): Ditto.
17445 (rs6000_elf_file_end): Ditto.
17446 (rs6000_function_value): Ditto.
17447 (rs6000_libcall_value): Ditto.
17448 * config/rs6000/rs6000.h: Ditto.
17449 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
17450 (TARGET_MINMAX): ... this. New.
17451 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
17452 * config/rs6000/rs6000.md: Remove Xilinx FP support.
17453 (*movsi_internal1_single): Delete.
17454 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
17455 mfpu=, mxilinx-fpu): Delete.
17456 * config/rs6000/singlefp.h: Delete.
17457 * config/rs6000/sysv4.h: Remove Xilinx FP support.
17458 * config/rs6000/t-rs6000: Ditto.
17459 * config/rs6000/t-xilinx: Delete.
17460 * config/rs6000/titan.md: Adjust for fp_type removal.
17461 * config/rs6000/vsx.md: Remove Xilinx FP support.
17462 (VStype_simple): Delete.
17463 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
17464 * config/rs6000/xfpu.h: Delete.
17465 * config/rs6000/xfpu.md: Delete.
17466 * config/rs6000/xilinx.h: Delete.
17467 * config/rs6000/xilinx.opt: Delete.
17468 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
17469 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
17471 2018-05-04 Tom de Vries <tom@codesourcery.com>
17474 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
17477 2018-05-04 Richard Biener <rguenther@suse.de>
17479 PR middle-end/85627
17480 * tree-complex.c (update_complex_assignment): We are always in SSA form.
17481 (expand_complex_div_wide): Likewise.
17482 (expand_complex_operations_1): Likewise.
17483 (expand_complex_libcall): Preserve EH info of the original stmt.
17484 (tree_lower_complex): Handle removed blocks.
17485 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
17486 on complex multiplication and division libcall builtins.
17488 2018-05-04 Richard Biener <rguenther@suse.de>
17490 PR middle-end/85574
17491 * fold-const.c (negate_expr_p): Restrict negation of operand
17492 zero of a division to when we know that can happen without
17494 (fold_negate_expr_1): Likewise.
17496 2018-05-04 Jakub Jelinek <jakub@redhat.com>
17499 * real.h (real_nextafter): Declare.
17500 * real.c (real_nextafter): New function.
17501 * fold-const-call.c (fold_const_nextafter): New function.
17502 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
17503 CASE_CFN_NEXTTOWARD.
17504 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
17505 even when arg1_mode is different from arg0_mode.
17507 2018-05-03 Nathan Sidwell <nathan@acm.org>
17509 * doc/extend.texi (Deprecated Features): Remove
17510 -ffriend-injection.
17511 (Backwards Compatibility): Likewise.
17512 * doc/invoke.texi (C++ Language Options): Likewise.
17513 (C++ Dialect Options): Likewise.
17515 2018-05-03 Jakub Jelinek <jakub@redhat.com>
17518 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
17519 _mm512_mask_mullox_epi64): New intrinsics.
17521 2018-05-03 Tom de Vries <tom@codesourcery.com>
17524 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
17525 dump files): Add offload-tree.
17527 2018-05-03 Richard Biener <rguenther@suse.de>
17529 PR tree-optimization/85615
17530 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
17531 to loops not nested in BBs loop father to avoid creating multi-entry
17534 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17536 PR tree-optimization/70291
17537 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
17538 arguments. Change return type to tree. Emit libcall as a new
17539 statement rather than replacing existing one when inplace_p is true.
17540 (expand_complex_multiplication_components): New function.
17541 (expand_complex_multiplication): Expand floating-point complex
17542 multiplication using the above.
17543 (expand_complex_division): Rename inner_type parameter to type.
17544 Update expand_complex_libcall call-site.
17545 (expand_complex_operations_1): Update expand_complex_multiplication
17546 and expand_complex_division call-sites.
17548 2018-05-02 Jakub Jelinek <jakub@redhat.com>
17551 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
17552 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
17553 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
17554 the highest significant bit of the shift count mask is clear. In
17555 check whether and[sq]i3 is needed verify that all significant bits
17556 of the shift count other than the highest are set.
17558 2018-05-02 Tom de Vries <tom@codesourcery.com>
17561 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
17562 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
17563 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
17564 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
17565 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
17566 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
17567 __builtin_goacc_parlevel_size.
17569 2018-05-02 Richard Biener <rguenther@suse.de>
17571 PR tree-optimization/85597
17572 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
17573 do not use split vect_get_vec_defs call but call vect_get_slp_defs
17576 2018-05-02 Tom de Vries <tom@codesourcery.com>
17579 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
17580 dump files): Add ltrans-tree.
17582 2018-05-02 Tom de Vries <tom@codesourcery.com>
17585 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
17586 dump files): Add wpa-ipa.
17588 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
17590 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
17591 powerpc*-*-linux*paired* target.
17592 * config/rs6000/750cl.h: Delete.
17593 * config/rs6000/paired.h: Delete.
17594 * config/rs6000/paired.md: Delete.
17595 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
17597 * config/rs6000/rs6000-builtin.def: Remove paired float support.
17598 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
17599 comment. Remove paired float support.
17600 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
17601 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
17603 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
17604 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
17606 * config/rs6000/rs6000.c: Remove paired float support.
17607 (paired_expand_vector_init, paired_expand_vector_move,
17608 paired_emit_vector_compare, paired_emit_vector_cond_expr,
17609 (paired_expand_lv_builtin, paired_expand_stv_builtin,
17610 paired_expand_builtin, paired_expand_predicate_builtin,
17611 paired_init_builtins): Delete.
17612 * config/rs6000/rs6000.h: Remove paired float support.
17613 * config/rs6000/rs6000.md: Remove paired float support.
17614 (move_from_CR_ov_bit): Delete.
17615 * config/rs6000/rs6000.opt (mpaired): Delete.
17616 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
17617 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
17619 2018-05-02 Richard Biener <rguenther@suse.de>
17621 PR middle-end/85567
17622 * gimplify.c (gimplify_save_expr): When in SSA form allow
17623 SAVE_EXPRs to compute to SSA vars.
17625 2018-05-02 Jakub Jelinek <jakub@redhat.com>
17628 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
17629 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
17630 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
17631 clobber operands[2], instead use a new pseudo. Formatting fixes.
17633 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
17635 PR tree-optimization/85586
17636 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
17637 exit early for statements in the same group if the accesses are
17640 2018-05-02 Tom de Vries <tom@codesourcery.com>
17643 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
17646 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
17648 PR tree-optimization/85143
17649 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
17651 2018-05-01 Tom de Vries <tom@codesourcery.com>
17654 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
17655 not found" error message.
17657 2018-05-01 Tom de Vries <tom@codesourcery.com>
17660 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
17661 * vec.c (test_ordered_remove_if): New function.
17662 (vec_c_tests): Call test_ordered_remove_if.
17663 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
17664 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
17665 * tree-vect-patterns.c (vect_pattern_recog_1): Use
17666 VEC_ORDERED_REMOVE_IF.
17668 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17670 PR tree-optimization/82665
17671 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
17672 pointer subtraction where arguments come from a memchr call.
17674 2018-05-01 Jakub Jelinek <jakub@redhat.com>
17676 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
17677 --push-state --as-needed and --pop-state instead of --as-needed and
17678 --no-as-needed if ld supports it.
17679 * configure: Regenerated.
17682 * doc/install.texi2html: Replace _002d with - and _002a with * in
17683 generated html files using sed.
17685 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17688 * gcc-rich-location.c (blank_line_before_p): New function.
17689 (use_new_line): New function.
17690 (gcc_rich_location::add_fixit_insert_formatted): New function.
17691 * gcc-rich-location.h
17692 (gcc_rich_location::add_fixit_insert_formatted): New function.
17694 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17696 * selftest.c (assert_streq): Rename "expected" and "actual" to
17697 "val1" and "val2". Extend NULL-handling to cover both inputs
17698 symmetrically, while still requiring both to be non-NULL for a pass.
17699 * selftest.h (assert_streq): Rename "expected" and "actual" to
17701 (ASSERT_EQ): Likewise.
17702 (ASSERT_EQ_AT): Likewise.
17703 (ASSERT_KNOWN_EQ): Likewise.
17704 (ASSERT_KNOWN_EQ_AT): Likewise.
17705 (ASSERT_NE): Likewise.
17706 (ASSERT_MAYBE_NE): Likewise.
17707 (ASSERT_MAYBE_NE_AT): Likewise.
17708 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
17709 the assertion to pass.
17710 (ASSERT_STREQ_AT): Likewise.
17712 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
17714 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
17715 interaction with -pie.
17717 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17719 * selftest.h: Fix alphabetization of per-source-file selftest
17722 2018-04-30 Jason Merrill <jason@redhat.com>
17724 PR c++/61982 - dead stores to destroyed objects.
17725 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
17728 2018-04-30 Jason Merrill <jason@redhat.com>
17730 * tree.c (build_clobber): New.
17731 * tree.h: Declare it.
17732 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
17734 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17736 * diagnostic-show-locus.c (layout::layout): Update for
17737 location_get_source_line returning a char_span.
17738 (struct char_span): Move to input.h.
17739 (struct correction): Update for fields in char_span becoming
17741 (struct source_line): Update for location_get_source_line
17742 returning a char_span.
17743 (layout::print_line): Likewise.
17744 * edit-context.c (edited_file::print_content): Likewise.
17745 (edited_file::print_diff_hunk): Likewise.
17746 (edited_file::print_run_of_changed_lines): Likewise.
17747 (edited_file::get_num_lines): Likewise.
17748 (edited_line::edited_line): Likewise.
17749 * final.c (asm_show_source): Likewise.
17750 * input.c (location_get_source_line): Convert return type
17751 from const char * to char_span, losing the final "line_len"
17753 (dump_location_info): Update for the above.
17754 (get_substring_ranges_for_loc): Likewise. Use a char_span
17755 when handling the literal within the line.
17756 (test_reading_source_line): Update for location_get_source_line
17757 returning a char_span.
17758 * input.h (class char_span): Move here from
17759 diagnostic-show-locus.c, converting from a struct to a class.
17760 Make data members private.
17761 (char_span::operator bool): New.
17762 (char_span::length): New.
17763 (char_span::get_buffer): New.
17764 (char_span::operator[]): New.
17765 (char_span::subspan): Make const.
17766 (char_span::xstrdup): New.
17767 (location_get_source_line): Convert return type from const char *
17768 to char_span, losing the final "line_size" param.
17770 2018-04-30 Jan Hubicka <jh@suse.cz>
17772 * lto-wrapper.c (ltrans_priorities): New static var.
17773 (cmp_priority): New.
17774 (run_gcc): Read priorities and if doing parallel build order
17775 the Makefile by them.
17777 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17779 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
17781 2018-04-30 Richard Biener <rguenther@suse.de>
17783 * tree-cfg.c (verify_address): Remove base argument, add
17784 flag whether to check TREE_ADDRESSABLE and do that.
17785 (verify_expr): Remove.
17786 (verify_types_in_gimple_reference): Add pieces from verify_expr.
17787 (verify_gimple_assign_single): Likewise.
17788 (verify_gimple_switch): Likewise.
17789 (verify_expr_location_1): Dereference tp once. Add (disabled)
17790 piece from verify_expr.
17791 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
17793 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
17795 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
17797 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
17799 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
17800 (small_data_pattern): Likewise.
17801 (arc_rewrite_small_data): Likewise.
17802 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
17803 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
17804 (get_symbol_alignment): New function.
17805 (legitimate_small_data_address_p): Likewise.
17806 (legitimate_scaled_address): Update, call
17807 legitimate_small_data_address_p.
17808 (output_sdata): New static variable.
17809 (arc_print_operand): Update how we handle small data operands.
17810 (arc_print_operand_address): Likewise.
17811 (arc_legitimate_address_p): Update, use
17812 legitimate_small_data_address_p.
17813 (arc_rewrite_small_data_p): Remove.
17814 (arc_rewrite_small_data_1): Likewise.
17815 (arc_rewrite_small_data): Likewise.
17816 (small_data_pattern): Likewise.
17817 (compact_sda_memory_operand): Update to use
17818 legitimate_small_data_address_p and get_symbol_alignment.
17819 (prepare_move_operands): Don't rewite sdata pattern.
17820 (prepare_extend_operands): Remove.
17821 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
17823 (zero_extendqisi2): Likewise.
17824 (zero_extendhisi2): Likewise.
17825 (extendqihi2): Likewise.
17826 (extendqisi2): Likewise.
17827 (extendhisi2): Likewise.
17828 (addsi3): Likewise.
17829 (subsi3): Likewise.
17830 (andsi3): Likewise.
17831 * config/arc/constraints.md (Usd): Change it to memory constraint.
17833 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
17835 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
17836 as source of std instructions.
17837 * config/arc/arc.md (movsi_insn): Update pattern predicate to
17838 allow 6-bit constants as source for store instructions.
17839 (movdi_insn): Update instruction pattern to allow 6-bit constants
17840 as source for store instructions.
17842 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
17844 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
17846 2018-04-30 Nathan Sidwell <nathan@acm.org>
17847 Sandra Loosemore <sandra@codesourcery.com>
17849 * dumpfile.c (dump_open): Allow '-' for stdout.
17850 * doc/invoke.texi (Developer Options): Document dump filename
17851 determination early. Document stdin/stdout selection.
17853 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
17855 Microblaze Target: PIC data text relative
17857 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
17858 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
17860 * config/microblaze/microblaze.h (microblaze_constant_address_p):
17861 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
17862 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
17863 New addressing mode for data-text relative position indepenedent code.
17864 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
17865 'ADDRESS_SYMBOLIC_TXT_REL'.
17866 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
17867 (microblaze_legitimate_pic_operand): Exclude function calls from
17868 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
17869 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
17871 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
17872 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
17873 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
17874 for 'address + offset'.
17875 (microblaze_expand_prologue): Add new function prologue call for
17877 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
17878 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
17879 table in case of TARGET_PIC_DATA_TEXT_REL.
17880 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
17881 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
17882 Add new macros 'UNSPEC_TEXT',
17883 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
17884 + exclude function calls from 'UNSPEC_PLT' in case of data text
17886 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
17887 new target hook for generating address diff vector tables in case of
17889 * doc/tm.texi : Regenerate.
17890 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
17891 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
17892 of addr diff vector generation.
17893 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
17894 target hook definition.
17895 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
17896 Add default function for generate_pic_addr_diff_vec -> flag_pic.
17897 * doc/invoke.texi (Add new pic option): Add new microblaze pic
17898 option for data text relative.
17900 2018-04-30 Richard Biener <rguenther@suse.de>
17902 * tree-chrec.h (evolution_function_is_constant_p): Remove
17904 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
17906 2018-04-30 Richard Biener <rguenther@suse.de>
17909 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
17911 2018-04-30 Richard Biener <rguenther@suse.de>
17913 PR tree-optimization/28364
17914 PR tree-optimization/85275
17915 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
17916 copying first exit test.
17918 2018-04-28 Mark Wielaard <mark@klomp.org>
17920 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
17921 dwarf_version >= 5.
17922 (dwarf_AT): Handle DW_AT_addr_base.
17923 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
17925 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
17928 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
17929 (*ashl<dwi>3_doubleword_mask_1): Ditto.
17930 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
17931 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
17933 2018-04-28 Richard Biener <rguenther@suse.de>
17935 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
17936 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
17937 to reflect use. Only add interesting stmts.
17939 2018-04-27 Martin Jambor <mjambor@suse.cz>
17942 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
17943 the jump function allows for passing through aggregate values.
17945 2018-04-27 David Malcolm <dmalcolm@redhat.com>
17947 * input.h (in_system_header_at): Convert from macro to inline
17949 (from_macro_expansion_at): Likewise.
17950 (from_macro_definition_at): Likewise.
17952 2018-04-27 Jeff Law <law@redhat.com>
17954 * config.gcc: Mark tile* targets as deprecated/obsolete.
17956 2018-04-27 Richard Biener <rguenther@suse.de>
17958 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
17961 2018-04-27 Richard Biener <rguenther@suse.de>
17963 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
17965 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
17967 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
17968 with Yd constraint. Set "preferred_for_speed" attribute from
17969 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
17970 with Yd constraint.
17971 (*movdi_internal): Ditto.
17972 (movti_interunit splitters): Remove
17973 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
17974 (movdi_interunit splitters): Ditto.
17975 * config/i386/constraints.md (Ye): Remove.
17976 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
17978 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17981 * config/aarch64/constraints.md (Usg): Limit to 31.
17982 (Usj): Limit to 63.
17984 2018-04-27 Jakub Jelinek <jakub@redhat.com>
17986 PR tree-optimization/85529
17987 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
17988 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
17989 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
17990 zero extension or masking of the MSB bit.
17991 (optimize_range_tests): Add FIRST_BB argument, pass it through
17992 to optimize_range_tests_var_bound.
17993 (maybe_optimize_range_tests, reassociate_bb): Adjust
17994 optimize_range_tests callers.
17996 2018-04-26 Richard Biener <rguenther@suse.de>
17997 Jakub Jelinek <jakub@redhat.com>
17999 * cgraph.h (symbol_table): Just declare debug method here.
18000 * symtab.c (symbol_table::debug): Define.
18002 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
18004 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
18006 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
18008 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
18009 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
18010 (*movdi_internal): Substitute Yi and Yj constraint with x
18011 and Ym and Yn constraint with y constraint. Update "isa"
18012 attribute and set "preferred_for_speed" attribute from
18013 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
18014 (*movsi_internal): Ditto.
18015 (*movdf_internal): Ditto.
18016 (*movsf_internal): Ditto.
18017 (*zero_extendsidi2): Ditto.
18018 * config/i386/sse.md (vec_set<mode>_0): Ditto.
18019 (sse2_loadld): Ditto.
18020 (*vec_extract<ssevecmodelower>_0): Ditto.
18021 (*vec_extractv4si_0_zext_sse4): Ditto.
18022 (vec_concatv2di): Ditto.
18023 (*vec_dup<mode>): Ditto.
18024 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
18025 * config/i386/constraints.md (Yi): Remove.
18030 2018-04-26 Nathan Sidwell <nathan@acm.org>
18032 * dumpfile.c (dump_open): New.
18033 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
18034 (dump_finish): Detect stdio/stderr by value not name.
18036 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
18038 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
18040 2018-04-26 Tom de Vries <tom@codesourcery.com>
18043 * config/nvptx/nvptx.c (verify_neutering_jumps)
18044 (verify_neutering_labels): New function
18045 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
18047 2018-04-26 Tom de Vries <tom@codesourcery.com>
18050 * config/nvptx/nvptx.c (needs_neutering_p): New function.
18051 (nvptx_single): Use needs_neutering_p to skip over insns that do not
18054 2018-04-26 Richard Biener <rguenther@suse.de>
18055 Tom de Vries <tom@codesourcery.com>
18058 * lto-streamer-out.c (output_function): Fixup loops if required to match
18059 discovery done in the reader.
18061 2018-04-26 Richard Biener <rguenther@suse.de>
18063 PR tree-optimization/85116
18064 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
18065 have a loop exit from the single latch predecessor. Remove
18066 case of header with just condition.
18067 (ch_base::copy_headers): Exclude infinite loops from any
18069 (pass_ch::execute): Record exits.
18071 2018-04-26 Richard Biener <rguenther@suse.de>
18073 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
18074 prologue cost vector and pass it to vect_get_load_cost.
18075 (vect_get_peeling_costs_all_drs): Likewise.
18076 (vect_peeling_hash_get_lowest_cost): Likewise.
18077 (vect_enhance_data_refs_alignment): Likewise.
18079 2018-04-26 Richard Biener <rguenther@suse.de>
18081 PR middle-end/85450
18082 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
18083 checking of integer<->pointer conversions.
18084 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
18085 sign-/zero-extending pointer types.
18086 (expand_omp_for_static_chunk): Likewise.
18088 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
18089 Jean Lee <xiaoyur347@gmail.com>
18091 * config/mips/mips.c (mips_asan_shadow_offset): New function.
18092 (TARGET_ASAN_SHADOW_OFFSET): Define.
18093 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
18094 true for -fsanitize=address.
18096 2018-04-25 Mark Wielaard <mark@klomp.org>
18098 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
18101 2018-04-25 Jakub Jelinek <jakub@redhat.com>
18103 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
18104 than "alu", remove explicit "memory" and "imm_disp" attributes.
18105 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
18107 PR middle-end/85414
18108 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
18109 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
18110 gen_lowpart_no_emit.
18112 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
18115 * config/i386/i386.c (ix86_expand_builtin): Change memory
18116 operand to XI, extend p0 to Pmode.
18117 * config/i386/i386.md: Change unspec volatile and operand
18118 1 mode to XI, change operand 0 mode to P.
18120 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18122 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
18123 GET_MODE_MASK before any checking.
18124 (nds32_can_use_bset_p): Likewise.
18125 (nds32_can_use_btgl_p): Likewise.
18127 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18129 * config/nds32/nds32-doubleword.md: New define_split pattern for
18130 illegal register number.
18132 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18134 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
18136 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18138 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
18140 2018-04-25 Richard Biener <rguenther@suse.de>
18142 * lto-streamer.h (LTO_major_version): Bump to 8.
18144 2018-04-25 Jakub Jelinek <jakub@redhat.com>
18146 * BASE-VER: Set to 9.0.0.
18148 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
18150 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
18151 in __abskf2 and __powikf2.
18153 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18156 * config/aarch64/constraints.md (Usg, Usj): New constraints.
18157 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
18158 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
18159 Use the above on operand 2. Reindent.
18160 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
18162 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
18165 * common/config/i386/i386-common.c (ix86_handle_option): Don't
18167 * config/i386/i386.opt (mcet): Removed.
18168 * doc/install.texi: Remove -mcet documentation.
18169 * doc/invoke.texi: Likewise.
18171 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
18174 * doc/install.texi: Remove -mcet from bootstrap-cet.
18176 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18179 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
18180 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
18184 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
18185 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
18186 containing a CONST_VECTOR.
18188 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
18190 * doc/install.texi: Update newlib dependency for nvptx.
18192 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18195 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
18196 instead of INTVAL when shifting x left.
18198 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
18200 PR tree-optimization/85478
18201 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
18202 vect_grouped_store_supported for single element vectors.
18204 2018-04-24 Richard Biener <rguenther@suse.de>
18207 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
18208 load cost increase to the case of non-constant step.
18210 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18213 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
18214 destination if any_malformed_asm.
18216 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
18218 PR middle-end/85496
18219 * expr.c (store_field): In the bitfield case, if the value comes from
18220 a function call and is returned in registers by means of a PARALLEL,
18221 do not change the mode of the temporary unless BLKmode and VOIDmode.
18223 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
18225 PR rtl-optimization/85423
18226 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
18227 dependencies to debug insns when the previous insn is non-debug.
18229 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
18231 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
18232 enums into a single definition.
18233 (fls): Fix predicates and printing.
18236 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
18238 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
18239 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
18240 and short u6 immediate.
18241 (check_if_valid_sleep_operand): Remove.
18242 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
18244 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18246 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
18247 flag_always_save_lp condition.
18248 * config/nds32/nds32.opt (malways-save-lp): New option.
18250 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18252 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
18253 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
18254 * config/nds32/nds32.h
18255 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
18256 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
18258 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18260 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
18261 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
18263 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18264 Chung-Ju Wu <jasonwucj@gmail.com>
18266 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
18268 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
18269 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
18271 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18273 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
18275 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18277 * config/nds32/nds32-protos.h (nds32_data_alignment,
18278 nds32_local_alignment): Declare.
18279 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
18280 nds32_local_alignment): New functions.
18281 (TARGET_CONSTANT_ALIGNMENT): Define.
18282 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
18284 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18286 * config/nds32/nds32.c
18287 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
18288 (TARGET_MODES_TIEABLE_P): Likewise.
18290 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18292 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
18293 level Ofast and Og.
18295 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
18296 Chung-Ju Wu <jasonwucj@gmail.com>
18298 * config/nds32/constants.md (unspec_volatile_element): Add enum values
18299 for unaligned access.
18300 * config/nds32/nds32-intrinsic.c: Implementation of expanding
18302 * config/nds32/nds32-intrinsic.md: Likewise.
18303 * config/nds32/nds32_intrinsic.h: Likewise.
18304 * config/nds32/nds32.h (nds32_builtins): Likewise.
18305 * config/nds32/nds32.opt (munaligned-access): New option.
18306 * config/nds32/nds32.c (nds32_asm_file_start): Display
18307 flag_unaligned_access status.
18309 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
18311 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
18312 -mno-relax is present.
18313 * config/riscv/linux.h (LINK_SPEC): Ditto.
18315 2018-04-20 Martin Sebor <msebor@redhat.com>
18318 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
18320 (gimple_fold_builtin_stxcpy_chk): Same.
18321 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
18323 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
18326 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
18327 __powikf2 when long double is IEEE 128-bit.
18329 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
18331 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
18332 step to make sure stack always aligned.
18334 2018-04-20 Carl Love <cel@us.ibm.com>
18337 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
18338 size check for arg0.
18340 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
18341 Tom de Vries <tom@codesourcery.com>
18344 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
18345 Emit insns for calls too.
18346 (nvptx_find_par): Always look for worker-level predecessor insn.
18347 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
18349 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
18350 (nvptx_process_pars): Propagate frames for calls.
18352 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
18355 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
18357 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
18358 (ix86_handle_option): Don't handle OPT_mibt.
18359 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
18361 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
18363 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
18364 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
18365 (ix86_target_macros): Define __CET__ with flag_cf_protection
18366 for -fcf-protection.
18367 * config/i386/i386.c (isa2_opts): Remove -mibt.
18368 * config/i386/i386.h (TARGET_IBT): Removed.
18369 (TARGET_IBT_P): Likewise.
18370 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
18371 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
18372 * config/i386/i386.opt (mcet): Update help message.
18373 (mshstk): Likewise.
18375 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
18376 -mcet as an alias for -mshstk.
18378 2018-04-20 Richard Biener <rguenther@suse.de>
18380 PR middle-end/85475
18381 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
18382 complexity by forcing a single use of the multiply operand.
18384 2018-04-20 Martin Jambor <mjambor@suse.cz>
18387 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
18388 recursion dependency to only apply to non-clones.
18390 2018-04-20 Martin Jambor <mjambor@suse.cz>
18393 * ipa-cp.c (create_specialized_node): Check that clones of
18394 self-recursive edges exist during IPA-CP.
18396 2018-04-19 Toon Moene <toon@moene.org>
18398 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
18401 2018-04-19 Jakub Jelinek <jakub@redhat.com>
18403 PR tree-optimization/85467
18404 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
18405 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
18406 VECTOR_CST element to type.
18408 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
18411 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
18412 * config/i386/i386.md (builtin_setjmp_setup): Removed.
18413 (builtin_longjmp): Likewise.
18414 (save_stack_nonlocal): New pattern.
18415 (restore_stack_nonlocal): Likewise.
18417 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
18420 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
18421 Replace ASM_OUTPUT_LABEL with fprintf.
18423 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
18426 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
18427 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
18428 * config/i386/i386-c.c (ix86_target_macros_internal): Also
18429 define __IBT__ and __SHSTK__ for -fcf-protection.
18430 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
18432 (ix86_trampoline_init): Likewise.
18433 (x86_output_mi_thunk): Likewise.
18434 (ix86_notrack_prefixed_insn_p): Likewise.
18435 (ix86_option_override_internal): Don't disallow -fcf-protection.
18436 * config/i386/i386.md (rdssp<mode>): Also enable for
18438 (incssp<mode>): Likewise.
18439 (nop_endbr): Likewise.
18440 * config/i386/i386.opt (mcet): Change help message to built-in
18443 (mshstk): Likewise.
18444 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
18445 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
18446 enable CET built-in functions.
18448 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
18450 * common/config/i386/i386-common.c
18451 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
18452 OPTION_MASK_ISA_MOVDIRI_UNSET,
18453 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
18454 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
18455 * config.gcc (movdirintrin.h): New header.
18456 * config/i386/cpuid.h (bit_MOVDIRI,
18457 bit_MOVDIR64B): New bits.
18458 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
18460 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
18461 (VOID, PVOID, PCVOID)): New function types.
18462 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
18463 __builtin_ia32_directstoreu_u64,
18464 __builtin_ia32_movdir64b): New builtins.
18465 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
18466 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
18468 (ix86_valid_target_attribute_inner_p): Ditto.
18469 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
18470 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
18471 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
18472 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
18473 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
18474 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
18475 (movdiri<mode>, movdir64b_<mode>): New.
18476 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
18477 * config/i386/immintrin.h: Include movdirintrin.h.
18478 * config/i386/movdirintrin.h: New file.
18479 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
18481 2018-04-19 Richard Biener <rguenther@suse.de>
18483 PR middle-end/85455
18484 * cfg.c (clear_bb_flags): When loop state says we have
18485 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
18487 2018-04-19 Richard Biener <rguenther@suse.de>
18489 PR tree-optimization/84737
18490 * tree-vect-data-refs.c (vect_copy_ref_info): New function
18491 copying restrict info.
18492 (vect_setup_realignment): Use it.
18493 * tree-vectorizer.h (vect_copy_ref_info): Declare.
18494 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
18495 the first DR to all generated stores.
18496 (vectorizable_load): Likewise for loads.
18498 2018-04-19 Jakub Jelinek <jakub@redhat.com>
18500 PR tree-optimization/85446
18501 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
18502 the integral and pointer types to have the same precision.
18504 * doc/install.texi: Document --disable-cet being the default and
18507 2018-04-18 Martin Liska <mliska@suse.cz>
18509 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
18512 2018-04-18 Martin Liska <mliska@suse.cz>
18515 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
18518 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
18519 arguments if they are comparable.
18521 2018-04-18 Martin Liska <mliska@suse.cz>
18524 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
18527 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
18530 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
18533 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
18534 ENDBR after calling __morestack.
18536 2018-04-18 David Malcolm <dmalcolm@redhat.com>
18539 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
18540 by using gcc_base_ver to generate a gcc_driver_version, and use
18541 it when generating GCC_DRIVER_NAME.
18542 * configure: Regenerate.
18544 2018-04-18 Jakub Jelinek <jakub@redhat.com>
18547 * config.gcc: Obsolete powerpc*-*-*spe*.
18549 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18552 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
18553 (stabstr_D): Change type of unum from unsigned int to
18554 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
18557 2018-04-17 Jim Wilson <jimw@sifive.com>
18560 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
18561 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
18562 Set arg_pointer_offset after using pretend_args_size.
18564 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18566 PR rtl-optimization/85431
18567 * dse.c (record_store): Ignore zero width stores.
18570 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
18571 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
18572 __builtin_stack_restore rather than after it.
18573 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
18574 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
18575 argument instead of virtual_dynamic_stack_rtx.
18577 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
18579 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
18581 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18582 Add note to error message to explain internal mapping of overloaded
18583 built-in function name to non-overloaded built-in function name.
18584 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
18587 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
18590 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
18591 where the inputs overlap with the output.
18593 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18595 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
18596 (=v, v) alternative and explicit "memory" attribute.
18597 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
18598 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
18600 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
18601 "sselog1" type instead of "sselog".
18602 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
18603 "sselog". Remove explicit "memory" attribute.
18604 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
18605 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
18607 (vec_extract_hi_v32hi): Merge all alternatives into one, use
18608 "sselog1" type instead of "sselog". Remove explicit "memory"
18610 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
18611 use "sselog1" type instead of "sselog". Remove explicit "memory"
18613 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
18614 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
18616 (vec_extract_hi_v64qi): Merge all alternatives into one, use
18617 "sselog1" type instead of "sselog". Remove explicit "memory"
18619 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
18620 use "sselog1" type instead of "sselog". Remove explicit "memory"
18624 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
18626 PR middle-end/85414
18627 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
18630 2018-04-17 Martin Jambor <mjambor@suse.cz>
18633 * ipa-cp.c (create_specialized_node): Call
18634 expand_all_artificial_thunks if necessary.
18636 2018-04-17 Martin Liska <mliska@suse.cz>
18639 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
18640 in message, remote space in between '_G' and '('.
18642 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18645 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
18646 avx512f_vmcmp<mode>3<round_saeonly_name>,
18647 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
18648 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
18649 avx512f_rndscale<mode><round_saeonly_name>,
18650 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
18651 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
18652 Use %<iptr>2 instead of %2 for -masm=intel.
18653 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
18654 avx512f_vcvttss2usi<round_saeonly_name>,
18655 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
18657 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
18658 avx512f_vcvttsd2usi<round_saeonly_name>,
18659 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
18660 Use %q1 instead of %1 for -masm=intel.
18661 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
18662 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
18663 of %3 for -masm=intel.
18664 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
18666 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
18668 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
18670 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
18671 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
18673 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
18675 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
18676 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
18677 %g1 and one with %0 and %1.
18678 (avx512er_vmrcp28<mode><round_saeonly_name>,
18679 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
18680 %1 for -masm=intel.
18681 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
18682 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
18683 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
18684 of %0 and %{%4%} for -masm=intel.
18685 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
18686 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
18687 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
18688 order of %0 and %{%5%}%{z%} for -masm=intel.
18690 2018-04-17 Jan Hubicka <jh@suse.cz>
18693 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
18695 2018-04-17 Martin Liska <mliska@suse.cz>
18698 * multiple_target.c (create_dispatcher_calls): Set apostrophes
18699 for target_clone error message. Make default implementation
18700 clone to be a local declaration.
18701 (separate_attrs): Add new argument and check for an empty
18703 (expand_target_clones): Handle it.
18704 (ipa_target_clone): Make redirection just for target_clones
18707 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
18708 Tom de Vries <tom@codesourcery.com>
18710 PR middle-end/84955
18711 * omp-expand.c (expand_oacc_for): Add dummy false branch for
18712 tiled basic blocks without omp continue statements.
18714 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
18717 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
18718 vec_extract expression as having side effects to make sure it gets
18721 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
18724 * config/i386/i386.c (get_builtin_code_for_version): Check
18727 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
18730 * config.gcc: Support "skylake".
18731 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
18733 * config/i386/i386.c (m_SKYLAKE): Define.
18734 (processor_target_table): Add "skylake".
18735 (ix86_option_override_internal): Add "skylake".
18736 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
18737 PROCESSOR_CANNONLAKE.
18738 (get_builtin_code_for_version): Fix priority for
18739 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
18740 PROCESSOR_SKYLAKE-AVX512.
18741 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
18742 (processor_type): Add PROCESSOR_SKYLAKE.
18744 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
18745 Jason Merrill <jason@redhat.com>
18748 * convert.c (convert_to_integer_1): Use direct recursion for
18749 enumeral types and types with a precision less than the number
18750 of bits in their mode.
18752 2018-04-16 Julia Koval <julia.koval@intel.com>
18755 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
18756 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
18758 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
18761 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
18762 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
18763 and -mno-direct-move.
18765 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
18768 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
18769 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
18770 Ensure negative shifts result in {0}.
18772 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
18774 PR rtl-optimization/79916
18775 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
18776 regs (if any) to define how to gnerate SD moves when LRA is in
18779 2018-04-13 Jakub Jelinek <jakub@redhat.com>
18781 PR rtl-optimization/85393
18782 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
18783 * except.c (expand_dw2_landing_pad_for_region): Make static.
18784 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
18785 a label and unconditional jump to old_bb, rather than
18786 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
18789 PR rtl-optimization/85376
18790 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
18791 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
18792 instead of a specific value.
18794 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
18795 Bin Cheng <bin.cheng@arm.com>
18797 PR tree-optimization/82965
18798 PR tree-optimization/83991
18799 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
18800 by_profile_only parameter.
18801 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
18802 information if the loop was predicted to iterate too many times.
18803 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
18805 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
18808 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
18811 2018-04-13 Martin Liska <mliska@suse.cz>
18812 Jakub Jelinek <jakub@redhat.com>
18814 PR middle-end/81657
18815 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
18816 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
18817 * builtins.c (expand_builtin_memory_copy_args): Use
18818 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
18819 handle dest_addr == pc_rtx.
18821 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
18824 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
18825 asked to not generate direct moves.
18826 (fix_trunc<mode>si2_stfiwx): Similar.
18827 (fix_trunc<mode>si2_internal): Similar.
18829 2018-04-12 Jakub Jelinek <jakub@redhat.com>
18832 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
18833 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
18834 lookup if dest in some wider mode is known to be const0_rtx and
18835 if so, record permanent equivalence for it to be ZERO_EXTEND of
18836 the narrower mode destination.
18838 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
18840 * lto-streamer-out.c (output_function): Revert 259346.
18841 * omp-expand.c (expand_oacc_for): Likewise.
18843 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
18845 PR rtl-optimization/85354
18846 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
18847 * sel-sched.c (sel_global_init): ... here.
18849 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
18852 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
18853 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
18854 mode for PE-COFF targets.
18855 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
18856 (i386_pe_asm_lto_end): Likewise.
18857 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
18858 (TARGET_ASM_LTO_END): Likewise.
18859 * config/i386/winnt.c (saved_debug_info_level): New static variable.
18860 (i386_pe_asm_lto_start): New function.
18861 (i386_pe_asm_lto_end): Likewise.
18863 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
18864 Richard Biener <rguenther@suse.de>
18866 PR middle-end/84955
18867 * lto-streamer-out.c (output_function): Fix CFG loop state before
18869 * omp-expand.c (expand_oacc_for): Handle calls to internal
18870 functions like regular functions.
18872 2018-04-12 Richard Biener <rguenther@suse.de>
18875 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
18876 for the early LTO debug to properly generate references to it
18877 during DIE emission. Do not re-use that for the skeleton for
18879 (dwarf2out_early_finish): Likewise.
18881 2018-04-12 Jakub Jelinek <jakub@redhat.com>
18884 * config/i386/sse.md
18885 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
18886 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
18887 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
18888 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
18889 and output is a reg, avoid creating invalid lowpart subreg, but
18890 instead split into a 512-bit move. Don't split if not AVX512VL,
18891 input is xmm16+ reg and output is a mem.
18892 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
18893 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
18894 xmm16+ reg and output is a mem.
18896 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18898 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
18899 also for flag_dwarf2_cfi_asm.
18901 2018-04-12 Jakub Jelinek <jakub@redhat.com>
18903 PR rtl-optimization/85342
18904 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
18905 a bool scalar var inside of the loop instead. Don't try to update
18906 recog_data.operand after failed apply_change_group.
18908 2018-04-12 Tom de Vries <tom@codesourcery.com>
18911 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
18912 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
18913 array with flexible array member as array without given dimension.
18914 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
18915 argument for undefined param to true.
18917 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
18920 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
18921 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
18922 from PowerPC section.
18923 * config/rs6000/sysv4.opt (mcall-): Improve help text.
18924 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
18925 help text that is too long.
18926 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
18927 help text that is too long.
18928 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
18929 help text that is too long.
18931 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
18933 * config/alpha/alpha.md (stack_probe_internal): Rename
18934 from "probe_stack". Update all callers.
18936 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
18938 PR rtl-optimization/84566
18939 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
18940 sched_macro_fuse_insns.
18942 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
18945 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
18946 (compute_block_dependences): ... from here.
18948 2018-04-11 Jakub Jelinek <jakub@redhat.com>
18950 PR tree-optimization/85331
18951 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
18952 from int to HOST_WIDE_INT.
18954 2018-04-11 Martin Jambor <mjambor@suse.cz>
18957 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
18958 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
18959 not the same as the source val.
18960 (cgraph_edge_brings_value_p): New parameter.
18961 (gather_edges_for_value): Pass destination value to
18962 cgraph_edge_brings_value_p.
18963 (perhaps_add_new_callers): Likewise.
18964 (get_info_about_necessary_edges): Likewise and exclude values brought
18965 only by self-recursive edges.
18966 (create_specialized_node): Redirect only clones of self-calling edges.
18967 (+self_recursive_pass_through_p): New function.
18968 (find_more_scalar_values_for_callers_subset): Use it.
18969 (find_aggregate_values_for_callers_subset): Likewise.
18970 (known_aggs_to_agg_replacement_list): Removed.
18971 (decide_whether_version_node): Re-calculate known constants for all
18972 remaining context clones.
18974 2018-04-11 Richard Biener <rguenther@suse.de>
18977 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
18978 from early DWARF output.
18979 (dwarf2out_early_finish): Output line info unconditionally into
18980 early DWARF and add reference to it.
18982 2018-04-11 Jakub Jelinek <jakub@redhat.com>
18985 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
18986 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
18987 other than V2DFmode using iptr mode attribute.
18988 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
18990 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
18992 PR rtl-optimization/84659
18993 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
18995 2018-04-11 Jakub Jelinek <jakub@redhat.com>
18998 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
19000 (output_loc_list): Pass address of a dummy size variable even in the
19001 locview handling loop.
19002 (index_location_lists): Add comment on why skip_loc_list_entry can't
19005 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
19008 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
19011 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
19014 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
19015 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
19016 and -mstring-compare-inline-limit.
19018 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19021 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
19022 for stack clash protection in a register whenever we need it to be in
19025 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19027 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
19028 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
19030 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19033 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
19035 (mlong-double-): Ditto.
19036 * config/rs6000/sysv4.opt (msdata=): Ditto.
19037 (mtls-size=): Ditto.
19039 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19041 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
19042 erroneous entries for
19043 "vector int vec_ldl (int, long int *)", and
19044 "vector unsigned int vec_ldl (int, unsigned long int *)".
19045 Add comments and entries for
19046 "vector bool char vec_ldl (int, bool char *)",
19047 "vector bool short vec_ldl (int, bool short *)",
19048 "vector bool int vec_ldl (int, bool int *)",
19049 "vector bool long long vec_ldl (int, bool long long *)",
19050 "vector pixel vec_ldl (int, pixel *)",
19051 "vector long long vec_ldl (int, long long *)",
19052 "vector unsigned long long vec_ldl (int, unsigned long long *)".
19053 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
19054 type tree bool_long_long_type_node and correct definition of
19055 bool_V2DI_type_node to make reference to this new type tree.
19056 (rs6000_mangle_type): Replace erroneous reference to
19057 bool_long_type_node with bool_long_long_type_node.
19058 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
19059 comments to emphasize sign distinctions for char and int types and
19060 replace RS6000_BTI_bool_long constant with
19061 RS6000_BTI_bool_long_long constant. Also add comment to restrict
19062 use of RS6000_BTI_pixel.
19063 (bool_long_type_node): Remove this macro definition.
19064 (bool_long_long_type_node): New macro definition
19066 2018-04-10 Jakub Jelinek <jakub@redhat.com>
19068 PR rtl-optimization/85300
19069 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
19070 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
19071 simplify_unary_operation fails.
19073 2018-04-10 Martin Liska <mliska@suse.cz>
19075 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
19076 cgraph_edge and ipa_ref.
19078 2018-04-10 Jakub Jelinek <jakub@redhat.com>
19082 * config/i386/sse.md
19083 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
19084 computation of the VEC_MERGE selector from mask.
19085 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
19086 Fix decoding of the VEC_MERGE selector into mask.
19088 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
19090 PR tree-optimization/85286
19091 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
19093 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
19095 * final.c (final_1): Set insn_last_address as well as
19096 insn_current_address.
19098 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19101 * explow.c (emit_stack_probe): Call validize_mem on memory location
19102 before passing it to gen_probe_stack. Create address operand and
19103 legitimize it for the probe_stack_address case.
19105 2018-04-09 Jan Hubicka <jh@suse.cz>
19108 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
19109 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
19110 * tree.c (free_lang_data_in_type): Fix handling of binfos;
19112 (free_lang_data): Rebuild type inheritance graph.
19114 2018-04-09 Martin Sebor <msebor@redhat.com>
19116 * invoke.texi (-finline-small-functions): Mention other optimization
19118 (-findirect-inlining, -fpartial-inlining): Same.
19119 (-finline-functions-called-once): Same.
19120 (-freorder-blocks-and-partition): Same.
19122 2018-04-09 Jan Hubicka <jh@suse.cz>
19125 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
19126 jumps; choose last target that matches the criteria (i.e.
19127 no partition changes for non-crossing jumps).
19128 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
19129 support for redirecting crossing jumps to non-crossing.
19131 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
19133 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
19134 also for naked functions.
19136 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
19138 * config/arc/arc.md (add_shift): New pattern.
19139 (add_shift2): Likewise.
19140 (sub_shift): Likewise.
19141 (sub_shift_cmp0_noout): Likewise.
19142 (compare_si_ashiftsi): Likewise.
19143 (xbfu_cmp0_noout): New combine pattern.
19144 (xbfu_cmp0"): Likewise.
19145 (movsi_set_cc_insn): Place the predicable variant first.
19146 (commutative_binary_cmp0_noout): Remove clobber.
19147 (commutative_binary_cmp0): New pattern.
19148 (noncommutative_binary_cmp0): Likewise.
19149 (noncommutative_binary_cmp0_noout): Likewise.
19150 (noncommutative_binary_comparison_result_used): Removed.
19151 (rsub_cmp0): New pattern.
19152 (rsub_cmp0_noout): Likewise.
19153 (extzvsi): Changed, keep only meaningful variants.
19154 (SQH, SEZ): New iterators.
19155 (SQH_postfix): New mode attribute.
19156 (SEZ_prefix): New code attribute.
19157 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
19158 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
19159 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
19160 of numerical value.
19161 (noncommutative_operator): Check the availability of barrel
19164 2018-04-09 Richard Biener <rguenther@suse.de>
19166 PR tree-optimization/85284
19167 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
19168 Only use the niter constraining form of simple_iv when the exit
19169 is always executed.
19171 2018-04-09 Tom de Vries <tom@codesourcery.com>
19174 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
19175 (define_expand "*memory_barrier"): New define_expand.
19176 (define_insn "memory_barrier"): New insn.
19178 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19180 PR rtl-optimization/80463
19181 PR rtl-optimization/83972
19182 PR rtl-optimization/83480
19184 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
19185 correct producer for the insn.
19186 (tidy_control_flow): Fixup seqnos in case of debug insns.
19188 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19190 PR rtl-optimization/83913
19192 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
19193 different sched-times when merging exprs.
19195 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19197 PR rtl-optimization/83962
19199 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
19200 tidy_fallthru_edge and tidy_control_flow.
19202 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19204 PR rtl-optimization/83530
19206 * sel-sched.c (force_next_insn): New global variable.
19207 (remove_insn_for_debug): When force_next_insn is true, also leave only
19208 next insn in the ready list.
19209 (sel_sched_region): When the region wasn't scheduled, make another pass
19210 over it with force_next_insn set to 1.
19212 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
19214 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
19216 * config/nds32/constants.md (unspec_volatile_element): Add enum values
19217 for interrupt control.
19218 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
19219 functions for interrupt control.
19220 * config/nds32/nds32-intrinsic.md: Likewise.
19221 * config/nds32/nds32_intrinsic.h: Likewise.
19222 * config/nds32/nds32.h (nds32_builtins): Likewise.
19224 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
19226 * config/nds32/nds32.c (nds32_init_machine_status,
19227 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
19228 strict_aligned_p field.
19229 (nds32_expand_to_rtl_hook): New function.
19230 (TARGET_EXPAND_TO_RTL_HOOK): Define.
19231 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
19233 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19234 Chung-Ju Wu <jasonwucj@gmail.com>
19236 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
19237 * config/nds32/nds32-n7.md: New file.
19238 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
19239 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
19241 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
19242 * config/nds32/nds32.md (pipeline_model): Add n7.
19243 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
19244 * config/nds32/pipelines.md: Include n7 settings.
19246 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19247 Chung-Ju Wu <jasonwucj@gmail.com>
19249 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
19250 * config/nds32/nds32-e8.md: New file.
19251 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
19252 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
19254 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
19255 * config/nds32/nds32.md (pipeline_model): Add e8.
19256 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
19257 * config/nds32/pipelines.md: Include e8 settings.
19259 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19260 Chung-Ju Wu <jasonwucj@gmail.com>
19262 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
19263 * config/nds32/nds32-n8.md: New file.
19264 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
19265 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
19267 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
19268 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
19269 * config/nds32/nds32.md (pipeline_model): Add n8.
19270 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
19271 * config/nds32/pipelines.md: Include n8 settings.
19273 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19274 Chung-Ju Wu <jasonwucj@gmail.com>
19276 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
19277 * config/nds32/nds32-n9-2r1w.md: New file.
19278 * config/nds32/nds32-n9-3r2w.md: New file.
19279 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
19280 nds32_register_ports): New or modify for cpu n9.
19281 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
19283 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
19284 * config/nds32/nds32-utils.c: New file.
19285 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
19286 TARGET_MUL_SLOW): Define.
19287 * config/nds32/nds32.md (pipeline_model): New attribute.
19288 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
19289 New options that support cpu n9.
19290 * config/nds32/pipelines.md: Include n9 settings.
19291 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
19293 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
19295 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
19296 information if necessary.
19297 (output_cond_branch_compare_zero): Likewise.
19298 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
19299 (nds32_target_alignment): Refine for alignment.
19300 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
19301 (FUNCTION_BOUNDARY): Modify.
19302 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
19304 * config/nds32/nds32.opt (malways-align, malign-functions): New.
19306 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
19308 * config/nds32/constants.md (unspec_volatile_element): Add values for
19309 TLB operation and data prefetch.
19310 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
19311 functions for TLB operation and data prefetch.
19312 * config/nds32/nds32-intrinsic.md: Likewise.
19313 * config/nds32/nds32_intrinsic.h: Likewise.
19314 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
19315 (nds32_print_operand): Likewise.
19316 * config/nds32/nds32.h (nds32_builtins): Likewise.
19318 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
19319 Andrew Pinski <pinsika@gcc.gnu.org>
19321 PR middle-end/82976
19322 * match.pd: Use constant_boolean_node of correct type instead of
19323 boolean_true_node or boolean_false_node for simplifying
19324 pointer comparisons to zero.
19326 2018-04-07 Jakub Jelinek <jakub@redhat.com>
19328 PR tree-optimization/80021
19329 * tree.c (verify_type_variant): Make error call in verify_variant_match
19330 translatable and remove final full stop.
19332 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19334 * config/nds32/constants.md (unspec_volatile_element): Add
19335 UNSPEC_VOLATILE_EH_RETURN.
19336 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
19337 nds32_output_stack_pop): Support dwarf exception handling process.
19338 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
19339 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
19340 exception handling process.
19341 (nds32_compute_stack_frame): Likewise.
19342 (nds32_return_addr_rtx): Likewise.
19343 (nds32_initial_elimination_offset): Likewise.
19344 (nds32_expand_prologue): Likewise.
19345 (nds32_expand_epilogue): Likewise.
19346 (nds32_dynamic_chain_address): New function.
19347 * config/nds32/nds32.h (machine_function): Add fields for dwarf
19348 exception handling.
19349 (DYNAMIC_CHAIN_ADDRESS): Define.
19350 (EH_RETURN_DATA_REGNO): Define.
19351 (EH_RETURN_STACKADJ_RTX): Define.
19352 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
19353 patterns for dwarf exception handling.
19355 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19357 * config/nds32/nds32.h: Clean up obsolete macros.
19359 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19361 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
19362 Add enum values for particular instructions.
19363 * config/nds32/nds32-intrinsic.c: Implementation of expanding
19364 particular intrinsic functions.
19365 * config/nds32/nds32-intrinsic.md: Likewise.
19366 * config/nds32/nds32_intrinsic.h: Likewise.
19367 * config/nds32/nds32.h (nds32_builtins): Likewise.
19368 * config/nds32/nds32.md (type): Add pbsad and pbsada.
19369 (btst, ave): New patterns for particular instructions.
19371 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19373 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
19374 Add enum values for atomic load/store and memory sync.
19375 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
19377 * config/nds32/nds32-intrinsic.md: Likewise.
19378 * config/nds32/nds32_intrinsic.h: Likewise.
19379 * config/nds32/nds32.h (nds32_builtins): Likewise.
19381 2018-04-07 Jakub Jelinek <jakub@redhat.com>
19383 PR tree-optimization/85257
19384 * fold-const.c (native_encode_vector): If not all elts could fit
19385 and off is -1, return 0 rather than offset.
19386 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
19387 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
19388 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
19389 adjust buffer in native_interpret_expr call.
19391 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19393 * config/nds32/constants.md (unspec_volatile_element): Add cache
19394 control enum values.
19395 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
19396 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
19397 * config/nds32/nds32.c (nds32_cctl_names): New.
19398 (nds32_print_operand): Handle cache control register names.
19399 * config/nds32/nds32.h (nds32_builtins): New enum values.
19400 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
19402 * config/nds32/nds32.md (type): Add mmu.
19403 * config/nds32/pipelines.md (simple_insn): Add mmu.
19405 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19407 * config/nds32/nds32.md (type): Remove call.
19408 * config/nds32/pipelines.md (simple_insn): Likewise.
19410 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19412 * config/nds32/constants.md (unspec_volatile_element): Add
19413 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
19414 UNSPEC_VOLATILE_FMFCFG.
19415 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
19416 description for fmfcfg and fmfcsr.
19417 (bdesc_1arg): Add fmtcsr.
19418 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
19419 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
19420 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
19421 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
19422 unspec_fmfcfg): New patterns.
19423 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
19424 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
19425 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
19426 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
19427 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
19428 __nds32__fmfcfg): Define.
19430 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19432 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
19433 intrinsic register names.
19434 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
19435 intrinsic register enum values and macros.
19437 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19439 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
19440 for load/store addressing form.
19441 (nds32_print_operand_address): Likewise.
19443 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
19446 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
19447 based on LABEL_REF. Remove useless assertion.
19448 (pic_address_needs_scratch): Fix formatting.
19449 (sparc_legitimize_pic_address): Minor tweaks.
19450 (sparc_delegitimize_address): Adjust assertion accordingly.
19451 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
19452 into symbolic_operand.
19453 (movsi_high_pic_label_ref): Likewise.
19454 (movsi_lo_sum_pic_label_ref): Likewise.
19455 (movdi_pic_label_ref): Likewise.
19456 (movdi_high_pic_label_ref): Likewise.
19457 (movdi_lo_sum_pic_label_ref): Likewise.
19459 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
19461 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
19462 custom LIB_SPEC setup.
19464 2018-04-06 Ruslan Bukin <br@bsdpad.com>
19465 Kito Cheng <kito.cheng@gmail.com>
19467 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
19468 * config/riscv/freebsd.h: New.
19470 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
19472 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
19473 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
19476 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
19477 Kito Cheng <kito.cheng@gmail.com>
19479 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
19480 nds32_output_call, nds32_symbol_binds_local_p): New functions.
19481 * config/nds32/nds32-protos.h (nds32_output_call,
19482 nds32_output_return): Declare.
19483 * config/nds32/nds32.md: Refine all the call and return patterns.
19485 2018-04-06 Jakub Jelinek <jakub@redhat.com>
19488 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
19489 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
19491 PR rtl-optimization/84872
19492 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
19493 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
19494 EDGE_CROSSING edge.
19496 2018-04-06 Tamar Christina <tamar.christina@arm.com>
19498 * expr.c (copy_blkmode_to_reg): Revert 254862.
19499 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
19501 2018-04-06 Richard Biener <rguenther@suse.de>
19503 PR middle-end/85244
19504 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
19505 after seeing a component reference with an adjacent field. Treat
19506 refs to arrays at struct end of external decls similar to
19507 refs to unconstrained commons.
19509 2018-04-06 Jakub Jelinek <jakub@redhat.com>
19512 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
19513 look through SAVE_EXPRs with non-side-effects argument. Adjust
19515 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
19518 2018-04-06 Richard Biener <rguenther@suse.de>
19520 PR middle-end/85180
19521 * alias.c (find_base_term): New wrapper around find_base_term
19522 unwinding CSELIB_VAL_PTR changes.
19523 (find_base_term): Do not restore CSELIB_VAL_PTR during the
19526 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19528 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
19530 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
19531 constant definitions.
19532 ("nop"): lr 0,0 -> nopr r0
19533 ("nop_lr0", "nop_lr1"): New insn definitions.
19535 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
19537 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
19538 NDS32_V3PUSH_AVAILABLE_P macro.
19540 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
19541 Chung-Ju Wu <jasonwucj@gmail.com>
19543 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
19544 (nds32*-*-*): Add float and fpu_config into supported_defaults.
19545 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
19546 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
19547 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
19548 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
19549 * config/nds32/constraints.md: New constraints and checking for hard
19550 float configuration.
19551 * config/nds32/iterators.md: New mode iterator and attribute for hard
19552 float configuration.
19553 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
19555 * config/nds32/nds32-fpu.md: New file.
19556 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
19557 deal with hard float code generation.
19558 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
19560 (abi_type, float_reg_number): New enum type.
19561 * config/nds32/nds32-predicates.c: New predicates for hard float.
19562 * config/nds32/nds32-protos.h: Declare functions for hard float.
19563 * config/nds32/nds32.c: Implementation for hard float configuration.
19564 * config/nds32/nds32.h: Definitions for hard float configuration.
19565 * config/nds32/nds32.md: Include hard float machine description and
19566 modify patterns for hard float configuration.
19567 * config/nds32/nds32.opt: New options for hard float configuration.
19568 * config/nds32/predicates.md: New predicates for hard float
19571 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
19573 * common/config/nds32/nds32-common.c
19574 (nds32_option_optimization_table): Enable -mreleax-hint by default.
19576 2018-04-05 Jakub Jelinek <jakub@redhat.com>
19578 PR middle-end/85195
19579 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
19580 CONSTRUCTOR_ELT (ctor, ...)->value.
19582 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
19585 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
19587 2018-04-05 Tom de Vries <tom@codesourcery.com>
19590 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
19593 2018-04-05 Shiva Chen <shiva0217@gmail.com>
19594 Kito Cheng <kito.cheng@gmail.com>
19596 * config/nds32/constraints.md (U33): Fine-tune checking condition.
19597 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
19598 * config/nds32/nds32.h (nds32_16bit_address_type): Add
19599 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
19601 2018-04-05 Shiva Chen <shiva0217@gmail.com>
19602 Kito Cheng <kito.cheng@gmail.com>
19604 * config/nds32/constraints.md (Ufe): New memory constraint.
19605 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
19606 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
19607 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
19609 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
19610 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
19612 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19614 * config/nds32/nds32.md: Use optimize_size in the condition for
19615 alu-shift instructions.
19617 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19619 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
19621 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19623 * config/nds32/nds32.md (negsi2): Refine pattern.
19625 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
19626 Chung-Ju Wu <jasonwucj@gmail.com>
19628 * config/nds32/iterators.md (shift_rotate): New code iterator.
19629 (shift): New code attribute.
19630 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
19631 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
19632 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
19633 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
19634 bit-wise operations.
19635 (andsi3, *andsi3): Ditto.
19636 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
19637 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
19638 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
19639 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
19640 nds32_ior_operand, nds32_xor_operand): New predicates.
19642 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19644 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
19645 (addsi3, subsi3): ... this.
19647 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19649 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
19651 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19653 * config/nds32/nds32.md: Adjust indention.
19655 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
19657 * config/nds32/nds32.md (feature): New attribute.
19659 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19661 * config/nds32/nds32.md (subtype): New attribute.
19663 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19666 * config/arm/arm-builtins.c (arm_expand_builtin): Change
19667 expansion to perform a bitwise AND of the argument followed by a
19668 boolean negation of the result.
19670 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
19672 PR rtl-optimization/84878
19673 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
19674 the basic block. Assert the use reference is not artificial and that
19675 it has an associated insn.
19677 2018-04-04 Michael Matz <matz@suse.de>
19679 * builtins.c (compute_objsize): Pass correct operand
19680 to array_at_struct_end_p.
19682 2018-04-04 Richard Biener <rguenther@suse.de>
19685 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
19686 from contexts for DINFO_LEVEL_TERSE and below.
19688 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
19690 * config/nds32/nds32-doubleword.md (move_<mode>): Require
19691 resiter_operand condition.
19692 * config/nds32/nds32.md (*move<mode>): Ditto.
19694 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
19695 Monk Chiang <sh.chiang04@gmail.com>
19697 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
19699 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
19701 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
19703 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
19704 Kito Cheng <kito.cheng@gmail.com>
19706 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
19707 nds32_cond_code_str, output_cond_branch,
19708 output_cond_branch_compare_zero, nds32_expand_cbranch,
19709 nds32_expand_cstore, nds32_expand_movcc,
19710 nds32_output_cbranchsi4_equality_zero,
19711 nds32_output_cbranchsi4_equality_reg,
19712 nds32_output_cbranchsi4_equality_reg_or_const_int,
19713 nds32_output_cbranchsi4_greater_less_zero: New functions.
19714 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
19715 nds32_expand_cstore, nds32_expand_movcc,
19716 nds32_output_cbranchsi4_equality_zero,
19717 nds32_output_cbranchsi4_equality_reg,
19718 nds32_output_cbranchsi4_equality_reg_or_const_int,
19719 nds32_output_cbranchsi4_greater_less_zero): Declare.
19720 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
19721 nds32_rimm11s_operand): New predicates.
19722 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
19723 * config/nds32/nds32.md: Rewrite all the branch and conditional move
19726 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
19728 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
19729 * config/nds32/nds32.md: Ditto.
19730 * config/nds32/pipelines.md: Ditto.
19732 2018-04-04 Richard Biener <rguenther@suse.de>
19734 PR tree-optimization/85168
19735 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
19736 propagating abnormals.
19738 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
19740 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
19742 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
19743 Kito Cheng <kito.cheng@gmail.com>
19745 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
19746 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
19747 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
19748 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
19749 * config/nds32/nds32.md (sibcall_internal): New.
19750 (sibcall_register): Remove.
19751 (sibcall_immediate): Remove.
19752 (sibcall_value_internal): New.
19753 (sibcall_value_register): Remove.
19754 (sibcall_value_immediate): Remove.
19755 * config/nds32/predicates.md (nds32_general_register_operand): New.
19756 (nds32_call_address_operand): New.
19758 2018-04-03 Jakub Jelinek <jakub@redhat.com>
19760 PR rtl-optimization/85167
19761 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
19762 bb_defs if *split_p, instead preinitialize it to NULL.
19764 PR tree-optimization/85156
19765 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
19766 evaluating the argument multiple times.
19768 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
19770 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
19772 (_mm_cvtpd_ps): Likewise.
19773 (_mm_cvttpd_epi32): Likewise.
19774 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
19775 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
19776 vector, pixel, and bool following altivec.h include.
19778 2018-04-03 Martin Sebor <msebor@redhat.com>
19780 * doc/extend.texi (Common Function Attributes): Clarify.
19781 (const attribute): Likewise.
19782 (pure attribute): Likewise.
19784 2018-04-03 Jakub Jelinek <jakub@redhat.com>
19787 * config/i386/i386.c (ix86_expand_vector_set): Use
19788 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
19790 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
19792 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
19793 instructions when changing rounding bits to preserve precision bits
19794 in the x87 control word.
19796 2018-04-03 Martin Liska <mliska@suse.cz>
19798 PR tree-optimization/82491
19799 * rtl.h (strip_offset_and_add): Replace += suboffset with
19800 poly_uint64 () + suboffset.
19802 2018-03-29 Martin Liska <mliska@suse.cz>
19803 Martin Jambor <mjambor@suse.cz>
19806 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
19807 param_type is not an integral or pointer type.
19809 2018-04-03 Richard Biener <rguenther@suse.de>
19811 * sese.h (recompute_all_dominators): Remove.
19813 2018-04-02 Martin Sebor <msebor@redhat.com>
19815 * doc/invoke.texi (-Wrestrict): Fix typos.
19817 2018-04-02 Jim Wilson <jimw@sifive.com>
19819 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
19820 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
19821 (<optab>di3, <optab>si3_extend): Likewise.
19822 (<optab>si3_mask, <optab>si3_mask_1): New.
19823 (<optab>di3_mask, <optab>di3_mask_1): New.
19824 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
19825 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
19826 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
19828 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
19830 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
19833 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
19835 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
19836 (nds32_canonicalize_comparison): New function.
19838 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
19839 Kito Cheng <kito.cheng@gmail.com>
19840 Kuan-Lin Chen <kuanlinchentw@gmail.com>
19842 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
19843 * config/nds32/constants.md (unspec_volatile_element): Add
19844 UNSPEC_VOLATILE_RELAX_GROUP.
19845 * config/nds32/nds32-relax-opt.c: New file.
19846 * config/nds32/nds32-predicates.c
19847 (nds32_symbol_load_store_p): New function.
19848 * config/nds32/nds32-protos.h
19849 (nds32_symbol_load_store_p): Declare function.
19850 (make_pass_nds32_relax_opt): Declare new rtl pass function.
19851 * config/nds32/nds32.c
19852 (nds32_register_pass): New function to register pass.
19853 (nds32_register_passes): New function to register passes.
19854 * config/nds32/nds32.md (relax_group): New pattern.
19855 * config/nds32/nds32.opt (mrelax-hint): New option.
19856 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
19858 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
19860 * config/nds32/t-nds32: Modify files dependency.
19862 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
19864 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
19865 (PROFILE_HOOK): Define its implementation.
19867 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
19869 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
19870 type and 32-bit size.
19872 2018-04-01 Jakub Jelinek <jakub@redhat.com>
19874 PR middle-end/85090
19875 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
19876 (V_128_256): New mode iterator.
19877 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
19878 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
19879 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
19881 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
19882 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
19884 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
19887 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
19888 NaN inputs correctly.
19890 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
19893 * config/rs6000/vsx.md (??r): New mode attribute.
19894 (*vsx_mov<mode>_64bit): Use it.
19895 (*vsx_mov<mode>_32bit): Likewise.
19897 2018-03-30 Martin Sebor <msebor@redhat.com>
19899 PR tree-optimization/84818
19900 * builtins.c (check_access): Use warning_n.
19902 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19905 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
19907 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
19910 2018-03-30 Julia Koval <julia.koval@intel.com>
19913 * x86-tune.def (movx, partial_reg_dependency): Enable for
19916 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
19918 PR inline-asm/84985
19919 * lra-constraints.c (process_alt_operands): Move setting
19920 this_alternative_matches below.
19922 2018-03-29 Martin Liska <mliska@suse.cz>
19925 * doc/invoke.texi: Document how LTO works with debug info.
19926 Describe auto-load support of binutils. Mention 'x86-64'
19927 as valid option value of -march option.
19929 2018-03-29 Jakub Jelinek <jakub@redhat.com>
19931 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
19934 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
19935 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
19936 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
19939 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
19942 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
19943 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
19944 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
19945 for RS6000_BTM_POWERPC64.
19946 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
19947 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
19948 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
19951 (DIVDEU): Likewise.
19953 2018-03-28 Carl Love <cel@us.ibm.com>
19956 2017-09-27 Carl Love <cel@us.ibm.com>
19958 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
19959 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
19960 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
19963 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19965 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
19966 instead of __vector bool.
19967 (_mm_max_pu8): Likewise.
19968 (_mm_min_pi16): Likewise.
19970 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
19973 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
19974 (DIVWEUO): Likewise.
19975 (DIVDEO): Likewise.
19976 (DIVDEUO): Likewise.
19977 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
19978 DIVWEUO and DIVDEUO.
19979 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
19980 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
19981 (div_extend): Likewise.
19982 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
19984 (__builtin_divweuo): Likewise.
19985 (__builtin_divdeo): Likewise.
19986 (__builtin_divdeuo): Likewise.
19988 2018-03-28 Jakub Jelinek <jakub@redhat.com>
19991 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
19992 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
19994 PR tree-optimization/82004
19995 * gimple-match-head.c (optimize_pow_to_exp): New function.
19996 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
19997 Don't fold to exp if optimize_pow_to_exp is false.
19999 2018-03-28 Martin Liska <mliska@suse.cz>
20002 * calls.c (initialize_argument_information): Fix trailing space.
20003 * common.opt: Fix typo and provide better explanation for
20004 -fsanitize-coverage option.
20005 * config/i386/i386.opt: Fix typo.
20007 2018-03-28 Jakub Jelinek <jakub@redhat.com>
20008 Martin Liska <mliska@suse.cz>
20011 * gimplify.c (asan_poison_variable): Don't do the check for
20012 gimplify_omp_ctxp here.
20013 (gimplify_decl_expr): Do it here.
20014 (gimplify_target_expr): Likewise.
20016 2018-03-28 Martin Liska <mliska@suse.cz>
20019 * config/i386/i386.c (ix86_function_arg_advance): Do not call
20020 chkp_type_bounds_count if MPX is not enabled.
20022 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
20024 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
20026 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20029 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
20030 function to create the function decl for complex long double
20031 multiply and divide for -mabi=ieeelongdouble.
20032 (init_float128_ieee): Call it.
20034 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
20037 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
20038 -fcf-protection=branch -mibt.
20039 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
20041 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20044 * config/arm/arm.c (arm_valid_symbolic_address): Handle
20045 arm_word_relocations.
20047 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
20050 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
20051 extern array declarations.
20053 2018-03-27 Richard Biener <rguenther@suse.de>
20055 PR middle-end/84067
20056 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
20057 explicit single_use checks.
20059 2018-03-27 Richard Biener <rguenther@suse.de>
20061 PR tree-optimization/85082
20062 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
20065 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20067 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
20068 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
20069 Turn on fasynchronous-unwind-tables and funwind-tables.
20071 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
20074 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
20075 (*bmi_blsr_<mode>_ccz): Ditto.
20077 2018-03-26 Tom de Vries <tom@codesourcery.com>
20079 PR tree-optimization/85063
20080 * omp-general.c (offloading_function_p): New function. Factor out
20082 * omp-offload.c (pass_omp_target_link::gate): ... here.
20083 * omp-general.h (offloading_function_p): Declare.
20084 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
20085 with attribute omp declare target for offloading functions.
20087 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
20089 PR tree-optimization/84005
20090 * tree-data-ref.h (get_base_for_alignment): Declare.
20091 * tree-data-ref.c (get_base_for_alignment_1): New function.
20092 (get_base_for_alignment): Likewise.
20093 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
20094 get_base_for_alignment to find a suitable base object, instead
20095 of always using drb->base_address.
20097 2018-03-23 Jakub Jelinek <jakub@redhat.com>
20099 PR inline-asm/85022
20100 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
20101 known size by default.
20103 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
20105 PR inline-asm/85030
20106 * lra-constraints.c (process_alt_operands): Don't match BLKmode
20107 and non BLKmode operands.
20109 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20112 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
20113 Clean up attributes.
20115 2018-03-23 Richard Biener <rguenther@suse.de>
20118 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
20119 we are going to emit early debug for LTO.
20121 2018-03-23 Jakub Jelinek <jakub@redhat.com>
20123 PR inline-asm/85034
20124 * function.c (match_asm_constraints_1): Don't optimize if input
20125 doesn't satisfy general_operand predicate for output's mode.
20127 PR inline-asm/85022
20128 * alias.c (write_dependence_p): Don't require for x_canonicalized
20129 non-VOIDmode if x has VOIDmode.
20132 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
20133 just don't try to optimize it rather than assert it never happens.
20135 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
20137 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
20138 macro expansions for definition of ST_INTERNAL_<mode> and
20139 LD_INTERNAL_<mode> builtins.
20140 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
20142 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
20144 (altivec_expand_st_builtin): Likewise.
20145 (altivec_expand_builtin): Remove calls to deleted functions.
20146 (rs6000_address_for_altivec): Delete this function.
20147 * config/rs6000/vector.md: Remove expands for
20148 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
20150 2018-03-22 Sudakshina Das <sudi.das@arm.com>
20153 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
20154 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
20155 re-computing once computed.
20156 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
20157 (arm_init_machine_status): Initialize
20158 machine->static_chain_stack_bytes.
20160 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
20163 * doc/extend.texi: Add four new prototypes for vec_ld.
20164 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
20165 definitions for more logical presentation.
20166 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
20167 entries for V1TI variants of __builtin_altivec_ld builtin.
20168 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
20169 handling of V1TI variant of LVX icode pattern.
20170 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
20171 (rs6000_gimple_fold_builtin): Likewise.
20172 (altivec_init_builtins): Add code to define
20173 __builtin_altivec_lvx_v1ti function.
20175 2018-03-22 Jakub Jelinek <jakub@redhat.com>
20177 PR inline-asm/84941
20178 * function.c (match_asm_constraints_1): Don't do the optimization
20179 if input isn't a REG, SUBREG, MEM or constant.
20181 2018-03-22 Tom de Vries <tom@codesourcery.com>
20183 PR tree-optimization/84956
20184 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
20185 bb_has_abnormal_pred.
20187 2018-03-22 Jakub Jelinek <jakub@redhat.com>
20190 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
20191 DECL_INITIAL (decl) to decl at the end.
20192 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
20193 adjust the comment.
20195 2018-03-21 Joseph Myers <joseph@codesourcery.com>
20197 * doc/extend.texi (__builtin_tgmath): Document when complex
20198 integer types are treated as _Complex _Float64.
20200 2018-03-21 Tom de Vries <tom@codesourcery.com>
20202 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
20204 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20206 PR tree-optimization/84960
20207 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
20208 if it is ENTRY block, move them into single succ of ENTRY in that case.
20210 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
20212 PR tree-optimization/84811
20213 * poly-int.h (poly_span_traits): Remove the T3 parameter and
20214 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
20215 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
20216 (known_subrange_p): Update accordingly. Cast each value involved
20217 in the size comparison, rather than casting the result of the
20220 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20222 PR tree-optimization/84982
20223 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
20224 by flipping the least significant bit rather than all bits from
20225 bitpos to bitpos + bitsize - 1.
20227 2018-03-21 Nathan Sidwell <nathan@acm.org>
20229 * doc/extend.texi (Deprecated Features): Remove mention of
20230 long-deleted deprecations.
20232 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20235 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
20236 * configure: Regenerate.
20238 2018-03-21 Tom de Vries <tom@codesourcery.com>
20240 PR tree-optimization/83126
20241 * tree-parloops.c (num_phis): New function.
20242 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
20244 2018-03-21 Nathan Sidwell <nathan@acm.org>
20246 * doc/extend.texi (Deprecated Features): Update deprecated flags,
20247 mention anon-struct/union members and trailing attributes.
20249 2018-03-21 Bin Cheng <bin.cheng@arm.com>
20251 PR tree-optimization/84969
20252 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
20253 builtin memset partitions if they set different rhs values.
20255 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20257 PR rtl-optimization/84989
20258 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
20259 VEC_DUPLICATE with scalar result mode.
20261 2018-03-21 Martin Liska <mliska@suse.cz>
20264 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
20265 not intended return statement.
20267 2018-03-21 Martin Liska <mliska@suse.cz>
20270 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
20271 (chkp_find_bound_slots_1): Limit number of iterations.
20273 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
20276 * Minor grammar fixes for x86 options.
20278 2018-03-20 Jakub Jelinek <jakub@redhat.com>
20281 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
20282 holding REG_CFA_RESTORE notes, instead turn them into a USE.
20284 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
20287 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
20288 (altivec_lvx_<mode>_1op): Likewise.
20289 (altivec_stvx_<mode>_2op): Likewise.
20290 (altivec_stvx_<mode>_1op): Likewise.
20291 (altivec_lvx_<VM2:mode>): New define_expand.
20292 (altivec_stvx_<VM2:mode>): Likewise.
20293 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
20294 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
20295 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
20296 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
20297 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
20298 (rs6000_gen_lvx): Likewise.
20299 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
20300 (altivec_expand_stv_builtin): Likewise.
20301 (altivec_expand_builtin): Likewise.
20302 * config/rs6000/vector.md: Likewise.
20304 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20307 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
20310 2018-03-20 Richard Biener <rguenther@suse.de>
20313 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
20314 sign-conversions as zero, fall back to standard scalar_stmt
20317 2018-03-20 Martin Liska <mliska@suse.cz>
20320 * predict.c (rebuild_frequencies): Handle case when we have
20321 PROFILE_ABSENT, but flag_guess_branch_prob is false.
20323 2018-03-20 Jakub Jelinek <jakub@redhat.com>
20326 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
20327 flag_section_anchors.
20328 * varasm.c (use_blocks_for_decl_p): Remove hack for
20329 dw2_force_const_mem.
20332 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
20334 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
20335 be created, use lowpart_subreg of operands[0] rather than operands[0]
20337 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
20338 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
20339 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
20340 and n constraint instead of aarch64_shift_imm_di and Usd.
20341 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
20342 (*aarch64_<optab>_reg_minus<mode>3): ... this.
20344 2018-03-20 Sudakshina Das <sudi.das@arm.com>
20347 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
20348 to favor GPR over NEON registers.
20349 (<shift>di3_neon): Likewise.
20351 2018-03-20 Tom de Vries <tom@codesourcery.com>
20354 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
20355 (nvptx_process_pars): Emit bar.sync asap and alap.
20357 2018-03-20 Tom de Vries <tom@codesourcery.com>
20360 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
20361 seen_label if seen_label is already set.
20363 2018-03-20 Jakub Jelinek <jakub@redhat.com>
20366 * config/i386/i386.c (fold_builtin_cpu): For features above 31
20367 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
20368 Use 1U instead of 1. Formatting fixes.
20371 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
20372 instead of TREE_TYPE (s1) for the return value.
20374 2018-03-19 Jakub Jelinek <jakub@redhat.com>
20376 PR tree-optimization/84946
20377 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
20378 bitsize + bitsize in poly_uint64 rather than poly_int64.
20381 * dwarf2asm.c: Include fold-const.c.
20382 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
20383 of decl rather than decl itself.
20385 PR rtl-optimization/84643
20386 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
20388 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
20391 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
20392 calling assemble_variable.
20394 2018-03-19 Sudakshina Das <sudi.das@arm.com>
20397 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
20398 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
20400 2018-03-19 Jim Wilson <jimw@sifive.com>
20403 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
20404 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
20405 (riscv_first_stack_step): Likewise.
20406 (riscv_option_override): Use STACK_BOUNDARY instead of
20407 MIN_STACK_BOUNDARY.
20408 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
20409 MIN_STACK_BOUNDARY.
20410 (BIGGEST_ALIGNMENT): Set to 128.
20411 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
20412 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
20415 2018-03-19 Richard Biener <rguenther@suse.de>
20417 PR tree-optimization/84933
20418 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
20419 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
20421 2018-03-19 Richard Biener <rguenther@suse.de>
20423 PR tree-optimization/84859
20424 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
20425 (cond_if_else_store_replacement): Perform sinking operation on
20426 single-store BBs regardless of MAX_STORES_TO_SINK setting.
20427 Generalize what a BB with a single eligible store is.
20429 2018-03-19 Richard Biener <rguenther@suse.de>
20431 PR tree-optimization/84929
20432 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
20433 chrec_is_positive against non-chrec arg.
20435 2018-03-19 Tamar Christina <tamar.christina@arm.com>
20438 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
20440 2018-03-18 Martin Liska <mliska@suse.cz>
20442 PR rtl-optimization/84635
20443 * regrename.c (build_def_use): Use matches_mode only when
20446 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
20448 PR tree-optimization/84913
20449 * tree-vect-loop.c (vectorizable_reduction): Don't try to
20450 vectorize chains of COND_EXPRs.
20452 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
20454 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
20456 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
20458 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
20460 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
20462 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
20464 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
20465 Kito Cheng <kito.cheng@gmail.com>
20467 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
20468 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
20469 (nds32_adjust_reg_alloc_order): New function.
20470 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
20472 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
20474 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
20475 nds32_print_operand, nds32_print_operand_address): Use
20476 HOST_WIDE_INT_PRINT_DEC instead.
20478 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
20480 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
20482 2018-03-17 Jakub Jelinek <jakub@redhat.com>
20485 * config/i386/i386.c (initial_ix86_tune_features,
20486 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
20487 unsigned long long.
20488 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
20489 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
20490 rather than 1u << ix86_tune. Formatting fix.
20491 (ix86_option_override_internal): Change ix86_arch_mask from
20492 unsigned int to unsigned HOST_WIDE_INT, initialize to
20493 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
20494 (ix86_function_specific_restore): Likewise.
20496 2018-03-16 Jakub Jelinek <jakub@redhat.com>
20499 * postreload.c (reload_combine_recognize_pattern): Perform
20500 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
20501 truncate_int_for_mode the result for the destination's mode.
20504 * hsa-gen.c (mem_type_for_type): Fix comment typo.
20505 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
20507 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
20510 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
20513 * lra-assigns.c (lra_split_hard_reg_for): Don't use
20514 regno_allocno_class_array and sorted_pseudos.
20515 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
20516 insns where regno is used.
20518 2018-03-16 Martin Liska <mliska@suse.cz>
20521 * multiple_target.c (create_dispatcher_calls): Redirect
20522 reference in the symbol table.
20524 2018-03-16 Martin Liska <mliska@suse.cz>
20527 * multiple_target.c (create_dispatcher_calls): Redirect also
20530 2018-03-16 Jakub Jelinek <jakub@redhat.com>
20534 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
20535 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
20536 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
20538 2018-03-16 Julia Koval <julia.koval@intel.com>
20540 * doc/invoke.texi (Skylake Server): Add CLWB.
20541 Cannonlake): Remove CLWB.
20543 2018-03-16 Jakub Jelinek <jakub@redhat.com>
20545 PR tree-optimization/84841
20546 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
20548 (FLOAT_ONE_CONST_TYPE): Define.
20549 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
20550 (sort_by_operand_rank): Put entries with higher constant_type last
20551 rather than first to match comments.
20553 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
20555 * config/nios2/nios2.md (movsi_internal): Fix thinko in
20558 2018-03-15 Jakub Jelinek <jakub@redhat.com>
20561 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
20562 check and use address of target always.
20564 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
20567 * config/i386/i386.c (indirect_thunk_needed): Update comments.
20568 (indirect_thunk_bnd_needed): Likewise.
20569 (indirect_thunks_used): Likewise.
20570 (indirect_thunks_bnd_used): Likewise.
20571 (indirect_return_needed): New.
20572 (indirect_return_bnd_needed): Likewise.
20573 (output_indirect_thunk_function): Add a bool argument for
20575 (output_indirect_thunk_function): Don't generate alias for
20576 function return thunk.
20577 (ix86_code_end): Call output_indirect_thunk_function to generate
20578 function return thunks.
20579 (ix86_output_function_return): Set indirect_return_bnd_needed
20580 and indirect_return_needed instead of indirect_thunk_bnd_needed
20581 and indirect_thunk_needed.
20583 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
20585 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
20586 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
20587 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
20589 2018-03-15 David Malcolm <dmalcolm@redhat.com>
20590 Paul Hua <paul.hua.gm@gmail.com>
20593 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
20595 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
20597 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
20598 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
20599 resp. SFmode cases.
20601 2018-03-15 Tamar Christina <tamar.christina@arm.com>
20604 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
20605 instead of GET_MODE_SIZE when comparing Units.
20607 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
20610 * varasm.c (hash_section): Return an unchangeble hash value
20611 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
20612 Return !aarch64_can_use_per_function_literal_pools_p ().
20614 2018-03-15 Jakub Jelinek <jakub@redhat.com>
20617 * optabs.c (emit_conditional_move): Pass address of cmode's copy
20618 rather than address of cmode as last argument to prepare_cmp_insn.
20620 2018-03-15 Julia Koval <julia.koval@intel.com>
20622 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
20623 F_AVX512VNNI, F_AVX512BITALG): New.
20625 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
20628 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
20629 insn for floating-point loads and stores.
20631 2018-03-14 Carl Love <cel@us.ibm.com>
20633 * config/rs6000/rs6000-c.c: Add macro definitions for
20634 ALTIVEC_BUILTIN_VEC_PERMXOR.
20635 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
20636 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
20637 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
20638 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
20640 * config/doc/extend.texi: Add prototypes for vec_permxor.
20642 2018-03-14 David Malcolm <dmalcolm@redhat.com>
20645 * diagnostic-show-locus.c (class layout_point): Convert m_line
20646 from int to linenum_type.
20647 (line_span::comparator): Use linenum "compare" function when
20648 comparing line numbers.
20649 (test_line_span): New function.
20650 (layout_range::contains_point): Convert param "row" from int to
20652 (layout_range::intersects_line_p): Likewise.
20653 (layout::will_show_line_p): Likewise.
20654 (layout::print_source_line): Likewise.
20655 (layout::should_print_annotation_line_p): Likewise.
20656 (layout::print_annotation_line): Likewise.
20657 (layout::print_leading_fixits): Likewise.
20658 (layout::annotation_line_showed_range_p): Likewise.
20659 (struct line_corrections): Likewise for field m_row.
20660 (line_corrections::line_corrections): Likewise for param "row".
20661 (layout::print_trailing_fixits): Likewise.
20662 (layout::get_state_at_point): Likewise.
20663 (layout::get_x_bound_for_row): Likewise.
20664 (layout::print_line): Likewise.
20665 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
20666 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
20667 * input.c (selftest::test_linenum_comparisons): New function.
20668 (selftest::input_c_tests): Call it.
20669 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
20670 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
20671 * selftest.h (ASSERT_GT): New macro.
20672 (ASSERT_GT_AT): New macro.
20673 (ASSERT_LT): New macro.
20674 (ASSERT_LT_AT): New macro.
20676 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
20678 PR rtl-optimization/84780
20679 * combine.c (distribute_links): Don't make a link based on pc_rtx.
20681 2018-03-14 Martin Liska <mliska@suse.cz>
20683 * tree.c (record_node_allocation_statistics): Use
20684 get_stats_node_kind.
20685 (get_stats_node_kind): New function extracted from
20686 record_node_allocation_statistics.
20687 (free_node): Use get_stats_node_kind.
20689 2018-03-14 Richard Biener <rguenther@suse.de>
20691 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
20692 that the value-set of ANTIC_IN doesn't grow.
20695 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
20697 (BB_VISITED_WITH_VISITED_SUCCS): New define.
20698 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
20700 2018-03-14 Julia Koval <julia.koval@intel.com>
20702 * config.gcc (icelake-client, icelake-server): New.
20704 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
20705 (initial_ix86_arch_features): Ditto.
20706 (PTA_SKYLAKE): Add SGX.
20707 (PTA_ICELAKE): Remove.
20708 (PTA_ICELAKE_CLIENT): New.
20709 (PTA_ICELAKE_SERVER): New.
20710 (ix86_option_override_internal): Split up icelake on icelake client and
20712 (get_builtin_code_for_version): Ditto.
20713 (fold_builtin_cpu): Ditto.
20714 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
20715 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
20716 * config/i386/i386.h (processor_type): Ditto.
20717 * doc/invoke.texi: Ditto.
20719 2018-03-14 Jakub Jelinek <jakub@redhat.com>
20722 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
20723 INTEGER_CST offset, add it together with bitpos / 8 and
20724 sign extend based on POINTER_SIZE.
20728 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20731 * config/i386/constraints.md (Yc): New register constraint.
20732 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
20733 Use Yc constraint for alternative 2 of operand 0. Remove
20734 preferred_for_speed attribute.
20736 2018-03-14 Richard Biener <rguenther@suse.de>
20738 PR tree-optimization/84830
20739 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
20740 with the old one to avoid oscillations.
20742 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
20745 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
20747 (assign_by_spills): Return a flag of reload assignment failure.
20748 Do not process the reload assignment failures. Do not spill other
20749 reload pseudos if they has the same reg class. Update n if
20751 (lra_assign): Add a return arg. Set up from the result of
20752 assign_by_spills call.
20753 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
20754 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
20755 usage_insns if it is not NULL.
20756 (spill_hard_reg_in_range): New function.
20757 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
20758 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
20759 function prototypes.
20760 (lra_assign): Change prototype.
20761 * lra.c (lra): Add code to deal with fails by splitting hard reg
20764 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
20766 * config/riscv/riscv.opt (mrelax): New option.
20767 * config/riscv/riscv.c (riscv_file_start): Emit ".option
20768 "norelax" when riscv_mrelax is disabled.
20769 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
20771 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20774 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
20775 reassociation for int modes.
20777 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
20779 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
20780 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
20782 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
20783 * config/aarch64/aarch64-sve.md
20784 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
20785 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
20786 (*extend<mode><Vwide>2): Rename to...
20787 (aarch64_sve_extend<mode><Vwide>2): ...this.
20788 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
20789 renaming the old pattern to...
20790 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
20792 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
20793 define_expand, renaming the old pattern to...
20794 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
20795 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
20796 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
20797 account when deciding which SVE instruction the optab should use.
20798 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
20800 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
20802 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
20803 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
20804 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
20805 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
20806 (tlsdesc_small_<mode>): Turn a define_expand and use
20807 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
20808 (tlsdesc_small_advsimd_<mode>): ...this.
20809 (tlsdesc_small_sve_<mode>): New pattern.
20811 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
20813 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
20814 (UNSPEC_UMUL_HIGHPART): New constants.
20815 (MUL_HIGHPART): New int iteraor.
20816 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
20817 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
20819 (*<su>mul<mode>3_highpart): New define_insn.
20821 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
20824 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
20827 2018-03-13 Martin Liska <mliska@suse.cz>
20830 * (sem_item_optimizer::sem_item_optimizer): Initialize new
20832 (sem_item_optimizer::~sem_item_optimizer): Release it.
20833 (sem_item_optimizer::merge_classes): Register variable aliases.
20834 (sem_item_optimizer::fixup_pt_set): New function.
20835 (sem_item_optimizer::fixup_points_to_sets): Likewise.
20836 * ipa-icf.h: Declare new variables and functions.
20838 2018-03-13 Jakub Jelinek <jakub@redhat.com>
20840 PR middle-end/84834
20841 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
20842 integer_pow2p@2 and test integer_pow2p in condition.
20843 (A < 0 ? C : 0): Similarly for @1.
20845 PR middle-end/84831
20846 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
20847 characters starting at p contain '\0' character, don't look beyond
20851 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
20852 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
20855 * reg-stack.c (change_stack): Change update_end var from int to
20856 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
20857 also call set_block_for_insn on the newly added insns and rescan.
20860 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
20861 on the last operand.
20864 * tree.c (stabilize_reference_1): Return save_expr (e) for
20865 STATEMENT_LIST even if it doesn't have side-effects.
20867 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
20869 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
20871 2018-03-12 Renlin Li <renlin.li@arm.com>
20873 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
20874 aarch64_output_scalar_simd_mov_immediate.
20876 2018-03-12 Martin Sebor <msebor@redhat.com>
20878 PR tree-optimization/83456
20879 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
20880 for perfectly overlapping calls to memcpy.
20881 (gimple_fold_builtin_memory_chk): Same.
20882 (gimple_fold_builtin_strcpy): Handle no-warning.
20883 (gimple_fold_builtin_stxcpy_chk): Same.
20884 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
20886 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
20888 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
20889 parameter. Use it for SFmode.
20890 (rs6000_function_arg_advance_1): Adjust.
20891 (rs6000_function_arg): Adjust.
20892 (rs6000_gimplify_va_arg): Pass false for that new parameter.
20894 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
20896 PR rtl-optimization/84169
20897 PR rtl-optimization/84780
20898 * combine.c (can_combine_p): Check for a 2-insn combination whether
20899 the destination register is used between the two insns, too.
20901 2018-03-12 Richard Biener <rguenther@suse.de>
20903 PR tree-optimization/84803
20904 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
20905 for refs DR analysis didn't process.
20907 2018-03-12 Richard Biener <rguenther@suse.de>
20909 PR tree-optimization/84777
20910 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
20911 force-vectorize loops ignore whether we are optimizing for size.
20913 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
20915 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
20916 (TARGET_MD_ASM_ADJUST): Define.
20918 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
20919 Kito Cheng <kito.cheng@gmail.com>
20920 Chung-Ju Wu <jasonwucj@gmail.com>
20922 * config/nds32/nds32.c (nds32_compute_stack_frame,
20923 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
20924 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
20925 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
20926 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
20927 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
20928 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
20929 * config/nds32/nds32.md (prologue, epilogue): Use macro
20930 NDS32_V3PUSH_AVAILABLE_P to do checking.
20932 2018-03-11 Jakub Jelinek <jakub@redhat.com>
20935 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
20936 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
20937 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
20938 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
20939 addition of most attributes on !orig_type_die or the attribute not
20940 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
20942 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
20943 Chung-Ju Wu <jasonwucj@gmail.com>
20945 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
20946 __NDS32_VH__ macro.
20947 * config/nds32/nds32.opt (mvh): New option.
20949 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
20950 Chung-Ju Wu <jasonwucj@gmail.com>
20952 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
20954 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
20955 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
20958 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
20959 Chung-Ju Wu <jasonwucj@gmail.com>
20961 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
20963 * config/nds32/nds32-multiple.md (strlensi): New pattern.
20964 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
20966 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
20967 Kito Cheng <kito.cheng@gmail.com>
20968 Chung-Ju Wu <jasonwucj@gmail.com>
20970 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
20971 UNSPEC_FFMISM and UNSPEC_FLMISM.
20972 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
20973 for ffb, ffmism and flmism.
20974 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
20975 (unspec_ffmism): Ditto.
20976 (unspec_flmism): Ditto.
20977 (nds32_expand_builtin_impl): Check if string extension is available.
20978 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
20979 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
20981 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
20984 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
20987 * lra-assigns.c (assign_by_spills): Return a flag of reload
20988 assignment failure. Do not process the reload assignment
20989 failures. Do not spill other reload pseudos if they has the same
20991 (lra_assign): Add a return arg. Set up from the result of
20992 assign_by_spills call.
20993 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
20994 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
20995 usage_insns if it is not NULL.
20996 (spill_hard_reg_in_range): New function.
20997 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
20998 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
20999 function prototypes.
21000 (lra_assign): Change prototype.
21001 * lra.c (lra): Add code to deal with fails by splitting hard reg
21004 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
21007 * config/i386/i386.opt: Replace Enforcment with Enforcement.
21009 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
21012 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
21013 (dw_val_node): Add val_symbolic_view.
21014 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
21015 (symview_upper_bound): New.
21016 (new_line_info_table): Initialize symviews_since_reset.
21017 (dwarf2out_source_line): Count symviews_since_reset and set
21018 symview_upper_bound.
21019 (dw_val_equal_p): Handle symview.
21020 (add_AT_symview): New.
21021 (print_dw_val): Handle symview.
21022 (attr_checksum, attr_checksum_ordered): Likewise.
21023 (same_dw_val_p, size_of_die): Likewise.
21024 (value_format, output_die): Likewise.
21025 (add_high_low_attributes): Use add_AT_symview for entry_view.
21026 (dwarf2out_finish): Reset symview_upper_bound, clear
21029 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
21032 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
21033 Add strict argument and use it.
21034 (rs6000_split_multireg_move): Update for new strict argument.
21035 (mem_operand_gpr): Disallow all non-offsettable addresses.
21036 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
21038 2018-03-09 Jakub Jelinek <jakub@redhat.com>
21041 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
21042 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
21043 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
21046 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
21047 decl, use remap_type if we want to use the type.
21049 2018-03-09 Martin Sebor <msebor@redhat.com>
21051 PR tree-optimization/84526
21052 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
21054 (builtin_access::generic_overlap): Be prepared to handle non-array
21057 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
21059 PR rtl-optimization/84682
21060 * lra-constraints.c (process_address_1): Check is_address flag
21061 for address constraints.
21062 (process_alt_operands): Likewise.
21063 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
21064 preprocess_constraints.
21065 * recog.h (preprocess_constraints): Add oploc parameter.
21067 * recog.c (preprocess_constraints): Test address_operand for
21068 CT_ADDRESS constraints.
21070 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
21073 * lra-assigns.c (assign_by_spills): Return a flag of reload
21074 assignment failure. Do not process the reload assignment
21075 failures. Do not spill other reload pseudos if they has the same
21077 (lra_assign): Add a return arg. Set up from the result of
21078 assign_by_spills call.
21079 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
21080 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
21081 usage_insns if it is not NULL.
21082 (spill_hard_reg_in_range): New function.
21083 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
21084 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
21085 function prototypes.
21086 (lra_assign): Change prototype.
21087 * lra.c (lra): Add code to deal with fails by splitting hard reg
21090 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21093 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
21094 Accept complain bool parameter. Only emit errors if it is true.
21095 (arm_parse_cpu_option_name): Likewise.
21096 (arm_target_thumb_only): Adjust callers of the above.
21097 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
21098 prototype to take a default true bool parameter.
21099 (arm_parse_arch_option_name): Likewise.
21101 2018-03-09 David Malcolm <dmalcolm@redhat.com>
21102 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
21106 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
21107 * configure: Regenerate.
21108 * configure.ac ("linker --version-script option"): New.
21109 ("linker soname option"): New.
21111 2018-03-09 Richard Biener <rguenther@suse.de>
21113 PR tree-optimization/84775
21114 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
21115 immediate uses of predicate stmts and mark them modified.
21118 PR tree-optimization/84178
21119 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
21121 (version_loop_for_if_conversion): Delay update_ssa call.
21122 (tree_if_conversion): Delay update_ssa until after predicate
21125 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
21128 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
21129 when the function accesses prior frames.
21131 2018-03-08 Jakub Jelinek <jakub@redhat.com>
21134 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
21135 gen_llsym, otherwise call maybe_gen_llsym.
21137 PR inline-asm/84742
21138 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
21139 has ',' character inside of it.
21141 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21144 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
21145 as clobbering CC_REGNUM.
21147 2018-03-08 Richard Biener <rguenther@suse.de>
21149 PR middle-end/84552
21150 * tree-scalar-evolution.c: Include tree-into-ssa.h.
21151 (follow_copies_to_constant): Do not follow SSA names registered
21154 2018-03-08 Richard Biener <rguenther@suse.de>
21156 PR tree-optimization/84178
21157 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
21159 (version_loop_for_if_conversion): Delay update_ssa call.
21160 (tree_if_conversion): Delay update_ssa until after predicate
21163 2018-03-08 David Malcolm <dmalcolm@redhat.com>
21165 PR tree-optimization/84178
21166 * tree-if-conv.c (release_bb_predicate): Remove the
21167 the assertion that the stmts have NULL use_ops.
21168 Discard the statements, asserting that they haven't
21169 yet been added to a BB.
21171 2018-03-08 Richard Biener <rguenther@suse.de>
21173 PR tree-optimization/84746
21174 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
21175 (phi_translate): Pass in destination ANTIC_OUT set.
21176 (phi_translate_1): Likewise. For a simplified result lookup
21177 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
21178 (phi_translate_set): Adjust.
21179 (do_pre_regular_insertion): Likewise.
21180 (do_pre_partial_partial_insertion): Likewise.
21182 2018-03-08 Martin Liska <mliska@suse.cz>
21184 PR gcov-profile/84735
21185 * doc/gcov.texi: Document usage of profile files.
21186 * gcov-io.h: Document changes in the format.
21188 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
21192 * dwarf2out.c (struct dw_line_info_table): Update comments for
21194 (FORCE_RESET_NEXT_VIEW): New.
21195 (FORCE_RESETTING_VIEW_P): New.
21196 (RESETTING_VIEW_P): Check for -1 too.
21197 (ZERO_VIEW_P): Likewise.
21198 (new_line_info_table): Force-reset next view.
21199 (dwarf2out_begin_function): Likewise.
21200 (dwarf2out_source_line): Simplify zero_view_p initialization.
21201 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
21202 view directly. Omit view when omitting .loc at line 0.
21204 2018-03-08 Jakub Jelinek <jakub@redhat.com>
21206 PR tree-optimization/84740
21207 * tree-switch-conversion.c (process_switch): Call build_constructors
21208 only if info.phi_count is non-zero.
21210 PR tree-optimization/84739
21211 * tree-tailcall.c (find_tail_calls): Check call arguments against
21212 DECL_ARGUMENTS (current_function_decl) rather than
21213 DECL_ARGUMENTS (func) when checking for tail recursion.
21215 2018-03-07 Jakub Jelinek <jakub@redhat.com>
21217 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
21218 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
21219 Volker Reichelt's entry and add entries for people that perform
21220 GCC fuzzy testing and report numerous bugs.
21222 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21225 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
21226 readonly data in sdata, if that is disabled.
21227 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
21228 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
21229 -mreadonly-in-sdata option.
21231 2018-03-07 Martin Sebor <msebor@redhat.com>
21233 PR tree-optimization/84468
21234 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
21235 basic block when looking for nul assignment.
21237 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
21240 * except.h (output_function_exception_table): Adjust prototype.
21241 * except.c (output_function_exception_table): Remove FNNAME parameter
21242 and add SECTION parameter. Ouput one part of the table at a time.
21243 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
21244 the first part of the exception table and emit unwind directives.
21245 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
21246 (i386_pe_seh_cold_init): Likewise.
21247 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
21248 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
21249 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
21250 (ix86_output_call_insn): Emit a nop in one more case for SEH.
21251 * config/i386/winnt.c: Include except.h.
21252 (struct seh_frame_state): Add reg_offset, after_prologue and
21253 in_cold_section fields.
21254 (i386_pe_seh_end_prologue): Set seh->after_prologue.
21255 (i386_pe_seh_cold_init): New function.
21256 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
21257 to seh->in_cold_section.
21258 (seh_emit_push): Record the offset of the push.
21259 (seh_emit_save): Record the offet of the save.
21260 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
21261 Test seh->after_prologue to disregard the epilogue.
21262 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
21263 (i386_pe_end_cold_function): New function.
21265 2018-03-07 Jakub Jelinek <jakub@redhat.com>
21268 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
21269 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
21272 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
21274 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
21275 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
21277 PR middle-end/84723
21278 * multiple_target.c: Include tree-inline.h and intl.h.
21279 (expand_target_clones): Diagnose and fail if node->definition and
21280 !tree_versionable_function_p (node->decl).
21282 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
21284 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
21286 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
21287 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
21288 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
21290 2018-03-06 Jakub Jelinek <jakub@redhat.com>
21293 * combine.c (try_combine): Use reg_or_subregno instead of handling
21294 just paradoxical SUBREGs and REGs.
21296 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
21298 * config/arc/arc.c (arc_finalize_pic): Remove function.
21299 (arc_must_save_register): We use single base PIC register, remove
21300 checks to save/restore the PIC register.
21301 (arc_expand_prologue): Likewise.
21302 * config/arc/arc-protos.h (arc_set_default_type_attributes):
21304 (arc_verify_short): Likewise.
21305 (arc_attr_type): Likewise.
21306 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
21307 (walk_stores): Likewise.
21308 (arc_address_cost): Make it static.
21309 (arc_verify_short): Likewise.
21310 (branch_dest): Likewise.
21311 (arc_attr_type): Likewise.
21312 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
21313 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
21314 (arc_final_prescan_insn): Remove inserting the nops due to
21315 hardware hazards. It is done in reorg step.
21316 (insn_length_variant_t): Remove.
21317 (insn_length_parameters_t): Likewise.
21318 (arc_insn_length_parameters): Likewise.
21319 (arc_get_insn_variants): Likewise.
21320 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
21322 2018-03-06 Jakub Jelinek <jakub@redhat.com>
21324 PR inline-asm/84683
21325 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
21328 PR tree-optimization/84687
21329 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
21331 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
21333 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21335 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
21336 Rename to ppc_speculation_barrier.
21337 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
21338 __builtin_ppc_speculation_barrier.
21340 2018-03-05 Jakub Jelinek <jakub@redhat.com>
21343 * combine.c (combine_simplify_rtx): Don't try to simplify if
21344 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
21347 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
21349 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
21350 to 32 bytes when compiling for POWER9.
21352 2018-03-05 Jakub Jelinek <jakub@redhat.com>
21355 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
21356 regparm >= 3 with no arg reg available also for calls with
21357 flag_force_indirect_call. Pass decl to ix86_function_regparm.
21360 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
21362 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
21364 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
21367 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
21369 2018-03-05 Richard Biener <rguenther@suse.de>
21371 PR tree-optimization/84486
21372 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
21373 When inserting a __builtin_assume_aligned call set the LHS
21374 SSA name alignment info accordingly.
21376 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
21378 PR tree-optimization/84114
21379 * config/aarch64/aarch64.c (aarch64_reassociation_width)
21380 Avoid reassociation of FLOAT_MODE addition.
21382 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
21384 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
21385 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
21386 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
21387 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
21388 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
21389 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
21390 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
21392 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
21393 __builtin_ia32_wbinvd): New builtins.
21394 (SPECIAL_ARGS2): New.
21395 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
21396 (SPECIAL_ARGS2): New.
21397 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
21398 (ix86_valid_target_attribute_inner_p): Ditto.
21399 (ix86_init_mmx_sse_builtins): Add special_args2.
21400 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
21401 TARGET_WBNOINVD_P): New.
21402 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
21403 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
21404 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
21405 * config/i386/immintrin.h (_wbinvd): New intrinsic.
21406 * config/i386/pconfigintrin.h: New file.
21407 * config/i386/wbnoinvdintrin.h: Ditto.
21408 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
21410 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
21412 2018-03-05 Richard Biener <rguenther@suse.de>
21414 PR tree-optimization/84670
21415 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
21417 (BB_VISITED_WITH_VISITED_SUCCS): New define.
21418 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
21419 (compute_antic_aux): Only assert the number of values in ANTIC_IN
21420 doesn't grow if all successors (recursively) were visited at least
21423 2018-03-05 Richard Biener <rguenther@suse.de>
21425 PR tree-optimization/84650
21426 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
21427 if executed in the loop pipeline.
21429 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
21431 * doc/configfiles.texi (Configuration Files): Move info about
21432 conditionalizing $target-protos.h to...
21433 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
21434 differs from $target-protos.h.
21436 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
21437 Chung-Ju Wu <jasonwucj@gmail.com>
21439 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
21440 * config/nds32/nds32-multiple.md (setmemsi): Define.
21441 * config/nds32/nds32-memory-manipulation.c
21442 (nds32_gen_dup_4_byte_to_word_value): New.
21443 (emit_setmem_word_loop): New.
21444 (emit_setmem_byte_loop): New.
21445 (nds32_expand_setmem_loop): New.
21446 (nds32_expand_setmem_loop_v3m): New.
21447 (nds32_expand_setmem_unroll): New.
21448 (nds32_expand_setmem): New.
21450 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
21451 Chung-Ju Wu <jasonwucj@gmail.com>
21453 * config/nds32/nds32-memory-manipulation.c
21454 (nds32_emit_load_store): New.
21455 (nds32_emit_post_inc_load_store): New.
21456 (nds32_emit_mem_move): New.
21457 (nds32_emit_mem_move_block): New.
21458 (nds32_expand_movmemsi_loop_unknown_size): New.
21459 (nds32_expand_movmemsi_loop_known_size): New.
21460 (nds32_expand_movmemsi_loop): New.
21461 (nds32_expand_movmemsi_unroll): New.
21462 (nds32_expand_movmemqi): Rename ...
21463 (nds32_expand_movmemsi): ... to this.
21464 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
21465 (movmemsi): ... to this.
21466 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
21467 (nds32_expand_movmemsi): ... to this.
21469 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
21470 Monk Chiang <sh.chiang04@gmail.com>
21471 Chung-Ju Wu <jasonwucj@gmail.com>
21473 * config/nds32/nds32-protos.h
21474 (nds32_expand_load_multiple): New arguments.
21475 (nds32_expand_store_multiple): Ditto.
21476 (nds32_valid_multiple_load_store): Rename ...
21477 (nds32_valid_multiple_load_store_p): ... to this.
21478 * config/nds32/nds32-memory-manipulation.c
21479 (nds32_expand_load_multiple): Refine implementation.
21480 (nds32_expand_store_multiple): Ditto.
21481 * config/nds32/nds32-multiple.md
21482 (load_multiple): Update nds32_expand_load_multiple interface.
21483 (store_multiple): Update nds32_expand_store_multiple interface.
21484 * config/nds32/nds32-predicates.c
21485 (nds32_valid_multiple_load_store): Rename ...
21486 (nds32_valid_multiple_load_store_p): ... to this and refine
21488 * config/nds32/predicates.md
21489 (nds32_load_multiple_and_update_address_operation): New predicate.
21490 (nds32_store_multiple_and_update_address_operation): New predicate.
21492 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
21493 Chung-Ju Wu <jasonwucj@gmail.com>
21495 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
21496 (combo): New attribute.
21497 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
21499 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
21501 * config/nds32/nds32.opt: Change -mcmodel= default value.
21503 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
21504 Monk Chiang <sh.chiang04@gmail.com>
21505 Chung-Ju Wu <jasonwucj@gmail.com>
21507 * config/nds32/constants.md (unspec_element): New enum.
21508 * config/nds32/constraints.md (Umw): New constraint.
21509 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
21510 * config/nds32/nds32-intrinsic.md: Likewise.
21511 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
21512 (nds32_valid_smw_lwm_base_p): New.
21513 (nds32_output_smw_single_word): New.
21514 (nds32_output_lmw_single_word): New.
21515 (nds32_expand_unaligned_load): New.
21516 (nds32_expand_unaligned_store): New.
21517 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
21518 (nds32_output_smw_single_word): Declare.
21519 (nds32_output_lmw_single_word): Declare.
21520 (nds32_expand_unaligned_load): Declare.
21521 (nds32_expand_unaligned_store): Declare.
21522 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
21523 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
21524 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
21525 NDS32_BUILTIN_UASTORE_DW.
21526 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
21529 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
21530 Kito Cheng <kito.cheng@gmail.com>
21531 Chung-Ju Wu <jasonwucj@gmail.com>
21533 * config/nds32/nds32-intrinsic.c
21534 (nds32_expand_builtin_null_ftype_reg): Delete.
21535 (nds32_expand_builtin_reg_ftype_imm): Ditto.
21536 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
21537 (nds32_read_argument): New.
21538 (nds32_legitimize_target): Ditto.
21539 (nds32_legitimize_argument): Ditto.
21540 (nds32_check_constant_argument): Ditto.
21541 (nds32_expand_unop_builtin): Ditto.
21542 (nds32_expand_unopimm_builtin): Ditto.
21543 (nds32_expand_binop_builtin): Ditto.
21544 (nds32_builtin_decl_impl): Ditto.
21545 (builtin_description): Ditto.
21546 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
21547 (nds32_init_builtins_impl): Ditto.
21548 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
21549 (nds32_builtin_decl): New.
21550 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
21551 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
21553 2018-03-02 Jeff Law <law@redhat.com>
21555 * reorg.c (stop_search_p): Handle DEBUG_INSN.
21556 (redundant_insn, fill_simple_delay_slots): Likewise.
21557 (fill_slots_from_thread): Likewise.
21558 * resource.c (mark_referenced_resources): Likewise.
21559 (mark_set_resources, find_dead_or_set_registers): Likewise.
21561 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21563 * substring-locations.h (format_warning_va): Formatting fix for
21564 ATTRIBUTE_GCC_DIAG.
21565 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
21567 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
21568 * substring-locations.c: Include intl.h.
21569 (format_warning_va): Turned into small wrapper around
21570 format_warning_n_va, renamed to ...
21571 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
21572 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
21574 (format_warning_at_substring_n): New function.
21575 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
21576 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
21577 format_warning_at_substring with just a shorter name instead of
21578 const function pointer.
21579 (fmtwarn_n): New function.
21580 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
21581 appropriate, get rid of all the fmtstr temporaries, move conditionals
21582 with G_() wrapped string literals directly into fmtwarn arguments,
21583 cast dir.len to (int), formatting fixes.
21585 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
21587 * doc/invoke.texi: Remove "Cilk Plus" references.
21589 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21590 Richard Biener <rguenther@suse.de>
21593 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
21594 for error or warning attributes if CALL_FROM_THUNK_P is set.
21597 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21600 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
21601 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
21604 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
21605 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
21607 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
21608 instead of -1U in last predictors element's probability member.
21610 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
21613 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
21614 arguments if they are comparable.
21616 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
21618 PR tree-optimization/84634
21619 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
21620 masks and masked_loop_p with a single loop_masks, making sure it's
21621 null for bb vectorization.
21623 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
21625 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
21626 (vect_analyze_data_ref_access): Use loop->safe_len rather than
21627 loop->force_vectorize to check whether there is no alias.
21629 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21632 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
21634 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
21636 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
21637 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
21638 instead of a loop around prev_real_insn.
21639 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
21642 PR inline-asm/84625
21643 * config/i386/i386.c (ix86_print_operand): Use conditional
21644 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
21647 2018-03-02 Richard Biener <rguenther@suse.de>
21649 PR tree-optimization/84427
21650 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
21651 (bitmap_set_subtract_values): Rewrite to handle multiple
21654 (prune_clobbered_mems): Likewise.
21655 (phi_translate): Take edge instead of pred/phiblock.
21656 (phi_translate_1): Likewise.
21657 (phi_translate_set): Likewise. Insert all translated
21658 exprs for a value into the set, keeping possibly multiple
21659 expressions per value.
21660 (compute_antic_aux): Adjust for phi_translate changes.
21661 When intersecting union the expressions and prune those
21662 not in the final value set, keeping possibly multiple
21663 expressions per value. Do not use value-insertion
21664 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
21665 all expressions. Add verification that the value-sets
21666 only shrink during iteration.
21667 (compute_partial_antic_aux): Adjust for the phi_translate changes.
21668 (do_pre_regular_insertion): Likewise.
21669 (do_pre_partial_partial_insertion): Likewise.
21671 2018-03-02 Richard Biener <rguenther@suse.de>
21674 * config/darwin.c (saved_debug_info_level): New static global.
21675 (darwin_asm_lto_start): Disable debug info generation for LTO out.
21676 (darwin_asm_lto_end): Restore debug info generation settings.
21678 2018-03-01 Martin Liska <mliska@suse.cz>
21681 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
21682 volatile arguments.
21684 2018-03-01 Richard Biener <rguenther@suse.de>
21687 * dwarf2out.c (gen_variable_die): Properly handle late VLA
21688 type annotation with LTO when debug was disabled at compile-time.
21690 2018-03-01 Matthew Fortune <mfortune@gmail.com>
21692 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
21694 (mips_final_postscan_insn): Likewise.
21696 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
21698 PR rtl-optimization/84528
21699 * alias.c (init_alias_target): Add commentary.
21700 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
21701 a unique base value if the frame pointer is not eliminated
21702 to the stack pointer.
21704 2018-03-01 Tom de Vries <tom@codesourcery.com>
21706 PR rtl-optimization/83327
21707 * lra-int.h (hard_regs_spilled_into): Declare.
21708 * lra.c (hard_regs_spilled_into): Define.
21709 (init_reg_info): Init hard_regs_spilled_into.
21710 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
21711 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
21712 (process_bb_lives): Handle hard_regs_spilled_into.
21713 (lra_create_live_ranges_1): Before doing liveness propagation, clear
21714 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
21716 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
21718 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
21719 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
21720 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
21721 * config/rs6000/aix72.h: New file.
21723 2018-02-28 Jakub Jelinek <jakub@redhat.com>
21725 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
21726 instead of warning_at with conditional singular and plural messages
21730 * stor-layout.c (update_alignment_for_field): For
21731 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
21732 && !DECL_PACKED (field), do the alignment update, just use
21733 only desired_align instead of MAX (type_align, desired_align)
21735 (place_field): Don't do known_align < desired_align handling
21736 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
21737 is non-NULL, instead do it after rli->prev_field handling and
21738 only if not within a bitfield word. For DECL_PACKED (field)
21739 use type_align of BITS_PER_UNIT.
21741 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
21743 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
21744 superfluous parentheses and trailing spaces.
21746 2018-02-28 Richard Biener <rguenther@suse.de>
21748 PR tree-optimization/84584
21749 * graphite-scop-detection.c (scop_detection::add_scop): Discard
21750 SCoPs with fake exit edge.
21752 2018-02-28 Martin Liska <mliska@suse.cz>
21755 * timevar.c (timer::print): Fix format to properly print 100%
21758 2018-02-28 Richard Biener <rguenther@suse.de>
21760 PR middle-end/84607
21761 * genmatch.c (capture_info::walk_match): Do not mark
21762 captured expressions without operands as expr_p given
21763 they act more like predicates and should be subject to
21764 "lost tail" side-effect preserving.
21766 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
21768 PR rtl-optimization/81611
21769 * auto-inc-dec.c (attempt_change): Move dead note from
21770 mem_insn if it's the next use of regno
21771 (find_address): Take address use of reg holding
21772 non-incremented value. Add parm to limit search to the named
21774 (merge_in_block): Attempt to use a mem insn that is the next
21775 use of the original regno.
21777 2018-02-27 Martin Sebor <msebor@redhat.com>
21780 * doc/invoke.texi (-Wmissing-attributes): New option.
21781 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
21783 2018-02-27 Martin Sebor <msebor@redhat.com>
21785 PR translation/84207
21786 * diagnostic-core.h (warning_n, error_n, inform_n): Change
21787 n argument to unsigned HOST_WIDE_INT.
21788 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
21789 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
21790 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
21791 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
21793 2018-02-27 Richard Biener <rguenther@suse.de>
21795 PR tree-optimization/84512
21796 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
21797 Do not use the estimate returned from record_stmt_cost for
21798 the scalar iteration cost but sum properly using add_stmt_cost.
21800 2018-02-27 Richard Biener <rguenther@suse.de>
21802 PR tree-optimization/84466
21803 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
21804 Adjust last change to less strictly validate use operands.
21806 2018-02-27 Martin Liska <mliska@suse.cz>
21808 PR gcov-profile/84548
21809 * gcov.c (process_file): Allow partial overlap and consider it
21810 also as group functions.
21811 (output_lines): Properly calculate range of lines for a group.
21813 2018-02-27 Martin Liska <mliska@suse.cz>
21815 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
21816 'ggc' suffixes. Change first column width.
21817 (timer::print): Fix formatting of the column.
21819 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
21821 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
21822 preserve inline entry blocks for the sake of debug inline
21823 entry point markers alone.
21824 (remove_unused_locals): Suggest in comments a better place to
21825 force the preservation of inline entry blocks that are
21826 otherwise unused, but do not preserve them.
21828 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
21830 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
21832 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
21835 * config/i386/constraints.md (Bs): Replace
21836 ix86_indirect_branch_register with
21837 TARGET_INDIRECT_BRANCH_REGISTER.
21839 * config/i386/i386.md (indirect_jump): Likewise.
21840 (tablejump): Likewise.
21841 (*sibcall_memory): Likewise.
21842 (*sibcall_value_memory): Likewise.
21843 Peepholes of indirect call and jump via memory: Likewise.
21844 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
21845 (*sibcall_value_GOT_32): Likewise.
21846 * config/i386/predicates.md (indirect_branch_operand): Likewise.
21847 (GOT_memory_operand): Likewise.
21848 (call_insn_operand): Likewise.
21849 (sibcall_insn_operand): Likewise.
21850 (GOT32_symbol_operand): Likewise.
21851 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
21853 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
21855 PR rtl-optimization/83496
21856 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
21857 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
21858 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
21859 redundant insn, if any.
21860 (relax_delay_slots): Likewise.
21861 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
21863 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
21865 PR tree-optimization/83965
21866 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
21867 that grouped statements are part of a reduction chain. Return
21868 true if the statement is not marked as a reduction itself but
21869 is part of a group.
21870 (vect_recog_dot_prod_pattern): Don't check whether the statement
21871 is part of a group here.
21872 (vect_recog_sad_pattern): Likewise.
21873 (vect_recog_widen_sum_pattern): Likewise.
21875 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
21878 * final.c (rest_of_clean_state): Also look for calls inside sequences.
21880 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
21883 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
21885 (ix86_output_indirect_function_return): New prototype.
21886 (ix86_split_simple_return_pop_internal): Likewise.
21887 * config/i386/i386.c (indirect_return_via_cx): New.
21888 (indirect_return_via_cx_bnd): Likewise.
21889 (indirect_thunk_name): Handle return va CX_REG.
21890 (output_indirect_thunk_function): Create alias for
21891 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
21892 (ix86_output_indirect_jmp): Remove the bool argument.
21893 (ix86_output_indirect_function_return): New function.
21894 (ix86_split_simple_return_pop_internal): Likewise.
21895 * config/i386/i386.md (*indirect_jump): Don't pass false
21896 to ix86_output_indirect_jmp.
21897 (*tablejump_1): Likewise.
21898 (simple_return_pop_internal): Change it to define_insn_and_split.
21899 Call ix86_split_simple_return_pop_internal to split it for
21900 -mfunction-return=.
21901 (simple_return_indirect_internal): Call
21902 ix86_output_indirect_function_return instead of
21903 ix86_output_indirect_jmp.
21905 2018-02-26 Jakub Jelinek <jakub@redhat.com>
21908 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
21909 memset and value initialization afterwards.
21911 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
21913 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
21915 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21918 * common/config/aarch64/aarch64-common.c
21919 (aarch_option_optimization_table[]): Switch
21920 off fomit-frame-pointer
21922 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
21923 Chung-Ju Wu <jasonwucj@gmail.com>
21925 * config/nds32/nds32-multiple.md (load_multiple): Disallow
21927 (store_multiple): Ditto.
21929 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
21931 * config.gcc: Add --with-cpu support for nds32 target.
21932 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
21933 * config/nds32/nds32.opt: Add -mcpu= option.
21935 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
21937 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
21938 isel=yes): Warn for these deprecated options.
21940 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
21942 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
21943 ISA_2_5_MASKS_EMBEDDED.
21945 2018-02-23 Jakub Jelinek <jakub@redhat.com>
21947 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
21948 p->max as pointers rather than using iterative_hash_expr.
21950 2018-02-23 Carl Love <cel@us.ibm.com>
21952 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
21953 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
21955 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
21956 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
21957 P8V_BUILTIN_VEC_VUNSIGNED2.
21959 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
21962 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
21963 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
21965 * lra-constraints.c (curr_insn_transform): Set up
21966 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
21968 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
21969 * lra-eliminations.c (spill_pseudos): Ditto.
21970 (process_insn_for_elimination): Ditto.
21971 * lra-lives.c (reg_early_clobber_p): Use the new macros.
21972 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
21973 LRA_NON_CLOBBERED_ALT.
21975 2018-02-22 Martin Sebor <msebor@redhat.com>
21977 PR tree-optimization/84480
21978 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
21979 to maybe_diag_stxncpy_trunc. Call it.
21980 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
21981 from gimple_fold_builtin_strcpy. Print inlining stack.
21982 (handle_builtin_stxncpy): Print inlining stack.
21983 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
21985 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
21988 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
21989 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
21990 and -fcheck-pointer-bounds are used together.
21991 (indirect_thunk_prefix): New enum.
21992 (indirect_thunk_need_prefix): New function.
21993 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
21994 "_nt" instead of "_bnd" for NOTRACK prefix.
21995 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
21996 (output_indirect_thunk_function): Likewise.
21998 (ix86_code_end): Update output_indirect_thunk_function calls.
21999 (ix86_output_indirect_branch_via_reg): Replace
22000 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
22001 (ix86_output_indirect_branch_via_push): Likewise.
22002 (ix86_output_function_return): Likewise.
22003 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
22004 incompatible with -fcf-protection=branch and
22005 -fcheck-pointer-bounds.
22007 2018-02-22 Steve Ellcey <sellcey@cavium.com>
22010 * config/aarch64/aarch64.c (aarch64_print_address_internal):
22011 Change gcc_assert call to output_operand_lossage.
22013 2018-02-22 Steve Ellcey <sellcey@cavium.com>
22015 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
22017 2018-02-22 DJ Delorie <dj@redhat.com>
22018 Sebastian Perta <sebastian.perta@renesas.com>
22019 Oleg Endo <olegendo@gcc.gnu.org>
22021 * config/rx/rx.c (rx_rtx_costs): New function.
22022 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
22024 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
22026 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
22028 2018-02-22 Martin Liska <mliska@suse.cz>
22031 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
22032 Add "native" as a possible value.
22034 2018-02-22 Martin Liska <mliska@suse.cz>
22037 * config/i386/i386.c (ix86_option_override_internal):
22038 Add "native" as a possible value for -march and -mtune.
22040 2018-02-22 Jakub Jelinek <jakub@redhat.com>
22043 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
22044 to all type variants.
22046 PR tree-optimization/84503
22047 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
22048 width as info->bitpos + info->bitsize - start.
22049 (merged_store_group::merge_overlapping): Simplify width computation.
22050 (check_no_overlap): New function.
22051 (imm_store_chain_info::try_coalesce_bswap): Compute expected
22052 start + width and last_order of the group, fail if check_no_overlap
22054 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
22055 to group if check_no_overlap fails.
22057 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
22059 * config/rs6000/altivec.md: Delete contraint arguments to
22060 define_expand, define_split, and define_peephole2, and in
22061 define_insn_and_split if always unused.
22062 * config/rs6000/darwin.md: Ditto.
22063 * config/rs6000/dfp.md: Ditto.
22064 * config/rs6000/rs6000.md: Ditto.
22065 * config/rs6000/sync.md: Ditto.
22066 * config/rs6000/vector.md: Ditto.
22067 * config/rs6000/vsx.md: Ditto.
22069 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
22071 * config/rs6000/altivec.md: Write output control strings as braced
22072 blocks instead of double-quoted strings.
22073 * config/rs6000/darwin.md: Ditto.
22074 * config/rs6000/rs6000.md: Ditto.
22075 * config/rs6000/vector.md: Ditto.
22076 * config/rs6000/vsx.md: Ditto.
22078 2018-02-21 Jason Merrill <jason@redhat.com>
22080 PR c++/84314 - ICE with templates and fastcall attribute.
22081 * attribs.c (build_type_attribute_qual_variant): Remove assert.
22083 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
22085 * ipa-cp.c (determine_versionability): Fix comment typos.
22087 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
22090 * ipa-cp.c (determine_versionability): Do not version functions caling
22093 2018-02-21 Martin Liska <mliska@suse.cz>
22096 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
22097 Add "native" as a possible value.
22098 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
22099 the macro when native cpu detection is available.
22101 2018-02-21 Martin Liska <mliska@suse.cz>
22104 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
22105 Add "native" as a possible value.
22106 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
22107 when native cpu detection is available.
22109 2018-02-21 Jakub Jelinek <jakub@redhat.com>
22110 Martin Sebor <msebor@redhat.com>
22112 PR tree-optimization/84478
22113 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
22115 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
22116 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
22117 support which is conservatively correct, for 2 only stay conservative
22118 for maxlen. Formatting and comment capitalization fixes. Add STRICT
22119 argument to the 2 argument get_range_strlen, adjust 6 arg
22120 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
22122 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
22123 (gimple_fold_builtin_strlen): Pass true as last argument to
22126 2018-02-20 Martin Sebor <msebor@redhat.com>
22128 PR middle-end/84095
22129 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
22130 (builtin_memref::set_base_and_offset): Same. Handle inner references.
22131 (builtin_memref::builtin_memref): Factor out parts into
22132 set_base_and_offset and call it.
22134 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
22136 PR middle-end/84406
22137 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
22138 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
22139 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
22140 search at the associated MODE_INT.
22142 2018-02-20 Jeff Law <law@redhat.com>
22144 PR middle-end/82123
22145 PR tree-optimization/81592
22146 PR middle-end/79257
22147 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
22148 for range data rather than using global data.
22149 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
22150 range data rather than using global data.
22151 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
22152 pass it to children as needed.
22153 (struct directive::fmtresult): Similarly.
22154 (struct directive::set_width): Similarly.
22155 (struct directive::set_precision): Similarly.
22156 (format_integer, format_directive, parse_directive): Similarly.
22157 (format_none): Accept unnamed vr_values parameter.
22158 (format_percent, format_floating, format_character): Similarly.
22159 (format_string, format_plain): Similarly.
22160 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
22161 the EVRP range analyzer for range data rather than using global data.
22162 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
22163 gimple-ssa-evrp-analyze.h
22164 (class sprintf_dom_walker): Add after_dom_children member function.
22165 Add evrp_range_analyzer member.
22166 (sprintf_dom_walker::before_dom_children): Call into the EVRP
22167 range analyzer as needed.
22168 (sprintf_dom_walker::after_dom_children): New member function.
22169 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
22171 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
22172 (evrp_range_analyzer::pop_to_marker): Likewise.
22174 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
22176 PR tree-optimization/84419
22177 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
22178 with the required type if its current type is compatible but
22181 2018-02-20 Jakub Jelinek <jakub@redhat.com>
22183 PR middle-end/82004
22184 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
22185 after vectorization.
22187 2018-02-20 Martin Liska <mliska@suse.cz>
22190 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
22191 possible values if we don't have a hint.
22193 2018-02-20 Martin Liska <mliska@suse.cz>
22197 * final.c (shorten_branches): Build align_tab array with one
22199 * opts.c (finish_options): Add alignment option limit check.
22200 (MAX_CODE_ALIGN): Likewise.
22201 (MAX_CODE_ALIGN_VALUE): Likewise.
22202 * doc/invoke.texi: Document maximum allowed option value for
22203 all -falign-* options.
22205 2018-02-19 Jakub Jelinek <jakub@redhat.com>
22208 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
22209 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
22210 * var-tracking.c (emit_note_insn_var_location): Remove all references
22211 to NOTE_INSN_CALL_ARG_LOCATION.
22212 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
22213 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
22214 Use copy_rtx_if_shared.
22215 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
22216 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
22217 (dwarf2out_var_location): Remove handling of
22218 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
22220 * final.c (final_scan_insn): Remove all references to
22221 NOTE_INSN_CALL_ARG_LOCATION.
22222 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
22223 before dumping final insns.
22224 * except.c (emit_note_eh_region_end): Remove all references to
22225 NOTE_INSN_CALL_ARG_LOCATION.
22226 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
22227 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
22228 * config/arc/arc.c (hwloop_optimize): Likewise.
22229 * config/arm/arm.c (create_fix_barrier): Likewise.
22230 * config/s390/s390.c (s390_chunkify_start): Likewise.
22231 * config/sh/sh.c (find_barrier): Likewise.
22232 * config/i386/i386.c (rest_of_insert_endbranch,
22233 ix86_seh_fixup_eh_fallthru): Likewise.
22234 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
22235 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
22236 * config/frv/frv.c (frv_function_prologue): Likewise.
22237 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
22239 (note_outside_basic_block_p): Remove all references to
22240 NOTE_INSN_CALL_ARG_LOCATION.
22241 * gengtype.c (adjust_field_rtx_def): Likewise.
22242 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
22244 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
22245 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
22248 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
22251 PR tree-optimization/84452
22252 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
22253 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
22256 2018-02-19 Martin Liska <mliska@suse.cz>
22259 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
22261 2018-02-19 Martin Liska <mliska@suse.cz>
22262 Richard Sandiford <richard.sandiford@linaro.org>
22264 PR tree-optimization/82491
22265 * gimple-fold.c (get_base_constructor): Make earlier bail out
22268 2018-02-19 Carl Love <cel@us.ibm.com>
22270 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
22271 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
22273 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
22274 P8V_BUILTIN_VEC_NEG.
22276 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
22278 * config/rl78/rl78.md (movdf): New define expand.
22280 2018-02-19 Martin Liska <mliska@suse.cz>
22283 * doc/invoke.texi: Fix typo.
22284 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
22286 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
22288 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
22289 handle rs6000_single_float and rs6000_double_float specially for
22292 2018-02-16 Jeff Law <law@redhat.com>
22294 * config/rx/rx.c (add_pop_cfi_notes): New function.;
22295 (pop_regs): Use it.
22297 2018-02-16 Jakub Jelinek <jakub@redhat.com>
22300 * ipa-inline.c (inline_small_functions): Fix a typo.
22302 2018-02-16 Nathan Sidwell <nathan@acm.org>
22304 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
22306 2018-02-16 Carl Love <cel@us.ibm.com>
22308 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
22309 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
22310 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
22311 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
22312 expansion to P8V_BUILTIN_VEC_FLOAT2.
22314 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
22316 PR rtl-optimization/70023
22317 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
22318 src_regno into account.
22320 2018-02-16 Carl Love <cel@us.ibm.com>
22322 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
22323 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
22324 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
22325 * config/rs6000/rs6000.c: Remove case statements for
22326 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
22327 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
22328 and P9V_BUILTIN_VEC_VINSERT4B.
22329 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
22330 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
22331 * config/rs6000/vsx.md:
22332 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
22335 2018-02-16 Carl Love <cel@us.ibm.com>
22337 * config/rs6000/altivec.h: Add builtin names vec_extract4b
22339 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
22341 * config/rs6000/rs6000-c.c: Add the definitions for
22342 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
22343 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
22344 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
22345 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
22346 definition for insert4b and define insn *insert3b_internal.
22347 * doc/extend.texi: Add documentation for vec_extract4b.
22349 2018-02-16 Nathan Sidwell <nathan@acm.org>
22351 * doc/extend.texi (Backwards Compatibility): Mention friend
22352 injection. Note for-scope is deprecated.
22353 * doc/invoke.texi (-ffriend-injection): Deprecate.
22355 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
22357 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
22358 that moved to I2, also allow destinations that are a paradoxical
22359 subreg (instead of a normal reg).
22361 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
22364 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
22367 2018-02-16 Richard Biener <rguenther@suse.de>
22369 PR tree-optimization/84037
22370 PR tree-optimization/84016
22372 * config/i386/i386.c (ix86_builtin_vectorization_cost):
22373 Adjust vec_construct for the fact we need additional higher latency
22374 128bit inserts for AVX256 and AVX512 vector builds.
22375 (ix86_add_stmt_cost): Scale vector construction cost for
22378 2018-02-16 Richard Biener <rguenther@suse.de>
22380 PR tree-optimization/84417
22381 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
22382 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
22383 (non_rewritable_lvalue_p): Likewise, use poly-ints.
22385 2018-02-16 Martin Liska <mliska@suse.cz>
22388 * internal-fn.def (ASAN_CHECK): Set proper flags.
22389 (ASAN_MARK): Likewise.
22391 2018-02-16 Julia Koval <julia.koval@intel.com>
22393 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
22394 from PTA_CANNONLAKE.
22396 2018-02-16 Jakub Jelinek <jakub@redhat.com>
22399 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
22400 Use ++iter rather than iter++ for std::list iterators.
22401 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
22402 defer deleting them until all nodes in the forest are processed. Do
22403 free even leaf nodes. Change to_process into auto_vec.
22406 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
22407 * vec.h (vec_default_construct): Use memset instead of placement new
22408 if BROKEN_VALUE_INITIALIZATION is defined.
22409 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
22410 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
22413 PR rtl-optimization/83723
22414 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
22415 * lra.c (lra_substitute_pseudo): Likewise. If true, use
22416 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
22418 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
22420 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
22422 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
22424 PR rtl-optimization/81443
22425 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
22426 from inner REGs to paradoxical SUBREGs.
22428 2018-02-16 Richard Biener <rguenther@suse.de>
22430 PR tree-optimization/84399
22431 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
22432 For operands we can analyze at their definition make sure we can
22433 analyze them at each use as well.
22435 2018-02-16 Richard Biener <rguenther@suse.de>
22437 PR tree-optimization/84190
22438 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
22439 volatile accesses if the decl isn't volatile.
22441 2018-02-15 Jason Merrill <jason@redhat.com>
22443 PR c++/84314 - ICE with templates and fastcall attribute.
22444 * attribs.c (build_type_attribute_qual_variant): Don't clobber
22445 TYPE_CANONICAL on an existing type.
22447 2018-02-15 Jakub Jelinek <jakub@redhat.com>
22449 PR tree-optimization/84383
22450 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
22451 dstoff nor call operand_equal_p if dstbase is NULL.
22453 PR tree-optimization/84334
22454 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
22455 also a CONSTANT_CLASS_P, punt.
22457 2018-02-14 Jim Wilson <jimw@sifive.com>
22459 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
22460 first SMALL_OPERAND check. New local min_second_step. Move assert
22461 to where locals are set. Add TARGET_RVC support.
22462 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
22464 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
22466 * doc/invoke.texi: Correct -Wformat-overflow code sample.
22468 2018-02-14 Martin Sebor <msebor@redhat.com>
22470 PR tree-optimization/83698
22471 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
22472 arrays constrain the offset range to their bounds.
22473 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
22474 (builtin_access::overlap): Avoid setting the size of overlap if it's
22476 (maybe_diag_overlap): Also consider arrays when deciding what values
22477 of offsets to include in diagnostics.
22479 2018-02-14 Martin Sebor <msebor@redhat.com>
22482 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
22483 that correspond to the kind of a declaration.
22485 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
22488 * config/pa/pa.md: Load address of PIC label using the linkage table
22489 if the label is nonlocal.
22491 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
22493 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
22494 warning message if user requests -maltivec=be.
22495 * doc/invoke.texi: Document deprecation of -maltivec=be.
22497 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
22500 * config/rs6000/rs6000-c.c: Update definitions for
22501 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
22502 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
22504 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
22507 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
22508 add _get_ssp intrinsics. Remove argument from
22509 __builtin_ia32_rdssp[d|q].
22510 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
22511 * config/i386/i386-builtin.def: Remove argument from
22512 __builtin_ia32_rdssp[d|q].
22513 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
22514 ix86_expand_special_args_builtin for _rdssp[d|q].
22515 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
22516 Clear register before usage.
22517 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
22518 Add documentation for new _get_ssp and _inc_ssp intrinsics.
22520 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
22522 PR tree-optimization/84357
22523 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
22524 operand 1 of an ARRAY_REF too.
22526 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
22529 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
22530 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
22532 (set_of_reg): New struct.
22533 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
22534 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
22535 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
22537 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
22538 Split into bitclr, bitset, bitinvert patterns if appropriate.
22539 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
22540 use rx_fuse_in_memory_bitop.
22541 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
22542 to named insn, correct maximum insn length.
22544 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
22547 * machmode.def: Define a complex mode for PARTIAL_INT.
22548 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
22550 * doc/rtl.texi: Document CSPImode.
22551 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
22553 (msp430_hard_regno_nregs_with_padding): Likewise.
22555 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
22558 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
22560 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
22562 PR rtl-optimization/84169
22563 * combine.c (try_combine): New variable split_i2i3. Set it to true if
22564 we generated a parallel as new i3 and we split that to new i2 and i3
22565 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
22566 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
22567 those to i2, not i1. Partially rewrite this scan code.
22569 2018-02-13 Jakub Jelinek <jakub@redhat.com>
22572 * stor-layout.c (place_field): For variable length fields, adjust
22573 offset_align afterwards not just based on the field's alignment,
22574 but also on the size.
22576 PR middle-end/84309
22577 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
22578 of exps and logs in the use_exp2 case.
22580 2018-02-13 Jeff Law <law@redhat.com>
22582 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
22583 entry for "vector".
22585 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
22588 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
22592 * common.opt (gas-loc-support, gas-locview-support): New.
22593 (ginline-points, ginternal-reset-location-views): New.
22594 * doc/invoke.texi: Document them. Use @itemx where intended.
22595 (gvariable-location-views): Adjust.
22596 * target.def (reset_location_view): New.
22597 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
22598 (TARGET_RESET_LOCATION_VIEW): New.
22599 * doc/tm.texi: Rebuilt.
22600 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
22601 (dwarf2out_default_as_locview_support): New.
22602 (output_asm_line_debug_info): Use option variables.
22603 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
22604 (output_loc_list): Likewise.
22605 (add_high_low_attributes): Check option variables.
22606 Don't output entry view attribute in strict mode.
22607 (gen_inlined_subroutine_die): Check option variables.
22608 (dwarf2out_inline_entry): Likewise.
22609 (init_sections_and_labels): Likewise.
22610 (dwarf2out_early_finish): Likewise.
22611 (maybe_reset_location_view): New, from...
22612 (dwarf2out_var_location): ... here. Call it.
22613 * debug.h (dwarf2out_default_as_loc_support): Declare.
22614 (dwarf2out_default_as_locview_support): Declare.
22615 * hooks.c (hook_int_rtx_insn_0): New.
22616 * hooks.h (hook_int_rtx_insn_0): Declare.
22617 * toplev.c (process_options): Take -gas-loc-support and
22618 -gas-locview-support from dwarf2out. Enable
22619 -gvariable-location-views by default only with locview
22620 assembler support. Enable -ginternal-reset-location-views by
22621 default only if the target defines the corresponding hook.
22622 Enable -ginline-points by default if location views are
22623 enabled; force it disabled if statement frontiers are
22625 * tree-inline.c (expand_call_inline): Check option variables.
22626 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
22628 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
22630 PR tree-optimization/84321
22631 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
22632 handling. Also check whether the anti-range contains any values
22633 that satisfy the mask; switch to a VR_RANGE if not.
22635 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
22638 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
22640 2018-02-13 Martin Jambor <mjambor@suse.cz>
22643 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
22644 of call statements, also set location of a load to a temporary.
22646 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
22648 * config/rl78/rl78.c (add_vector_labels): New function.
22649 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
22650 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
22651 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
22652 which checks that no arguments are passed.
22653 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
22654 * doc/extend.texi: Documentation for the new attribute.
22656 2018-02-13 Andreas Schwab <schwab@suse.de>
22658 * config/riscv/linux.h (CPP_SPEC): Define.
22660 2018-02-13 Jakub Jelinek <jakub@redhat.com>
22663 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
22664 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
22665 OPTION_MASK_ISA_AES as first argument to def_builtin_const
22666 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
22667 instead of OPTION_MASK_ISA_PCLMUL as first argument to
22668 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
22669 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
22670 temporarily for AES and PCLMUL builtins.
22672 PR tree-optimization/84339
22673 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
22674 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
22677 PR middle-end/84309
22678 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
22679 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
22680 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
22682 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
22684 * omp-simd-clone.h: New file.
22685 * omp-simd-clone.c: Include omp-simd-clone.h.
22686 (expand_simd_clones): No longer static.
22687 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
22688 cgraph.h and omp-simd-clone.h.
22689 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
22690 (vect_recog_widen_shift_pattern): Formatting fix.
22691 (vect_pattern_recog_1): Don't check optab for calls.
22694 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
22695 operands[2] into a REG before using gen_lowpart on it.
22697 2018-02-12 Jeff Law <law@redhat.com>
22700 * config/sh/sh.c (find_barrier): Consider a sibling call
22703 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
22704 successfully back substituting a reg.
22706 2018-02-12 Richard Biener <rguenther@suse.de>
22708 PR tree-optimization/84037
22709 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
22710 parameter, move visited init to caller.
22711 (vect_slp_analyze_operations): Separate cost from validity
22712 check, initialize visited once for all instances.
22713 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
22715 * tree-vect-stmts.c (vect_model_simple_cost): Make early
22717 (vect_model_promotion_demotion_cost): Likewise.
22718 (vectorizable_bswap): Guard cost modeling with !slp_node
22719 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
22721 (vectorizable_call): Likewise.
22722 (vectorizable_conversion): Likewise.
22723 (vectorizable_assignment): Likewise.
22724 (vectorizable_shift): Likewise.
22725 (vectorizable_operation): Likewise.
22726 (vectorizable_store): Likewise.
22727 (vectorizable_load): Likewise.
22728 (vectorizable_condition): Likewise.
22729 (vectorizable_comparison): Likewise.
22731 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
22734 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
22735 (ASAN_MARK): Fix fnspec to account for return value, change pointer
22736 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
22738 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
22740 PR middle-end/83665
22741 * params.def (inline-min-speedup): Increase from 8 to 15.
22742 (max-inline-insns-auto): Decrease from 40 to 30.
22743 * ipa-split.c (consider_split): Add some buffer for function to
22744 be considered inlining candidate.
22745 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
22748 2018-02-12 Richard Biener <rguenther@suse.de>
22750 PR tree-optimization/84037
22751 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
22752 matched stmts if we cannot swap the non-matched ones.
22754 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
22756 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
22757 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
22758 _mm_maskz_scalef_round_ss): New intrinsics.
22759 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
22760 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
22761 __builtin_ia32_scalefss_round): Remove.
22762 (__builtin_ia32_scalefsd_mask_round,
22763 __builtin_ia32_scalefss_mask_round): New intrinsics.
22764 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
22765 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
22766 ((match_operand:VF_128 2 "<round_nimm_predicate>"
22767 "<round_constraint>")): Changed to ...
22768 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
22769 "<round_scalar_constraint>")): ... this.
22770 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
22771 %0, %1, %2<round_op3>}"): Changed to ...
22772 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
22773 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
22774 %2<round_scalar_mask_op3>}"): ... this.
22775 * config/i386/subst.md (round_scalar_nimm_predicate): New.
22777 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
22779 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
22780 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
22781 (_mm_maskz_sqrt_round_ss): New intrinsics.
22782 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
22783 (__builtin_ia32_sqrtsd_mask_round)
22784 (__builtin_ia32_sqrtss_mask_round): New builtins.
22785 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
22786 (__builtin_ia32_sqrtss_round): Remove.
22787 (__builtin_ia32_sqrtsd_mask_round)
22788 (__builtin_ia32_sqrtss_mask_round): New builtins.
22789 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
22790 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
22791 ((match_operand:VF_128 1 "vector_operand"
22792 "xBm,<round_constraint>")): Changed to ...
22793 ((match_operand:VF_128 1 "vector_operand"
22794 "xBm,<round_scalar_constraint>")): ... this.
22795 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
22796 %0, %2, %<iptr>1<round_op3>}): Changed to ...
22797 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
22798 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
22799 %<iptr>1<round_scalar_mask_op3>}): ... this.
22800 ((set_attr "prefix" "<round_prefix>")): Changed to ...
22801 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
22803 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
22806 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
22807 Cast vec_cmpeq result to correct type.
22808 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
22809 Cast vec_cmpgt result to correct type.
22811 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
22813 * final.c (final_scan_insn_1): Renamed from...
22814 (final_scan_insn): ... this. New wrapper, to recover
22815 seen from the outermost call in recursive ones.
22816 * config/sparc/sparc.c (output_return): Drop seen from call.
22817 (output_sibcall): Likewise.
22818 * config/visium/visium.c (output_branch): Likewise.
22820 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
22822 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
22825 2018-02-10 Alan Modra <amodra@gmail.com>
22828 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
22829 Specify LR as an input.
22831 2018-02-10 Jakub Jelinek <jakub@redhat.com>
22834 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
22835 remove_member_access_dummy_vars): New functions.
22836 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
22837 lower_omp_1, execute_lower_omp): Use them.
22839 PR rtl-optimization/84308
22840 * shrink-wrap.c (spread_components): Release todo vector.
22842 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
22844 PR rtl-optimization/57193
22845 * ira-color.c (struct allocno_color_data): Add member
22846 conflict_allocno_hard_prefs.
22847 (update_conflict_allocno_hard_prefs): New.
22848 (bucket_allocno_compare_func): Add a preference based on
22849 conflict_allocno_hard_prefs.
22850 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
22851 (color_allocnos): Remove a dead code. Initiate
22852 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
22854 2018-02-09 Jakub Jelinek <jakub@redhat.com>
22857 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
22858 constraint from =wa to wa. Avoid a subreg on the output operand,
22859 instead use a pseudo and subreg it in a move.
22860 (p9_xxbrd_<mode>): Changed to ...
22861 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
22862 (p9_xxbrd_v2df): New expander.
22863 (p9_xxbrw_<mode>): Changed to ...
22864 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
22865 (p9_xxbrw_v4sf): New expander.
22867 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
22869 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
22871 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
22874 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
22875 multiply in 32-bit mode.
22876 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
22877 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
22880 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
22882 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
22883 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
22884 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
22885 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
22887 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
22890 * dwarf2out.c (is_trivial_indirect_ref): New function.
22891 (dwarf2out_late_global_decl): Do not generate a location
22892 attribute for variables that have a non-trivial DECL_VALUE_EXPR
22893 and that are not defined in the current unit.
22895 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
22897 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
22898 instead of a libcall for UNORDERED.
22900 2018-02-09 Tamar Christina <tamar.christina@arm.com>
22903 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
22904 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
22906 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22909 * config/s390/s390.c (s390_set_current_function): Invoke
22910 s390_indirect_branch_settings also if fndecl didn't change.
22912 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
22914 * config/rs6000/rs6000.md (blockage): Set length to zero.
22916 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
22918 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
22920 2018-02-09 Jakub Jelinek <jakub@redhat.com>
22923 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
22924 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
22928 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
22929 PARALLEL incoming that failed vt_get_decl_and_offset check.
22931 PR middle-end/84237
22932 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
22933 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
22935 (get_variable_section): For decls in named .bss* sections pass true as
22936 second argument to bss_initializer_p.
22938 2018-02-09 Marek Polacek <polacek@redhat.com>
22939 Jakub Jelinek <jakub@redhat.com>
22942 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
22943 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
22944 Sync some changes from cxx_fold_indirect_ref.
22946 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
22948 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
22950 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
22951 (BLOCK_INLINE_ENTRY_LABEL): New.
22952 (dwarf2out_var_location): Disregard inline entry markers.
22953 (inline_entry_data): New struct.
22954 (inline_entry_data_hasher): New hashtable type.
22955 (inline_entry_data_hasher::hash): New.
22956 (inline_entry_data_hasher::equal): New.
22957 (inline_entry_data_table): New variable.
22958 (add_high_low_attributes): Add DW_AT_entry_pc and
22959 DW_AT_GNU_entry_view attributes if a pending entry is found
22960 in inline_entry_data_table. Add old entry_pc attribute only
22961 if debug nonbinding markers are disabled.
22962 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
22963 markers are enabled.
22964 (block_within_block_p, dwarf2out_inline_entry): New.
22965 (dwarf2out_finish): Check that no entries remained in
22966 inline_entry_data_table.
22967 * final.c (reemit_insn_block_notes): Handle inline entry notes.
22968 (final_scan_insn, notice_source_line): Likewise.
22969 (rest_of_clean_state): Skip inline entry markers.
22970 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
22972 * gimple.c (gimple_build_debug_inline_entry): New.
22973 * gimple.h (enum gimple_debug_subcode): Add
22974 GIMPLE_DEBUG_INLINE_ENTRY.
22975 (gimple_build_debug_inline_entry): Declare.
22976 (gimple_debug_inline_entry_p): New.
22977 (gimple_debug_nonbind_marker_p): Adjust.
22978 * insn-notes.def (INLINE_ENTRY): New.
22979 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
22980 inline entry marker notes.
22981 (print_insn): Likewise.
22982 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
22983 (INSN_DEBUG_MARKER_KIND): Likewise.
22984 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
22985 * tree-inline.c (expand_call_inline): Build and insert
22986 debug_inline_entry stmt.
22987 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
22988 inline entry blocks early, if nonbind markers are enabled.
22989 (dump_scope_block): Dump fragment info.
22990 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
22991 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
22992 (gimple_build_debug_inline_entry): New.
22993 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
22994 Enable/disable inline entry points too.
22995 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
22996 (DEBUG_INSN): Describe inline entry markers.
22998 * common.opt (gvariable-location-views): New.
22999 (gvariable-location-views=incompat5): New.
23000 * config.in: Rebuilt.
23001 * configure: Rebuilt.
23002 * configure.ac: Test assembler for view support.
23003 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
23004 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
23005 * dwarf2out.c (var_loc_view): New typedef.
23006 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
23007 (dwarf2out_locviews_in_attribute): New.
23008 (dwarf2out_locviews_in_loclist): New.
23009 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
23010 (enum dw_line_info_opcode): Add LI_adv_address.
23011 (struct dw_line_info_table): Add view.
23012 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
23013 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
23014 (zero_view_p): New variable.
23015 (ZERO_VIEW_P): New macro.
23016 (output_asm_line_debug_info): New.
23017 (struct var_loc_node): Add view.
23018 (add_AT_view_list, AT_loc_list): New.
23019 (add_var_loc_to_decl): Add view param. Test it against last.
23020 (new_loc_list): Add view params. Record them.
23021 (AT_loc_list_ptr): Handle loc and view lists.
23022 (view_list_to_loc_list_val_node): New.
23023 (print_dw_val): Handle dw_val_class_view_list.
23024 (size_of_die): Likewise.
23025 (value_format): Likewise.
23026 (loc_list_has_views): New.
23027 (gen_llsym): Set vl_symbol too.
23028 (maybe_gen_llsym, skip_loc_list_entry): New.
23029 (dwarf2out_maybe_output_loclist_view_pair): New.
23030 (output_loc_list): Output view list or entries too.
23031 (output_view_list_offset): New.
23032 (output_die): Handle dw_val_class_view_list.
23033 (output_dwarf_version): New.
23034 (output_compilation_unit_header): Use it.
23035 (output_skeleton_debug_sections): Likewise.
23036 (output_rnglists, output_line_info): Likewise.
23037 (output_pubnames, output_aranges): Update version comments.
23038 (output_one_line_info_table): Output view numbers in asm comments.
23039 (dw_loc_list): Determine current endview, pass it to new_loc_list.
23040 Call maybe_gen_llsym.
23041 (loc_list_from_tree_1): Adjust.
23042 (add_AT_location_description): Create view list attribute if
23043 needed, check it's absent otherwise.
23044 (convert_cfa_to_fb_loc_list): Adjust.
23045 (maybe_emit_file): Call output_asm_line_debug_info for test.
23046 (dwarf2out_var_location): Reset views as needed. Precompute
23047 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
23048 attribute. Set view.
23049 (new_line_info_table): Reset next view.
23050 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
23051 (dwarf2out_source_line): Likewise. Output view resets and labels to
23052 the assembler, or select appropriate line info opcodes.
23053 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
23054 (optimize_string_length): Catch it. Adjust.
23055 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
23056 dw_val_class_view_list, and remove it if no longer needed.
23057 (hash_loc_list): Hash view numbers.
23058 (loc_list_hasher::equal): Compare them.
23059 (optimize_location_lists): Check whether a view list symbol is
23060 needed, and whether the locview attribute is present, and
23061 whether they match. Remove the locview attribute if no longer
23063 (index_location_lists): Call skip_loc_list_entry for test.
23064 (dwarf2out_finish): Call output_asm_line_debug_info for test.
23065 Use output_dwarf_version.
23066 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
23067 (struct dw_val_node): Add val_view_list.
23068 * final.c (SEEN_NEXT_VIEW): New.
23069 (set_next_view_needed): New.
23070 (clear_next_view_needed): New.
23071 (maybe_output_next_view): New.
23072 (final_start_function): Rename to...
23073 (final_start_function_1): ... this. Take pointer to FIRST,
23074 add SEEN parameter. Emit param bindings in the initial view.
23075 (final_start_function): Reintroduce SEEN-less interface.
23076 (final): Rename to...
23077 (final_1): ... this. Take SEEN parameter. Output final pending
23078 next view at the end.
23079 (final): Reintroduce seen-less interface.
23080 (final_scan_insn): Output pending next view before switching
23081 sections or ending a block. Mark the next view as needed when
23082 outputting variable locations. Notify debug backend of section
23083 changes, and of location view changes.
23084 (rest_of_handle_final): Adjust.
23085 * toplev.c (process_options): Autodetect value for debug variable
23086 location views option. Warn on incompat5 without -gdwarf-5.
23087 * doc/invoke.texi (gvariable-location-views): New.
23088 (gvariable-location-views=incompat5): New.
23089 (gno-variable-location-views): New.
23091 2018-02-08 David Malcolm <dmalcolm@redhat.com>
23093 PR tree-optimization/84136
23094 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
23095 that the result of find_edge is non-NULL.
23097 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
23100 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
23101 storing integer register in SImode. Fix cost of 256 and 512
23102 byte aligned SSE register store.
23104 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
23106 * config/i386/i386.c (ix86_multiplication_cost): Fix
23107 multiplication cost for TARGET_AVX512DQ.
23109 2018-02-08 Marek Polacek <polacek@redhat.com>
23111 PR tree-optimization/84238
23112 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
23115 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23117 PR tree-optimization/84265
23118 * tree-vect-stmts.c (vectorizable_store): Don't treat
23119 VMAT_CONTIGUOUS accesses as grouped.
23120 (vectorizable_load): Likewise.
23122 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23124 PR tree-optimization/81635
23125 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
23126 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
23127 (test_round_for_mask): New functions.
23128 (wide_int_cc_tests): Call test_round_for_mask.
23129 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
23130 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
23131 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
23132 range returned by get_range_info.
23134 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
23137 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
23138 * symtab.c: Include builtins.h
23139 (symtab_node::output_to_lto_symbol_table_p): Move here
23140 from lto-streamer-out.c:output_symbol_p.
23141 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
23142 (output_symbol_p): Move all logic to symtab.c
23143 (produce_symtab): Update.
23145 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23147 * config/s390/s390-opts.h (enum indirect_branch): Define.
23148 * config/s390/s390-protos.h (s390_return_addr_from_memory)
23149 (s390_indirect_branch_via_thunk)
23150 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
23151 (enum s390_indirect_branch_type): Define.
23152 * config/s390/s390.c (struct s390_frame_layout, struct
23153 machine_function): Remove.
23154 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
23155 (indirect_branch_table_label_no, indirect_branch_table_name):
23157 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
23158 (enum s390_indirect_branch_option): Define.
23159 (s390_return_addr_from_memory): New function.
23160 (s390_handle_string_attribute): New function.
23161 (s390_attribute_table): Add new attribute handler.
23162 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
23163 (s390_indirect_branch_via_thunk): New function.
23164 (s390_indirect_branch_via_inline_thunk): New function.
23165 (s390_function_ok_for_sibcall): When jumping via thunk disallow
23166 sibling call optimization for non z10 compiles.
23167 (s390_emit_call): Force indirect branch target to be a single
23168 register. Add r1 clobber for non-z10 compiles.
23169 (s390_emit_epilogue): Emit return jump via return_use expander.
23170 (s390_reorg): Handle JUMP_INSNs as execute targets.
23171 (s390_option_override_internal): Perform validity checks for the
23172 new command line options.
23173 (s390_indirect_branch_attrvalue): New function.
23174 (s390_indirect_branch_settings): New function.
23175 (s390_set_current_function): Invoke s390_indirect_branch_settings.
23176 (s390_output_indirect_thunk_function): New function.
23177 (s390_code_end): Implement target hook.
23178 (s390_case_values_threshold): Implement target hook.
23179 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
23181 * config/s390/s390.h (struct s390_frame_layout)
23182 (struct machine_function): Move here from s390.c.
23183 (TARGET_INDIRECT_BRANCH_NOBP_RET)
23184 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
23185 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
23186 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
23187 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
23188 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
23189 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
23190 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
23191 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
23192 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
23193 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
23194 (mnemonic attribute): Add values which aren't recognized
23196 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
23197 pattern for branch conversion. Fix mnemonic attribute.
23198 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
23199 indirect branch via thunk if requested.
23200 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
23201 ("*indirect_jump"): Disable for branch conversion using out of
23203 ("indirect_jump_via_thunk<mode>_z10")
23204 ("indirect_jump_via_thunk<mode>")
23205 ("indirect_jump_via_inlinethunk<mode>_z10")
23206 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
23207 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
23208 ("casesi_jump_via_inlinethunk<mode>_z10")
23209 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
23210 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
23211 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
23212 ("*indirect2_jump"): Disable for branch conversion.
23213 ("casesi_jump"): Turn into expander and expand patterns for branch
23215 ("return_use"): New expander.
23216 ("*return"): Emit return via thunk and rename it to ...
23217 ("*return<mode>"): ... this one.
23218 * config/s390/s390.opt: Add new options and and enum for the
23221 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23223 * lra-constraints.c (match_reload): Unconditionally use
23224 gen_lowpart_SUBREG, rather than selecting between that
23225 and equivalent gen_rtx_SUBREG code.
23227 2018-02-08 Richard Biener <rguenther@suse.de>
23229 PR tree-optimization/84233
23230 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
23231 changed flag instead of boguously re-using phi_inserted.
23233 2018-02-08 Martin Jambor <mjambor@suse.cz>
23235 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
23236 static local variables.
23238 2018-02-08 Richard Biener <rguenther@suse.de>
23240 PR tree-optimization/84278
23241 * tree-vect-stmts.c (vectorizable_store): When looking for
23242 smaller vector types to perform grouped strided loads/stores
23243 make sure the mode is supported by the target.
23244 (vectorizable_load): Likewise.
23246 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
23248 * config/aarch64/aarch64.c (aarch64_components_for_bb):
23249 Increase LDP/STP opportunities by adding adjacent callee-saves.
23251 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
23253 PR rtl-optimization/84068
23254 PR rtl-optimization/83459
23255 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
23257 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
23259 PR tree-optimization/84224
23260 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
23261 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
23262 non-zero arguments.
23264 2018-02-07 Iain Sandoe <iain@codesourcery.com>
23267 * config/rs6000/altivec.md (*restore_world): Remove LR use.
23268 * config/rs6000/predicates.md (restore_world_operation): Adjust op
23269 count, remove one USE.
23271 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23273 * doc/install.texi (Configuration): Document the
23274 --with-long-double-format={ibm,ieee} PowerPC configuration
23278 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
23279 Convert from define_expand to be define_insn_and_split. Rework
23280 float/double/_Float128 conversions to QI/HI/SImode to work with
23281 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
23282 conversions to QI/HImode types did a store and then a load to
23283 truncate the value. For conversions to VSX registers, don't split
23284 the insn, instead emit the code directly. Use the code iterator
23285 any_fix to combine signed and unsigned conversions.
23286 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
23287 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
23288 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
23289 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
23290 (fix_<mode>di2_hw): Likewise.
23291 (fixuns_<mode>di2_hw): Likewise.
23292 (fix_<mode>si2_hw): Likewise.
23293 (fixuns_<mode>si2_hw): Likewise.
23294 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
23295 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
23296 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
23297 fix<uns>_trunc<SFDF:mode>si2_p8.
23298 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
23299 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23300 (fix<uns>_<mode>_mem): Likewise.
23301 (fctiw<u>z_<mode>_mem): Likewise.
23302 (fix<uns>_<mode>_mem): Likewise.
23303 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
23304 the register allocator from doing a direct move to the GPRs to do
23305 a store, and instead use the ISA 3.0 store byte/half-word from
23306 vector register instruction. For IEEE 128-bit floating point,
23307 also optimize stores of 32-bit ints.
23308 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
23310 2018-02-07 Alan Hayward <alan.hayward@arm.com>
23312 * genextract.c (push_pathstr_operand): New function to support
23314 (walk_rtx): Call push_pathstr_operand.
23315 (print_path): Support [a-zA-Z].
23317 2018-02-07 Richard Biener <rguenther@suse.de>
23319 PR tree-optimization/84037
23320 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
23321 (cse_and_gimplify_to_preheader): Declare.
23322 (vect_get_place_in_interleaving_chain): Likewise.
23323 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
23325 (_loop_vec_info::~_loop_vec_info): Delete it.
23326 (cse_and_gimplify_to_preheader): New function.
23327 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
23328 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
23329 (vectorizable_load): Likewise. For grouped stores always base
23330 the IV on the first element.
23331 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
23332 condition before gimplifying.
23334 2018-02-07 Jakub Jelinek <jakub@redhat.com>
23336 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
23337 *DIV_EXPR and *MOD_EXPR.
23339 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
23342 * config/i386/i386.c (ix86_option_override_internal): Mask out
23343 the CF_SET bit when checking -fcf-protection.
23345 2018-02-07 Tom de Vries <tom@codesourcery.com>
23348 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
23351 2018-02-07 Richard Biener <rguenther@suse.de>
23353 PR tree-optimization/84204
23354 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
23357 PR tree-optimization/84205
23358 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
23359 special-case isl_ast_op_zdiv_r.
23361 PR tree-optimization/84223
23362 * graphite-scop-detection.c (gather_bbs::before_dom_children):
23363 Only add conditions from within the region.
23364 (gather_bbs::after_dom_children): Adjust.
23366 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
23369 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
23370 * config/avr/avr.md: Only post-reload split REG-REG moves if
23371 either register is GENERAL_REG_P.
23373 2018-02-07 Jakub Jelinek <jakub@redhat.com>
23375 PR tree-optimization/84235
23376 * tree-ssa-scopedtables.c
23377 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
23378 if the subtraction is performed in floating point type where NaNs are
23379 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
23380 build 1. Formatting fix.
23382 2018-02-06 Jakub Jelinek <jakub@redhat.com>
23385 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
23386 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
23387 and skip it regardless of bb boundaries. Use CALL_P macro,
23388 don't test INSN_P (insn) together with CALL_P or JUMP_P check
23389 unnecessarily, formatting fix.
23391 2018-02-06 Michael Collison <michael.collison@arm.com>
23393 * config/arm/thumb2.md:
23394 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
23395 (*thumb_mov_notscc): Ditto.
23397 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
23400 * config/rs6000/rs6000.md (su code attribute): Use "u" for
23401 unsigned_fix, not "s".
23403 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23405 * configure.ac (gcc_fn_eh_frame_ro): New function.
23406 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
23407 correct .eh_frame permissions.
23408 * configure: Regenerate.
23410 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
23412 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
23413 irrelevant options.
23415 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23417 * config/rs6000/rs6000.c (rs6000_option_override_internal):
23418 Display warning message for -mno-speculate-indirect-jumps.
23420 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
23422 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
23424 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
23426 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
23428 PR tree-optimization/84225
23429 * tree-eh.c (find_trapping_overflow): Only call
23430 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
23432 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
23435 * config/i386/i386.c: Reimplement the check of possible options
23436 -mibt/-mshstk conbination. Change error messages.
23437 * doc/invoke.texi: Fix a typo: remove extra '='.
23439 2018-02-06 Marek Polacek <polacek@redhat.com>
23441 PR tree-optimization/84228
23442 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
23444 2018-02-06 Tamar Christina <tamar.christina@arm.com>
23447 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
23448 emitted arch directives.
23449 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
23450 __ARM_FEATURE_COPROC before changing architectures.
23452 2018-02-06 Richard Biener <rguenther@suse.de>
23454 * config/i386/i386.c (print_reg): Fix typo.
23455 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
23457 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
23459 * configure: Regenerate.
23461 2018-02-05 Martin Sebor <msebor@redhat.com>
23463 PR tree-optimization/83369
23464 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
23467 2018-02-05 Martin Liska <mliska@suse.cz>
23469 * doc/invoke.texi: Cherry-pick upstream r323995.
23471 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
23473 * ira.c (ira_init_register_move_cost): Adjust comment.
23475 2018-02-05 Martin Liska <mliska@suse.cz>
23477 PR gcov-profile/84137
23478 * doc/gcov.texi: Fix typo in documentation.
23480 2018-02-05 Martin Liska <mliska@suse.cz>
23482 PR gcov-profile/83879
23483 * doc/gcov.texi: Document necessity of --dynamic-list-data when
23484 using dlopen functionality.
23486 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
23488 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
23489 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
23490 _mm_maskz_range_ss, _mm_mask_range_round_ss,
23491 _mm_maskz_range_round_ss): New intrinsics.
23492 (__builtin_ia32_rangesd128_round)
23493 (__builtin_ia32_rangess128_round): Remove.
23494 (__builtin_ia32_rangesd128_mask_round,
23495 __builtin_ia32_rangess128_mask_round): New builtins.
23496 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
23497 __builtin_ia32_rangess128_round): Remove.
23498 (__builtin_ia32_rangesd128_mask_round,
23499 __builtin_ia32_rangess128_mask_round): New builtins.
23500 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
23501 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
23502 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
23503 "<round_saeonly_constraint>")): Changed to ...
23504 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
23505 "<round_saeonly_scalar_constraint>")): ... this.
23506 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
23507 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
23508 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
23509 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
23510 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
23512 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
23514 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
23516 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
23517 Remove all values except native, 8540 and 8548.
23519 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
23521 * config/i386/i386.c (ix86_output_function_return): Pass
23522 INVALID_REGNUM, instead of -1, as invalid register number to
23523 indirect_thunk_name and output_indirect_thunk.
23525 2018-02-02 Julia Koval <julia.koval@intel.com>
23527 * config.gcc: Add -march=icelake.
23528 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
23529 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
23530 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
23531 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
23532 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
23533 (processor_target_table): Add icelake.
23534 (ix86_option_override_internal): Handle new PTAs.
23535 (get_builtin_code_for_version): Handle icelake.
23536 (M_INTEL_COREI7_ICELAKE): New.
23537 (fold_builtin_cpu): Handle icelake.
23538 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
23539 * doc/invoke.texi: Add -march=icelake.
23541 2018-02-02 Julia Koval <julia.koval@intel.com>
23543 * config/i386/i386.c (ix86_option_override_internal): Change flags type
23544 to wide_int_bitmask.
23545 * wide-int-bitmask.h: New.
23547 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
23550 * config/i386/i386.md: Replace Pmode with word_mode in
23551 builtin_setjmp_setup and builtin_longjmp to support x32.
23553 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
23557 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
23559 (rs6000_supported_cpu_names): New static variable.
23560 (linux_cpu_translation_table): Likewise.
23561 (elf_platform) <cpu>: Define new static variable and use it.
23562 Translate kernel AT_PLATFORM name to canonical name if needed.
23563 Error if platform name is unknown.
23565 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
23568 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
23570 2018-02-01 Jeff Law <law@redhat.com>
23573 * config/i386/i386.c (release_scratch_register_on_entry): Add new
23574 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
23575 the scratch if RELEASE_VIA_POP is false.
23576 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
23577 If we have to save a temporary register, decrement SIZE appropriately.
23578 Pass new arguments to release_scratch_register_on_entry.
23579 (ix86_adjust_stack_and_probe): Likewise.
23580 (ix86_emit_probe_stack_range): Pass new arguments to
23581 release_scratch_register_on_entry.
23583 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
23585 PR rtl-optimization/84157
23586 * combine.c (change_zero_ext): Use REG_P predicate in
23587 front of HARD_REGISTER_P predicate.
23589 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
23591 * config/avr/avr.c (avr_option_override): Move disabling of
23592 -fdelete-null-pointer-checks to...
23593 * common/config/avr/avr-common.c (avr_option_optimization_table):
23596 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23598 PR tree-optimization/81635
23599 * tree-data-ref.c (split_constant_offset_1): For types that
23600 wrap on overflow, try to use range info to prove that wrapping
23603 2018-02-01 Renlin Li <renlin.li@arm.com>
23606 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
23607 TAILCALL_ADDR_REGS.
23608 (aarch64_register_move_cost): Likewise.
23609 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
23610 TAILCALL_ADDR_REGS.
23611 (REG_CLASS_NAMES): Likewise.
23612 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
23613 TAILCALL_ADDR_REGS. Remove IP registers.
23614 * config/aarch64/aarch64.md (Ucs): Update register constraint.
23616 2018-02-01 Richard Biener <rguenther@suse.de>
23618 * domwalk.h (dom_walker::dom_walker): Add additional constructor
23619 for specifying RPO order and allow NULL for that.
23620 * domwalk.c (dom_walker::dom_walker): Likewise.
23621 (dom_walker::walk): Handle NULL RPO order.
23622 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
23624 (rewrite_update_dom_walker): Likewise.
23625 (mark_def_dom_walker): Likewise.
23627 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23629 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
23630 (aarch64_maybe_expand_sve_subreg_move): Declare.
23631 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
23632 * config/aarch64/predicates.md (aarch64_any_register_operand): New
23634 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
23635 that are semantically a reverse operation.
23636 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
23637 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
23638 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
23640 (aarch64_can_change_mode_class): For big-endian, forbid changes
23641 between two SVE modes if they have different element sizes.
23643 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23645 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
23646 the TImode handling for big-endian targets.
23648 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23650 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
23651 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
23653 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
23654 Remove BSWAP handing for big-endian targets and use the form of
23655 LD1RQ appropariate for the mode.
23657 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23659 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
23660 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
23661 duplicated element.
23663 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23666 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
23667 check for operands that need to go through aarch64_sve_reload_be.
23669 2018-02-01 Jakub Jelinek <jakub@redhat.com>
23671 PR tree-optimization/81661
23672 PR tree-optimization/84117
23673 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
23674 * tree-eh.c: Include gimplify.h.
23675 (find_trapping_overflow, replace_trapping_overflow,
23676 rewrite_to_non_trapping_overflow): New functions.
23677 * tree-vect-loop.c: Include tree-eh.h.
23678 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
23679 * tree-data-ref.c: Include tree-eh.h.
23680 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
23682 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
23684 PR rtl-optimization/84123
23685 * combine.c (change_zero_ext): Check if hard register satisfies
23686 can_change_dest_mode before calling gen_lowpart_SUBREG.
23688 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
23691 * ira.c (ira_init_register_move_cost): Remove assert.
23693 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
23695 PR rtl-optimization/84071
23696 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
23697 * doc/tm.texi: Regenerate.
23699 2018-01-31 Richard Biener <rguenther@suse.de>
23701 PR tree-optimization/84132
23702 * tree-data-ref.c (analyze_miv_subscript): Properly
23703 check whether evolution_function_is_affine_multivariate_p
23704 before calling gcd_of_steps_may_divide_p.
23706 2018-01-31 Julia Koval <julia.koval@intel.com>
23709 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
23710 * config/i386/i386.md (rdpid_rex64) New.
23711 (rdpid): Make 32bit only.
23713 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
23716 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
23717 an IDENTIFIER_NODE for FUNCTION_TYPE's.
23719 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
23722 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
23724 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
23726 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
23728 PR rtl-optimization/84071
23729 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
23730 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
23732 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
23734 * config/arc/arc.c (arc_handle_aux_attribute): New function.
23735 (arc_attribute_table): Add 'aux' attribute.
23736 (arc_in_small_data_p): Consider aux like variables.
23737 (arc_is_aux_reg_p): New function.
23738 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
23739 (arc_get_aux_arg): New function.
23740 (prepare_move_operands): Handle aux-register access.
23741 (arc_handle_aux_attribute): New function.
23742 * doc/extend.texi (ARC Variable attributes): Add subsection.
23744 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
23746 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
23747 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
23748 (arc_attribute_table): Add 'uncached' attribute.
23749 (arc_print_operand): Print '.di' flag for uncached memory
23751 (arc_in_small_data_p): Do not consider for small data the uncached
23753 (arc_is_uncached_mem_p): New function.
23754 * config/arc/predicates.md (compact_store_memory_operand): Check
23755 for uncached memory accesses.
23756 (nonvol_nonimm_operand): Likewise.
23757 * doc/extend.texi (ARC Type Attribute): New subsection.
23759 2018-01-31 Jakub Jelinek <jakub@redhat.com>
23762 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
23763 falign-loops=): Add Optimization flag.
23765 2018-01-30 Jeff Law <law@redhat.com>
23768 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
23769 INT_REGISTERS_SAVED. Check it prior to calling
23770 get_scratch_register_on_entry.
23771 (ix86_adjust_stack_and_probe): Similarly.
23772 (ix86_emit_probe_stack_range): Similarly.
23773 (ix86_expand_prologue): Corresponding changes.
23775 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23778 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
23779 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
23781 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
23784 * lra-constraints.c (curr_insn_transform): Process AND in the
23787 2018-01-30 Jakub Jelinek <jakub@redhat.com>
23789 PR rtl-optimization/83986
23790 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
23791 dependence against last_pending_memory_flush in addition to
23792 pending_jump_insns.
23794 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
23796 PR tree-optimization/81611
23797 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
23800 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23803 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
23806 2018-01-30 Richard Biener <rguenther@suse.de>
23807 Jakub Jelinek <jakub@redhat.com>
23809 PR tree-optimization/84111
23810 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
23811 inner loops added during recursion, as they don't have up-to-date
23814 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
23817 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
23818 (can_inline_edge_by_limits_p): ... here.
23819 (can_early_inline_edge_p, check_callers,
23820 update_caller_keys, update_callee_keys, recursive_inlining,
23821 add_new_edges_to_heap, speculation_useful_p,
23822 inline_small_functions,
23823 inline_small_functions, flatten_function,
23824 inline_to_all_callers_1): Update.
23826 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
23828 * profile-count.c (profile_count::combine_with_ipa_count): Handle
23831 2018-01-30 Richard Biener <rguenther@suse.de>
23833 PR tree-optimization/83008
23834 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
23835 invariant and constant vector uses in stmts when they need
23836 more than one stmt.
23838 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23841 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
23842 * configure: Regenerate.
23844 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
23846 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
23848 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
23849 Use gen_rtx_REG rather than gen_lowpart.
23851 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
23853 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
23854 rather than 0 when creating partial subregs.
23856 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
23858 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
23861 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
23864 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
23865 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
23866 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
23867 flags. This restores the settings used before the 2017-07-24.
23868 Turning off pre increment/decrement/modify allows IVOPTS to
23869 optimize DF/SF loops where the index is an int.
23871 2018-01-29 Richard Biener <rguenther@suse.de>
23872 Kelvin Nilsen <kelvin@gcc.gnu.org>
23875 * tree-vect-stmts.c (vectorizable_call): Don't call
23876 targetm.vectorize_builtin_md_vectorized_function if callee is
23879 2018-01-22 Carl Love <cel@us.ibm.com>
23881 * doc/extend.tex: Fix typo in second arg in
23882 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
23884 2018-01-29 Richard Biener <rguenther@suse.de>
23886 PR tree-optimization/84086
23887 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
23888 (flush_ssaname_freelist): When SSA names were released reset
23889 the SCEV hash table.
23891 2018-01-29 Richard Biener <rguenther@suse.de>
23893 PR tree-optimization/84057
23894 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
23895 removed paths when removing edges.
23897 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
23899 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
23900 -mfunction-return=@var{choice}.
23902 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
23904 PR diagnostic/84034
23905 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
23906 Handle CR like TAB.
23907 (layout::print_source_line): Likewise.
23908 (test_get_line_width_without_trailing_whitespace): Add test cases.
23910 2018-01-27 Jakub Jelinek <jakub@redhat.com>
23912 PR middle-end/84040
23913 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
23916 2018-01-26 Jim Wilson <jimw@sifive.com>
23918 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
23920 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
23923 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23925 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
23926 and CMP + SUB-immediate -> SUBS.
23928 2018-01-26 Martin Sebor <msebor@redhat.com>
23930 PR tree-optimization/83896
23931 * tree-ssa-strlen.c (get_string_len): Rename...
23932 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
23933 Avoid assuming length is constant.
23934 (handle_char_store): Use HOST_WIDE_INT for string length.
23936 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
23939 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
23940 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
23942 2018-01-26 Richard Biener <rguenther@suse.de>
23944 PR rtl-optimization/84003
23945 * dse.c (record_store): Only record redundant stores when
23946 the earlier store aliases at least all accesses the later one does.
23948 2018-01-26 Jakub Jelinek <jakub@redhat.com>
23950 PR rtl-optimization/83985
23951 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
23952 REG_CFA_RESTORE insns.
23953 (delete_unmarked_insns): Don't ignore separate shrink wrapping
23954 REG_CFA_RESTORE insns here.
23957 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
23958 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
23960 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
23962 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
23963 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
23964 (arc_init): Likewise.
23965 (arc_override_options): Likewise.
23966 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
23968 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
23970 * config/arc/arc.h (TARGET_DBNZ): Define.
23971 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
23972 properly set the tune attribute.
23973 (dbnz): Use TARGET_DBNZ guard.
23974 * config/arc/arc.opt (mtune): Add core3 option.
23976 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
23978 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
23979 recognize new pic like addresses.
23980 (arc_delegitimize_address): Clean up.
23982 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
23984 * config/arc/arc-arches.def: Option mrf16 valid for all
23986 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
23987 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
23988 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
23989 * config/arc/arc-tables.opt: Regenerate.
23990 * config/arc/arc.c (arc_conditional_register_usage): Handle
23991 reduced register file case.
23992 (arc_file_start): Set must have build attributes.
23993 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
23994 mrf16 option value.
23995 * config/arc/arc.opt (mrf16): Add new option.
23996 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
23997 * config/arc/genmultilib.awk: Handle new mrf16 option.
23998 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
23999 * config/arc/t-multilib: Regenerate.
24000 * doc/invoke.texi (ARC Options): Document mrf16 option.
24002 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24004 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
24005 * config/arc/arc.c (arc_handle_secure_attribute): New function.
24006 (arc_attribute_table): Add 'secure_call' attribute.
24007 (arc_print_operand): Print secure call operand.
24008 (arc_function_ok_for_sibcall): Don't optimize tail calls when
24010 (arc_is_secure_call_p): New function. * config/arc/arc.md
24011 (call_i): Add support for sjli instruction.
24012 (call_value_i): Likewise.
24013 * config/arc/constraints.md (Csc): New constraint.
24015 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24016 John Eric Martin <John.Martin@emmicro-us.com>
24018 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
24019 * config/arc/arc.c (_arc_jli_section): New struct.
24020 (arc_jli_section): New type.
24021 (rc_jli_sections): New static variable.
24022 (arc_handle_jli_attribute): New function.
24023 (arc_attribute_table): Add jli_always and jli_fixed attribute.
24024 (arc_file_end): New function.
24025 (TARGET_ASM_FILE_END): Define.
24026 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
24027 (arc_add_jli_section): New function.
24028 (jli_call_scan): Likewise.
24029 (arc_reorg): Call jli_call_scan.
24030 (arc_output_addsi): Remove 'S' from printing asm operand.
24031 (arc_is_jli_call_p): New function.
24032 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
24034 (movhi_insn): Likewise.
24035 (movsi_insn): Likewise.
24036 (movsi_set_cc_insn): Likewise.
24037 (loadqi_update): Likewise.
24038 (load_zeroextendqisi_update): Likewise.
24039 (load_signextendqisi_update): Likewise.
24040 (loadhi_update): Likewise.
24041 (load_zeroextendhisi_update): Likewise.
24042 (load_signextendhisi_update): Likewise.
24043 (loadsi_update): Likewise.
24044 (loadsf_update): Likewise.
24045 (movsicc_insn): Likewise.
24046 (bset_insn): Likewise.
24047 (bxor_insn): Likewise.
24048 (bclr_insn): Likewise.
24049 (bmsk_insn): Likewise.
24050 (bicsi3_insn): Likewise.
24051 (cmpsi_cc_c_insn): Likewise.
24052 (movsi_ne): Likewise.
24053 (movsi_cond_exec): Likewise.
24054 (clrsbsi2): Likewise.
24055 (norm_f): Likewise.
24065 (call_i): Remove 'S' asm letter, add jli instruction.
24066 (call_value_i): Likewise.
24067 * config/arc/arc.op (mjli-always): New option.
24068 * config/arc/constraints.md (Cji): New constraint.
24069 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
24071 (subsf3_fpx): Likewise.
24072 (mulsf3_fpx): Likewise.
24073 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
24075 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
24076 function attrbutes.
24077 * doc/invoke.texi (ARC): Document mjli-always option.
24079 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
24081 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
24082 avoid addition with 0 and use incw and decw where possible.
24084 2018-01-26 Richard Biener <rguenther@suse.de>
24086 PR tree-optimization/81082
24087 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
24088 association if it requires casting to unsigned.
24089 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
24090 from fold_plusminus_mult_expr to catch important cases late when
24091 range info is available.
24093 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24095 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
24096 * configure.ac (hidden_linkonce): New test.
24097 * configure: Regenerate.
24098 * config.in: Regenerate.
24100 2018-01-26 Julia Koval <julia.koval@intel.com>
24102 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
24103 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
24104 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
24105 _mm_mask_bitshuffle_epi64_mask): Fix type.
24106 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
24107 USI_FTYPE_V4DI_V4DI_USI): Remove.
24108 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
24109 __builtin_ia32_vpshufbitqmb256_mask,
24110 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
24111 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
24112 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
24114 2018-01-26 Alan Modra <amodra@gmail.com>
24117 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
24118 UNSPEC_VBPERMQ. Sort other unspecs.
24120 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
24122 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
24124 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
24126 PR middle-end/83055
24127 * predict.c (drop_profile): Do not push/pop cfun; update also
24129 (handle_missing_profiles): Fix logic looking for zero profiles.
24131 2018-01-25 Jakub Jelinek <jakub@redhat.com>
24133 PR middle-end/83977
24134 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
24135 on functions with #pragma omp declare simd or functions with simd
24137 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
24138 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
24139 Remove trailing \n from warning_at calls.
24141 2018-01-25 Tom de Vries <tom@codesourcery.com>
24144 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
24145 for neutered workers.
24147 2018-01-24 Joseph Myers <joseph@codesourcery.com>
24150 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
24151 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
24153 2018-01-24 Jeff Law <law@redhat.com>
24156 * i386.c (get_probe_interval): Move to earlier point.
24157 (ix86_compute_frame_layout): If -fstack-clash-protection and
24158 the frame is larger than the probe interval, then use pushes
24159 to save registers rather than reg->mem moves.
24160 (ix86_expand_prologue): Remove conditional for int_registers_saved
24163 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
24166 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
24167 min/max for never referenced object.
24169 2018-01-24 Jakub Jelinek <jakub@redhat.com>
24171 PR middle-end/83977
24172 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
24174 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
24175 attributes from DECL_ATTRIBUTES (decl) without affecting
24176 DECL_ATTRIBUTES (current_function_decl).
24177 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
24178 functions with non-NULL DECL_ABSTRACT_ORIGIN.
24180 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
24182 PR tree-optimization/83979
24183 * fold-const.c (fold_comparison): Use constant_boolean_node
24184 instead of boolean_{true,false}_node.
24186 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
24188 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
24191 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24193 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
24194 Simplify the clause that sets the length attribute.
24195 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
24196 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
24197 clause that sets the length attribute.
24198 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
24200 2018-01-24 Tom de Vries <tom@codesourcery.com>
24203 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
24204 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
24205 Add strict parameter.
24206 (prevent_branch_around_nothing): Insert dummy insn between branch to
24207 label and label with no ptx insn inbetween.
24208 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
24210 2018-01-24 Tom de Vries <tom@codesourcery.com>
24213 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
24214 for neutered threads in warp.
24215 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
24217 2018-01-24 Richard Biener <rguenther@suse.de>
24219 PR tree-optimization/83176
24220 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
24223 2018-01-24 Richard Biener <rguenther@suse.de>
24225 PR tree-optimization/82819
24226 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
24227 code generating pluses that are no-ops in the target precision.
24229 2018-01-24 Richard Biener <rguenther@suse.de>
24231 PR middle-end/84000
24232 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
24234 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
24236 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
24237 to merge probabilities.
24238 * predict.c (probably_never_executed): Also mark as cold functions
24239 with global 0 profile and guessed local profile.
24240 * profile-count.c (profile_probability::combine_with_count): New
24242 * profile-count.h (profile_probability::operator*,
24243 profile_probability::operator*=, profile_probability::operator/,
24244 profile_probability::operator/=): Reduce precision to adjusted
24245 and set value to guessed on contradictory divisions.
24246 (profile_probability::combine_with_freq): Remove.
24247 (profile_probability::combine_wiht_count): Declare.
24248 (profile_count::force_nonzero):: Set to adjusted.
24249 (profile_count::probability_in):: Set quality to adjusted.
24250 * tree-ssa-tail-merge.c (replace_block_by): Use
24251 combine_with_count.
24253 2018-01-23 Andrew Waterman <andrew@sifive.com>
24254 Jim Wilson <jimw@sifive.com>
24256 * config/riscv/riscv.c (riscv_stack_boundary): New.
24257 (riscv_option_override): Set riscv_stack_boundary. Handle
24258 riscv_preferred_stack_boundary_arg.
24259 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
24260 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
24261 (STACK_BOUNDARY): Set to riscv_stack_boundary.
24262 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
24263 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
24264 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
24266 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
24269 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
24270 of struct ix86_frame.
24271 (ix86_expand_epilogue): Likewise. Add a local variable for
24272 the reg_save_offset field in struct ix86_frame.
24274 2018-01-23 Bin Cheng <bin.cheng@arm.com>
24276 PR tree-optimization/82604
24277 * tree-loop-distribution.c (enum partition_kind): New enum item
24278 PKIND_PARTIAL_MEMSET.
24279 (partition_builtin_p): Support above new enum item.
24280 (generate_code_for_partition): Ditto.
24281 (compute_access_range): Differentiate cases that equality can be
24282 proven at all loops, the innermost loops or no loops.
24283 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
24284 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
24285 (finalize_partitions, distribute_loop): Don't fuse partition of
24286 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
24287 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
24288 parloop is enabled.
24290 2018-01-23 Martin Liska <mliska@suse.cz>
24292 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
24293 order to ignore the predictor.
24294 (PRED_POLYMORPHIC_CALL): Likewise.
24295 (PRED_RECURSIVE_CALL): Likewise.
24297 2018-01-23 Martin Liska <mliska@suse.cz>
24299 * tree-profile.c (tree_profiling): Print function header to
24300 aware reader which function we are working on.
24301 * value-prof.c (gimple_find_values_to_profile): Do not print
24302 not interesting value histograms.
24304 2018-01-23 Martin Liska <mliska@suse.cz>
24306 * profile-count.h (enum profile_quality): Add
24307 profile_uninitialized as the first value. Do not number values
24308 as they are zero based.
24309 (profile_count::verify): Update sanity check.
24310 (profile_probability::verify): Likewise.
24312 2018-01-23 Nathan Sidwell <nathan@acm.org>
24314 * doc/invoke.texi (ffor-scope): Deprecate.
24316 2018-01-23 David Malcolm <dmalcolm@redhat.com>
24318 PR tree-optimization/83510
24319 * domwalk.c (set_all_edges_as_executable): New function.
24320 (dom_walker::dom_walker): Convert bool param
24321 "skip_unreachable_blocks" to enum reachability. Move setup of
24322 edge flags to set_all_edges_as_executable and only do it when
24323 reachability is REACHABLE_BLOCKS.
24324 * domwalk.h (enum dom_walker::reachability): New enum.
24325 (dom_walker::dom_walker): Convert bool param
24326 "skip_unreachable_blocks" to enum reachability.
24327 (set_all_edges_as_executable): New decl.
24328 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
24329 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
24331 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
24332 but converting true to REACHABLE_BLOCKS.
24333 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
24335 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
24336 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
24337 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
24339 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
24340 if check_all_array_refs will be called.
24342 2018-01-23 David Malcolm <dmalcolm@redhat.com>
24344 * tree.c (selftest::test_location_wrappers): Add more test
24347 2018-01-23 David Malcolm <dmalcolm@redhat.com>
24349 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
24350 (selftest::test_bit_in_range): Likewise.
24352 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
24355 * doc/sourcebuild.texi (vect_float): Say that the selector
24356 only describes the situation when -funsafe-math-optimizations is on.
24357 (vect_float_strict): Document.
24359 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
24361 PR tree-optimization/83965
24362 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
24363 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
24364 instead of checking only for a reduction.
24365 (vect_recog_widen_sum_pattern): Likewise.
24367 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
24369 * predict.c (probably_never_executed): Only use precise profile info.
24370 (compute_function_frequency): Skip after inlining hack since we now
24371 have quality checking.
24373 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
24375 * profile-count.h (profile_probability::very_unlikely,
24376 profile_probability::unlikely, profile_probability::even): Set
24377 precision to guessed.
24379 2018-01-23 Richard Biener <rguenther@suse.de>
24381 PR tree-optimization/83963
24382 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
24383 Properly terminate dominator walk when crossing the exit edge not
24384 when visiting its source block.
24386 2018-01-23 Jakub Jelinek <jakub@redhat.com>
24389 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
24390 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
24392 2018-01-22 Jakub Jelinek <jakub@redhat.com>
24394 PR tree-optimization/83957
24395 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
24396 semicolon after for body surrounded by braces.
24398 PR tree-optimization/83081
24399 * profile-count.h (profile_probability::split): New method.
24400 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
24401 Use profile_probability::split.
24402 (do_compare_rtx_and_jump): Fix adjustment of probabilities
24403 when splitting a single conditional jump into 2.
24405 2018-01-22 David Malcolm <dmalcolm@redhat.com>
24407 PR tree-optimization/69452
24408 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
24411 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24413 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
24414 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
24415 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
24417 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24419 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
24421 * config/rl78/rl78.md (movdi): New define_expand.
24422 * config/rl78/rl78.c (rl78_split_movdi): New function.
24424 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
24427 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
24429 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
24430 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
24431 128-bit to produce an UNSPEC move to get the double word with the
24432 signbit and then a shift directly to do signbit.
24433 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
24434 implementation with a new version that just does either a direct
24435 move or a regular move. Move memory interface to separate insns.
24436 Move insns so they are next to the expander.
24437 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
24438 with signbit move. Split big and little endian case.
24439 (signbit<mode>2_dm_mem_le): Likewise.
24440 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
24441 (signbit<mode>2_dm2): Likewise.
24443 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24445 * config/rl78/rl78.md (anddi3): New define_expand.
24447 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24449 * config/rl78/rl78.md (umindi3): New define_expand.
24451 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24453 * config/rl78/rl78.md (smindi3): New define_expand.
24455 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24457 * config/rl78/rl78.md (smaxdi3): New define_expand.
24459 2018-01-22 Carl Love <cel@us.ibm.com>
24461 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
24462 LVX_V1TI): Add macro expansion.
24463 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
24464 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
24465 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
24466 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
24467 Change check to determine if the instruction is a byte reversing
24468 entry. Fix typo in comment.
24469 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
24470 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
24471 Add def_builtin calls for new builtins.
24472 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
24473 Add define_insn expansion.
24475 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24477 * config/rl78/rl78.md (umaxdi3): New define_expand.
24479 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24481 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
24482 for non-QImode registers.
24484 2018-01-22 Richard Biener <rguenther@suse.de>
24486 PR tree-optimization/83963
24487 * graphite-scop-detection.c (scop_detection::get_sese): Delay
24488 including the loop exit block.
24489 (scop_detection::merge_sese): Likewise.
24490 (scop_detection::add_scop): Do it here instead.
24492 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24494 * doc/sourcebuild.texi (arm_softfloat): Document.
24496 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
24499 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
24500 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
24501 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
24503 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24504 David Edelsohn <dje.gcc@gmail.com>
24507 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
24508 Change "crset eq" to "crset 2".
24509 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
24510 (*call_indirect_aix<mode>_nospec): Likewise.
24511 (*call_value_indirect_aix<mode>_nospec): Likewise.
24512 (*call_indirect_elfv2<mode>_nospec): Likewise.
24513 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
24514 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
24515 change assembly output from . to $.
24516 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
24517 (indirect_jump<mode>_nospec): Change assembly output from . to $.
24518 (*tablejump<mode>_internal1_nospec): Likewise.
24520 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
24523 * config/sh/sh_optimize_sett_clrt.cc:
24524 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
24526 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
24528 PR tree-optimization/83940
24529 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
24530 offset_dt to vect_constant_def rather than vect_unknown_def_type.
24531 (vect_check_load_store_mask): Add a mask_dt_out parameter and
24532 use it to pass back the definition type.
24533 (vect_check_store_rhs): Likewise rhs_dt_out.
24534 (vect_build_gather_load_calls): Add a mask_dt argument and use
24535 it instead of a call to vect_is_simple_use.
24536 (vectorizable_store): Update calls to vect_check_load_store_mask
24537 and vect_check_store_rhs. Use the dt returned by the latter instead
24538 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
24539 instead of calls to vect_is_simple_use. Pass the scalar rather
24540 than the vector operand to vect_is_simple_use when handling
24541 second and subsequent copies of an rhs value.
24542 (vectorizable_load): Update calls to vect_check_load_store_mask
24543 and vect_build_gather_load_calls. Use the cached mask_dt and
24544 gs_info.offset_dt instead of calls to vect_is_simple_use.
24546 2018-01-20 Jakub Jelinek <jakub@redhat.com>
24548 PR middle-end/83945
24549 * tree-emutls.c: Include gimplify.h.
24550 (lower_emutls_2): New function.
24551 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
24552 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
24553 it before further processing.
24556 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
24557 UINTVAL (trueop1) instead of INTVAL (op1).
24559 2018-01-19 Jakub Jelinek <jakub@redhat.com>
24563 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
24564 INCOMING_FRAME_SP_OFFSET if not defined.
24565 (scan_trace): Add ENTRY argument. If true and
24566 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
24567 emit a note to adjust the CFA offset.
24568 (create_cfi_notes): Adjust scan_trace callers.
24569 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
24570 INCOMING_FRAME_SP_OFFSET in the CIE.
24571 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
24572 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
24574 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
24575 * doc/tm.texi: Regenerated.
24577 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24579 PR rtl-optimization/83147
24580 * lra-constraints.c (remove_inheritance_pseudos): Use
24581 lra_substitute_pseudo_within_insn.
24583 2018-01-19 Tom de Vries <tom@codesourcery.com>
24584 Cesar Philippidis <cesar@codesourcery.com>
24587 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
24589 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
24592 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
24593 spaces for function labels.
24595 2018-01-19 Martin Liska <mliska@suse.cz>
24597 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
24598 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
24599 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
24600 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
24601 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
24602 (PRED_CONST_RETURN): Change from 69 to 65.
24603 (PRED_NULL_RETURN): Change from 91 to 71.
24604 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
24605 (PRED_LOOP_GUARD): Change from 66 to 73.
24607 2018-01-19 Martin Liska <mliska@suse.cz>
24609 * predict.c (predict_insn_def): Add new assert.
24610 (struct branch_predictor): Change type to signed integer.
24611 (test_prediction_value_range): Amend test to cover
24612 PROB_UNINITIALIZED.
24613 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
24614 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
24615 (PRED_LOOP_ITERATIONS_MAX): Likewise.
24616 (PRED_LOOP_IV_COMPARE): Likewise.
24617 * predict.h (PROB_UNINITIALIZED): Define new constant.
24619 2018-01-19 Martin Liska <mliska@suse.cz>
24621 * predict.c (dump_prediction): Add new format for
24622 analyze_brprob.py script which is enabled with -details
24624 * profile-count.h (precise_p): New function.
24626 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
24628 PR tree-optimization/83922
24629 * tree-vect-loop.c (vect_verify_full_masking): Return false if
24630 there are no statements that need masking.
24631 (vect_active_double_reduction_p): New function.
24632 (vect_analyze_loop_operations): Use it when handling phis that
24633 are not in the loop header.
24635 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
24637 PR tree-optimization/83914
24638 * tree-vect-loop.c (vectorizable_induction): Don't convert
24639 init_expr or apply the peeling adjustment for inductions
24640 that are nested within the vectorized loop.
24642 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24644 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
24647 2018-01-18 Jakub Jelinek <jakub@redhat.com>
24651 * function.h (gimplify_parameters): Add gimple_seq * argument.
24652 * function.c: Include gimple.h and options.h.
24653 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
24654 for the added local temporaries if needed.
24655 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
24656 if there are any parameter cleanups, wrap whole body into a
24657 try/finally with the cleanups.
24659 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
24662 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
24663 Use GET_MODE_CLASS for scalar floating point.
24665 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
24669 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
24670 Fix call of call_cgraph_insertion_hooks.
24672 2018-01-18 Martin Sebor <msebor@redhat.com>
24674 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
24676 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
24679 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
24682 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
24685 * common.opt: (-ffile-prefix-map): New option.
24686 * opts.c (common_handle_option): Defer it.
24687 * opts-global.c (handle_common_deferred_options): Handle it.
24688 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
24689 * file-prefix-map.h: New file.
24690 (remap_debug_filename, add_debug_prefix_map): ...here.
24691 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
24692 * final.c (debug_prefix_map, add_debug_prefix_map
24693 remap_debug_filename): Move to...
24694 * file-prefix-map.c: New file.
24695 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
24696 generalize, get rid of alloca(), use strrchr() instead of strchr().
24697 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
24698 Implement in terms of add_prefix_map().
24699 (remap_macro_filename, remap_debug_filename): Implement in term of
24701 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
24702 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
24703 * dbxout.c: Include file-prefix-map.h.
24704 * varasm.c: Likewise.
24705 * vmsdbgout.c: Likewise.
24706 * xcoffout.c: Likewise.
24707 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
24708 * doc/cppopts.texi (-fmacro-prefix-map): Document.
24709 * doc/invoke.texi (-ffile-prefix-map): Document.
24710 (-fdebug-prefix-map): Update description.
24712 2018-01-18 Martin Liska <mliska@suse.cz>
24714 * config/i386/i386.c (indirect_thunk_name): Document that also
24716 (output_indirect_thunk): Document why both instructions
24717 (pause and lfence) are generated.
24719 2018-01-18 Richard Biener <rguenther@suse.de>
24721 PR tree-optimization/83887
24722 * graphite-scop-detection.c
24723 (scop_detection::get_nearest_dom_with_single_entry): Remove.
24724 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
24725 (scop_detection::merge_sese): Re-implement with a flood-fill
24726 algorithm that properly finds a SESE region if it exists.
24728 2018-01-18 Jakub Jelinek <jakub@redhat.com>
24731 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
24732 pointer_diff optimizations use view_convert instead of convert.
24734 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24736 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
24737 Generate different code for -mno-speculate-indirect-jumps.
24738 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
24739 (*call_indirect_aix<mode>): Disable for
24740 -mno-speculate-indirect-jumps.
24741 (*call_indirect_aix<mode>_nospec): New define_insn.
24742 (*call_value_indirect_aix<mode>): Disable for
24743 -mno-speculate-indirect-jumps.
24744 (*call_value_indirect_aix<mode>_nospec): New define_insn.
24745 (*sibcall_nonlocal_sysv<mode>): Generate different code for
24746 -mno-speculate-indirect-jumps.
24747 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
24749 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
24751 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
24752 long double type, set the flags for noting the default long double
24753 type, even if we don't pass or return a long double type.
24755 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
24758 * ipa-inline.c (flatten_function): Do not overwrite final inlining
24761 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
24763 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
24764 support for merge[hl].
24765 (fold_mergehl_helper): New helper function.
24766 (tree-vector-builder.h): New #include for tree_vector_builder usage.
24767 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
24768 (altivec_vmrglw_direct): Add xxmrglw insn.
24770 2018-01-17 Andrew Waterman <andrew@sifive.com>
24772 * config/riscv/riscv.c (riscv_conditional_register_usage): If
24773 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
24775 2018-01-17 David Malcolm <dmalcolm@redhat.com>
24778 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
24779 call the lto_location_cache before reading the
24780 DECL_SOURCE_LOCATION of the types.
24782 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
24783 Richard Sandiford <richard.sandiford@linaro.org>
24785 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
24786 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
24787 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
24788 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
24789 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
24791 * config/aarch64/constraints.md (aarch64_movti_operand):
24793 * config/aarch64/predicates.md (Uti): Add new constraint.
24795 2018-01-17 Carl Love <cel@us.ibm.com>
24797 * config/rs6000/vsx.md (define_expand xl_len_r,
24798 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
24799 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
24801 (define_expand, define_insn): Move the shift left from the
24802 define_insn to the define_expand for lxvl and stxvl instructions.
24803 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
24804 and XL_LEN_R definitions to PURE.
24806 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
24808 * config/i386/i386.c (indirect_thunk_name): Declare regno
24809 as unsigned int. Compare regno with INVALID_REGNUM.
24810 (output_indirect_thunk): Ditto.
24811 (output_indirect_thunk_function): Ditto.
24812 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
24813 in the call to output_indirect_thunk_function.
24815 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
24817 PR middle-end/83884
24818 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
24819 rather than the size of inner_type to determine the stack slot size
24820 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
24822 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
24825 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
24828 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
24830 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
24831 endian Linux systems to optionally enable multilibs for selecting
24832 the long double type if the user configured an explicit type.
24833 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
24834 have no long double multilibs if not defined.
24835 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
24836 warn if the user used -mabi={ieee,ibm}longdouble and we built
24837 multilibs for long double.
24838 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
24839 appropriate multilib option.
24840 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
24842 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
24843 for building long double multilibs.
24844 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
24846 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
24848 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
24851 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
24853 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
24856 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
24859 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
24862 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
24864 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
24865 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
24867 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
24869 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
24870 different rtl trees depending on TARGET_64BIT.
24871 (rs6000_gen_lvx): Likewise.
24873 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
24875 * config/visium/visium.md (nop): Tweak comment.
24876 (hazard_nop): Likewise.
24878 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24880 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
24881 -mspeculate-indirect-jumps.
24882 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
24883 for -mno-speculate-indirect-jumps.
24884 (*call_indirect_elfv2<mode>_nospec): New define_insn.
24885 (*call_value_indirect_elfv2<mode>): Disable for
24886 -mno-speculate-indirect-jumps.
24887 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
24888 (indirect_jump): Emit different RTL for
24889 -mno-speculate-indirect-jumps.
24890 (*indirect_jump<mode>): Disable for
24891 -mno-speculate-indirect-jumps.
24892 (*indirect_jump<mode>_nospec): New define_insn.
24893 (tablejump): Emit different RTL for
24894 -mno-speculate-indirect-jumps.
24895 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
24896 (tablejumpsi_nospec): New define_expand.
24897 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
24898 (tablejumpdi_nospec): New define_expand.
24899 (*tablejump<mode>_internal1): Disable for
24900 -mno-speculate-indirect-jumps.
24901 (*tablejump<mode>_internal1_nospec): New define_insn.
24902 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
24905 2018-01-16 Artyom Skrobov tyomitch@gmail.com
24907 * caller-save.c (insert_save): Drop unnecessary parameter. All
24910 2018-01-16 Jakub Jelinek <jakub@redhat.com>
24911 Richard Biener <rguenth@suse.de>
24914 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
24915 return early, inline manually is_gimple_sizepos. Make sure if we
24916 call gimplify_expr we don't end up with a gimple constant.
24917 * tree.c (variably_modified_type_p): Don't return true for
24918 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
24919 * gimplify.h (is_gimple_sizepos): Remove.
24921 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
24923 PR tree-optimization/83857
24924 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
24925 vectorizable_live_operation for pure SLP statements.
24926 (vectorizable_live_operation): Handle PHIs.
24928 2018-01-16 Richard Biener <rguenther@suse.de>
24930 PR tree-optimization/83867
24931 * tree-vect-stmts.c (vect_transform_stmt): Precompute
24932 nested_in_vect_loop_p since the scalar stmt may get invalidated.
24934 2018-01-16 Jakub Jelinek <jakub@redhat.com>
24937 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
24938 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
24939 If off is not INTEGER_CST, issue a may not be aligned warning
24940 rather than isn't aligned. Use isn%'t rather than isn't.
24941 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
24943 <case MULT_EXPR>: Improve the case when bottom and one of the
24944 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
24945 operand, in that case check if the other operand is multiple of
24946 bottom divided by the INTEGER_CST operand.
24948 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
24951 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
24952 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
24953 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
24954 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
24955 * config/pa/pa.c (pa_function_arg_advance): Likewise.
24956 (pa_function_arg, pa_arg_partial_bytes): Likewise.
24957 (pa_function_arg_size): New function.
24959 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
24961 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
24962 in a separate statement.
24964 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
24966 PR tree-optimization/83847
24967 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
24968 group gathers and scatters.
24970 2018-01-16 Jakub Jelinek <jakub@redhat.com>
24972 PR rtl-optimization/86620
24973 * params.def (max-sched-ready-insns): Bump minimum value to 1.
24975 PR rtl-optimization/83213
24976 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
24977 to last if both are JUMP_INSNs.
24979 PR tree-optimization/83843
24980 * gimple-ssa-store-merging.c
24981 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
24982 store_immediate_info for bswap/nop orig_stores.
24984 2018-01-15 Andrew Waterman <andrew@sifive.com>
24986 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
24988 <UDIV>: Increase cost if !TARGET_DIV.
24990 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24992 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
24993 (define_attr "cr_logical_3op"): New.
24994 (cceq_ior_compare): Adjust.
24995 (cceq_ior_compare_complement): Adjust.
24996 (*cceq_rev_compare): Adjust.
24997 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
24998 (is_cracked_insn): Adjust.
24999 (insn_must_be_first_in_group): Adjust.
25000 * config/rs6000/40x.md: Adjust.
25001 * config/rs6000/440.md: Adjust.
25002 * config/rs6000/476.md: Adjust.
25003 * config/rs6000/601.md: Adjust.
25004 * config/rs6000/603.md: Adjust.
25005 * config/rs6000/6xx.md: Adjust.
25006 * config/rs6000/7450.md: Adjust.
25007 * config/rs6000/7xx.md: Adjust.
25008 * config/rs6000/8540.md: Adjust.
25009 * config/rs6000/cell.md: Adjust.
25010 * config/rs6000/e300c2c3.md: Adjust.
25011 * config/rs6000/e500mc.md: Adjust.
25012 * config/rs6000/e500mc64.md: Adjust.
25013 * config/rs6000/e5500.md: Adjust.
25014 * config/rs6000/e6500.md: Adjust.
25015 * config/rs6000/mpc.md: Adjust.
25016 * config/rs6000/power4.md: Adjust.
25017 * config/rs6000/power5.md: Adjust.
25018 * config/rs6000/power6.md: Adjust.
25019 * config/rs6000/power7.md: Adjust.
25020 * config/rs6000/power8.md: Adjust.
25021 * config/rs6000/power9.md: Adjust.
25022 * config/rs6000/rs64.md: Adjust.
25023 * config/rs6000/titan.md: Adjust.
25025 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25027 * config/i386/predicates.md (indirect_branch_operand): Rewrite
25028 ix86_indirect_branch_register logic.
25030 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25032 * config/i386/constraints.md (Bs): Update
25033 ix86_indirect_branch_register check. Don't check
25034 ix86_indirect_branch_register with GOT_memory_operand.
25036 * config/i386/predicates.md (GOT_memory_operand): Don't check
25037 ix86_indirect_branch_register here.
25038 (GOT32_symbol_operand): Likewise.
25040 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25042 * config/i386/predicates.md (constant_call_address_operand):
25043 Rewrite ix86_indirect_branch_register logic.
25044 (sibcall_insn_operand): Likewise.
25046 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25048 * config/i386/constraints.md (Bs): Replace
25049 ix86_indirect_branch_thunk_register with
25050 ix86_indirect_branch_register.
25052 * config/i386/i386.md (indirect_jump): Likewise.
25053 (tablejump): Likewise.
25054 (*sibcall_memory): Likewise.
25055 (*sibcall_value_memory): Likewise.
25056 Peepholes of indirect call and jump via memory: Likewise.
25057 * config/i386/i386.opt: Likewise.
25058 * config/i386/predicates.md (indirect_branch_operand): Likewise.
25059 (GOT_memory_operand): Likewise.
25060 (call_insn_operand): Likewise.
25061 (sibcall_insn_operand): Likewise.
25062 (GOT32_symbol_operand): Likewise.
25064 2018-01-15 Jakub Jelinek <jakub@redhat.com>
25066 PR middle-end/83837
25067 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
25068 type rather than type addr's type points to.
25069 (expand_omp_atomic_mutex): Likewise.
25070 (expand_omp_atomic): Likewise.
25072 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25075 * config/i386/i386.c (output_indirect_thunk_function): Use
25076 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
25077 for __x86_return_thunk.
25079 2018-01-15 Richard Biener <rguenther@suse.de>
25081 PR middle-end/83850
25082 * expmed.c (extract_bit_field_1): Fix typo.
25084 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25087 * config/arm/iterators.md (VF): New mode iterator.
25088 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
25089 Remove integer-related logic from pattern.
25090 (neon_vabd<mode>_3): Likewise.
25092 2018-01-15 Jakub Jelinek <jakub@redhat.com>
25094 PR middle-end/82694
25095 * common.opt (fstrict-overflow): No longer an alias.
25096 (fwrapv-pointer): New option.
25097 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
25098 also for pointer types based on flag_wrapv_pointer.
25099 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
25100 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
25101 opts->x_flag_wrapv got set.
25102 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
25103 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
25104 POINTER_TYPE_OVERFLOW_UNDEFINED.
25105 * match.pd: Likewise in address comparison pattern.
25106 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
25108 2018-01-15 Richard Biener <rguenther@suse.de>
25111 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
25112 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
25113 Reset type names to their identifier if their TYPE_DECL doesn't
25114 have linkage (and thus is used for ODR and devirt).
25115 (save_debug_info_for_decl): Remove.
25116 (save_debug_info_for_type): Likewise.
25117 (add_tree_to_fld_list): Adjust.
25118 * tree-pretty-print.c (dump_generic_node): Make dumping of
25119 type names more robust.
25121 2018-01-15 Richard Biener <rguenther@suse.de>
25123 * BASE-VER: Bump to 8.0.1.
25125 2018-01-14 Martin Sebor <msebor@redhat.com>
25128 * builtins.c (check_access): Avoid warning when the no-warning bit
25131 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
25133 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
25134 * ira-color (allocno_hard_regs_compare): Likewise.
25136 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
25139 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
25140 Use .pushsection/.popsection.
25142 2018-01-14 Martin Sebor <msebor@redhat.com>
25145 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
25147 2018-01-14 Jakub Jelinek <jakub@redhat.com>
25149 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
25150 entry from extra_headers.
25151 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
25152 extra_headers, make the list bitwise identical to the i?86-*-* one.
25154 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25156 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
25157 -mcmodel=large with -mindirect-branch=thunk,
25158 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
25159 -mfunction-return=thunk-extern.
25160 * doc/invoke.texi: Document -mcmodel=large is incompatible with
25161 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
25162 -mfunction-return=thunk and -mfunction-return=thunk-extern.
25164 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25166 * config/i386/i386.c (print_reg): Print the name of the full
25167 integer register without '%'.
25168 (ix86_print_operand): Handle 'V'.
25169 * doc/extend.texi: Document 'V' modifier.
25171 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25173 * config/i386/constraints.md (Bs): Disallow memory operand for
25174 -mindirect-branch-register.
25176 * config/i386/predicates.md (indirect_branch_operand): Likewise.
25177 (GOT_memory_operand): Likewise.
25178 (call_insn_operand): Likewise.
25179 (sibcall_insn_operand): Likewise.
25180 (GOT32_symbol_operand): Likewise.
25181 * config/i386/i386.md (indirect_jump): Call convert_memory_address
25182 for -mindirect-branch-register.
25183 (tablejump): Likewise.
25184 (*sibcall_memory): Likewise.
25185 (*sibcall_value_memory): Likewise.
25186 Disallow peepholes of indirect call and jump via memory for
25187 -mindirect-branch-register.
25188 (*call_pop): Replace m with Bw.
25189 (*call_value_pop): Likewise.
25190 (*sibcall_pop_memory): Replace m with Bs.
25191 * config/i386/i386.opt (mindirect-branch-register): New option.
25192 * doc/invoke.texi: Document -mindirect-branch-register option.
25194 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25196 * config/i386/i386-protos.h (ix86_output_function_return): New.
25197 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
25198 set function_return_type.
25199 (indirect_thunk_name): Add ret_p to indicate thunk for function
25201 (output_indirect_thunk_function): Pass false to
25202 indirect_thunk_name.
25203 (ix86_output_indirect_branch_via_reg): Likewise.
25204 (ix86_output_indirect_branch_via_push): Likewise.
25205 (output_indirect_thunk_function): Create alias for function
25206 return thunk if regno < 0.
25207 (ix86_output_function_return): New function.
25208 (ix86_handle_fndecl_attribute): Handle function_return.
25209 (ix86_attribute_table): Add function_return.
25210 * config/i386/i386.h (machine_function): Add
25211 function_return_type.
25212 * config/i386/i386.md (simple_return_internal): Use
25213 ix86_output_function_return.
25214 (simple_return_internal_long): Likewise.
25215 * config/i386/i386.opt (mfunction-return=): New option.
25216 (indirect_branch): Mention -mfunction-return=.
25217 * doc/extend.texi: Document function_return function attribute.
25218 * doc/invoke.texi: Document -mfunction-return= option.
25220 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25222 * config/i386/i386-opts.h (indirect_branch): New.
25223 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
25224 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
25225 with local indirect jump when converting indirect call and jump.
25226 (ix86_set_indirect_branch_type): New.
25227 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
25228 (indirectlabelno): New.
25229 (indirect_thunk_needed): Likewise.
25230 (indirect_thunk_bnd_needed): Likewise.
25231 (indirect_thunks_used): Likewise.
25232 (indirect_thunks_bnd_used): Likewise.
25233 (INDIRECT_LABEL): Likewise.
25234 (indirect_thunk_name): Likewise.
25235 (output_indirect_thunk): Likewise.
25236 (output_indirect_thunk_function): Likewise.
25237 (ix86_output_indirect_branch_via_reg): Likewise.
25238 (ix86_output_indirect_branch_via_push): Likewise.
25239 (ix86_output_indirect_branch): Likewise.
25240 (ix86_output_indirect_jmp): Likewise.
25241 (ix86_code_end): Call output_indirect_thunk_function if needed.
25242 (ix86_output_call_insn): Call ix86_output_indirect_branch if
25244 (ix86_handle_fndecl_attribute): Handle indirect_branch.
25245 (ix86_attribute_table): Add indirect_branch.
25246 * config/i386/i386.h (machine_function): Add indirect_branch_type
25247 and has_local_indirect_jump.
25248 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
25250 (tablejump): Likewise.
25251 (*indirect_jump): Use ix86_output_indirect_jmp.
25252 (*tablejump_1): Likewise.
25253 (simple_return_indirect_internal): Likewise.
25254 * config/i386/i386.opt (mindirect-branch=): New option.
25255 (indirect_branch): New.
25258 (thunk-inline): Likewise.
25259 (thunk-extern): Likewise.
25260 * doc/extend.texi: Document indirect_branch function attribute.
25261 * doc/invoke.texi: Document -mindirect-branch= option.
25263 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
25266 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
25268 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
25270 * ipa-inline.c (want_inline_small_function_p): Return false if
25271 inlining has already failed with CIF_FINAL_ERROR.
25272 (update_caller_keys): Call want_inline_small_function_p before
25274 (update_callee_keys): Likewise.
25276 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25278 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
25280 (rs6000_quadword_masked_address_p): Likewise.
25281 (quad_aligned_load_p): Likewise.
25282 (quad_aligned_store_p): Likewise.
25283 (const_load_sequence_p): Add comment to describe the outer-most loop.
25284 (mimic_memory_attributes_and_flags): New function.
25285 (rs6000_gen_stvx): Likewise.
25286 (replace_swapped_aligned_store): Likewise.
25287 (rs6000_gen_lvx): Likewise.
25288 (replace_swapped_aligned_load): Likewise.
25289 (replace_swapped_load_constant): Capitalize argument name in
25290 comment describing this function.
25291 (rs6000_analyze_swaps): Add a third pass to search for vector loads
25292 and stores that access quad-word aligned addresses and replace
25293 with stvx or lvx instructions when appropriate.
25294 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
25295 New function prototype.
25296 (rs6000_quadword_masked_address_p): Likewise.
25297 (rs6000_gen_lvx): Likewise.
25298 (rs6000_gen_stvx): Likewise.
25299 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
25300 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
25301 when memory address is aligned.
25302 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
25303 this split to select lvx instruction when memory address is aligned.
25304 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
25305 instruction when memory address is aligned.
25306 (*vsx_le_perm_load_v16qi): Likewise.
25307 (four unnamed splitters): Modify to select the stvx instruction
25308 when memory is aligned.
25310 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
25312 * predict.c (determine_unlikely_bbs): Handle correctly BBs
25313 which appears in the queue multiple times.
25315 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25316 Alan Hayward <alan.hayward@arm.com>
25317 David Sherwood <david.sherwood@arm.com>
25319 * tree-vectorizer.h (vec_lower_bound): New structure.
25320 (_loop_vec_info): Add check_nonzero and lower_bounds.
25321 (LOOP_VINFO_CHECK_NONZERO): New macro.
25322 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
25323 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
25324 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
25325 fields. Make seg_len the distance travelled, not including the
25327 (dr_direction_indicator): Declare.
25328 (dr_zero_step_indicator): Likewise.
25329 (dr_known_forward_stride_p): Likewise.
25330 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
25332 (runtime_alias_check_p): Allow runtime alias checks with
25334 (operator ==): Compare access_size and align.
25335 (prune_runtime_alias_test_list): Rework for new distinction between
25336 the access_size and seg_len.
25337 (create_intersect_range_checks_index): Likewise. Cope with polynomial
25339 (get_segment_min_max): New function.
25340 (create_intersect_range_checks): Use it.
25341 (dr_step_indicator): New function.
25342 (dr_direction_indicator): Likewise.
25343 (dr_zero_step_indicator): Likewise.
25344 (dr_known_forward_stride_p): Likewise.
25345 * tree-loop-distribution.c (data_ref_segment_size): Return
25346 DR_STEP * (niters - 1).
25347 (compute_alias_check_pairs): Update call to the dr_with_seg_len
25349 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
25350 (vect_preserves_scalar_order_p): New function, split out from...
25351 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
25352 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
25353 (vect_vfa_access_size): New function.
25354 (vect_vfa_align): Likewise.
25355 (vect_compile_time_alias): Take access_size_a and access_b arguments.
25356 (dump_lower_bound): New function.
25357 (vect_check_lower_bound): Likewise.
25358 (vect_small_gap_p): Likewise.
25359 (vectorizable_with_step_bound_p): Likewise.
25360 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
25361 depencies if the vectorization factor is 1. Convert the checks
25362 for nonzero steps into checks on the bounds of DR_STEP. Try using
25363 a bunds check for variable steps if the minimum required step is
25364 relatively small. Update calls to the dr_with_seg_len
25365 constructor and to vect_compile_time_alias.
25366 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
25368 (vect_loop_versioning): Call it.
25369 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
25371 (vect_estimate_min_profitable_iters): Account for any bounds checks.
25373 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25374 Alan Hayward <alan.hayward@arm.com>
25375 David Sherwood <david.sherwood@arm.com>
25377 * doc/sourcebuild.texi (vect_scatter_store): Document.
25378 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
25380 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
25382 * genopinit.c (main): Add supports_vec_scatter_store and
25383 supports_vec_scatter_store_cached to target_optabs.
25384 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
25385 IFN_MASK_SCATTER_STORE.
25386 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
25388 * internal-fn.h (internal_store_fn_p): Declare.
25389 (internal_fn_stored_value_index): Likewise.
25390 * internal-fn.c (scatter_store_direct): New macro.
25391 (expand_scatter_store_optab_fn): New function.
25392 (direct_scatter_store_optab_supported_p): New macro.
25393 (internal_store_fn_p): New function.
25394 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
25395 IFN_MASK_SCATTER_STORE.
25396 (internal_fn_mask_index): Likewise.
25397 (internal_fn_stored_value_index): New function.
25398 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
25399 for scatter stores.
25400 * optabs-query.h (supports_vec_scatter_store_p): Declare.
25401 * optabs-query.c (supports_vec_scatter_store_p): New function.
25402 * tree-vectorizer.h (vect_get_store_rhs): Declare.
25403 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
25404 true for scatter stores.
25405 (vect_gather_scatter_fn_p): Handle scatter stores too.
25406 (vect_check_gather_scatter): Consider using scatter stores if
25407 supports_vec_scatter_store_p.
25408 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
25409 scatter stores too.
25410 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
25411 internal_fn_stored_value_index.
25412 (check_load_store_masking): Handle scatter stores too.
25413 (vect_get_store_rhs): Make public.
25414 (vectorizable_call): Use internal_store_fn_p.
25415 (vectorizable_store): Handle scatter store internal functions.
25416 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
25417 when deciding whether the end of the group has been reached.
25418 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
25419 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
25420 (mask_scatter_store<mode>): New insns.
25422 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25423 Alan Hayward <alan.hayward@arm.com>
25424 David Sherwood <david.sherwood@arm.com>
25426 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
25427 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
25428 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
25430 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
25431 Use vect_truncate_gather_scatter_offset if we can't treat the
25432 operation as a normal gather load or scatter store.
25433 (get_group_load_store_type): Take the gather_scatter_info
25434 as argument. Try using a gather load or scatter store for
25435 single-element groups.
25436 (get_load_store_type): Update calls to get_group_load_store_type
25437 and vect_use_strided_gather_scatters_p.
25439 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25440 Alan Hayward <alan.hayward@arm.com>
25441 David Sherwood <david.sherwood@arm.com>
25443 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
25444 optional tree argument.
25445 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
25447 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
25448 but continue to use the current value as a fallback.
25449 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
25450 to compare the updates.
25451 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
25452 (get_load_store_type): Use it when handling a strided access.
25453 (vect_get_strided_load_store_ops): New function.
25454 (vect_get_data_ptr_increment): Likewise.
25455 (vectorizable_load): Handle strided gather loads. Always pass
25456 a step to vect_create_data_ref_ptr and bump_vector_ptr.
25458 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25459 Alan Hayward <alan.hayward@arm.com>
25460 David Sherwood <david.sherwood@arm.com>
25462 * doc/md.texi (gather_load@var{m}): Document.
25463 (mask_gather_load@var{m}): Likewise.
25464 * genopinit.c (main): Add supports_vec_gather_load and
25465 supports_vec_gather_load_cached to target_optabs.
25466 * optabs-tree.c (init_tree_optimization_optabs): Use
25467 ggc_cleared_alloc to allocate target_optabs.
25468 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
25469 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
25471 * internal-fn.h (internal_load_fn_p): Declare.
25472 (internal_gather_scatter_fn_p): Likewise.
25473 (internal_fn_mask_index): Likewise.
25474 (internal_gather_scatter_fn_supported_p): Likewise.
25475 * internal-fn.c (gather_load_direct): New macro.
25476 (expand_gather_load_optab_fn): New function.
25477 (direct_gather_load_optab_supported_p): New macro.
25478 (direct_internal_fn_optab): New function.
25479 (internal_load_fn_p): Likewise.
25480 (internal_gather_scatter_fn_p): Likewise.
25481 (internal_fn_mask_index): Likewise.
25482 (internal_gather_scatter_fn_supported_p): Likewise.
25483 * optabs-query.c (supports_at_least_one_mode_p): New function.
25484 (supports_vec_gather_load_p): Likewise.
25485 * optabs-query.h (supports_vec_gather_load_p): Declare.
25486 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
25487 and memory_type field.
25488 (NUM_PATTERNS): Bump to 15.
25489 * tree-vect-data-refs.c: Include internal-fn.h.
25490 (vect_gather_scatter_fn_p): New function.
25491 (vect_describe_gather_scatter_call): Likewise.
25492 (vect_check_gather_scatter): Try using internal functions for
25493 gather loads. Recognize existing calls to a gather load function.
25494 (vect_analyze_data_refs): Consider using gather loads if
25495 supports_vec_gather_load_p.
25496 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
25497 (vect_get_gather_scatter_offset_type): Likewise.
25498 (vect_convert_mask_for_vectype): Likewise.
25499 (vect_add_conversion_to_patterm): Likewise.
25500 (vect_try_gather_scatter_pattern): Likewise.
25501 (vect_recog_gather_scatter_pattern): New pattern recognizer.
25502 (vect_vect_recog_func_ptrs): Add it.
25503 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
25504 internal_fn_mask_index and internal_gather_scatter_fn_p.
25505 (check_load_store_masking): Take the gather_scatter_info as an
25506 argument and handle gather loads.
25507 (vect_get_gather_scatter_ops): New function.
25508 (vectorizable_call): Check internal_load_fn_p.
25509 (vectorizable_load): Likewise. Handle gather load internal
25511 (vectorizable_store): Update call to check_load_store_masking.
25512 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
25513 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
25514 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
25515 (aarch64_gather_scale_operand_d): New predicates.
25516 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
25517 (mask_gather_load<mode>): New insns.
25519 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25520 Alan Hayward <alan.hayward@arm.com>
25521 David Sherwood <david.sherwood@arm.com>
25523 * optabs.def (fold_left_plus_optab): New optab.
25524 * doc/md.texi (fold_left_plus_@var{m}): Document.
25525 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
25526 * internal-fn.c (fold_left_direct): Define.
25527 (expand_fold_left_optab_fn): Likewise.
25528 (direct_fold_left_optab_supported_p): Likewise.
25529 * fold-const-call.c (fold_const_fold_left): New function.
25530 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
25531 * tree-parloops.c (valid_reduction_p): New function.
25532 (gather_scalar_reductions): Use it.
25533 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
25534 (vect_finish_replace_stmt): Declare.
25535 * tree-vect-loop.c (fold_left_reduction_fn): New function.
25536 (needs_fold_left_reduction_p): New function, split out from...
25537 (vect_is_simple_reduction): ...here. Accept reductions that
25538 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
25539 (vect_force_simple_reduction): Also store the reduction type in
25540 the assignment's STMT_VINFO_REDUC_TYPE.
25541 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
25542 (merge_with_identity): New function.
25543 (vect_expand_fold_left): Likewise.
25544 (vectorize_fold_left_reduction): Likewise.
25545 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
25546 scalar phi in place for it. Check for target support and reject
25547 cases that would reassociate the operation. Defer the transform
25548 phase to vectorize_fold_left_reduction.
25549 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
25550 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
25551 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
25553 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25555 * tree-if-conv.c (predicate_mem_writes): Remove redundant
25556 call to ifc_temp_var.
25558 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25559 Alan Hayward <alan.hayward@arm.com>
25560 David Sherwood <david.sherwood@arm.com>
25562 * target.def (legitimize_address_displacement): Take the original
25563 offset as a poly_int.
25564 * targhooks.h (default_legitimize_address_displacement): Update
25566 * targhooks.c (default_legitimize_address_displacement): Likewise.
25567 * doc/tm.texi: Regenerate.
25568 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
25569 as an argument, moving assert of ad->disp == ad->disp_term to...
25570 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
25571 Try calling targetm.legitimize_address_displacement before expanding
25572 the address rather than afterwards, and adjust for the new interface.
25573 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
25574 Match the new hook interface. Handle SVE addresses.
25575 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
25576 new hook interface.
25578 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25580 * Makefile.in (OBJS): Add early-remat.o.
25581 * target.def (select_early_remat_modes): New hook.
25582 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
25583 * doc/tm.texi: Regenerate.
25584 * targhooks.h (default_select_early_remat_modes): Declare.
25585 * targhooks.c (default_select_early_remat_modes): New function.
25586 * timevar.def (TV_EARLY_REMAT): New timevar.
25587 * passes.def (pass_early_remat): New pass.
25588 * tree-pass.h (make_pass_early_remat): Declare.
25589 * early-remat.c: New file.
25590 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
25592 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
25594 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25595 Alan Hayward <alan.hayward@arm.com>
25596 David Sherwood <david.sherwood@arm.com>
25598 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
25599 vfm1 with a bound_epilog parameter.
25600 (vect_do_peeling): Update calls accordingly, and move the prologue
25601 call earlier in the function. Treat the base bound_epilog as 0 for
25602 fully-masked loops and retain vf - 1 for other loops. Add 1 to
25603 this base when peeling for gaps.
25604 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
25605 with fully-masked loops.
25606 (vect_estimate_min_profitable_iters): Handle the single peeled
25607 iteration in that case.
25609 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25610 Alan Hayward <alan.hayward@arm.com>
25611 David Sherwood <david.sherwood@arm.com>
25613 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
25614 single-element interleaving even if the size is not a power of 2.
25615 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
25616 accesses for single-element interleaving if the group size is
25619 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25620 Alan Hayward <alan.hayward@arm.com>
25621 David Sherwood <david.sherwood@arm.com>
25623 * doc/md.texi (fold_extract_last_@var{m}): Document.
25624 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
25625 * optabs.def (fold_extract_last_optab): New optab.
25626 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
25627 * internal-fn.c (fold_extract_direct): New macro.
25628 (expand_fold_extract_optab_fn): Likewise.
25629 (direct_fold_extract_optab_supported_p): Likewise.
25630 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
25631 * tree-vect-loop.c (vect_model_reduction_cost): Handle
25632 EXTRACT_LAST_REDUCTION.
25633 (get_initial_def_for_reduction): Do not create an initial vector
25634 for EXTRACT_LAST_REDUCTION reductions.
25635 (vectorizable_reduction): Leave the scalar phi in place for
25636 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
25637 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
25638 epilogue code for EXTRACT_LAST_REDUCTION and defer the
25639 transform phase to vectorizable_condition.
25640 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
25642 (vect_finish_stmt_generation): ...here.
25643 (vect_finish_replace_stmt): New function.
25644 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
25645 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
25647 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
25649 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25650 Alan Hayward <alan.hayward@arm.com>
25651 David Sherwood <david.sherwood@arm.com>
25653 * doc/md.texi (extract_last_@var{m}): Document.
25654 * optabs.def (extract_last_optab): New optab.
25655 * internal-fn.def (EXTRACT_LAST): New internal function.
25656 * internal-fn.c (cond_unary_direct): New macro.
25657 (expand_cond_unary_optab_fn): Likewise.
25658 (direct_cond_unary_optab_supported_p): Likewise.
25659 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
25660 loops using EXTRACT_LAST.
25661 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
25662 (extract_last_<mode>): ...this optab.
25663 (vec_extract<mode><Vel>): Update accordingly.
25665 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25666 Alan Hayward <alan.hayward@arm.com>
25667 David Sherwood <david.sherwood@arm.com>
25669 * target.def (empty_mask_is_expensive): New hook.
25670 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
25671 * doc/tm.texi: Regenerate.
25672 * targhooks.h (default_empty_mask_is_expensive): Declare.
25673 * targhooks.c (default_empty_mask_is_expensive): New function.
25674 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
25675 if the target says that empty masks are expensive.
25676 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
25678 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
25680 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25681 Alan Hayward <alan.hayward@arm.com>
25682 David Sherwood <david.sherwood@arm.com>
25684 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
25685 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
25686 (vect_use_loop_mask_for_alignment_p): New function.
25687 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
25688 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
25689 niters_skip argument. Make sure that the first niters_skip elements
25690 of the first iteration are inactive.
25691 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
25692 Update call to vect_set_loop_masks_directly.
25693 (get_misalign_in_elems): New function, split out from...
25694 (vect_gen_prolog_loop_niters): ...here.
25695 (vect_update_init_of_dr): Take a code argument that specifies whether
25696 the adjustment should be added or subtracted.
25697 (vect_update_init_of_drs): Likewise.
25698 (vect_prepare_for_masked_peels): New function.
25699 (vect_do_peeling): Skip prologue peeling if we're using a mask
25700 instead. Update call to vect_update_inits_of_drs.
25701 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
25703 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
25704 alignment. Do not include the number of peeled iterations in
25705 the minimum threshold in that case.
25706 (vectorizable_induction): Adjust the start value down by
25707 LOOP_VINFO_MASK_SKIP_NITERS iterations.
25708 (vect_transform_loop): Call vect_prepare_for_masked_peels.
25709 Take the number of skipped iterations into account when calculating
25711 * tree-vect-stmts.c (vect_gen_while_not): New function.
25713 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25714 Alan Hayward <alan.hayward@arm.com>
25715 David Sherwood <david.sherwood@arm.com>
25717 * doc/sourcebuild.texi (vect_fully_masked): Document.
25718 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
25719 default value to 0.
25720 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
25722 (vect_analyze_loop_2): ...here. Don't check the vectorization
25723 factor against the number of loop iterations if the loop is
25726 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25727 Alan Hayward <alan.hayward@arm.com>
25728 David Sherwood <david.sherwood@arm.com>
25730 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
25731 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
25732 (dump_groups): Update accordingly.
25733 (iv_use::mem_type): New member variable.
25734 (address_p): New function.
25735 (record_use): Add a mem_type argument and initialize the new
25737 (record_group_use): Add a mem_type argument. Use address_p.
25738 Remove obsolete null checks of base_object. Update call to record_use.
25739 (find_interesting_uses_op): Update call to record_group_use.
25740 (find_interesting_uses_cond): Likewise.
25741 (find_interesting_uses_address): Likewise.
25742 (get_mem_type_for_internal_fn): New function.
25743 (find_address_like_use): Likewise.
25744 (find_interesting_uses_stmt): Try find_address_like_use before
25745 calling find_interesting_uses_op.
25746 (addr_offset_valid_p): Use the iv mem_type field as the type
25747 of the addressed memory.
25748 (add_autoinc_candidates): Likewise.
25749 (get_address_cost): Likewise.
25750 (split_small_address_groups_p): Use address_p.
25751 (split_address_groups): Likewise.
25752 (add_iv_candidate_for_use): Likewise.
25753 (autoinc_possible_for_pair): Likewise.
25754 (rewrite_groups): Likewise.
25755 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
25756 (determine_group_iv_cost): Update after split of USE_ADDRESS.
25757 (get_alias_ptr_type_for_ptr_address): New function.
25758 (rewrite_use_address): Rewrite address uses in calls that were
25759 identified by find_address_like_use.
25761 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25762 Alan Hayward <alan.hayward@arm.com>
25763 David Sherwood <david.sherwood@arm.com>
25765 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
25767 * gimple-expr.h (is_gimple_addressable: Likewise.
25768 * gimple-expr.c (is_gimple_address): Likewise.
25769 * internal-fn.c (expand_call_mem_ref): New function.
25770 (expand_mask_load_optab_fn): Use it.
25771 (expand_mask_store_optab_fn): Likewise.
25773 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25774 Alan Hayward <alan.hayward@arm.com>
25775 David Sherwood <david.sherwood@arm.com>
25777 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
25778 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
25779 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
25780 (cond_umax@var{mode}): Document.
25781 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
25782 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
25783 (cond_umin_optab, cond_umax_optab): New optabs.
25784 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
25785 (COND_IOR, COND_XOR): New internal functions.
25786 * internal-fn.h (get_conditional_internal_fn): Declare.
25787 * internal-fn.c (cond_binary_direct): New macro.
25788 (expand_cond_binary_optab_fn): Likewise.
25789 (direct_cond_binary_optab_supported_p): Likewise.
25790 (get_conditional_internal_fn): New function.
25791 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
25792 Cope with reduction statements that are vectorized as calls rather
25794 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
25795 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
25796 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
25797 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
25798 (UNSPEC_COND_EOR): New unspecs.
25799 (optab): Add mappings for them.
25800 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
25801 (sve_int_op, sve_fp_op): New int attributes.
25803 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25804 Alan Hayward <alan.hayward@arm.com>
25805 David Sherwood <david.sherwood@arm.com>
25807 * optabs.def (while_ult_optab): New optab.
25808 * doc/md.texi (while_ult@var{m}@var{n}): Document.
25809 * internal-fn.def (WHILE_ULT): New internal function.
25810 * internal-fn.h (direct_internal_fn_supported_p): New override
25811 that takes two types as argument.
25812 * internal-fn.c (while_direct): New macro.
25813 (expand_while_optab_fn): New function.
25814 (convert_optab_supported_p): Likewise.
25815 (direct_while_optab_supported_p): New macro.
25816 * wide-int.h (wi::udiv_ceil): New function.
25817 * tree-vectorizer.h (rgroup_masks): New structure.
25818 (vec_loop_masks): New typedef.
25819 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
25820 and fully_masked_p.
25821 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
25822 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
25823 (vect_max_vf): New function.
25824 (slpeel_make_loop_iterate_ntimes): Delete.
25825 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
25826 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
25827 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
25828 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
25829 internal-fn.h, stor-layout.h and optabs-query.h.
25830 (vect_set_loop_mask): New function.
25831 (add_preheader_seq): Likewise.
25832 (add_header_seq): Likewise.
25833 (interleave_supported_p): Likewise.
25834 (vect_maybe_permute_loop_masks): Likewise.
25835 (vect_set_loop_masks_directly): Likewise.
25836 (vect_set_loop_condition_masked): Likewise.
25837 (vect_set_loop_condition_unmasked): New function, split out from
25838 slpeel_make_loop_iterate_ntimes.
25839 (slpeel_make_loop_iterate_ntimes): Rename to..
25840 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
25841 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
25842 (vect_do_peeling): Update call accordingly.
25843 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
25845 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
25846 mask_compare_type, can_fully_mask_p and fully_masked_p.
25847 (release_vec_loop_masks): New function.
25848 (_loop_vec_info): Use it to free the loop masks.
25849 (can_produce_all_loop_masks_p): New function.
25850 (vect_get_max_nscalars_per_iter): Likewise.
25851 (vect_verify_full_masking): Likewise.
25852 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
25853 retries, and free the mask rgroups before retrying. Check loop-wide
25854 reasons for disallowing fully-masked loops. Make the final decision
25855 about whether use a fully-masked loop or not.
25856 (vect_estimate_min_profitable_iters): Do not assume that peeling
25857 for the number of iterations will be needed for fully-masked loops.
25858 (vectorizable_reduction): Disable fully-masked loops.
25859 (vectorizable_live_operation): Likewise.
25860 (vect_halve_mask_nunits): New function.
25861 (vect_double_mask_nunits): Likewise.
25862 (vect_record_loop_mask): Likewise.
25863 (vect_get_loop_mask): Likewise.
25864 (vect_transform_loop): Handle the case in which the final loop
25865 iteration might handle a partial vector. Call vect_set_loop_condition
25866 instead of slpeel_make_loop_iterate_ntimes.
25867 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
25868 (check_load_store_masking): New function.
25869 (prepare_load_store_mask): Likewise.
25870 (vectorizable_store): Handle fully-masked loops.
25871 (vectorizable_load): Likewise.
25872 (supportable_widening_operation): Use vect_halve_mask_nunits for
25874 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
25875 (vect_gen_while): New function.
25876 * config/aarch64/aarch64.md (umax<mode>3): New expander.
25877 (aarch64_uqdec<mode>): New insn.
25879 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25880 Alan Hayward <alan.hayward@arm.com>
25881 David Sherwood <david.sherwood@arm.com>
25883 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
25884 (reduc_xor_scal_optab): New optabs.
25885 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
25886 (reduc_xor_scal_@var{m}): Document.
25887 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
25888 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
25889 internal functions.
25890 * fold-const-call.c (fold_const_call): Handle them.
25891 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
25892 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
25893 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
25894 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
25895 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
25896 (UNSPEC_XORV): New unspecs.
25897 (optab): Add entries for them.
25898 (BITWISEV): New int iterator.
25899 (bit_reduc_op): New int attributes.
25901 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25902 Alan Hayward <alan.hayward@arm.com>
25903 David Sherwood <david.sherwood@arm.com>
25905 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
25906 * internal-fn.def (VEC_SHL_INSERT): New internal function.
25907 * optabs.def (vec_shl_insert_optab): New optab.
25908 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
25909 (duplicate_and_interleave): Likewise.
25910 * tree-vect-loop.c: Include internal-fn.h.
25911 (neutral_op_for_slp_reduction): New function, split out from
25912 get_initial_defs_for_reduction.
25913 (get_initial_def_for_reduction): Handle option 2 for variable-length
25914 vectors by loading the neutral value into a vector and then shifting
25915 the initial value into element 0.
25916 (get_initial_defs_for_reduction): Replace the code argument with
25917 the neutral value calculated by neutral_op_for_slp_reduction.
25918 Use gimple_build_vector for constant-length vectors.
25919 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
25920 but the first group_size elements have a neutral value.
25921 Use duplicate_and_interleave otherwise.
25922 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
25923 Update call to get_initial_defs_for_reduction. Handle SLP
25924 reductions for variable-length vectors by creating one vector
25925 result for each scalar result, with the elements associated
25926 with other scalar results stubbed out with the neutral value.
25927 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
25928 Require IFN_VEC_SHL_INSERT for double reductions on
25929 variable-length vectors, or SLP reductions that have
25930 a neutral value. Require can_duplicate_and_interleave_p
25931 support for variable-length unchained SLP reductions if there
25932 is no neutral value, such as for MIN/MAX reductions. Also require
25933 the number of vector elements to be a multiple of the number of
25934 SLP statements when doing variable-length unchained SLP reductions.
25935 Update call to vect_create_epilog_for_reduction.
25936 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
25937 and remove initial values.
25938 (duplicate_and_interleave): Make public.
25939 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
25940 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
25942 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25943 Alan Hayward <alan.hayward@arm.com>
25944 David Sherwood <david.sherwood@arm.com>
25946 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
25947 (can_duplicate_and_interleave_p): New function.
25948 (vect_get_and_check_slp_defs): Take the vector of statements
25949 rather than just the current one. Remove excess parentheses.
25950 Restriction rejectinon of vect_constant_def and vect_external_def
25951 for variable-length vectors to boolean types, or types for which
25952 can_duplicate_and_interleave_p is false.
25953 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
25954 (duplicate_and_interleave): New function.
25955 (vect_get_constant_vectors): Use gimple_build_vector for
25956 constant-length vectors and suitable variable-length constant
25957 vectors. Use duplicate_and_interleave for other variable-length
25958 vectors. Don't defer the update when inserting new statements.
25960 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25961 Alan Hayward <alan.hayward@arm.com>
25962 David Sherwood <david.sherwood@arm.com>
25964 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
25965 min_profitable_iters doesn't go negative.
25967 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25968 Alan Hayward <alan.hayward@arm.com>
25969 David Sherwood <david.sherwood@arm.com>
25971 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
25972 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
25973 * optabs.def (vec_mask_load_lanes_optab): New optab.
25974 (vec_mask_store_lanes_optab): Likewise.
25975 * internal-fn.def (MASK_LOAD_LANES): New internal function.
25976 (MASK_STORE_LANES): Likewise.
25977 * internal-fn.c (mask_load_lanes_direct): New macro.
25978 (mask_store_lanes_direct): Likewise.
25979 (expand_mask_load_optab_fn): Handle masked operations.
25980 (expand_mask_load_lanes_optab_fn): New macro.
25981 (expand_mask_store_optab_fn): Handle masked operations.
25982 (expand_mask_store_lanes_optab_fn): New macro.
25983 (direct_mask_load_lanes_optab_supported_p): Likewise.
25984 (direct_mask_store_lanes_optab_supported_p): Likewise.
25985 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
25987 (vect_load_lanes_supported): Likewise.
25988 * tree-vect-data-refs.c (strip_conversion): New function.
25989 (can_group_stmts_p): Likewise.
25990 (vect_analyze_data_ref_accesses): Use it instead of checking
25991 for a pair of assignments.
25992 (vect_store_lanes_supported): Take a masked_p parameter.
25993 (vect_load_lanes_supported): Likewise.
25994 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
25995 vect_store_lanes_supported and vect_load_lanes_supported.
25996 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
25997 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
25998 parameter. Don't allow gaps for masked accesses.
25999 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
26000 and vect_load_lanes_supported.
26001 (get_load_store_type): Take a masked_p parameter and update
26002 call to get_group_load_store_type.
26003 (vectorizable_store): Update call to get_load_store_type.
26004 Handle IFN_MASK_STORE_LANES.
26005 (vectorizable_load): Update call to get_load_store_type.
26006 Handle IFN_MASK_LOAD_LANES.
26008 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26009 Alan Hayward <alan.hayward@arm.com>
26010 David Sherwood <david.sherwood@arm.com>
26012 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
26014 * config/aarch64/aarch64-protos.h
26015 (aarch64_sve_struct_memory_operand_p): Declare.
26016 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
26017 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
26018 (VPRED, vpred): Handle SVE structure modes.
26019 * config/aarch64/constraints.md (Utx): New constraint.
26020 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
26021 (aarch64_sve_struct_nonimmediate_operand): New predicates.
26022 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
26023 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
26024 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
26025 structure modes. Split into pieces after RA.
26026 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
26027 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
26029 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
26030 SVE structure modes.
26031 (aarch64_classify_address): Likewise.
26032 (sizetochar): Move earlier in file.
26033 (aarch64_print_operand): Handle SVE register lists.
26034 (aarch64_array_mode): New function.
26035 (aarch64_sve_struct_memory_operand_p): Likewise.
26036 (TARGET_ARRAY_MODE): Redefine.
26038 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26039 Alan Hayward <alan.hayward@arm.com>
26040 David Sherwood <david.sherwood@arm.com>
26042 * target.def (array_mode): New target hook.
26043 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
26044 * doc/tm.texi: Regenerate.
26045 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
26046 * hooks.c (hook_optmode_mode_uhwi_none): New function.
26047 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
26048 targetm.array_mode.
26049 * stor-layout.c (mode_for_array): Likewise. Support polynomial
26052 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26053 Alan Hayward <alan.hayward@arm.com>
26054 David Sherwood <david.sherwood@arm.com>
26056 * fold-const.c (fold_binary_loc): Check the argument types
26057 rather than the result type when testing for a vector operation.
26059 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26061 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
26062 * doc/tm.texi: Regenerate.
26064 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26065 Alan Hayward <alan.hayward@arm.com>
26066 David Sherwood <david.sherwood@arm.com>
26068 * doc/invoke.texi (-msve-vector-bits=): Document new option.
26069 (sve): Document new AArch64 extension.
26070 * doc/md.texi (w): Extend the description of the AArch64
26071 constraint to include SVE vectors.
26072 (Upl, Upa): Document new AArch64 predicate constraints.
26073 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
26075 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
26076 (msve-vector-bits=): New option.
26077 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
26078 SVE when these are disabled.
26079 (sve): New extension.
26080 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
26081 modes. Adjust their number of units based on aarch64_sve_vg.
26082 (MAX_BITSIZE_MODE_ANY_MODE): Define.
26083 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
26084 aarch64_addr_query_type.
26085 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
26086 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
26087 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
26088 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
26089 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
26090 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
26091 (aarch64_simd_imm_zero_p): Delete.
26092 (aarch64_check_zero_based_sve_index_immediate): Declare.
26093 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
26094 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
26095 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
26096 (aarch64_sve_float_mul_immediate_p): Likewise.
26097 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
26098 rather than an rtx.
26099 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
26100 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
26101 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
26102 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
26103 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
26104 (aarch64_regmode_natural_size): Likewise.
26105 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
26106 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
26108 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
26109 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
26110 for VG and the SVE predicate registers.
26111 (V_ALIASES): Add a "z"-prefixed alias.
26112 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
26113 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
26114 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
26115 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
26116 (REG_CLASS_NAMES): Add entries for them.
26117 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
26118 and the predicate registers.
26119 (aarch64_sve_vg): Declare.
26120 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
26121 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
26122 (REGMODE_NATURAL_SIZE): Define.
26123 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
26125 * config/aarch64/aarch64.c: Include cfgrtl.h.
26126 (simd_immediate_info): Add a constructor for series vectors,
26127 and an associated step field.
26128 (aarch64_sve_vg): New variable.
26129 (aarch64_dbx_register_number): Handle VG and the predicate registers.
26130 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
26131 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
26132 (VEC_ANY_DATA, VEC_STRUCT): New constants.
26133 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
26134 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
26135 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
26136 (aarch64_get_mask_mode): New functions.
26137 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
26138 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
26139 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
26140 predicate modes and predicate registers. Explicitly restrict
26141 GPRs to modes of 16 bytes or smaller. Only allow FP registers
26142 to store a vector mode if it is recognized by
26143 aarch64_classify_vector_mode.
26144 (aarch64_regmode_natural_size): New function.
26145 (aarch64_hard_regno_caller_save_mode): Return the original mode
26147 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
26148 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
26149 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
26150 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
26152 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
26153 does not overlap dest if the function is frame-related. Handle
26155 (aarch64_split_add_offset): New function.
26156 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
26157 them aarch64_add_offset.
26158 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
26159 and update call to aarch64_sub_sp.
26160 (aarch64_add_cfa_expression): New function.
26161 (aarch64_expand_prologue): Pass extra temporary registers to the
26162 functions above. Handle the case in which we need to emit new
26163 DW_CFA_expressions for registers that were originally saved
26164 relative to the stack pointer, but now have to be expressed
26165 relative to the frame pointer.
26166 (aarch64_output_mi_thunk): Pass extra temporary registers to the
26168 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
26169 IP0 and IP1 values for SVE frames.
26170 (aarch64_expand_vec_series): New function.
26171 (aarch64_expand_sve_widened_duplicate): Likewise.
26172 (aarch64_expand_sve_const_vector): Likewise.
26173 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
26174 Handle SVE constants. Use emit_move_insn to move a force_const_mem
26175 into the register, rather than emitting a SET directly.
26176 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
26177 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
26178 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
26179 (offset_9bit_signed_scaled_p): New functions.
26180 (aarch64_replicate_bitmask_imm): New function.
26181 (aarch64_bitmask_imm): Use it.
26182 (aarch64_cannot_force_const_mem): Reject expressions involving
26183 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
26184 (aarch64_classify_index): Handle SVE indices, by requiring
26185 a plain register index with a scale that matches the element size.
26186 (aarch64_classify_address): Handle SVE addresses. Assert that
26187 the mode of the address is VOIDmode or an integer mode.
26188 Update call to aarch64_classify_symbol.
26189 (aarch64_classify_symbolic_expression): Update call to
26190 aarch64_classify_symbol.
26191 (aarch64_const_vec_all_in_range_p): New function.
26192 (aarch64_print_vector_float_operand): Likewise.
26193 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
26194 "vN" for FP registers with SVE modes. Handle (const ...) vectors
26195 and the FP immediates 1.0 and 0.5.
26196 (aarch64_print_address_internal): Handle SVE addresses.
26197 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
26198 (aarch64_regno_regclass): Handle predicate registers.
26199 (aarch64_secondary_reload): Handle big-endian reloads of SVE
26201 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
26202 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
26203 (aarch64_convert_sve_vector_bits): New function.
26204 (aarch64_override_options): Use it to handle -msve-vector-bits=.
26205 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
26206 rather than an rtx.
26207 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
26208 Handle SVE vector and predicate modes. Accept VL-based constants
26209 that need only one temporary register, and VL offsets that require
26210 no temporary registers.
26211 (aarch64_conditional_register_usage): Mark the predicate registers
26212 as fixed if SVE isn't available.
26213 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
26214 Return true for SVE vector and predicate modes.
26215 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
26216 rather than an unsigned int. Handle SVE modes.
26217 (aarch64_preferred_simd_mode): Update call accordingly. Handle
26219 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
26221 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
26222 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
26223 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
26224 (aarch64_sve_float_mul_immediate_p): New functions.
26225 (aarch64_sve_valid_immediate): New function.
26226 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
26227 Explicitly reject structure modes. Check for INDEX constants.
26228 Handle PTRUE and PFALSE constants.
26229 (aarch64_check_zero_based_sve_index_immediate): New function.
26230 (aarch64_simd_imm_zero_p): Delete.
26231 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
26232 vector modes. Accept constants in the range of CNT[BHWD].
26233 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
26234 ask for an Advanced SIMD mode.
26235 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
26236 (aarch64_simd_vector_alignment): Handle SVE predicates.
26237 (aarch64_vectorize_preferred_vector_alignment): New function.
26238 (aarch64_simd_vector_alignment_reachable): Use it instead of
26240 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
26241 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
26243 (MAX_VECT_LEN): Delete.
26244 (expand_vec_perm_d): Add a vec_flags field.
26245 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
26246 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
26247 (aarch64_evpc_ext): Don't apply a big-endian lane correction
26249 (aarch64_evpc_rev): Rename to...
26250 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
26251 (aarch64_evpc_rev_global): New function.
26252 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
26253 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
26255 (aarch64_evpc_sve_tbl): New function.
26256 (aarch64_expand_vec_perm_const_1): Update after rename of
26257 aarch64_evpc_rev. Handle SVE permutes too, trying
26258 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
26259 than aarch64_evpc_tbl.
26260 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
26261 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
26262 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
26263 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
26264 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
26265 (aarch64_expand_sve_vcond): New functions.
26266 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
26267 of aarch64_vector_mode_p.
26268 (aarch64_dwarf_poly_indeterminate_value): New function.
26269 (aarch64_compute_pressure_classes): Likewise.
26270 (aarch64_can_change_mode_class): Likewise.
26271 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
26272 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
26273 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
26274 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
26275 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
26276 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
26277 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
26278 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
26280 (Dn, Dl, Dr): Accept const as well as const_vector.
26281 (Dz): Likewise. Compare against CONST0_RTX.
26282 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
26283 of "vector" where appropriate.
26284 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
26285 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
26286 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
26287 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
26288 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
26289 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
26290 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
26291 (v_int_equiv): Extend to SVE modes.
26292 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
26294 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
26295 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
26296 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
26297 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
26298 (SVE_COND_FP_CMP): New int iterators.
26299 (perm_hilo): Handle the new unpack unspecs.
26300 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
26302 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
26303 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
26304 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
26305 (aarch64_equality_operator, aarch64_constant_vector_operand)
26306 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
26307 (aarch64_sve_nonimmediate_operand): Likewise.
26308 (aarch64_sve_general_operand): Likewise.
26309 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
26310 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
26311 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
26312 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
26313 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
26314 (aarch64_sve_float_arith_immediate): Likewise.
26315 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
26316 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
26317 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
26318 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
26319 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
26320 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
26321 (aarch64_sve_float_arith_operand): Likewise.
26322 (aarch64_sve_float_arith_with_sub_operand): Likewise.
26323 (aarch64_sve_float_mul_operand): Likewise.
26324 (aarch64_sve_vec_perm_operand): Likewise.
26325 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
26326 (aarch64_mov_operand): Accept const_poly_int and const_vector.
26327 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
26328 as well as const_vector.
26329 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
26330 in file. Use CONST0_RTX and CONSTM1_RTX.
26331 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
26332 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
26333 Use aarch64_simd_imm_zero.
26334 * config/aarch64/aarch64-sve.md: New file.
26335 * config/aarch64/aarch64.md: Include it.
26336 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
26337 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
26338 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
26339 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
26340 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
26341 (sve): New attribute.
26342 (enabled): Disable instructions with the sve attribute unless
26344 (movqi, movhi): Pass CONST_POLY_INT operaneds through
26345 aarch64_expand_mov_immediate.
26346 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
26347 CNT[BHSD] immediates.
26348 (movti): Split CONST_POLY_INT moves into two halves.
26349 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
26350 Split additions that need a temporary here if the destination
26351 is the stack pointer.
26352 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
26353 (*add<mode>3_poly_1): New instruction.
26354 (set_clobber_cc): New expander.
26356 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26358 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
26359 parameter and use it instead of GET_MODE_SIZE (innermode). Use
26360 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
26361 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
26362 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
26363 Change innermode from fixed_mode_size to machine_mode.
26364 (simplify_subreg): Update call accordingly. Handle a constant-sized
26365 subreg of a variable-length CONST_VECTOR.
26367 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26368 Alan Hayward <alan.hayward@arm.com>
26369 David Sherwood <david.sherwood@arm.com>
26371 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
26372 (add_offset_to_base): New function, split out from...
26373 (create_mem_ref): ...here. When handling a scale other than 1,
26374 check first whether the address is valid without the offset.
26375 Add it into the base if so, leaving the index and scale as-is.
26377 2018-01-12 Jakub Jelinek <jakub@redhat.com>
26380 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
26381 fold_for_warn before checking if arg2 is INTEGER_CST.
26383 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
26385 * config/rs6000/predicates.md (load_multiple_operation): Delete.
26386 (store_multiple_operation): Delete.
26387 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
26388 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
26389 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
26390 guarded by TARGET_STRING.
26391 (rs6000_output_load_multiple): Delete.
26392 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26393 OPTION_MASK_STRING / TARGET_STRING handling.
26394 (print_operand) <'N', 'O'>: Add comment that these are unused now.
26395 (const rs6000_opt_masks) <"string">: Change mask to 0.
26396 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
26397 (MASK_STRING): Delete.
26398 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
26400 (load_multiple): Delete.
26407 (store_multiple): Delete.
26414 (movmemsi_8reg): Delete.
26415 (corresponding unnamed define_insn): Delete.
26416 (movmemsi_6reg): Delete.
26417 (corresponding unnamed define_insn): Delete.
26418 (movmemsi_4reg): Delete.
26419 (corresponding unnamed define_insn): Delete.
26420 (movmemsi_2reg): Delete.
26421 (corresponding unnamed define_insn): Delete.
26422 (movmemsi_1reg): Delete.
26423 (corresponding unnamed define_insn): Delete.
26424 * config/rs6000/rs6000.opt (mno-string): New.
26425 (mstring): Replace by deprecation warning stub.
26426 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
26428 2018-01-12 Jakub Jelinek <jakub@redhat.com>
26430 * regrename.c (regrename_do_replace): If replacing the same
26431 reg multiple times, try to reuse last created gen_raw_REG.
26434 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
26435 main to workaround a bug in GDB.
26437 2018-01-12 Tom de Vries <tom@codesourcery.com>
26440 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
26442 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
26444 PR rtl-optimization/80481
26445 * ira-color.c (get_cap_member): New function.
26446 (allocnos_conflict_by_live_ranges_p): Use it.
26447 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
26448 (setup_slot_coalesced_allocno_live_ranges): Ditto.
26450 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
26453 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
26454 (*saddl_se_1): Ditto.
26455 (*ssubsi_1): Ditto.
26456 (*ssubl_se_1): Ditto.
26458 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
26460 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
26461 rather than wi::to_widest for DR_INITs.
26462 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
26463 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
26464 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
26466 (vect_analyze_group_access_1): Note that here.
26468 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
26470 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
26471 polynomial type sizes.
26473 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
26475 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
26476 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
26477 (gimple_add_tmp_var): Likewise.
26479 2018-01-12 Martin Liska <mliska@suse.cz>
26481 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
26482 (gimple_alloc_sizes): Likewise.
26483 (dump_gimple_statistics): Use PRIu64 in printf format.
26484 * gimple.h: Change uint64_t to int.
26486 2018-01-12 Martin Liska <mliska@suse.cz>
26488 * tree-core.h: Use uint64_t instead of int.
26489 * tree.c (tree_node_counts): Likewise.
26490 (tree_node_sizes): Likewise.
26491 (dump_tree_statistics): Use PRIu64 in printf format.
26493 2018-01-12 Martin Liska <mliska@suse.cz>
26495 * Makefile.in: As qsort_chk is implemented in vec.c, add
26496 vec.o to linkage of gencfn-macros.
26497 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
26498 passing the info to record_node_allocation_statistics.
26499 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
26501 * ggc-common.c (struct ggc_usage): Add operator== and use
26502 it in operator< and compare function.
26503 * mem-stats.h (struct mem_usage): Likewise.
26504 * vec.c (struct vec_usage): Remove operator< and compare
26505 function. Can be simply inherited.
26507 2018-01-12 Martin Jambor <mjambor@suse.cz>
26510 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
26511 * tree-ssa-math-opts.c: Include domwalk.h.
26512 (convert_mult_to_fma_1): New function.
26513 (fma_transformation_info): New type.
26514 (fma_deferring_state): Likewise.
26515 (cancel_fma_deferring): New function.
26516 (result_of_phi): Likewise.
26517 (last_fma_candidate_feeds_initial_phi): Likewise.
26518 (convert_mult_to_fma): Added deferring logic, split actual
26519 transformation to convert_mult_to_fma_1.
26520 (math_opts_dom_walker): New type.
26521 (math_opts_dom_walker::after_dom_children): New method, body moved
26522 here from pass_optimize_widening_mul::execute, added deferring logic
26524 (pass_optimize_widening_mul::execute): Moved most of code to
26525 math_opts_dom_walker::after_dom_children.
26526 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
26527 * config/i386/i386.c (ix86_option_override_internal): Added
26528 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
26530 2018-01-12 Richard Biener <rguenther@suse.de>
26533 * dwarf2out.c (gen_variable_die): Do not reset old_die for
26534 inline instance vars.
26536 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
26539 * config/rx/rx.c (rx_is_restricted_memory_address):
26540 Handle SUBREG case.
26542 2018-01-12 Richard Biener <rguenther@suse.de>
26544 PR tree-optimization/80846
26545 * target.def (split_reduction): New target hook.
26546 * targhooks.c (default_split_reduction): New function.
26547 * targhooks.h (default_split_reduction): Declare.
26548 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
26549 target requests first reduce vectors by combining low and high
26551 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
26552 (get_vectype_for_scalar_type_and_size): Export.
26553 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
26554 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
26555 * doc/tm.texi: Regenerate.
26556 * config/i386/i386.c (ix86_split_reduction): Implement
26557 TARGET_VECTORIZE_SPLIT_REDUCTION.
26559 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
26562 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
26563 in PIC mode except for TARGET_VXWORKS_RTP.
26564 * config/sparc/sparc.c: Include cfgrtl.h.
26565 (TARGET_INIT_PIC_REG): Define.
26566 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
26567 (sparc_pic_register_p): New predicate.
26568 (sparc_legitimate_address_p): Use it.
26569 (sparc_legitimize_pic_address): Likewise.
26570 (sparc_delegitimize_address): Likewise.
26571 (sparc_mode_dependent_address_p): Likewise.
26572 (gen_load_pcrel_sym): Remove 4th parameter.
26573 (load_got_register): Adjust call to above. Remove obsolete stuff.
26574 (sparc_expand_prologue): Do not call load_got_register here.
26575 (sparc_flat_expand_prologue): Likewise.
26576 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
26577 (sparc_use_pseudo_pic_reg): New function.
26578 (sparc_init_pic_reg): Likewise.
26579 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
26580 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
26582 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
26584 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26585 Add item for branch_cost.
26587 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
26589 PR rtl-optimization/83565
26590 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
26591 not extend the result to a larger mode for rotate operations.
26592 (num_sign_bit_copies1): Likewise.
26594 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26597 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
26599 Use values-Xc.o for -pedantic.
26600 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
26602 2018-01-12 Martin Liska <mliska@suse.cz>
26605 * ipa-devirt.c (final_warning_record::grow_type_warnings):
26607 (possible_polymorphic_call_targets): Use it.
26608 (ipa_devirt): Likewise.
26610 2018-01-12 Martin Liska <mliska@suse.cz>
26612 * profile-count.h (enum profile_quality): Use 0 as invalid
26613 enum value of profile_quality.
26615 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
26617 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
26618 -mext-string options.
26620 2018-01-12 Richard Biener <rguenther@suse.de>
26622 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
26623 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
26624 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
26626 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
26628 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
26630 * configure.ac (--with-long-double-format): Add support for the
26631 configuration option to change the default long double format on
26633 * config.gcc (powerpc*-linux*-*): Likewise.
26634 * configure: Regenerate.
26635 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
26636 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
26637 used without modification.
26639 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26641 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
26642 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
26643 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
26644 MISC_BUILTIN_SPEC_BARRIER.
26645 (rs6000_init_builtins): Likewise.
26646 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
26648 (speculation_barrier): New define_insn.
26649 * doc/extend.texi: Document __builtin_speculation_barrier.
26651 2018-01-11 Jakub Jelinek <jakub@redhat.com>
26654 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
26655 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
26656 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
26658 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
26659 integral modes instead of "ss" and "sd".
26660 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
26661 vectors with 32-bit and 64-bit elements.
26662 (vecdupssescalarmodesuffix): New mode attribute.
26663 (vec_dup<mode>): Use it.
26665 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
26668 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
26669 frame if argument is passed on stack.
26671 2018-01-11 Jakub Jelinek <jakub@redhat.com>
26674 * ree.c (combine_reaching_defs): Optimize also
26675 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
26676 reg2=any_extend(exp); reg1=reg2;, formatting fix.
26678 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
26680 PR middle-end/83189
26681 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
26683 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
26685 PR middle-end/83718
26686 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
26687 after they are computed.
26689 2018-01-11 Bin Cheng <bin.cheng@arm.com>
26691 PR tree-optimization/83695
26692 * gimple-loop-linterchange.cc
26693 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
26694 reset cached scev information after interchange.
26695 (pass_linterchange::execute): Remove call to scev_reset_htab.
26697 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26699 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
26700 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
26701 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
26702 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
26703 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
26704 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
26705 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
26706 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
26707 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
26708 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
26709 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
26710 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
26711 (V_lane_reg): Likewise.
26712 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
26714 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
26715 (vfmal_lane_low<mode>_intrinsic,
26716 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
26717 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
26718 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
26719 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
26720 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
26721 vfmsl_lane_high<mode>_intrinsic): New define_insns.
26723 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26725 * config/arm/arm-cpus.in (fp16fml): New feature.
26726 (ALL_SIMD): Add fp16fml.
26727 (armv8.2-a): Add fp16fml as an option.
26728 (armv8.3-a): Likewise.
26729 (armv8.4-a): Add fp16fml as part of fp16.
26730 * config/arm/arm.h (TARGET_FP16FML): Define.
26731 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
26733 * config/arm/arm-modes.def (V2HF): Define.
26734 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
26735 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
26736 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
26737 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
26738 vfmsl_low, vfmsl_high): New set of builtins.
26739 * config/arm/iterators.md (PLUSMINUS): New code iterator.
26740 (vfml_op): New code attribute.
26741 (VFMLHALVES): New int iterator.
26742 (VFML, VFMLSEL): New mode attributes.
26743 (V_reg): Define mapping for V2HF.
26744 (V_hi, V_lo): New mode attributes.
26745 (VF_constraint): Likewise.
26746 (vfml_half, vfml_half_selector): New int attributes.
26747 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
26749 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
26750 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
26752 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
26753 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
26754 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
26755 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
26757 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
26758 Document new effective target and option set.
26760 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26762 * config/arm/arm-cpus.in (armv8_4): New feature.
26763 (ARMv8_4a): New fgroup.
26764 (armv8.4-a): New arch.
26765 * config/arm/arm-tables.opt: Regenerate.
26766 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
26767 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
26768 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
26769 Add matching rules for -march=armv8.4-a and extensions.
26770 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
26772 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
26775 * config/rx/rx.md (BW): New mode attribute.
26776 (sync_lock_test_and_setsi): Add mode suffix to insn output.
26778 2018-01-11 Richard Biener <rguenther@suse.de>
26780 PR tree-optimization/83435
26781 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
26782 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
26783 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
26785 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
26786 Alan Hayward <alan.hayward@arm.com>
26787 David Sherwood <david.sherwood@arm.com>
26789 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
26791 (aarch64_classify_address): Initialize it. Track polynomial offsets.
26792 (aarch64_print_address_internal): Use it to check for a zero offset.
26794 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
26795 Alan Hayward <alan.hayward@arm.com>
26796 David Sherwood <david.sherwood@arm.com>
26798 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
26799 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
26800 Return a poly_int64 rather than a HOST_WIDE_INT.
26801 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
26802 rather than a HOST_WIDE_INT.
26803 * config/aarch64/aarch64.h (aarch64_frame): Protect with
26804 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
26805 hard_fp_offset, frame_size, initial_adjust, callee_offset and
26806 final_offset from HOST_WIDE_INT to poly_int64.
26807 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
26808 to_constant when getting the number of units in an Advanced SIMD
26810 (aarch64_builtin_vectorized_function): Check for a constant number
26812 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
26814 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
26815 attribute instead of GET_MODE_NUNITS.
26816 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
26817 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
26818 GET_MODE_SIZE for fixed-size registers.
26819 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
26820 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
26821 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
26822 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
26823 (aarch64_print_operand, aarch64_print_address_internal)
26824 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
26825 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
26826 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
26827 Handle polynomial GET_MODE_SIZE.
26828 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
26829 wider than SImode without modification.
26830 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
26831 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
26832 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
26833 passing and returning SVE modes.
26834 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
26835 rather than GEN_INT.
26836 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
26837 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
26838 (aarch64_allocate_and_probe_stack_space): Likewise.
26839 (aarch64_layout_frame): Cope with polynomial offsets.
26840 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
26841 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
26842 polynomial offsets.
26843 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
26844 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
26845 poly_int64 rather than a HOST_WIDE_INT.
26846 (aarch64_get_separate_components, aarch64_process_components)
26847 (aarch64_expand_prologue, aarch64_expand_epilogue)
26848 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
26849 (aarch64_anchor_offset): New function, split out from...
26850 (aarch64_legitimize_address): ...here.
26851 (aarch64_builtin_vectorization_cost): Handle polynomial
26852 TYPE_VECTOR_SUBPARTS.
26853 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
26855 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
26856 number of elements from the PARALLEL rather than the mode.
26857 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
26858 rather than GET_MODE_BITSIZE.
26859 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
26860 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
26861 (aarch64_expand_vec_perm_const_1): Handle polynomial
26862 d->perm.length () and d->perm elements.
26863 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
26864 Apply to_constant to d->perm elements.
26865 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
26866 polynomial CONST_VECTOR_NUNITS.
26867 (aarch64_move_pointer): Take amount as a poly_int64 rather
26869 (aarch64_progress_pointer): Avoid temporary variable.
26870 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
26871 the mode attribute instead of GET_MODE.
26873 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
26874 Alan Hayward <alan.hayward@arm.com>
26875 David Sherwood <david.sherwood@arm.com>
26877 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
26878 x exists before using it.
26879 (aarch64_add_constant_internal): Rename to...
26880 (aarch64_add_offset_1): ...this. Replace regnum with separate
26881 src and dest rtxes. Handle the case in which they're different,
26882 including when the offset is zero. Replace scratchreg with an rtx.
26883 Use 2 additions if there is no spare register into which we can
26884 move a 16-bit constant.
26885 (aarch64_add_constant): Delete.
26886 (aarch64_add_offset): Replace reg with separate src and dest
26887 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
26888 Use aarch64_add_offset_1.
26889 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
26890 an rtx rather than an int. Take the delta as a poly_int64
26891 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
26892 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
26893 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
26894 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
26895 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
26896 and aarch64_add_sp.
26897 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
26898 aarch64_add_constant.
26900 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
26902 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
26903 Use scalar_float_mode.
26905 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
26907 * config/aarch64/aarch64-simd.md
26908 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
26909 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
26910 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
26911 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
26912 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
26913 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
26914 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
26915 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
26916 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
26917 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
26919 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26922 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
26923 targ_options->x_arm_arch_string is non NULL.
26925 2018-01-11 Tamar Christina <tamar.christina@arm.com>
26927 * config/aarch64/aarch64.h
26928 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
26930 2018-01-11 Sudakshina Das <sudi.das@arm.com>
26933 * expmed.c (emit_store_flag_force): Swap if const op0
26934 and change VOIDmode to mode of op0.
26936 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
26938 PR rtl-optimization/83761
26939 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
26940 than bytes to mode_for_size.
26942 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
26944 PR middle-end/83189
26945 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
26946 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
26949 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
26951 PR middle-end/83575
26952 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
26953 when in layout mode.
26954 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
26955 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
26958 2018-01-10 Michael Collison <michael.collison@arm.com>
26960 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
26961 * config/aarch64/aarch64-option-extension.def: Add
26962 AARCH64_OPT_EXTENSION of 'fp16fml'.
26963 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
26964 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
26965 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
26966 * config/aarch64/constraints.md (Ui7): New constraint.
26967 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
26968 (VFMLA_SEL_W): Ditto.
26971 (VFMLA16_LOW): New int iterator.
26972 (VFMLA16_HIGH): Ditto.
26973 (UNSPEC_FMLAL): New unspec.
26974 (UNSPEC_FMLSL): Ditto.
26975 (UNSPEC_FMLAL2): Ditto.
26976 (UNSPEC_FMLSL2): Ditto.
26977 (f16mac): New code attribute.
26978 * config/aarch64/aarch64-simd-builtins.def
26979 (aarch64_fmlal_lowv2sf): Ditto.
26980 (aarch64_fmlsl_lowv2sf): Ditto.
26981 (aarch64_fmlalq_lowv4sf): Ditto.
26982 (aarch64_fmlslq_lowv4sf): Ditto.
26983 (aarch64_fmlal_highv2sf): Ditto.
26984 (aarch64_fmlsl_highv2sf): Ditto.
26985 (aarch64_fmlalq_highv4sf): Ditto.
26986 (aarch64_fmlslq_highv4sf): Ditto.
26987 (aarch64_fmlal_lane_lowv2sf): Ditto.
26988 (aarch64_fmlsl_lane_lowv2sf): Ditto.
26989 (aarch64_fmlal_laneq_lowv2sf): Ditto.
26990 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
26991 (aarch64_fmlalq_lane_lowv4sf): Ditto.
26992 (aarch64_fmlsl_lane_lowv4sf): Ditto.
26993 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
26994 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
26995 (aarch64_fmlal_lane_highv2sf): Ditto.
26996 (aarch64_fmlsl_lane_highv2sf): Ditto.
26997 (aarch64_fmlal_laneq_highv2sf): Ditto.
26998 (aarch64_fmlsl_laneq_highv2sf): Ditto.
26999 (aarch64_fmlalq_lane_highv4sf): Ditto.
27000 (aarch64_fmlsl_lane_highv4sf): Ditto.
27001 (aarch64_fmlalq_laneq_highv4sf): Ditto.
27002 (aarch64_fmlsl_laneq_highv4sf): Ditto.
27003 * config/aarch64/aarch64-simd.md:
27004 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
27005 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
27006 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
27007 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
27008 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
27009 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
27010 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
27011 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
27012 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
27013 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
27014 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
27015 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
27016 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
27017 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
27018 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
27019 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
27020 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
27021 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
27022 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
27023 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
27024 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
27025 (vfmlsl_low_u32): Ditto.
27026 (vfmlalq_low_u32): Ditto.
27027 (vfmlslq_low_u32): Ditto.
27028 (vfmlal_high_u32): Ditto.
27029 (vfmlsl_high_u32): Ditto.
27030 (vfmlalq_high_u32): Ditto.
27031 (vfmlslq_high_u32): Ditto.
27032 (vfmlal_lane_low_u32): Ditto.
27033 (vfmlsl_lane_low_u32): Ditto.
27034 (vfmlal_laneq_low_u32): Ditto.
27035 (vfmlsl_laneq_low_u32): Ditto.
27036 (vfmlalq_lane_low_u32): Ditto.
27037 (vfmlslq_lane_low_u32): Ditto.
27038 (vfmlalq_laneq_low_u32): Ditto.
27039 (vfmlslq_laneq_low_u32): Ditto.
27040 (vfmlal_lane_high_u32): Ditto.
27041 (vfmlsl_lane_high_u32): Ditto.
27042 (vfmlal_laneq_high_u32): Ditto.
27043 (vfmlsl_laneq_high_u32): Ditto.
27044 (vfmlalq_lane_high_u32): Ditto.
27045 (vfmlslq_lane_high_u32): Ditto.
27046 (vfmlalq_laneq_high_u32): Ditto.
27047 (vfmlslq_laneq_high_u32): Ditto.
27048 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
27049 (AARCH64_FL_FOR_ARCH8_4): New.
27050 (AARCH64_ISA_F16FML): New ISA flag.
27051 (TARGET_F16FML): New feature flag for fp16fml.
27052 (doc/invoke.texi): Document new fp16fml option.
27054 2018-01-10 Michael Collison <michael.collison@arm.com>
27056 * config/aarch64/aarch64-builtins.c:
27057 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
27058 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27059 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
27060 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
27061 (AARCH64_ISA_SHA3): New ISA flag.
27062 (TARGET_SHA3): New feature flag for sha3.
27063 * config/aarch64/iterators.md (sha512_op): New int attribute.
27064 (CRYPTO_SHA512): New int iterator.
27065 (UNSPEC_SHA512H): New unspec.
27066 (UNSPEC_SHA512H2): Ditto.
27067 (UNSPEC_SHA512SU0): Ditto.
27068 (UNSPEC_SHA512SU1): Ditto.
27069 * config/aarch64/aarch64-simd-builtins.def
27070 (aarch64_crypto_sha512hqv2di): New builtin.
27071 (aarch64_crypto_sha512h2qv2di): Ditto.
27072 (aarch64_crypto_sha512su0qv2di): Ditto.
27073 (aarch64_crypto_sha512su1qv2di): Ditto.
27074 (aarch64_eor3qv8hi): Ditto.
27075 (aarch64_rax1qv2di): Ditto.
27076 (aarch64_xarqv2di): Ditto.
27077 (aarch64_bcaxqv8hi): Ditto.
27078 * config/aarch64/aarch64-simd.md:
27079 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
27080 (aarch64_crypto_sha512su0qv2di): Ditto.
27081 (aarch64_crypto_sha512su1qv2di): Ditto.
27082 (aarch64_eor3qv8hi): Ditto.
27083 (aarch64_rax1qv2di): Ditto.
27084 (aarch64_xarqv2di): Ditto.
27085 (aarch64_bcaxqv8hi): Ditto.
27086 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
27087 (vsha512h2q_u64): Ditto.
27088 (vsha512su0q_u64): Ditto.
27089 (vsha512su1q_u64): Ditto.
27090 (veor3q_u16): Ditto.
27091 (vrax1q_u64): Ditto.
27092 (vxarq_u64): Ditto.
27093 (vbcaxq_u16): Ditto.
27094 * config/arm/types.md (crypto_sha512): New type attribute.
27095 (crypto_sha3): Ditto.
27096 (doc/invoke.texi): Document new sha3 option.
27098 2018-01-10 Michael Collison <michael.collison@arm.com>
27100 * config/aarch64/aarch64-builtins.c:
27101 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
27102 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27103 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
27104 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
27105 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
27106 (AARCH64_ISA_SM4): New ISA flag.
27107 (TARGET_SM4): New feature flag for sm4.
27108 * config/aarch64/aarch64-simd-builtins.def
27109 (aarch64_sm3ss1qv4si): Ditto.
27110 (aarch64_sm3tt1aq4si): Ditto.
27111 (aarch64_sm3tt1bq4si): Ditto.
27112 (aarch64_sm3tt2aq4si): Ditto.
27113 (aarch64_sm3tt2bq4si): Ditto.
27114 (aarch64_sm3partw1qv4si): Ditto.
27115 (aarch64_sm3partw2qv4si): Ditto.
27116 (aarch64_sm4eqv4si): Ditto.
27117 (aarch64_sm4ekeyqv4si): Ditto.
27118 * config/aarch64/aarch64-simd.md:
27119 (aarch64_sm3ss1qv4si): Ditto.
27120 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
27121 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
27122 (aarch64_sm4eqv4si): Ditto.
27123 (aarch64_sm4ekeyqv4si): Ditto.
27124 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
27125 (sm3part_op): Ditto.
27126 (CRYPTO_SM3TT): Ditto.
27127 (CRYPTO_SM3PART): Ditto.
27128 (UNSPEC_SM3SS1): New unspec.
27129 (UNSPEC_SM3TT1A): Ditto.
27130 (UNSPEC_SM3TT1B): Ditto.
27131 (UNSPEC_SM3TT2A): Ditto.
27132 (UNSPEC_SM3TT2B): Ditto.
27133 (UNSPEC_SM3PARTW1): Ditto.
27134 (UNSPEC_SM3PARTW2): Ditto.
27135 (UNSPEC_SM4E): Ditto.
27136 (UNSPEC_SM4EKEY): Ditto.
27137 * config/aarch64/constraints.md (Ui2): New constraint.
27138 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
27139 * config/arm/types.md (crypto_sm3): New type attribute.
27140 (crypto_sm4): Ditto.
27141 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
27142 (vsm3tt1aq_u32): Ditto.
27143 (vsm3tt1bq_u32): Ditto.
27144 (vsm3tt2aq_u32): Ditto.
27145 (vsm3tt2bq_u32): Ditto.
27146 (vsm3partw1q_u32): Ditto.
27147 (vsm3partw2q_u32): Ditto.
27148 (vsm4eq_u32): Ditto.
27149 (vsm4ekeyq_u32): Ditto.
27150 (doc/invoke.texi): Document new sm4 option.
27152 2018-01-10 Michael Collison <michael.collison@arm.com>
27154 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
27155 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
27156 (AARCH64_FL_FOR_ARCH8_4): New.
27157 (AARCH64_FL_V8_4): New flag.
27158 (doc/invoke.texi): Document new armv8.4-a option.
27160 2018-01-10 Michael Collison <michael.collison@arm.com>
27162 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27163 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
27164 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
27165 * config/aarch64/aarch64-option-extension.def: Add
27166 AARCH64_OPT_EXTENSION of 'sha2'.
27167 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
27168 (crypto): Disable sha2 and aes if crypto disabled.
27169 (crypto): Enable aes and sha2 if enabled.
27170 (simd): Disable sha2 and aes if simd disabled.
27171 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
27173 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
27174 (TARGET_SHA2): New feature flag for sha2.
27175 (TARGET_AES): New feature flag for aes.
27176 * config/aarch64/aarch64-simd.md:
27177 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
27178 conditional on TARGET_AES.
27179 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
27180 (aarch64_crypto_sha1hsi): Make pattern conditional
27182 (aarch64_crypto_sha1hv4si): Ditto.
27183 (aarch64_be_crypto_sha1hv4si): Ditto.
27184 (aarch64_crypto_sha1su1v4si): Ditto.
27185 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
27186 (aarch64_crypto_sha1su0v4si): Ditto.
27187 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
27188 (aarch64_crypto_sha256su0v4si): Ditto.
27189 (aarch64_crypto_sha256su1v4si): Ditto.
27190 (doc/invoke.texi): Document new aes and sha2 options.
27192 2018-01-10 Martin Sebor <msebor@redhat.com>
27194 PR tree-optimization/83781
27195 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
27198 2018-01-11 Martin Sebor <msebor@gmail.com>
27199 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27201 PR tree-optimization/83501
27202 PR tree-optimization/81703
27204 * tree-ssa-strlen.c (get_string_cst): Rename...
27205 (get_string_len): ...to this. Handle global constants.
27206 (handle_char_store): Adjust.
27208 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
27209 Jim Wilson <jimw@sifive.com>
27211 * config/riscv/riscv-protos.h (riscv_output_return): New.
27212 * config/riscv/riscv.c (struct machine_function): New naked_p field.
27213 (riscv_attribute_table, riscv_output_return),
27214 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
27215 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
27216 (riscv_compute_frame_info): Only compute frame->mask if not a naked
27218 (riscv_expand_prologue): Add early return for naked function.
27219 (riscv_expand_epilogue): Likewise.
27220 (riscv_function_ok_for_sibcall): Return false for naked function.
27221 (riscv_set_current_function): New.
27222 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
27223 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
27224 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
27225 * doc/extend.texi (RISC-V Function Attributes): New.
27227 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
27229 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
27230 check for 128-bit long double before checking TCmode.
27231 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
27232 128-bit long doubles before checking TFmode or TCmode.
27233 (FLOAT128_IBM_P): Likewise.
27235 2018-01-10 Martin Sebor <msebor@redhat.com>
27237 PR tree-optimization/83671
27238 * builtins.c (c_strlen): Unconditionally return zero for the empty
27240 Use -Warray-bounds for warnings.
27241 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
27242 for non-constant array indices with COMPONENT_REF, arrays of
27243 arrays, and pointers to arrays.
27244 (gimple_fold_builtin_strlen): Determine and set length range for
27245 non-constant character arrays.
27247 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
27249 PR middle-end/81897
27250 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
27253 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
27255 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
27257 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
27260 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
27261 VECTOR_MEM_ALTIVEC_OR_VSX_P.
27262 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
27263 indexed_or_indirect_operand predicate.
27264 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
27265 (*vsx_le_perm_load_v8hi): Likewise.
27266 (*vsx_le_perm_load_v16qi): Likewise.
27267 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
27268 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
27269 (*vsx_le_perm_store_v8hi): Likewise.
27270 (*vsx_le_perm_store_v16qi): Likewise.
27271 (eight unnamed splitters): Likewise.
27273 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
27275 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
27276 * config/rs6000/emmintrin.h: Likewise.
27277 * config/rs6000/mmintrin.h: Likewise.
27278 * config/rs6000/xmmintrin.h: Likewise.
27280 2018-01-10 David Malcolm <dmalcolm@redhat.com>
27283 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
27285 * tree.c (tree_nop_conversion): Return true for location wrapper
27287 (maybe_wrap_with_location): New function.
27288 (selftest::check_strip_nops): New function.
27289 (selftest::test_location_wrappers): New function.
27290 (selftest::tree_c_tests): Call it.
27291 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
27292 (maybe_wrap_with_location): New decl.
27293 (EXPR_LOCATION_WRAPPER_P): New macro.
27294 (location_wrapper_p): New inline function.
27295 (tree_strip_any_location_wrapper): New inline function.
27297 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
27300 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
27301 stack_realign_offset for the largest alignment of stack slot
27303 (ix86_find_max_used_stack_alignment): New function.
27304 (ix86_finalize_stack_frame_flags): Use it. Set
27305 max_used_stack_alignment if we don't realign stack.
27306 * config/i386/i386.h (machine_function): Add
27307 max_used_stack_alignment.
27309 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
27311 * config/arm/arm.opt (-mbranch-cost): New option.
27312 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
27315 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
27318 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
27319 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
27321 2018-01-10 Richard Biener <rguenther@suse.de>
27324 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
27325 early out so it also covers the case where we have a non-NULL
27328 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
27330 PR tree-optimization/83753
27331 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
27332 for non-strided grouped accesses if the number of elements is 1.
27334 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
27337 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
27338 * i386.h (TARGET_USE_GATHER): Define.
27339 * x86-tune.def (X86_TUNE_USE_GATHER): New.
27341 2018-01-10 Martin Liska <mliska@suse.cz>
27344 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
27345 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
27347 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
27348 CLEANUP_NO_PARTITIONING is not set.
27350 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
27352 * doc/rtl.texi: Remove documentation of (const ...) wrappers
27353 for vectors, as a partial revert of r254296.
27354 * rtl.h (const_vec_p): Delete.
27355 (const_vec_duplicate_p): Don't test for vector CONSTs.
27356 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
27357 * expmed.c (make_tree): Likewise.
27360 * common.md (E, F): Use CONSTANT_P instead of checking for
27362 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
27363 checking for CONST_VECTOR.
27365 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
27367 PR middle-end/83575
27368 * predict.c (force_edge_cold): Handle in more sane way edges
27369 with no prediction.
27371 2018-01-09 Carl Love <cel@us.ibm.com>
27373 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
27375 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
27376 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
27377 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
27378 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
27379 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
27380 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
27381 * config/rs6000/rs6000-protos.h: Add extern defition for
27382 rs6000_generate_float2_double_code.
27383 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
27385 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
27386 (float2_v2df): Add define_expand.
27388 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
27391 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
27392 op_mode in the force_to_mode call.
27394 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
27396 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
27397 instead of checking each element individually.
27398 (aarch64_evpc_uzp): Likewise.
27399 (aarch64_evpc_zip): Likewise.
27400 (aarch64_evpc_ext): Likewise.
27401 (aarch64_evpc_rev): Likewise.
27402 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
27403 instead of checking each element individually. Return true without
27405 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
27406 whether all selected elements come from the same input, instead of
27407 checking each element individually. Remove calls to gen_rtx_REG,
27408 start_sequence and end_sequence and instead assert that no rtl is
27411 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
27413 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
27414 order of HIGH and CONST checks.
27416 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
27418 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
27419 if the destination isn't an SSA_NAME.
27421 2018-01-09 Richard Biener <rguenther@suse.de>
27423 PR tree-optimization/83668
27424 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
27426 (canonicalize_loop_form): ... here, renamed from ...
27427 (canonicalize_loop_closed_ssa_form): ... this and amended to
27428 swap successor edges for loop exit blocks to make us use
27429 the RPO order we need for initial schedule generation.
27431 2018-01-09 Joseph Myers <joseph@codesourcery.com>
27433 PR tree-optimization/64811
27434 * match.pd: When optimizing comparisons with Inf, avoid
27435 introducing or losing exceptions from comparisons with NaN.
27437 2018-01-09 Martin Liska <mliska@suse.cz>
27440 * asan.c (shadow_mem_size): Add gcc_assert.
27442 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
27444 Don't save registers in main().
27447 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
27448 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
27449 * config/avr/avr.c (avr_set_current_function): Don't error if
27450 naked, OS_task or OS_main are specified at the same time.
27451 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
27453 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
27455 * common/config/avr/avr-common.c (avr_option_optimization_table):
27456 Switch on -mmain-is-OS_task for optimizing compilations.
27458 2018-01-09 Richard Biener <rguenther@suse.de>
27460 PR tree-optimization/83572
27461 * graphite.c: Include cfganal.h.
27462 (graphite_transform_loops): Connect infinite loops to exit
27463 and remove fake edges at the end.
27465 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
27467 * ipa-inline.c (edge_badness): Revert accidental checkin.
27469 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
27472 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
27473 symbols; not inline clones.
27475 2018-01-09 Jakub Jelinek <jakub@redhat.com>
27478 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
27479 hard registers. Formatting fixes.
27481 PR preprocessor/83722
27482 * gcc.c (try_generate_repro): Pass
27483 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
27484 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
27487 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
27488 Kito Cheng <kito.cheng@gmail.com>
27490 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
27491 (riscv_leaf_function_p): Delete.
27492 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
27494 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27496 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
27498 (do_ifelse): New function.
27499 (do_isel): New function.
27500 (do_sub3): New function.
27501 (do_add3): New function.
27502 (do_load_mask_compare): New function.
27503 (do_overlap_load_compare): New function.
27504 (expand_compare_loop): New function.
27505 (expand_block_compare): Call expand_compare_loop() when appropriate.
27506 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
27507 option description.
27508 (-mblock-compare-inline-loop-limit): New option.
27510 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27513 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
27514 Reverse order of second and third operands in first alternative.
27515 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
27516 of first and second elements in UNSPEC_VPERMR vector.
27517 (altivec_expand_vec_perm_le): Likewise.
27519 2018-01-08 Jeff Law <law@redhat.com>
27521 PR rtl-optimizatin/81308
27522 * tree-switch-conversion.c (cfg_altered): New file scoped static.
27523 (process_switch): If group_case_labels makes a change, then set
27525 (pass_convert_switch::execute): If a switch is converted, then
27526 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
27528 PR rtl-optimization/81308
27529 * recog.c (split_all_insns): Conditionally cleanup the CFG after
27532 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
27534 PR target/83663 - Revert r255946
27535 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
27536 generation for cases where splatting a value is not useful.
27537 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
27538 across a vec_duplicate and a paradoxical subreg forming a vector
27539 mode to a vec_concat.
27541 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27543 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
27544 -march=armv8.3-a variants.
27545 * config/arm/t-multilib: Likewise.
27546 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
27548 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27550 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
27552 (cceq_ior_compare_complement): Give it a name so I can use it, and
27553 change boolean_or_operator predicate to boolean_operator so it can
27554 be used to generate a crand.
27555 (eqne): New code iterator.
27556 (bd/bd_neg): New code_attrs.
27557 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
27558 a single define_insn.
27559 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
27560 decrement (bdnzt/bdnzf/bdzt/bdzf).
27561 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
27562 with the new names of the branch decrement patterns, and added the
27563 names of the branch decrement conditional patterns.
27565 2018-01-08 Richard Biener <rguenther@suse.de>
27567 PR tree-optimization/83563
27568 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
27571 2018-01-08 Richard Biener <rguenther@suse.de>
27573 PR middle-end/83713
27574 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
27576 2018-01-08 Richard Biener <rguenther@suse.de>
27578 PR tree-optimization/83685
27579 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
27580 references to abnormals.
27582 2018-01-08 Richard Biener <rguenther@suse.de>
27585 * dwarf2out.c (output_indirect_strings): Handle empty
27586 skeleton_debug_str_hash.
27587 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
27589 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
27591 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
27592 (emit_store_direct): Likewise.
27593 (arc_trampoline_adjust_address): Likewise.
27594 (arc_asm_trampoline_template): New function.
27595 (arc_initialize_trampoline): Use asm_trampoline_template.
27596 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
27597 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
27598 * config/arc/arc.md (flush_icache): Delete pattern.
27600 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
27602 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
27603 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
27606 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
27609 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
27610 by not USED_FOR_TARGET.
27611 (make_pass_resolve_sw_modes): Likewise.
27613 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
27615 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
27618 2018-01-08 Richard Biener <rguenther@suse.de>
27620 PR middle-end/83580
27621 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
27623 2018-01-08 Richard Biener <rguenther@suse.de>
27625 PR middle-end/83517
27626 * match.pd ((t * 2) / 2) -> t): Add missing :c.
27628 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
27630 PR middle-end/81897
27631 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
27632 basic blocks with a small number of successors.
27633 (convert_control_dep_chain_into_preds): Improve handling of
27635 (dump_predicates): Split apart into...
27636 (dump_pred_chain): ...here...
27637 (dump_pred_info): ...and here.
27638 (can_one_predicate_be_invalidated_p): Add debugging printfs.
27639 (can_chain_union_be_invalidated_p): Improve check for invalidation
27641 (uninit_uses_cannot_happen): Avoid unnecessary if
27642 convert_control_dep_chain_into_preds yielded nothing.
27644 2018-01-06 Martin Sebor <msebor@redhat.com>
27646 PR tree-optimization/83640
27647 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
27648 subtracting negative offset from size.
27649 (builtin_access::overlap): Adjust offset bounds of the access to fall
27650 within the size of the object if possible.
27652 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
27654 PR rtl-optimization/83699
27655 * expmed.c (extract_bit_field_1): Restrict the vector usage of
27656 extract_bit_field_as_subreg to cases in which the extracted
27657 value is also a vector.
27659 * lra-constraints.c (process_alt_operands): Test for the equivalence
27660 substitutions when detecting a possible reload cycle.
27662 2018-01-06 Jakub Jelinek <jakub@redhat.com>
27665 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
27666 by default if flag_selective_schedling{,2}. Formatting fixes.
27668 PR rtl-optimization/83682
27669 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
27670 if it has non-VECTOR_MODE element mode.
27671 (vec_duplicate_p): Likewise.
27673 PR middle-end/83694
27674 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
27675 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
27677 2018-01-05 Jakub Jelinek <jakub@redhat.com>
27680 * config/i386/i386-builtin.def
27681 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
27682 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
27683 Require also OPTION_MASK_ISA_AVX512F in addition to
27684 OPTION_MASK_ISA_GFNI.
27685 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
27686 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
27687 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
27688 to OPTION_MASK_ISA_GFNI.
27689 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
27690 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
27691 OPTION_MASK_ISA_AVX512BW.
27692 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
27693 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
27694 addition to OPTION_MASK_ISA_GFNI.
27695 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
27696 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
27697 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
27698 to OPTION_MASK_ISA_GFNI.
27699 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
27700 a requirement for all ISAs rather than any of them with a few
27702 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
27704 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
27705 bitmasks to be enabled with 3 exceptions, instead of requiring any
27706 enabled ISA with lots of exceptions.
27707 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
27708 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
27709 Change avx512bw in isa attribute to avx512f.
27710 * config/i386/sgxintrin.h: Add license boilerplate.
27711 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
27712 to __AVX512F__ and __AVX512VL to __AVX512VL__.
27713 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
27714 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
27716 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
27717 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
27718 temporarily sse2 rather than sse if not enabled already.
27721 * config/i386/sse.md (VI248_VLBW): Rename to ...
27722 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
27723 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
27724 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
27725 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
27726 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
27727 mode iterator instead of VI248_VLBW.
27729 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
27731 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
27732 (record_modified): Skip clobbers; add debug output.
27733 (param_change_prob): Use sreal frequencies.
27735 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
27737 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
27738 punt for user-aligned variables.
27740 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
27742 * tree-chrec.c (chrec_contains_symbols): Return true for
27745 2018-01-05 Sudakshina Das <sudi.das@arm.com>
27748 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
27749 of (x|y) == x for BICS pattern.
27751 2018-01-05 Jakub Jelinek <jakub@redhat.com>
27753 PR tree-optimization/83605
27754 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
27755 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
27758 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
27760 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
27761 * config/epiphany/rtems.h: New file.
27763 2018-01-04 Jakub Jelinek <jakub@redhat.com>
27764 Uros Bizjak <ubizjak@gmail.com>
27767 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
27768 QIreg_operand instead of register_operand predicate.
27769 * config/i386/i386.c (ix86_rop_should_change_byte_p,
27770 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
27771 comments instead of -fmitigate[-_]rop.
27773 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27776 * cgraphunit.c (symbol_table::compile): Switch to text_section
27777 before calling assembly_start debug hook.
27778 * run-rtl-passes.c (run_rtl_passes): Likewise.
27781 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27783 * tree-vrp.c (extract_range_from_binary_expr_1): Check
27784 range_int_cst_p rather than !symbolic_range_p before calling
27785 extract_range_from_multiplicative_op_1.
27787 2018-01-04 Jeff Law <law@redhat.com>
27789 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
27790 redundant test in assertion.
27792 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27794 * doc/rtl.texi: Document machine_mode wrapper classes.
27796 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27798 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
27799 using tree_to_uhwi.
27801 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27803 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
27804 the VEC_PERM_EXPR fold to fail.
27806 2018-01-04 Jakub Jelinek <jakub@redhat.com>
27809 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
27810 to switched_sections.
27812 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27815 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
27816 test for d.testing.
27818 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
27821 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
27822 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
27824 2018-01-04 Jakub Jelinek <jakub@redhat.com>
27827 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
27828 is BLKmode and bitpos not zero or mode change is needed.
27830 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27833 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
27836 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
27839 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
27840 instead of MULT rtx. Update all corresponding splitters.
27841 (*saddl_se): Ditto.
27842 (*ssub<modesuffix>): Ditto.
27843 (*ssubl_se): Ditto.
27844 (*cmp_sadd_di): Update split patterns.
27845 (*cmp_sadd_si): Ditto.
27846 (*cmp_sadd_sidi): Ditto.
27847 (*cmp_ssub_di): Ditto.
27848 (*cmp_ssub_si): Ditto.
27849 (*cmp_ssub_sidi): Ditto.
27850 * config/alpha/predicates.md (const23_operand): New predicate.
27851 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
27852 Look for ASHIFT, not MULT inner operand.
27853 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
27855 2018-01-04 Martin Liska <mliska@suse.cz>
27857 PR gcov-profile/83669
27858 * gcov.c (output_intermediate_file): Add version to intermediate
27860 * doc/gcov.texi: Document new field 'version' in intermediate
27861 file format. Fix location of '-k' option of gcov command.
27863 2018-01-04 Martin Liska <mliska@suse.cz>
27866 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
27868 2018-01-04 Jakub Jelinek <jakub@redhat.com>
27870 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
27872 2018-01-03 Martin Sebor <msebor@redhat.com>
27874 PR tree-optimization/83655
27875 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
27876 checking calls with invalid arguments.
27878 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27880 * tree-vect-stmts.c (vect_get_store_rhs): New function.
27881 (vectorizable_mask_load_store): Delete.
27882 (vectorizable_call): Return false for masked loads and stores.
27883 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
27884 instead of gimple_assign_rhs1.
27885 (vectorizable_load): Handle IFN_MASK_LOAD.
27886 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
27888 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27890 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
27892 (vectorizable_mask_load_store): ...here.
27893 (vectorizable_load): ...and here.
27895 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27897 * tree-vect-stmts.c (vect_build_all_ones_mask)
27898 (vect_build_zero_merge_argument): New functions, split out from...
27899 (vectorizable_load): ...here.
27901 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27903 * tree-vect-stmts.c (vect_check_store_rhs): New function,
27905 (vectorizable_mask_load_store): ...here.
27906 (vectorizable_store): ...and here.
27908 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27910 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
27912 (vectorizable_mask_load_store): ...here.
27914 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27916 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
27917 (vect_model_store_cost): Take a vec_load_store_type instead of a
27919 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
27920 (vect_model_store_cost): Take a vec_load_store_type instead of a
27922 (vectorizable_mask_load_store): Update accordingly.
27923 (vectorizable_store): Likewise.
27924 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
27926 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27928 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
27929 IFN_MASK_LOAD calls here rather than...
27930 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
27932 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27933 Alan Hayward <alan.hayward@arm.com>
27934 David Sherwood <david.sherwood@arm.com>
27936 * expmed.c (extract_bit_field_1): For vector extracts,
27937 fall back to extract_bit_field_as_subreg if vec_extract
27940 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27941 Alan Hayward <alan.hayward@arm.com>
27942 David Sherwood <david.sherwood@arm.com>
27944 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
27945 they are variable or constant sized.
27946 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
27947 slots for constant-sized data.
27949 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27950 Alan Hayward <alan.hayward@arm.com>
27951 David Sherwood <david.sherwood@arm.com>
27953 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
27954 handling COND_EXPRs with boolean comparisons, try to find a better
27955 basis for the mask type than the boolean itself.
27957 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27959 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
27960 is calculated and how it can be overridden.
27961 * genmodes.c (max_bitsize_mode_any_mode): New variable.
27962 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
27964 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
27967 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27968 Alan Hayward <alan.hayward@arm.com>
27969 David Sherwood <david.sherwood@arm.com>
27971 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
27972 Remove the mode argument.
27973 (aarch64_simd_valid_immediate): Remove the mode and inverse
27975 * config/aarch64/iterators.md (bitsize): New iterator.
27976 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
27977 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
27978 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
27979 aarch64_simd_valid_immediate.
27980 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
27981 (aarch64_reg_or_bic_imm): Likewise.
27982 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
27983 with an insn_type enum and msl with a modifier_type enum.
27984 Replace element_width with a scalar_mode. Change the shift
27985 to unsigned int. Add constructors for scalar_float_mode and
27986 scalar_int_mode elements.
27987 (aarch64_vect_float_const_representable_p): Delete.
27988 (aarch64_can_const_movi_rtx_p)
27989 (aarch64_simd_scalar_immediate_valid_for_move)
27990 (aarch64_simd_make_constant): Update call to
27991 aarch64_simd_valid_immediate.
27992 (aarch64_advsimd_valid_immediate_hs): New function.
27993 (aarch64_advsimd_valid_immediate): Likewise.
27994 (aarch64_simd_valid_immediate): Remove mode and inverse
27995 arguments. Rewrite to use the above. Use const_vec_duplicate_p
27996 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
27997 and aarch64_float_const_representable_p on the result.
27998 (aarch64_output_simd_mov_immediate): Remove mode argument.
27999 Update call to aarch64_simd_valid_immediate and use of
28000 simd_immediate_info.
28001 (aarch64_output_scalar_simd_mov_immediate): Update call
28004 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28005 Alan Hayward <alan.hayward@arm.com>
28006 David Sherwood <david.sherwood@arm.com>
28008 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
28009 (mode_nunits): Likewise CONST_MODE_NUNITS.
28010 * machmode.def (ADJUST_NUNITS): Document.
28011 * genmodes.c (mode_data::need_nunits_adj): New field.
28012 (blank_mode): Update accordingly.
28013 (adj_nunits): New variable.
28014 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
28016 (emit_mode_size_inline): Set need_bytesize_adj for all modes
28017 listed in adj_nunits.
28018 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
28019 listed in adj_nunits. Don't emit case statements for such modes.
28020 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
28021 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
28022 nothing if adj_nunits is nonnull.
28023 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
28024 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
28025 (emit_mode_fbit): Update use of print_maybe_const_decl.
28026 (emit_move_size): Likewise. Treat the array as non-const
28028 (emit_mode_adjustments): Handle adj_nunits.
28030 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28032 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
28033 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
28034 (VECTOR_MODES): Use it.
28035 (make_vector_modes): Take the prefix as an argument.
28037 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28038 Alan Hayward <alan.hayward@arm.com>
28039 David Sherwood <david.sherwood@arm.com>
28041 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
28042 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
28043 for MODE_VECTOR_BOOL.
28044 * machmode.def (VECTOR_BOOL_MODE): Document.
28045 * genmodes.c (VECTOR_BOOL_MODE): New macro.
28046 (make_vector_bool_mode): New function.
28047 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
28049 * lto-streamer-in.c (lto_input_mode_table): Likewise.
28050 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
28052 * stor-layout.c (int_mode_for_mode): Likewise.
28053 * tree.c (build_vector_type_for_mode): Likewise.
28054 * varasm.c (output_constant_pool_2): Likewise.
28055 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
28056 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
28057 for MODE_VECTOR_BOOL.
28058 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
28059 of mode class checks.
28060 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
28061 instead of a list of mode class checks.
28062 (expand_vector_scalar_condition): Likewise.
28063 (type_for_widest_vector_mode): Handle BImode as an inner mode.
28065 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28066 Alan Hayward <alan.hayward@arm.com>
28067 David Sherwood <david.sherwood@arm.com>
28069 * machmode.h (mode_size): Change from unsigned short to
28071 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
28072 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
28073 or if measurement_type is not polynomial.
28074 (fixed_size_mode::includes_p): Check for constant-sized modes.
28075 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
28076 return a poly_uint16 rather than an unsigned short.
28077 (emit_mode_size): Change the type of mode_size from unsigned short
28078 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
28079 (emit_mode_adjustments): Cope with polynomial vector sizes.
28080 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
28082 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
28084 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
28085 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
28086 * caller-save.c (setup_save_areas): Likewise.
28087 (replace_reg_with_saved_mem): Likewise.
28088 * calls.c (emit_library_call_value_1): Likewise.
28089 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
28090 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
28091 (gen_lowpart_for_combine): Likewise.
28092 * convert.c (convert_to_integer_1): Likewise.
28093 * cse.c (equiv_constant, cse_insn): Likewise.
28094 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
28095 (cselib_subst_to_values): Likewise.
28096 * dce.c (word_dce_process_block): Likewise.
28097 * df-problems.c (df_word_lr_mark_ref): Likewise.
28098 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
28099 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
28100 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
28101 (rtl_for_decl_location): Likewise.
28102 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
28103 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
28104 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
28105 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
28106 (expand_expr_real_1): Likewise.
28107 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
28108 (pad_below): Likewise.
28109 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
28110 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
28111 * ira.c (get_subreg_tracking_sizes): Likewise.
28112 * ira-build.c (ira_create_allocno_objects): Likewise.
28113 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
28114 (ira_sort_regnos_for_alter_reg): Likewise.
28115 * ira-costs.c (record_operand_costs): Likewise.
28116 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
28117 (resolve_simple_move): Likewise.
28118 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
28119 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
28120 (lra_constraints): Likewise.
28121 (CONST_POOL_OK_P): Reject variable-sized modes.
28122 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
28123 (add_pseudo_to_slot, lra_spill): Likewise.
28124 * omp-low.c (omp_clause_aligned_alignment): Likewise.
28125 * optabs-query.c (get_best_extraction_insn): Likewise.
28126 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
28127 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
28128 (expand_mult_highpart, valid_multiword_target_p): Likewise.
28129 * recog.c (offsettable_address_addr_space_p): Likewise.
28130 * regcprop.c (maybe_mode_change): Likewise.
28131 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
28132 * regrename.c (build_def_use): Likewise.
28133 * regstat.c (dump_reg_info): Likewise.
28134 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
28135 (find_reloads, find_reloads_subreg_address): Likewise.
28136 * reload1.c (eliminate_regs_1): Likewise.
28137 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
28138 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
28139 (simplify_binary_operation_1, simplify_subreg): Likewise.
28140 * targhooks.c (default_function_arg_padding): Likewise.
28141 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
28142 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
28143 (verify_gimple_assign_ternary): Likewise.
28144 * tree-inline.c (estimate_move_cost): Likewise.
28145 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
28146 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
28147 (get_address_cost_ainc): Likewise.
28148 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28149 (vect_supportable_dr_alignment): Likewise.
28150 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28151 (vectorizable_reduction): Likewise.
28152 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
28153 (vectorizable_operation, vectorizable_load): Likewise.
28154 * tree.c (build_same_sized_truth_vector_type): Likewise.
28155 * valtrack.c (cleanup_auto_inc_dec): Likewise.
28156 * var-tracking.c (emit_note_insn_var_location): Likewise.
28157 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
28158 (ADDR_VEC_ALIGN): Likewise.
28160 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28161 Alan Hayward <alan.hayward@arm.com>
28162 David Sherwood <david.sherwood@arm.com>
28164 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
28166 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
28167 or if measurement_type is polynomial.
28168 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
28169 * combine.c (make_extraction): Likewise.
28170 * dse.c (find_shift_sequence): Likewise.
28171 * dwarf2out.c (mem_loc_descriptor): Likewise.
28172 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
28173 (extract_bit_field, extract_low_bits): Likewise.
28174 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
28175 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
28176 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
28177 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
28178 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
28179 * reload.c (find_reloads): Likewise.
28180 * reload1.c (alter_reg): Likewise.
28181 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
28182 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
28183 * tree-if-conv.c (predicate_mem_writes): Likewise.
28184 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
28185 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
28186 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
28187 * valtrack.c (dead_debug_insert_temp): Likewise.
28188 * varasm.c (mergeable_constant_section): Likewise.
28189 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
28191 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28192 Alan Hayward <alan.hayward@arm.com>
28193 David Sherwood <david.sherwood@arm.com>
28195 * expr.c (expand_assignment): Cope with polynomial mode sizes
28196 when assigning to a CONCAT.
28198 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28199 Alan Hayward <alan.hayward@arm.com>
28200 David Sherwood <david.sherwood@arm.com>
28202 * machmode.h (mode_precision): Change from unsigned short to
28204 (mode_to_precision): Return a poly_uint16 rather than an unsigned
28206 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
28207 or if measurement_type is not polynomial.
28208 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
28209 in which the mode is already known to be a scalar_int_mode.
28210 * genmodes.c (emit_mode_precision): Change the type of mode_precision
28211 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
28213 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
28214 for GET_MODE_PRECISION.
28215 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
28216 for GET_MODE_PRECISION.
28217 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
28219 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
28220 (expand_field_assignment, make_extraction): Likewise.
28221 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
28222 (get_last_value): Likewise.
28223 * convert.c (convert_to_integer_1): Likewise.
28224 * cse.c (cse_insn): Likewise.
28225 * expr.c (expand_expr_real_1): Likewise.
28226 * lra-constraints.c (simplify_operand_subreg): Likewise.
28227 * optabs-query.c (can_atomic_load_p): Likewise.
28228 * optabs.c (expand_atomic_load): Likewise.
28229 (expand_atomic_store): Likewise.
28230 * ree.c (combine_reaching_defs): Likewise.
28231 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
28232 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
28233 * tree.h (type_has_mode_precision_p): Likewise.
28234 * ubsan.c (instrument_si_overflow): Likewise.
28236 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28237 Alan Hayward <alan.hayward@arm.com>
28238 David Sherwood <david.sherwood@arm.com>
28240 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
28241 polynomial numbers of units.
28242 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
28243 (valid_vector_subparts_p): New function.
28244 (build_vector_type): Remove temporary shim and take the number
28245 of units as a poly_uint64 rather than an int.
28246 (build_opaque_vector_type): Take the number of units as a
28247 poly_uint64 rather than an int.
28248 * tree.c (build_vector_from_ctor): Handle polynomial
28249 TYPE_VECTOR_SUBPARTS.
28250 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
28251 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
28252 (build_vector_from_val): If the number of units is variable,
28253 use build_vec_duplicate_cst for constant operands and
28254 VEC_DUPLICATE_EXPR otherwise.
28255 (make_vector_type): Remove temporary is_constant ().
28256 (build_vector_type, build_opaque_vector_type): Take the number of
28257 units as a poly_uint64 rather than an int.
28258 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
28260 * cfgexpand.c (expand_debug_expr): Likewise.
28261 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
28262 (store_constructor, expand_expr_real_1): Likewise.
28263 (const_scalar_mask_from_tree): Likewise.
28264 * fold-const-call.c (fold_const_reduction): Likewise.
28265 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
28266 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
28267 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
28268 (fold_relational_const): Likewise.
28269 (native_interpret_vector): Likewise. Change the size from an
28270 int to an unsigned int.
28271 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
28272 TYPE_VECTOR_SUBPARTS.
28273 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
28274 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
28275 duplicating a non-constant operand into a variable-length vector.
28276 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
28277 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
28278 * ipa-icf.c (sem_variable::equals): Likewise.
28279 * match.pd: Likewise.
28280 * omp-simd-clone.c (simd_clone_subparts): Likewise.
28281 * print-tree.c (print_node): Likewise.
28282 * stor-layout.c (layout_type): Likewise.
28283 * targhooks.c (default_builtin_vectorization_cost): Likewise.
28284 * tree-cfg.c (verify_gimple_comparison): Likewise.
28285 (verify_gimple_assign_binary): Likewise.
28286 (verify_gimple_assign_ternary): Likewise.
28287 (verify_gimple_assign_single): Likewise.
28288 * tree-pretty-print.c (dump_generic_node): Likewise.
28289 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
28290 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
28291 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
28292 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
28293 (vect_shift_permute_load_chain): Likewise.
28294 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
28295 (expand_vector_condition, optimize_vector_constructor): Likewise.
28296 (lower_vec_perm, get_compute_type): Likewise.
28297 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28298 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
28299 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
28300 (vect_recog_mask_conversion_pattern): Likewise.
28301 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
28302 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
28303 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
28304 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
28305 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
28306 (vectorizable_shift, vectorizable_operation, vectorizable_store)
28307 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
28308 (supportable_widening_operation): Likewise.
28309 (supportable_narrowing_operation): Likewise.
28310 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
28312 * varasm.c (output_constant): Likewise.
28314 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28315 Alan Hayward <alan.hayward@arm.com>
28316 David Sherwood <david.sherwood@arm.com>
28318 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
28319 so that both the length == 3 and length != 3 cases set up their
28320 own permute vectors. Add comments explaining why we know the
28321 number of elements is constant.
28322 (vect_permute_load_chain): Likewise.
28324 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28325 Alan Hayward <alan.hayward@arm.com>
28326 David Sherwood <david.sherwood@arm.com>
28328 * machmode.h (mode_nunits): Change from unsigned char to
28330 (ONLY_FIXED_SIZE_MODES): New macro.
28331 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
28332 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
28333 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
28335 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
28336 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
28337 or if measurement_type is not polynomial.
28338 * genmodes.c (ZERO_COEFFS): New macro.
28339 (emit_mode_nunits_inline): Make mode_nunits_inline return a
28341 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
28342 Use ZERO_COEFFS when emitting initializers.
28343 * data-streamer.h (bp_pack_poly_value): New function.
28344 (bp_unpack_poly_value): Likewise.
28345 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
28346 for GET_MODE_NUNITS.
28347 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
28348 for GET_MODE_NUNITS.
28349 * tree.c (make_vector_type): Remove temporary shim and make
28350 the real function take the number of units as a poly_uint64
28351 rather than an int.
28352 (build_vector_type_for_mode): Handle polynomial nunits.
28353 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
28354 * emit-rtl.c (const_vec_series_p_1): Likewise.
28355 (gen_rtx_CONST_VECTOR): Likewise.
28356 * fold-const.c (test_vec_duplicate_folding): Likewise.
28357 * genrecog.c (validate_pattern): Likewise.
28358 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
28359 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
28360 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
28361 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
28362 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
28363 * rtlanal.c (subreg_get_info): Likewise.
28364 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
28365 (vect_grouped_load_supported): Likewise.
28366 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
28367 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
28368 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
28369 (simplify_const_unary_operation, simplify_binary_operation_1)
28370 (simplify_const_binary_operation, simplify_ternary_operation)
28371 (test_vector_ops_duplicate, test_vector_ops): Likewise.
28372 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
28373 instead of CONST_VECTOR_NUNITS.
28374 * varasm.c (output_constant_pool_2): Likewise.
28375 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
28376 explicit-encoded elements in the XVEC for variable-length vectors.
28378 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28380 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
28382 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28383 Alan Hayward <alan.hayward@arm.com>
28384 David Sherwood <david.sherwood@arm.com>
28386 * coretypes.h (fixed_size_mode): Declare.
28387 (fixed_size_mode_pod): New typedef.
28388 * builtins.h (target_builtins::x_apply_args_mode)
28389 (target_builtins::x_apply_result_mode): Change type to
28390 fixed_size_mode_pod.
28391 * builtins.c (apply_args_size, apply_result_size, result_vector)
28392 (expand_builtin_apply_args_1, expand_builtin_apply)
28393 (expand_builtin_return): Update accordingly.
28395 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28397 * cse.c (hash_rtx_cb): Hash only the encoded elements.
28398 * cselib.c (cselib_hash_rtx): Likewise.
28399 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
28400 CONST_VECTOR encoding.
28402 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28403 Jeff Law <law@redhat.com>
28406 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
28407 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
28408 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
28409 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
28412 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
28413 explicitly probe *sp in a noreturn function if there were any callee
28414 register saves or frame pointer is needed.
28416 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28419 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
28420 BLKmode for ternary, binary or unary expressions.
28423 * var-tracking.c (delete_vta_debug_insn): New inline function.
28424 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
28425 insns from get_insns () to NULL instead of each bb separately.
28426 Use delete_vta_debug_insn. No longer static.
28427 (vt_debug_insns_local, variable_tracking_main_1): Adjust
28428 delete_vta_debug_insns callers.
28429 * rtl.h (delete_vta_debug_insns): Declare.
28430 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
28431 instead of variable_tracking_main.
28433 2018-01-03 Martin Sebor <msebor@redhat.com>
28435 PR tree-optimization/83603
28436 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
28437 arguments past the endof the argument list in functions declared
28438 without a prototype.
28439 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
28440 Avoid checking when arguments are null.
28442 2018-01-03 Martin Sebor <msebor@redhat.com>
28445 * doc/extend.texi (attribute const): Fix a typo.
28446 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
28447 issuing -Wsuggest-attribute for void functions.
28449 2018-01-03 Martin Sebor <msebor@redhat.com>
28451 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
28452 offset_int::from instead of wide_int::to_shwi.
28453 (maybe_diag_overlap): Remove assertion.
28454 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
28455 * gimple-ssa-sprintf.c (format_directive): Same.
28456 (parse_directive): Same.
28457 (sprintf_dom_walker::compute_format_length): Same.
28458 (try_substitute_return_value): Same.
28460 2018-01-03 Jeff Law <law@redhat.com>
28462 PR middle-end/83654
28463 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
28464 non-constant residual for zero at runtime and avoid probing in
28465 that case. Reorganize code for trailing problem to mirror handling
28468 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28470 PR tree-optimization/83501
28471 * tree-ssa-strlen.c (get_string_cst): New.
28472 (handle_char_store): Call get_string_cst.
28474 2018-01-03 Martin Liska <mliska@suse.cz>
28476 PR tree-optimization/83593
28477 * tree-ssa-strlen.c: Include tree-cfg.h.
28478 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
28479 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
28480 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
28482 (strlen_dom_walker::before_dom_children): Call
28483 gimple_purge_dead_eh_edges. Dump tranformation with details
28485 (strlen_dom_walker::before_dom_children): Update call by adding
28486 new argument cleanup_eh.
28487 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
28489 2018-01-03 Martin Liska <mliska@suse.cz>
28492 * cif-code.def (VARIADIC_THUNK): New enum value.
28493 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
28496 2018-01-03 Jan Beulich <jbeulich@suse.com>
28498 * sse.md (mov<mode>_internal): Tighten condition for when to use
28499 vmovdqu<ssescalarsize> for TI and OI modes.
28501 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28503 Update copyright years.
28505 2018-01-03 Martin Liska <mliska@suse.cz>
28508 * ipa-visibility.c (function_and_variable_visibility): Skip
28509 functions with noipa attribure.
28511 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28513 * gcc.c (process_command): Update copyright notice dates.
28514 * gcov-dump.c (print_version): Ditto.
28515 * gcov.c (print_version): Ditto.
28516 * gcov-tool.c (print_version): Ditto.
28517 * gengtype.c (create_file): Ditto.
28518 * doc/cpp.texi: Bump @copying's copyright year.
28519 * doc/cppinternals.texi: Ditto.
28520 * doc/gcc.texi: Ditto.
28521 * doc/gccint.texi: Ditto.
28522 * doc/gcov.texi: Ditto.
28523 * doc/install.texi: Ditto.
28524 * doc/invoke.texi: Ditto.
28526 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28528 * vector-builder.h (vector_builder::m_full_nelts): Change from
28529 unsigned int to poly_uint64.
28530 (vector_builder::full_nelts): Update prototype accordingly.
28531 (vector_builder::new_vector): Likewise.
28532 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
28533 (vector_builder::operator ==): Likewise.
28534 (vector_builder::finalize): Likewise.
28535 * int-vector-builder.h (int_vector_builder::int_vector_builder):
28536 Take the number of elements as a poly_uint64 rather than an
28538 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
28539 from unsigned int to poly_uint64.
28540 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
28541 (vec_perm_indices::new_vector): Likewise.
28542 (vec_perm_indices::length): Likewise.
28543 (vec_perm_indices::nelts_per_input): Likewise.
28544 (vec_perm_indices::input_nelts): Likewise.
28545 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
28546 number of elements per input as a poly_uint64 rather than an
28547 unsigned int. Use the original encoding for variable-length
28548 vectors, rather than clamping each individual element.
28549 For the second and subsequent elements in each pattern,
28550 clamp the step and base before clamping their sum.
28551 (vec_perm_indices::series_p): Handle polynomial element counts.
28552 (vec_perm_indices::all_in_range_p): Likewise.
28553 (vec_perm_indices_to_tree): Likewise.
28554 (vec_perm_indices_to_rtx): Likewise.
28555 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
28556 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
28557 (tree_vector_builder::new_binary_operation): Handle polynomial
28558 element counts. Return false if we need to know the number
28559 of elements at compile time.
28560 * fold-const.c (fold_vec_perm): Punt if the number of elements
28561 isn't known at compile time.
28563 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28565 * vec-perm-indices.h (vec_perm_builder): Change element type
28566 from HOST_WIDE_INT to poly_int64.
28567 (vec_perm_indices::element_type): Update accordingly.
28568 (vec_perm_indices::clamp): Handle polynomial element_types.
28569 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
28570 (vec_perm_indices::all_in_range_p): Likewise.
28571 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
28573 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
28574 polynomial vec_perm_indices element types.
28575 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
28576 * fold-const.c (fold_vec_perm): Likewise.
28577 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
28578 * tree-vect-generic.c (lower_vec_perm): Likewise.
28579 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
28580 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
28581 element type to HOST_WIDE_INT.
28583 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28584 Alan Hayward <alan.hayward@arm.com>
28585 David Sherwood <david.sherwood@arm.com>
28587 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
28588 rather than an int. Use plus_constant.
28589 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
28590 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
28592 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28593 Alan Hayward <alan.hayward@arm.com>
28594 David Sherwood <david.sherwood@arm.com>
28596 * calls.c (emit_call_1, expand_call): Change struct_value_size from
28597 a HOST_WIDE_INT to a poly_int64.
28599 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28600 Alan Hayward <alan.hayward@arm.com>
28601 David Sherwood <david.sherwood@arm.com>
28603 * calls.c (load_register_parameters): Cope with polynomial
28604 mode sizes. Require a constant size for BLKmode parameters
28605 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
28606 forces a parameter to be padded at the lsb end in order to
28607 fill a complete number of words, require the parameter size
28608 to be ordered wrt UNITS_PER_WORD.
28610 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28611 Alan Hayward <alan.hayward@arm.com>
28612 David Sherwood <david.sherwood@arm.com>
28614 * reload1.c (spill_stack_slot_width): Change element type
28615 from unsigned int to poly_uint64_pod.
28616 (alter_reg): Treat mode sizes as polynomial.
28618 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28619 Alan Hayward <alan.hayward@arm.com>
28620 David Sherwood <david.sherwood@arm.com>
28622 * reload.c (complex_word_subreg_p): New function.
28623 (reload_inner_reg_of_subreg, push_reload): Use it.
28625 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28626 Alan Hayward <alan.hayward@arm.com>
28627 David Sherwood <david.sherwood@arm.com>
28629 * lra-constraints.c (process_alt_operands): Reject matched
28630 operands whose sizes aren't ordered.
28631 (match_reload): Refer to this check here.
28633 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28634 Alan Hayward <alan.hayward@arm.com>
28635 David Sherwood <david.sherwood@arm.com>
28637 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
28638 that the mode size is in the set {1, 2, 4, 8, 16}.
28640 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28641 Alan Hayward <alan.hayward@arm.com>
28642 David Sherwood <david.sherwood@arm.com>
28644 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
28645 Use plus_constant instead of gen_rtx_PLUS.
28647 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28648 Alan Hayward <alan.hayward@arm.com>
28649 David Sherwood <david.sherwood@arm.com>
28651 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
28652 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
28653 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
28654 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
28655 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
28656 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
28657 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
28658 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
28659 * config/i386/i386.c (ix86_push_rounding): ...this new function.
28660 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
28662 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
28663 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
28664 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
28665 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
28666 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
28667 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
28668 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
28669 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
28670 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
28671 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
28673 * expr.c (emit_move_resolve_push): Treat the input and result
28674 of PUSH_ROUNDING as a poly_int64.
28675 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
28676 (emit_push_insn): Likewise.
28677 * lra-eliminations.c (mark_not_eliminable): Likewise.
28678 * recog.c (push_operand): Likewise.
28679 * reload1.c (elimination_effects): Likewise.
28680 * rtlanal.c (nonzero_bits1): Likewise.
28681 * calls.c (store_one_arg): Likewise. Require the padding to be
28682 known at compile time.
28684 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28685 Alan Hayward <alan.hayward@arm.com>
28686 David Sherwood <david.sherwood@arm.com>
28688 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
28689 Use plus_constant instead of gen_rtx_PLUS.
28691 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28692 Alan Hayward <alan.hayward@arm.com>
28693 David Sherwood <david.sherwood@arm.com>
28695 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
28696 rather than an int.
28698 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28699 Alan Hayward <alan.hayward@arm.com>
28700 David Sherwood <david.sherwood@arm.com>
28702 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
28703 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
28704 via stack temporaries. Treat the mode size as polynomial too.
28706 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28707 Alan Hayward <alan.hayward@arm.com>
28708 David Sherwood <david.sherwood@arm.com>
28710 * expr.c (expand_expr_real_2): When handling conversions involving
28711 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
28712 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
28713 as a poly_uint64 too.
28715 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28716 Alan Hayward <alan.hayward@arm.com>
28717 David Sherwood <david.sherwood@arm.com>
28719 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
28721 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28722 Alan Hayward <alan.hayward@arm.com>
28723 David Sherwood <david.sherwood@arm.com>
28725 * combine.c (can_change_dest_mode): Handle polynomial
28726 REGMODE_NATURAL_SIZE.
28727 * expmed.c (store_bit_field_1): Likewise.
28728 * expr.c (store_constructor): Likewise.
28729 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
28730 and polynomial REGMODE_NATURAL_SIZE.
28731 (gen_lowpart_common): Likewise.
28732 * reginfo.c (record_subregs_of_mode): Likewise.
28733 * rtlanal.c (read_modify_subreg_p): Likewise.
28735 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28736 Alan Hayward <alan.hayward@arm.com>
28737 David Sherwood <david.sherwood@arm.com>
28739 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
28740 numbers of elements.
28742 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28743 Alan Hayward <alan.hayward@arm.com>
28744 David Sherwood <david.sherwood@arm.com>
28746 * match.pd: Cope with polynomial numbers of vector elements.
28748 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28749 Alan Hayward <alan.hayward@arm.com>
28750 David Sherwood <david.sherwood@arm.com>
28752 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
28753 in a POINTER_PLUS_EXPR.
28755 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28756 Alan Hayward <alan.hayward@arm.com>
28757 David Sherwood <david.sherwood@arm.com>
28759 * omp-simd-clone.c (simd_clone_subparts): New function.
28760 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
28761 (ipa_simd_modify_function_body): Likewise.
28763 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28764 Alan Hayward <alan.hayward@arm.com>
28765 David Sherwood <david.sherwood@arm.com>
28767 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
28768 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
28769 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
28770 (expand_vector_condition, vector_element): Likewise.
28771 (subparts_gt): New function.
28772 (get_compute_type): Use subparts_gt.
28773 (count_type_subparts): Delete.
28774 (expand_vector_operations_1): Use subparts_gt instead of
28775 count_type_subparts.
28777 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28778 Alan Hayward <alan.hayward@arm.com>
28779 David Sherwood <david.sherwood@arm.com>
28781 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
28782 (vect_compile_time_alias): ...this new function. Do the calculation
28783 on poly_ints rather than trees.
28784 (vect_prune_runtime_alias_test_list): Update call accordingly.
28786 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28787 Alan Hayward <alan.hayward@arm.com>
28788 David Sherwood <david.sherwood@arm.com>
28790 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
28792 (vect_schedule_slp_instance): Likewise.
28794 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28795 Alan Hayward <alan.hayward@arm.com>
28796 David Sherwood <david.sherwood@arm.com>
28798 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
28799 constant and extern definitions for variable-length vectors.
28800 (vect_get_constant_vectors): Note that the number of units
28801 is known to be constant.
28803 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28804 Alan Hayward <alan.hayward@arm.com>
28805 David Sherwood <david.sherwood@arm.com>
28807 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
28808 of units as polynomial. Choose between WIDE and NARROW based
28811 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28812 Alan Hayward <alan.hayward@arm.com>
28813 David Sherwood <david.sherwood@arm.com>
28815 * tree-vect-stmts.c (simd_clone_subparts): New function.
28816 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
28818 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28819 Alan Hayward <alan.hayward@arm.com>
28820 David Sherwood <david.sherwood@arm.com>
28822 * tree-vect-stmts.c (vectorizable_call): Treat the number of
28823 vectors as polynomial. Use build_index_vector for
28824 IFN_GOMP_SIMD_LANE.
28826 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28827 Alan Hayward <alan.hayward@arm.com>
28828 David Sherwood <david.sherwood@arm.com>
28830 * tree-vect-stmts.c (get_load_store_type): Treat the number of
28831 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
28832 for variable-length vectors.
28833 (vectorizable_mask_load_store): Treat the number of units as
28834 polynomial, asserting that it is constant if the condition has
28835 already been enforced.
28836 (vectorizable_store, vectorizable_load): Likewise.
28838 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28839 Alan Hayward <alan.hayward@arm.com>
28840 David Sherwood <david.sherwood@arm.com>
28842 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
28843 of units as polynomial. Punt if we can't tell at compile time
28844 which vector contains the final result.
28846 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28847 Alan Hayward <alan.hayward@arm.com>
28848 David Sherwood <david.sherwood@arm.com>
28850 * tree-vect-loop.c (vectorizable_induction): Treat the number
28851 of units as polynomial. Punt on SLP inductions. Use an integer
28852 VEC_SERIES_EXPR for variable-length integer reductions. Use a
28853 cast of such a series for variable-length floating-point
28856 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28857 Alan Hayward <alan.hayward@arm.com>
28858 David Sherwood <david.sherwood@arm.com>
28860 * tree.h (build_index_vector): Declare.
28861 * tree.c (build_index_vector): New function.
28862 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
28863 of units as polynomial, forcibly converting it to a constant if
28864 vectorizable_reduction has already enforced the condition.
28865 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
28866 to create a {1,2,3,...} vector.
28867 (vectorizable_reduction): Treat the number of units as polynomial.
28868 Choose vectype_in based on the largest scalar element size rather
28869 than the smallest number of units. Enforce the restrictions
28872 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28873 Alan Hayward <alan.hayward@arm.com>
28874 David Sherwood <david.sherwood@arm.com>
28876 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
28877 number of units as polynomial.
28879 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28880 Alan Hayward <alan.hayward@arm.com>
28881 David Sherwood <david.sherwood@arm.com>
28883 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
28884 * target.def (autovectorize_vector_sizes): Return the vector sizes
28885 by pointer, using vector_sizes rather than a bitmask.
28886 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
28887 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
28888 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
28890 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
28891 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
28892 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
28893 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
28894 * omp-general.c (omp_max_vf): Likewise.
28895 * omp-low.c (omp_clause_aligned_alignment): Likewise.
28896 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
28897 * tree-vect-loop.c (vect_analyze_loop): Likewise.
28898 * tree-vect-slp.c (vect_slp_bb): Likewise.
28899 * doc/tm.texi: Regenerate.
28900 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
28902 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
28903 the vector size as a poly_uint64 rather than an unsigned int.
28904 (current_vector_size): Change from an unsigned int to a poly_uint64.
28905 (get_vectype_for_scalar_type): Update accordingly.
28906 * tree.h (build_truth_vector_type): Take the size and number of
28907 units as a poly_uint64 rather than an unsigned int.
28908 (build_vector_type): Add a temporary overload that takes
28909 the number of units as a poly_uint64 rather than an unsigned int.
28910 * tree.c (make_vector_type): Likewise.
28911 (build_truth_vector_type): Take the number of units as a poly_uint64
28912 rather than an unsigned int.
28914 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28915 Alan Hayward <alan.hayward@arm.com>
28916 David Sherwood <david.sherwood@arm.com>
28918 * target.def (get_mask_mode): Take the number of units and length
28919 as poly_uint64s rather than unsigned ints.
28920 * targhooks.h (default_get_mask_mode): Update accordingly.
28921 * targhooks.c (default_get_mask_mode): Likewise.
28922 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
28923 * doc/tm.texi: Regenerate.
28925 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28926 Alan Hayward <alan.hayward@arm.com>
28927 David Sherwood <david.sherwood@arm.com>
28929 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
28930 * omp-general.c (omp_max_vf): Likewise.
28931 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
28932 (expand_omp_simd): Handle polynomial safelen.
28933 * omp-low.c (omplow_simd_context): Add a default constructor.
28934 (omplow_simd_context::max_vf): Change from int to poly_uint64.
28935 (lower_rec_simd_input_clauses): Update accordingly.
28936 (lower_rec_input_clauses): Likewise.
28938 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28939 Alan Hayward <alan.hayward@arm.com>
28940 David Sherwood <david.sherwood@arm.com>
28942 * tree-vectorizer.h (vect_nunits_for_cost): New function.
28943 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
28944 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
28945 (vect_analyze_slp_cost): Likewise.
28946 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
28947 (vect_model_load_cost): Likewise.
28949 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28950 Alan Hayward <alan.hayward@arm.com>
28951 David Sherwood <david.sherwood@arm.com>
28953 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
28954 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
28955 from an unsigned int * to a poly_uint64_pod *.
28956 (calculate_unrolling_factor): New function.
28957 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
28959 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28960 Alan Hayward <alan.hayward@arm.com>
28961 David Sherwood <david.sherwood@arm.com>
28963 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
28964 from an unsigned int to a poly_uint64.
28965 (_loop_vec_info::slp_unrolling_factor): Likewise.
28966 (_loop_vec_info::vectorization_factor): Change from an int
28968 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
28969 (vect_get_num_vectors): New function.
28970 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
28971 (vect_get_num_copies): Use vect_get_num_vectors.
28972 (vect_analyze_data_ref_dependences): Change max_vf from an int *
28973 to an unsigned int *.
28974 (vect_analyze_data_refs): Change min_vf from an int * to a
28976 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
28977 than an unsigned HOST_WIDE_INT.
28978 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
28979 (vect_analyze_data_ref_dependence): Change max_vf from an int *
28980 to an unsigned int *.
28981 (vect_analyze_data_ref_dependences): Likewise.
28982 (vect_compute_data_ref_alignment): Handle polynomial vf.
28983 (vect_enhance_data_refs_alignment): Likewise.
28984 (vect_prune_runtime_alias_test_list): Likewise.
28985 (vect_shift_permute_load_chain): Likewise.
28986 (vect_supportable_dr_alignment): Likewise.
28987 (dependence_distance_ge_vf): Take the vectorization factor as a
28988 poly_uint64 rather than an unsigned HOST_WIDE_INT.
28989 (vect_analyze_data_refs): Change min_vf from an int * to a
28991 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
28992 vfm1 as a poly_uint64 rather than an int. Make the same change
28993 for the returned bound_scalar.
28994 (vect_gen_vector_loop_niters): Handle polynomial vf.
28995 (vect_do_peeling): Likewise. Update call to
28996 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
28997 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
28999 * tree-vect-loop.c (vect_determine_vectorization_factor)
29000 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
29001 (vect_get_known_peeling_cost): Likewise.
29002 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
29003 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
29004 (vect_transform_loop): Likewise. Use the lowest possible VF when
29005 updating the upper bounds of the loop.
29006 (vect_min_worthwhile_factor): Make static. Return an unsigned int
29007 rather than an int.
29008 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
29009 polynomial unroll factors.
29010 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
29011 (vect_make_slp_decision): Likewise.
29012 (vect_supported_load_permutation_p): Likewise, and polynomial
29014 (vect_analyze_slp_cost): Handle polynomial vf.
29015 (vect_slp_analyze_node_operations): Likewise.
29016 (vect_slp_analyze_bb_1): Likewise.
29017 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
29018 than an unsigned HOST_WIDE_INT.
29019 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
29020 (vectorizable_load): Handle polynomial vf.
29021 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
29023 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
29025 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29026 Alan Hayward <alan.hayward@arm.com>
29027 David Sherwood <david.sherwood@arm.com>
29029 * match.pd: Handle bit operations involving three constants
29030 and try to fold one pair.
29032 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29034 * tree-vect-loop-manip.c: Include gimple-fold.h.
29035 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
29036 niters_maybe_zero parameters. Handle other cases besides a step of 1.
29037 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
29038 Add a path that uses a step of VF instead of 1, but disable it
29040 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
29041 and niters_no_overflow parameters. Update calls to
29042 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
29043 Create a new SSA name if the latter choses to use a ste other
29044 than zero, and return it via niters_vector_mult_vf_var.
29045 * tree-vect-loop.c (vect_transform_loop): Update calls to
29046 vect_do_peeling, vect_gen_vector_loop_niters and
29047 slpeel_make_loop_iterate_ntimes.
29048 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
29049 (vect_gen_vector_loop_niters): Update declarations after above changes.
29051 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
29053 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
29054 128-bit round to integer instructions.
29055 (ceil<mode>2): Likewise.
29056 (btrunc<mode>2): Likewise.
29057 (round<mode>2): Likewise.
29059 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29061 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
29062 unaligned VSX load/store on P8/P9.
29063 (expand_block_clear): Allow the use of unaligned VSX
29064 load/store on P8/P9.
29066 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29068 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
29070 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
29071 swap associated with both a load and a store.
29073 2018-01-02 Andrew Waterman <andrew@sifive.com>
29075 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
29076 * config/riscv/riscv.md (clear_cache): Use it.
29078 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
29080 * web.c: Remove out-of-date comment.
29082 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29084 * expr.c (fixup_args_size_notes): Check that any existing
29085 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
29086 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
29087 (emit_single_push_insn): ...here.
29089 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29091 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
29092 (const_vector_encoded_nelts): New function.
29093 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
29094 (const_vector_int_elt, const_vector_elt): Declare.
29095 * emit-rtl.c (const_vector_int_elt_1): New function.
29096 (const_vector_elt): Likewise.
29097 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
29098 of CONST_VECTOR_ELT.
29100 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29102 * expr.c: Include rtx-vector-builder.h.
29103 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
29104 directly on the tree encoding.
29105 (const_vector_from_tree): Likewise.
29106 * optabs.c: Include rtx-vector-builder.h.
29107 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
29108 sequence of "u" values.
29109 * vec-perm-indices.c: Include rtx-vector-builder.h.
29110 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
29111 directly on the vec_perm_indices encoding.
29113 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29115 * doc/rtl.texi (const_vector): Describe new encoding scheme.
29116 * Makefile.in (OBJS): Add rtx-vector-builder.o.
29117 * rtx-vector-builder.h: New file.
29118 * rtx-vector-builder.c: Likewise.
29119 * rtl.h (rtx_def::u2): Add a const_vector field.
29120 (CONST_VECTOR_NPATTERNS): New macro.
29121 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
29122 (CONST_VECTOR_DUPLICATE_P): Likewise.
29123 (CONST_VECTOR_STEPPED_P): Likewise.
29124 (CONST_VECTOR_ENCODED_ELT): Likewise.
29125 (const_vec_duplicate_p): Check for a duplicated vector encoding.
29126 (unwrap_const_vec_duplicate): Likewise.
29127 (const_vec_series_p): Check for a non-duplicated vector encoding.
29128 Say that the function only returns true for integer vectors.
29129 * emit-rtl.c: Include rtx-vector-builder.h.
29130 (gen_const_vec_duplicate_1): Delete.
29131 (gen_const_vector): Call gen_const_vec_duplicate instead of
29132 gen_const_vec_duplicate_1.
29133 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
29134 (gen_const_vec_duplicate): Use rtx_vector_builder.
29135 (gen_const_vec_series): Likewise.
29136 (gen_rtx_CONST_VECTOR): Likewise.
29137 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
29138 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
29139 Build a new vector rather than modifying a CONST_VECTOR in-place.
29140 (handle_special_swappables): Update call accordingly.
29141 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
29142 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
29143 Build a new vector rather than modifying a CONST_VECTOR in-place.
29144 (handle_special_swappables): Update call accordingly.
29146 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29148 * simplify-rtx.c (simplify_const_binary_operation): Use
29149 CONST_VECTOR_ELT instead of XVECEXP.
29151 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29153 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
29154 the selector elements to be different from the data elements
29155 if the selector is a VECTOR_CST.
29156 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
29157 ssizetype for the selector.
29159 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29161 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
29162 before testing each element individually.
29163 * tree-vect-generic.c (lower_vec_perm): Likewise.
29165 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29167 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
29168 * selftest-run-tests.c (selftest::run_tests): Call it.
29169 * vector-builder.h (vector_builder::operator ==): New function.
29170 (vector_builder::operator !=): Likewise.
29171 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
29172 (vec_perm_indices::all_from_input_p): New function.
29173 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
29174 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
29175 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
29176 instead of reading the VECTOR_CST directly. Detect whether both
29177 vector inputs are the same before constructing the vec_perm_indices,
29178 and update the number of inputs argument accordingly. Use the
29179 utility functions added above. Only construct sel2 if we need to.
29181 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29183 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
29184 the broadcast of the low byte.
29185 (expand_mult_highpart): Use an explicit encoding for the permutes.
29186 * optabs-query.c (can_mult_highpart_p): Likewise.
29187 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
29188 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29189 (vectorizable_bswap): Likewise.
29190 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
29191 explicit encoding for the power-of-2 permutes.
29192 (vect_permute_store_chain): Likewise.
29193 (vect_grouped_load_supported): Likewise.
29194 (vect_permute_load_chain): Likewise.
29196 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29198 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
29199 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
29200 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
29201 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
29202 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
29203 (vect_gen_perm_mask_any): Likewise.
29205 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29207 * int-vector-builder.h: New file.
29208 * vec-perm-indices.h: Include int-vector-builder.h.
29209 (vec_perm_indices): Redefine as an int_vector_builder.
29210 (auto_vec_perm_indices): Delete.
29211 (vec_perm_builder): Redefine as a stand-alone class.
29212 (vec_perm_indices::vec_perm_indices): New function.
29213 (vec_perm_indices::clamp): Likewise.
29214 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
29215 (vec_perm_indices::new_vector): New function.
29216 (vec_perm_indices::new_expanded_vector): Update for new
29217 vec_perm_indices class.
29218 (vec_perm_indices::rotate_inputs): New function.
29219 (vec_perm_indices::all_in_range_p): Operate directly on the
29220 encoded form, without computing elided elements.
29221 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
29222 encoding. Update for new vec_perm_indices class.
29223 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
29224 the given vec_perm_builder.
29225 (expand_vec_perm_var): Update vec_perm_builder constructor.
29226 (expand_mult_highpart): Use vec_perm_builder instead of
29227 auto_vec_perm_indices.
29228 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
29229 vec_perm_indices instead of auto_vec_perm_indices. Use a single
29230 or double series encoding as appropriate.
29231 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
29232 vec_perm_indices instead of auto_vec_perm_indices.
29233 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
29234 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
29235 (vect_permute_store_chain): Likewise.
29236 (vect_grouped_load_supported): Likewise.
29237 (vect_permute_load_chain): Likewise.
29238 (vect_shift_permute_load_chain): Likewise.
29239 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
29240 (vect_transform_slp_perm_load): Likewise.
29241 (vect_schedule_slp_instance): Likewise.
29242 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29243 (vectorizable_mask_load_store): Likewise.
29244 (vectorizable_bswap): Likewise.
29245 (vectorizable_store): Likewise.
29246 (vectorizable_load): Likewise.
29247 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
29248 vec_perm_indices instead of auto_vec_perm_indices. Use
29249 tree_to_vec_perm_builder to read the vector from a tree.
29250 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
29251 vec_perm_builder instead of a vec_perm_indices.
29252 (have_whole_vector_shift): Use vec_perm_builder and
29253 vec_perm_indices instead of auto_vec_perm_indices. Leave the
29254 truncation to calc_vec_perm_mask_for_shift.
29255 (vect_create_epilog_for_reduction): Likewise.
29256 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
29257 from auto_vec_perm_indices to vec_perm_indices.
29258 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
29259 instead of changing individual elements.
29260 (aarch64_vectorize_vec_perm_const): Use new_vector to install
29261 the vector in d.perm.
29262 * config/arm/arm.c (expand_vec_perm_d::perm): Change
29263 from auto_vec_perm_indices to vec_perm_indices.
29264 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
29265 instead of changing individual elements.
29266 (arm_vectorize_vec_perm_const): Use new_vector to install
29267 the vector in d.perm.
29268 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
29269 Update vec_perm_builder constructor.
29270 (rs6000_expand_interleave): Likewise.
29271 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
29272 (rs6000_expand_interleave): Likewise.
29274 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29276 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
29277 to qimode could truncate the indices.
29278 * optabs.c (expand_vec_perm_var): Likewise.
29280 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29282 * Makefile.in (OBJS): Add vec-perm-indices.o.
29283 * vec-perm-indices.h: New file.
29284 * vec-perm-indices.c: Likewise.
29285 * target.h (vec_perm_indices): Replace with a forward class
29287 (auto_vec_perm_indices): Move to vec-perm-indices.h.
29288 * optabs.h: Include vec-perm-indices.h.
29289 (expand_vec_perm): Delete.
29290 (selector_fits_mode_p, expand_vec_perm_var): Declare.
29291 (expand_vec_perm_const): Declare.
29292 * target.def (vec_perm_const_ok): Replace with...
29293 (vec_perm_const): ...this new hook.
29294 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
29295 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
29296 * doc/tm.texi: Regenerate.
29297 * optabs.def (vec_perm_const): Delete.
29298 * doc/md.texi (vec_perm_const): Likewise.
29299 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
29300 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
29301 expand_vec_perm for constant permutation vectors. Assert that
29302 the mode of variable permutation vectors is the integer equivalent
29303 of the mode that is being permuted.
29304 * optabs-query.h (selector_fits_mode_p): Declare.
29305 * optabs-query.c: Include vec-perm-indices.h.
29306 (selector_fits_mode_p): New function.
29307 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
29308 is defined, instead of checking whether the vec_perm_const_optab
29309 exists. Use targetm.vectorize.vec_perm_const instead of
29310 targetm.vectorize.vec_perm_const_ok. Check whether the indices
29311 fit in the vector mode before using a variable permute.
29312 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
29313 vec_perm_indices instead of an rtx.
29314 (expand_vec_perm): Replace with...
29315 (expand_vec_perm_const): ...this new function. Take the selector
29316 as a vec_perm_indices rather than an rtx. Also take the mode of
29317 the selector. Update call to shift_amt_for_vec_perm_mask.
29318 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
29319 Use vec_perm_indices::new_expanded_vector to expand the original
29320 selector into bytes. Check whether the indices fit in the vector
29321 mode before using a variable permute.
29322 (expand_vec_perm_var): Make global.
29323 (expand_mult_highpart): Use expand_vec_perm_const.
29324 * fold-const.c: Includes vec-perm-indices.h.
29325 * tree-ssa-forwprop.c: Likewise.
29326 * tree-vect-data-refs.c: Likewise.
29327 * tree-vect-generic.c: Likewise.
29328 * tree-vect-loop.c: Likewise.
29329 * tree-vect-slp.c: Likewise.
29330 * tree-vect-stmts.c: Likewise.
29331 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
29333 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
29334 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
29335 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
29336 (aarch64_vectorize_vec_perm_const): ...this new function.
29337 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
29338 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29339 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
29340 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
29341 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
29342 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29343 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
29345 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
29346 check for NEON modes.
29347 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
29348 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
29349 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
29350 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
29352 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
29353 the old VEC_PERM_CONST conditions.
29354 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
29355 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
29356 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
29357 (ia64_vectorize_vec_perm_const_ok): Merge into...
29358 (ia64_vectorize_vec_perm_const): ...this new function.
29359 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
29360 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
29361 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
29362 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
29363 * config/mips/mips.c (mips_expand_vec_perm_const)
29364 (mips_vectorize_vec_perm_const_ok): Merge into...
29365 (mips_vectorize_vec_perm_const): ...this new function.
29366 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
29367 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
29368 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
29369 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
29370 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
29371 (rs6000_expand_vec_perm_const): Delete.
29372 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
29374 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29375 (altivec_expand_vec_perm_const_le): Take each operand individually.
29376 Operate on constant selectors rather than rtxes.
29377 (altivec_expand_vec_perm_const): Likewise. Update call to
29378 altivec_expand_vec_perm_const_le.
29379 (rs6000_expand_vec_perm_const): Delete.
29380 (rs6000_vectorize_vec_perm_const_ok): Delete.
29381 (rs6000_vectorize_vec_perm_const): New function.
29382 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
29383 an element count and rtx array.
29384 (rs6000_expand_extract_even): Update call accordingly.
29385 (rs6000_expand_interleave): Likewise.
29386 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
29387 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
29388 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
29389 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
29390 (rs6000_expand_vec_perm_const): Delete.
29391 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
29392 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29393 (altivec_expand_vec_perm_const_le): Take each operand individually.
29394 Operate on constant selectors rather than rtxes.
29395 (altivec_expand_vec_perm_const): Likewise. Update call to
29396 altivec_expand_vec_perm_const_le.
29397 (rs6000_expand_vec_perm_const): Delete.
29398 (rs6000_vectorize_vec_perm_const_ok): Delete.
29399 (rs6000_vectorize_vec_perm_const): New function. Remove stray
29400 reference to the SPE evmerge intructions.
29401 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
29402 an element count and rtx array.
29403 (rs6000_expand_extract_even): Update call accordingly.
29404 (rs6000_expand_interleave): Likewise.
29405 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
29406 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
29408 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29410 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29412 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
29413 vector mode and that that mode matches the mode of the data
29415 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
29416 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
29417 directly using expand_vec_perm_1 when forcing selectors into
29419 (expand_vec_perm_var): New function, split out from expand_vec_perm.
29421 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29423 * optabs-query.h (can_vec_perm_p): Delete.
29424 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
29425 * optabs-query.c (can_vec_perm_p): Split into...
29426 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
29427 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
29428 particular selector is valid.
29429 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
29430 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
29431 (vect_grouped_load_supported): Likewise.
29432 (vect_shift_permute_load_chain): Likewise.
29433 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
29434 (vect_transform_slp_perm_load): Likewise.
29435 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29436 (vectorizable_bswap): Likewise.
29437 (vect_gen_perm_mask_checked): Likewise.
29438 * fold-const.c (fold_ternary_loc): Likewise. Don't take
29439 implementations of variable permutation vectors into account
29440 when deciding which selector to use.
29441 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
29442 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
29443 with a false third argument.
29444 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
29445 to test whether the constant selector is valid and can_vec_perm_var_p
29446 to test whether a variable selector is valid.
29448 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29450 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
29451 * optabs-query.c (can_vec_perm_p): Likewise.
29452 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
29453 instead of vec_perm_indices.
29454 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
29455 (vect_gen_perm_mask_checked): Likewise,
29456 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
29457 (vect_gen_perm_mask_checked): Likewise,
29459 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29461 * optabs-query.h (qimode_for_vec_perm): Declare.
29462 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
29463 (qimode_for_vec_perm): ...this new function.
29464 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
29466 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29468 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
29469 does not have a conditional at the top.
29471 2018-01-02 Richard Biener <rguenther@suse.de>
29473 * ipa-inline.c (big_speedup_p): Fix expression.
29475 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
29478 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
29481 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
29485 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
29486 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
29487 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
29488 cond_taken_branch_cost 3->4.
29490 2018-01-01 Jakub Jelinek <jakub@redhat.com>
29492 PR tree-optimization/83581
29493 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
29494 TODO_cleanup_cfg if any changes have been made.
29496 PR middle-end/83608
29497 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
29498 convert_modes if target mode has the right side, but different mode
29501 PR middle-end/83609
29502 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
29503 last argument when extracting from CONCAT. If either from_real or
29504 from_imag is NULL, use expansion through memory. If result is not
29505 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
29506 the parts directly to inner mode, if even that fails, use expansion
29509 PR middle-end/83623
29510 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
29511 check for bswap in mode rather than HImode and use that in expand_unop
29514 Copyright (C) 2018 Free Software Foundation, Inc.
29516 Copying and distribution of this file, with or without modification,
29517 are permitted in any medium without royalty provided the copyright
29518 notice and this notice are preserved.