1 2018-10-30 Aaron Sawdey <acsawdey@linux.ibm.com>
3 * config/rs6000/rs6000.md (bswapdi2): Force address into register
4 if not in indexed or indirect form.
5 (bswapdi2_load): Change predicate to indexed_or_indirect_operand.
6 (bswapdi2_store): Ditto.
7 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem): New
9 * config/rs6000/rs6000-protos.h (rs6000_force_indexed_or_indirect_mem):
10 Prototype for helper function.
12 2018-10-30 Martin Sebor <msebor@redhat.com>
14 * doc/extend.texi (optimize): Clarify/expand attribute documentation.
15 (target, pragma GCC optimize, pragma GCC target): Ditto.
17 2018-10-30 Jonathan Wakely <jwakely@redhat.com>
19 * doc/extend.texi: Fix prototype and description of
20 __builtin_expect_with_probability.
22 2018-10-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
24 * cgraph.h (clone_function_name_1): Replaced by new
25 clone_function_name_numbered that takes name as string; for
26 privatize_symbol_name_1 use only.
27 (clone_function_name): Renamed to
28 clone_function_name_numbered to be explicit about numbering.
29 (clone_function_name): New two-argument function that does
30 not number its output.
31 (clone_function_name): New three-argument function that
32 takes a number to append to its output.
33 * cgraphclones.c (duplicate_thunk_for_node):
34 (clone_function_name_1): Renamed.
35 (clone_function_name_numbered): Two new functions.
36 (clone_function_name): Improved documentation.
37 (cgraph_node::create_virtual_clone): Use clone_function_name_numbered.
38 * config/rs6000/rs6000.c (make_resolver_func): Ditto.
39 * final.c (final_scan_insn_1): Use the new clone_function_name
41 * multiple_target.c (create_dispatcher_calls): Ditto.
42 (create_target_clone): Ditto.
43 * omp-expand.c (grid_expand_target_grid_body): Ditto.
44 * omp-low.c (create_omp_child_function_name): Ditto.
45 * omp-simd-clone.c (simd_clone_create): Ditto.
46 * symtab.c (simd_symtab_node::noninterposable_alias): Use the
47 new clone_function_name without numbering.
49 2018-10-30 Richard Earnshaw <rearnsha@arm.com>
51 * alloc-pool.h (base_pool_allocator <TBlockAllocator>::initialize):
52 Assert that the allocation size is not zero.
54 2018-10-30 Richard Biener <rguenther@suse.de>
56 PR tree-optimization/87800
57 * tree-vect-slp.c (vect_build_slp_tree_2): Reject any
58 non-induction or reduction PHIs.
60 2018-10-30 Sameera Deshpande <sameera.deshpande@linaro.org
62 * config/aarch64/falkor-tag-collision-avoidance.c
63 (execute_tag_collision_avoidance): Call df_note_add_problem.
65 2018-10-30 Martin Liska <mliska@suse.cz>
67 * doc/extend.texi: Fix typo in documentation
68 of __builtin_expect_with_probability.
70 2018-10-29 David Malcolm <dmalcolm@redhat.com>
73 * input.c (get_substring_ranges_for_loc): Detect if
74 linemap_resolve_location gives us a NULL map, and reject
77 2018-10-29 Iain Buclaw <ibuclaw@gdcproject.org>
79 * config.gcc (xstormy16-*-elf): Set tm_d_file.
81 2018-10-29 David Malcolm <dmalcolm@redhat.com>
82 Martin Sebor <msebor@redhat.com>
83 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
85 * doc/ux.texi (Quoting): New subsection, adapted from material at
86 https://gcc.gnu.org/wiki/DiagnosticsGuidelines written by
87 MartinSebor and ManuelLopezIbanez.
88 (Fix-it hints): Note that fix-it hints shouldn't be marked for
91 2018-10-29 Kugan Vivekanandarajah <kuganv@linaro.org>
94 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Fix niter
97 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
99 * gcc.target/powerpc/ssse3-check.h: Remove inaccurate comment.
101 2018-10-29 Paul Koning <ni1d@arrl.net>
103 * doc/rtl.texi (CONST_WIDE_INT_ELT): Give correct macro name.
105 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
107 * gcc/config/rs6000/mmintrin.h (_mm_packs_pi16, _mm_packs_pi32,
108 _mm_packs_pu16, _mm_unpackhi_pi8, _mm_unpacklo_pi8, _mm_add_pi8,
109 _mm_add_pi16, _mm_add_pi32, _mm_sub_pi8, _mm_sub_pi16, _mm_sub_pi32,
110 _mm_cmpgt_pi8, _mm_cmpeq_pi16, _mm_cmpgt_pi16, _mm_cmpeq_pi32,
111 _mm_cmpgt_pi32, _mm_adds_pi8, _mm_adds_pi16, _mm_adds_pu8,
112 _mm_adds_pu16, _mm_subs_pi8, _mm_subs_pi16, _mm_subs_pu8,
113 _mm_subs_pu16, _mm_madd_pi16, _mm_mulhi_pi16, _mm_mullo_pi16,
114 _mm_sll_pi16, _mm_sra_pi16, _mm_srl_pi16, _mm_set1_pi16, _mm_set1_pi8):
115 Change 'vector' to '__vector'.
116 * gcc/config/rs6000/xmmintrin.h (_mm_cvtps_pi32, _mm_cvttps_pi32,
117 _mm_cvtps_pi16, _mm_cvtps_pi8, _mm_max_pi16, _mm_max_pu8, _mm_min_pi16,
118 _mm_min_pu8, _mm_mulhi_pu16, _mm_shuffle_pi16, _mm_avg_pu8,
119 _mm_avg_pu16): Likewise. And, whitespace corrections.
121 2018-10-29 Richard Biener <rguenther@suse.de>
123 PR tree-optimization/87785
124 * tree-vect-slp.c (vect_gather_slp_loads): Only gather
127 2018-10-29 Olivier Hainque <hainque@adacore.com>
129 * VXWORKS_NET_LIBS_RTP: New macro, network part of VXWORKS_LIBS_RTP.
131 2018-10-29 Olivier Hainque <hainque@adacore.com>
133 * config/vxworks.h (VXWORKS_PERSONALITY): New VxWorks
134 ports configuration macro, defaults to "gnu".
135 (VXWORKS_OS_CPP_BUILTINS): Use it. Feed TOOL and
136 TOOL_FAMILY instead of _VX_TOOL/_VXTOOL_FAMILY.
138 2018-10-29 Olivier Hainque <hainque@adacore.com>
140 * config/vxworks.h (CLEAR_INSN_CACHE): #undef before #define.
142 2018-10-29 Richard Biener <rguenther@suse.de>
144 PR tree-optimization/87790
145 * tree-vect-slp.c (vect_mark_slp_stmts): Simplify.
146 (vect_make_slp_decision): Adjust.
147 (vect_slp_analyze_bb_1): Likewise.
148 (vect_detect_hybrid_slp_stmts): Properly union SLP type over
151 2018-10-29 Richard Biener <rguenther@suse.de>
153 PR tree-optimization/87785
154 * tree-vect-slp.c (vect_build_slp_tree_2): Remove loads argument
156 (vect_build_slp_tree): Likewise.
157 (vect_gather_slp_loads): New function.
158 (vect_analyze_slp_instance): Gather loads separately from the
161 2018-10-29 Martin Liska <mliska@suse.cz>
163 * Makefile.in: Make dependency to json.o.
164 * doc/gcov.texi: Document new JSON format, remove
165 old intermediate format documentation.
166 * gcov.c (struct function_info): Come up with m_name and
168 (function_info::function_info): Initialize it.
169 (function_info::~function_info): Release it.
170 (main): Rename flag_intermediate_format to flag_json_format.
171 (print_usage): Describe --json-format.
172 (process_args): Set flag_json_format.
173 (output_intermediate_line): Remove.
174 (output_intermediate_json_line): Likewise.
175 (get_gcov_intermediate_filename): Return new extension
177 (output_intermediate_file): Implement JSON emission.
178 (output_json_intermediate_file): Implement JSON emission.
179 (generate_results): Use ::get_name for function name.
180 Handle JSON output file.
181 (read_graph_file): Use ::get_name instead of cplus_demangle.
182 (read_count_file): Likewise.
183 (solve_flow_graph): Likewise.
184 (add_line_counts): Likewise.
185 (accumulate_line_counts): Use new flag_json_format.
186 (output_function_details): Use ::get_name instead of cplus_demangle.
187 (output_lines): Likewise.
188 * json.cc (test_writing_literals): Add new tests.
189 * json.h (class literal): Add new boolean constructor.
191 2018-10-29 Segher Boessenkool <segher@kernel.crashing.org>
193 PR rtl-optimization/87701
194 PR rtl-optimization/87780
195 * combine.c (make_more_copies): Rewrite.
197 2018-10-28 Kugan Vivekanandarajah <kuganv@linaro.org>
199 * doc/generic.texi (ABSU_EXPR): Document.
200 * match.pd (absu(x)*absu(x) -> x*x): Handle.
201 (absu(absu(X)) -> absu(X)): Likewise.
202 (absu(-X) -> absu(X)): Likewise.
203 (absu(X) where X is nonnegative -> X): Likewise.
205 2018-10-28 Iain Buclaw <ibuclaw@gdcproject.org>
207 * Makefile.in (tm_d_file_list, tm_d_include_list): New variables.
208 (TM_D_H, D_TARGET_DEF, D_TARGET_H, D_TARGET_OBJS): New variables.
209 (tm_d.h, cs-tm_d.h, default-d.o): New rules.
210 (d/d-target-hooks-def.h, s-d-target-hooks-def-h): New rules.
211 (s-tm-texi): Also check timestamp on d-target.def.
212 (generated_files): Add TM_D_H and d-target-hooks-def.h.
213 (build/genhooks.o): Also depend on D_TARGET_DEF.
214 * config.gcc (tm_d_file, d_target_objs, target_has_targetdm): New
216 * config/aarch64/aarch64-d.c: New file.
217 * config/aarch64/aarch64-linux.h (GNU_USER_TARGET_D_CRITSEC_SIZE):
219 * config/aarch64/aarch64-protos.h (aarch64_d_target_versions): New
221 * config/aarch64/aarch64.h (TARGET_D_CPU_VERSIONS): Define.
222 * config/aarch64/t-aarch64 (aarch64-d.o): New rule.
223 * config/arm/arm-d.c: New file.
224 * config/arm/arm-protos.h (arm_d_target_versions): New prototype.
225 * config/arm/arm.h (TARGET_D_CPU_VERSIONS): Define.
226 * config/arm/linux-eabi.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
227 * config/arm/t-arm (arm-d.o): New rule.
228 * config/default-d.c: New file.
229 * config/glibc-d.c: New file.
230 * config/gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
231 * config/i386/i386-d.c: New file.
232 * config/i386/i386-protos.h (ix86_d_target_versions): New prototype.
233 * config/i386/i386.h (TARGET_D_CPU_VERSIONS): Define.
234 * config/i386/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
235 (GNU_USER_TARGET_D_CRITSEC_SIZE): Define.
236 * config/i386/t-i386 (i386-d.o): New rule.
237 * config/kfreebsd-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
238 * config/kopensolaris-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
239 * config/linux-android.h (ANDROID_TARGET_D_OS_VERSIONS): Define.
240 * config/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
241 * config/mips/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
242 * config/mips/mips-d.c: New file.
243 * config/mips/mips-protos.h (mips_d_target_versions): New prototype.
244 * config/mips/mips.h (TARGET_D_CPU_VERSIONS): Define.
245 * config/mips/t-mips (mips-d.o): New rule.
246 * config/powerpcspe/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
247 * config/powerpcspe/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
248 * config/powerpcspe/powerpcspe-d.c: New file.
249 * config/powerpcspe/powerpcspe-protos.h (rs6000_d_target_versions):
251 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
252 Support GNU D by using 0 as the language type.
253 * config/powerpcspe/powerpcspe.h (TARGET_D_CPU_VERSIONS): Define.
254 * config/powerpcspe/t-powerpcspe (powerpcspe-d.o): New rule.
255 * config/riscv/riscv-d.c: New file.
256 * config/riscv/riscv-protos.h (riscv_d_target_versions): New
258 * config/riscv/riscv.h (TARGET_D_CPU_VERSIONS): Define.
259 * config/riscv/t-riscv (riscv-d.o): New rule.
260 * config/rs6000/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
261 * config/rs6000/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
262 * config/rs6000/rs6000-d.c: New file.
263 * config/rs6000/rs6000-protos.h (rs6000_d_target_versions): New
265 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
266 Support GNU D by using 0 as the language type.
267 * config/rs6000/rs6000.h (TARGET_D_CPU_VERSIONS): Define.
268 * config/rs6000/t-rs6000 (rs6000-d.o): New rule.
269 * config/s390/s390-d.c: New file.
270 * config/s390/s390-protos.h (s390_d_target_versions): New prototype.
271 * config/s390/s390.h (TARGET_D_CPU_VERSIONS): Define.
272 * config/s390/t-s390 (s390-d.o): New rule.
273 * config/sparc/sparc-d.c: New file.
274 * config/sparc/sparc-protos.h (sparc_d_target_versions): New
276 * config/sparc/sparc.h (TARGET_D_CPU_VERSIONS): Define.
277 * config/sparc/t-sparc (sparc-d.o): New rule.
278 * config/t-glibc (glibc-d.o): New rule.
279 * configure: Regenerated.
280 * configure.ac (tm_d_file): New variable.
281 (tm_d_file_list, tm_d_include_list, d_target_objs): Add substitutes.
282 * doc/contrib.texi (Contributors): Add self for the D frontend.
283 * doc/frontends.texi (G++ and GCC): Mention D as a supported language.
284 * doc/install.texi (Configuration): Mention libphobos as an option for
285 --enable-shared. Mention d as an option for --enable-languages.
286 (Testing): Mention check-d as a target.
287 * doc/invoke.texi (Overall Options): Mention .d, .dd, and .di as file
288 name suffixes. Mention d as a -x option.
289 * doc/sourcebuild.texi (Top Level): Mention libphobos.
290 * doc/standards.texi (Standards): Add section on D language.
291 * doc/tm.texi: Regenerated.
292 * doc/tm.texi.in: Add @node for D language and ABI, and @hook for
293 TARGET_CPU_VERSIONS, TARGET_D_OS_VERSIONS, and TARGET_D_CRITSEC_SIZE.
294 * dwarf2out.c (is_dlang): New function.
295 (gen_compile_unit_die): Use DW_LANG_D for D.
296 (declare_in_namespace): Return module die for D, instead of adding
297 extra declarations into the namespace.
298 (gen_namespace_die): Generate DW_TAG_module for D.
299 (gen_decl_die): Handle CONST_DECLSs for D.
300 (dwarf2out_decl): Likewise.
301 (prune_unused_types_walk_local_classes): Handle DW_tag_interface_type.
302 (prune_unused_types_walk): Handle DW_tag_interface_type same as other
304 * gcc.c (default_compilers): Add entries for .d, .dd and .di.
305 * genhooks.c: Include d/d-target.def.
307 2018-10-28 Iain Sandoe <iain@sandoe.co.uk>
310 * config/rs6000/darwin.h (STACK_BOUNDARY): New.
311 (RS6000_STARTING_FRAME_OFFSET): Adjust to preserve 16byte alignment.
312 (STACK_DYNAMIC_OFFSET): Likewise.
314 2018-10-27 Sandra Loosemore <sandra@codesourcery.com>
317 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Fix
320 2018-10-26 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
322 * doc/extend.texi (PowerPC builtins): Fix __builtin_unpack_ibm128
323 return type and other typos.
325 2018-10-26 Aaron Sawdey <acsawdey@linux.ibm.com>
327 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Change to
328 a shorter sequence with fewer branches.
329 (emit_final_str_compare_gpr): Ditto.
331 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
333 * config/rs6000/tmmintrin.h: New file.
334 * config.gcc (powerpc*-*-*): Add tmmintrin.h to extra_headers.
336 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
338 * config/rs6000/mmintrin.h: Enable 32bit compilation.
339 * config/rs6000/xmmintrin.h: Likewise.
341 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
343 * config/rs6000/xmmintrin.h (_mm_extract_pi16): Fix for big-endian.
345 2018-10-26 Richard Biener <rguenther@suse.de>
347 * tree-vect-slp.c (vect_mark_slp_stmts): Add visited hash_set
349 (vect_mark_slp_stmts_relevant): Likewise.
350 (vect_detect_hybrid_slp_stmts): Likewise.
351 (vect_bb_slp_scalar_cost): Likewise.
352 (vect_remove_slp_scalar_calls): Likewise.
354 2018-10-26 Jan Hubicka <jh@suse.cz>
356 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix recursion.
357 (warn_types_mismatch): Fix walk of DECL_NAME.
358 (odr_types_equivalent_p): Fix overactive assert.
360 2018-10-26 Richard Biener <rguenther@suse.de>
362 PR tree-optimization/87105
363 * tree-vectorizer.h (_slp_tree::refcnt): New member.
364 * tree-vect-slp.c (vect_free_slp_tree): Decrement and honor
366 (vect_create_new_slp_node): Initialize refcnt to one.
368 (scalar_stmts_set_t, bst_fail): Remove.
369 (vect_build_slp_tree_2): Add bst_map argument and adjust calls.
370 (vect_build_slp_tree): Add bst_map argument and lookup
371 already created SLP nodes.
372 (vect_print_slp_tree): Handle a SLP graph, print SLP node
374 (vect_slp_rearrange_stmts): Handle a SLP graph.
375 (vect_analyze_slp_instance): Adjust and free SLP nodes from
376 the CSE map. Fix indenting.
377 (vect_schedule_slp_instance): Add short-cut.
379 2018-10-26 Martin Liska <mliska@suse.cz>
382 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Hash with
383 addr_expr and not with pointers.
385 2018-10-26 Jan Hubicka <jh@suse.cz>
387 * tree.c (free_lang_data_in_type): Only check main variants.
388 * ipa-devirt.c (warn_odr): Make static.
389 (types_same_for_odr): Drop strict variant.
390 (types_odr_comparable): Likewise.
391 (odr_or_derived_type_p): Look for main variants.
392 (odr_name_hasher::equal): Cleanup comment.
393 (odr_subtypes_equivalent): Add warn and warned arguments; check main
395 (type_variants_equivalent_p): break out from ...
396 (odr_types_equivalent): ... here; go for main variants where needed.
397 (warn_odr): ... here; turn static.
398 (warn_types_mismatch): Compare mangled names of main variants.
399 * ipa-utils.h (types_odr_comparable): Drop strict parameter.
400 (type_with_linkage_p): Sanity check that we look at main variant.
401 * lto.c (lto_read_decls): Only consider main variant to be ODR type.
402 * tree.h (types_same_for_odr): Drop strict argument.
404 2018-10-26 Richard Biener <rguenther@suse.de>
406 PR tree-optimization/87746
407 * tree-vect-data-refs.c (vect_update_misalignment_for_peel):
408 Simplify and fix WRT strided store groups with size not
409 equal to step in element count.
410 (vect_analyze_group_access_1): Dump the whole group.
412 2018-10-25 Carl Love <cel@us.ibm.com>
414 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_VSCEDPGT,
415 P9V_BUILTIN_VEC_VSCEDPLT, P9V_BUILTIN_VEC_VSCEDPEQ,
416 P9V_BUILTIN_VEC_VSCEDPUO): Rename base overloaded name. Add quad
417 precicion entry for each overloaded builtin.
418 * config/rs6000/rs6000-builtin.def (VSCEDPGT, VSCEDPLT, VSCEDPEQ,
419 VSCEDPUO): Rename overloaded name.
420 (VSCEDPGT, VSCEQPGT, VSCEDPLT, VSCEQPLT, VSCEDPEQ, VSCEQPEQ,
421 VSCEDPUO, VSCEQPUO): Add defitions for overloaded builtins.
422 * config/rs6000/vsx.md (xscmpexpqp_<code>_<mode>): Add
423 define_expand for xscmpexqp instruction.
424 (*xscmpexpqp): Add define_insn for the xscmpexqp instruction.
426 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
427 Jinsong Ji <jji@us.ibm.com>
429 * config/rs6000/emmintrin.h (_mm_slli_epi16): Replace deprecated
430 function with vec_sl.
431 (_mm_slli_epi32): Likewise.
432 (_mm_slli_epi64): Likewise.
433 (_mm_srai_epi16): Replace deprecated function with vec_sra.
434 (_mm_srai_epi32): Likewise.
435 (_mm_srli_epi16): Replace deprecated function with vec_sr.
436 (_mm_srli_epi32): Likewise.
437 (_mm_srli_epi64): Likewise.
438 (_mm_sll_epi16): Replace deprecated function with vec_sl.
439 (_mm_sll_epi32): Likewise.
440 (_mm_sll_epi64): Likewise.
441 (_mm_sra_epi16): Replace deprecated function with vec_sra.
442 (_mm_sra_epi32): Likewise.
443 (_mm_srl_epi16): Replace deprecated function with vec_sr.
444 (_mm_srl_epi32): Likewise.
445 (_mm_srl_epi64): Likewise.
447 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
448 Jinsong Ji <jji@us.ibm.com>
450 * gcc/config/rs6000/emmintrin.h (_mm_sll_epi16): Replace
451 comparison operators with vec_cmp* for compatibility due to
452 unfortunate history; clean up formatting and use types more
454 (_mm_sll_epi32): Likewise.
455 (_mm_sll_epi64): Likewise.
456 (_mm_srl_epi16): Likewise.
457 (_mm_srl_epi32): Likewise.
458 (_mm_srl_epi64): Likewise.
460 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
461 Jinsong Ji <jji@us.ibm.com>
463 * config/rs6000/emmintrin.h (_mm_sll_epi64): Remove wrong cast.
464 * config/rs6000/xmmintrin.h (_mm_min_ps): Change m's type to
465 __vector __bool int. Use vec_cmpgt in preference to deprecated
466 function vec_vcmpgtfp.
467 (_mm_max_ps): Likewise.
469 2018-10-25 Jeff Law <law@redhat.com>
471 * config/rl78/rl78.c (insn_ok_now): Always re-recognize the insn
474 2018-10-25 Martin Sebor <msebor@redhat.com>
476 * doc/extend.texi (aligned): Expand attribute description.
477 (Alignment): Rename section. Discuss function arguments.
479 2018-10-25 Jan Hubicka <jh@suse.cz>
481 * ipa-devirt.c (main_odr_variant): Remove.
482 (hash_odr_name, types_same_for_odr, types_odr_comparable,
483 odr_name_hasher::equal, odr_subtypes_equivalent_p):
484 Drop use of main_odr_variant.
485 (add_type_duplicate): Silence confused warnings on integer types.
486 (get_odr_type): Always look for main variant.
487 (register_odr_type): Simplify.
489 2018-10-25 Richard Biener <rguenther@suse.de>
491 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses):
492 Initialize ng to silence error with release checking bootstrap.
494 2018-10-25 Richard Biener <rguenther@suse.de>
496 * tree-if-conv.c: Include tree-ssa-sccvn.h.
497 (tree_if_conversion): Run CSE on the if-converted loop body.
499 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
501 * config/s390/constraints.md (ZL): New constraint.
502 * config/s390/s390.c (legitimate_pic_operand_p): Accept LARL
504 * config/s390/s390.md (movdi_larl): Remove.
505 (movdi_64): Add the LARL alternative.
507 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
510 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
511 (RTX_CODE_HWINT_P): New macro.
512 (rtx_code_size): Use RTX_CODE_HWINT_P ().
514 2018-10-25 Jan Hubicka <jh@suse.cz>
516 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
519 2018-10-25 Richard Biener <rguenther@suse.de>
521 PR tree-optimization/87665
522 PR tree-optimization/87745
523 * tree-vectorizer.h (get_earlier_stmt): Remove.
524 (get_later_stmt): Pick up UID from the original non-pattern stmt.
526 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
528 * options.texi (Deprecated): Move list to Var section.
530 2018-10-24 Bill Schmidt <wschmidt@linux.ibm.com>
531 Jinsong Ji <jji@us.ibm.com>
533 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Change deprecated
534 __vector long to __vector long long.
535 (_mm_cvtpd_ps): Likewise.
536 (_mm_cvttpd_epi32): Likewise.
537 (_mm_cvtpi32_pd): Likewise.
538 (_mm_unpackhi_epi64): Likewise.
539 (_mm_unpacklo_epi64): Likewise.
541 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
543 PR rtl-optimization/87720
544 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
546 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
548 * gimple-ssa-isolate-paths.c
549 (find_implicit_erroneous_behavior): Do not change code if the
550 pass is running for warnings only.
551 (find_explicit_erroneous_behavior): Likewise.
553 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
555 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
556 Define as rs6000_mangle_decl_assembler_name.
557 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
558 long double to IEEE long double, switch the names of the long
559 double built-in functions to be <func>f128 instead of <func>l.
561 2018-10-24 Martin Sebor <msebor@redhat.com>
563 * doc/extend.texi (nonnull): List no-argument form. Reference
564 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
566 2018-10-24 Richard Biener <rguenther@suse.de>
568 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
570 2018-10-24 Martin Liska <mliska@suse.cz>
572 PR tree-optimization/84436
573 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
575 (switch_conversion::contains_linear_function_p): New.
576 (switch_conversion::build_one_array): Support linear
577 transformation on input.
578 * tree-switch-conversion.h (struct switch_conversion): Add
579 contains_linear_function_p declaration.
581 2018-10-24 Richard Biener <rguenther@suse.de>
583 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
584 if its argument is CONSTANT_CLASS_P.
586 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
588 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
589 it is wrong for forward declarations.
591 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
593 * config/s390/s390.c (s390_check_qrst_address): Add the missing
594 SYMBOL_REF_P () check.
596 2018-10-24 Richard Biener <rguenther@suse.de>
598 PR tree-optimization/87105
599 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
601 (vect_analyze_data_ref_accesses): Handle duplicate loads and
602 stores by splitting the affected group after the fact.
603 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
604 fail the SLP build because of size constraints.
606 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
608 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
609 * configure: Regenerate.
610 * config.in: Regenerate.
611 * varasm.c (mergeable_string_section): Use readonly_data_section
612 if linker doesn't support SHF_MERGE with alignment > 8.
613 (mergeable_constant_section): Likewise.
615 2018-10-24 Richard Biener <rguenther@suse.de>
617 PR tree-optimization/84013
618 * tree-ssa-structalias.c (struct msdi_data): New struct for
619 marshalling data to walk_stmt_load_store_ops.
620 (maybe_set_dependence_info): Refactor as callback for
621 walk_stmt_load_store_ops.
622 (compute_dependence_clique): Set restrict info on all stmt kinds.
624 2018-10-24 Martin Liska <mliska@suse.cz>
626 * cgraph.c (cgraph_node::dump):
627 Remove reduntant dumps and make tp_first_run dump more compact.
629 2018-10-24 Richard Biener <rguenther@suse.de>
631 PR tree-optimization/87665
632 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
635 2018-10-12 Jeff Law <law@redhat.com>
637 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
640 2018-10-23 Richard Biener <rguenther@suse.de>
642 * tree-vrp.c (add_assert_info): Guard dump_printf with
644 * gimple-ssa-evrp-analyze.c
645 (evrp_range_analyzer::record_ranges_from_incoming_edge):
646 Use value_range::ignore_equivs_equal_p.
648 2018-10-23 Richard Biener <rguenther@suse.de>
650 PR tree-optimization/87105
651 PR tree-optimization/87608
652 * passes.def (pass_all_early_optimizations): Add early phi-opt
654 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
655 addition to debug stmts.
656 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
657 and abs replacement early.
658 * tree-cfg.c (gimple_empty_block_p): Likewise.
660 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
663 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
664 specified to configure.
665 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
667 2018-10-23 Richard Biener <rguenther@suse.de>
669 PR tree-optimization/87700
670 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
672 2018-10-23 Jakub Jelinek <jakub@redhat.com>
675 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
676 second argument from __mmask16 to __mmask8.
677 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
678 _mm_mask_packs_epi32): Likewise.
679 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
681 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
683 2018-10-23 Richard Biener <rguenther@suse.de>
685 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
687 2018-10-23 Richard Biener <rguenther@suse.de>
689 PR tree-optimization/86144
690 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
693 2018-10-23 Richard Biener <rguenther@suse.de>
695 PR tree-optimization/87693
696 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
697 the case we do not find the taken edge.
699 2018-10-22 Bill Schmidt <wschmidt@linux.ibm.com>
700 Jinsong Ji <jji@us.ibm.com>
702 * config/rs6000/emmintrin.h (_MM_SHUFFLE2): Comment cleanup.
703 (_mm_store_pd): Use unaligned vector type for pointer cast.
704 (_mm_maskmoveu_si128): Likewise.
705 * config/rs6000/xmmintrin.h (__m128_u): New typedef.
706 (_mm_store_ps): Use unaligned vector type for pointer cast.
708 2018-10-22 Paul Koning <ni1d@arrl.net>
710 * symtab.c (symtab_node::increase_alignment): Correct max
713 2018-10-22 Yury Gribov <tetra2005@gmail.com>
715 PR tree-optimization/87633
716 * match.pd: Do not generate unordered integer comparisons.
718 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
720 PR rtl-optimization/87600
721 * combine.c: Add include of expr.h.
722 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
723 register to a pseudo.
724 (make_more_copies): New function, add a copy to a new pseudo after
725 the moves from hard registers into pseudos.
726 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
727 later. Call make_more_copies.
729 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
731 * lra-constraints.c (process_alt_operands): New local array,
732 matching_early_clobber. Check matching_early_clobber before
733 decrementing reject, and set matching_early_clobber after.
735 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
738 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
739 call output_addr_const and hope for the best.
741 2018-10-22 Richard Biener <rguenther@suse.de>
743 * gimple-ssa-evrp-analyze.c
744 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
745 smarter about what ranges to use.
746 * tree-vrp.c (add_assert_info): Dump here.
747 (register_edge_assert_for_2): Instead of here at multiple but
750 * gcc.dg/tree-ssa/evrp12.c: New testcase.
751 * gcc.dg/predict-6.c: Adjust.
752 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
753 * gcc.dg/tree-ssa/vrp02.c: Likewise.
754 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
756 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
757 Richard Biener <rguenther@suse.de>
759 * bitmap.h: Update data structure documentation, including a
760 description of bitmap views as either linked-lists or splay trees.
761 (struct bitmap_element_def): Update comments for splay tree bitmaps.
762 (struct bitmap_head_def): Likewise.
763 (bitmap_list_view, bitmap_tree_view): New prototypes.
764 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
766 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
767 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
768 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
769 released bitmap element here.
770 (bitmap_element_free): Remove.
771 (bitmap_elt_clear_from): Work on splay tree bitmaps.
772 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
773 this function similar ones such that linked-list bitmap implementation
774 functions are grouped.
775 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
776 and moved for grouping.
777 (bitmap_list_insert_element_after): Renamed from
778 bitmap_elt_insert_after, and moved for grouping.
779 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
780 (bitmap_tree_link_left, bitmap_tree_link_right,
781 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
782 bitmap_tree_link_element, bitmap_tree_unlink_element,
783 bitmap_tree_find_element): New functions for splay-tree bitmap
785 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
786 Renamed and moved, see above entries.
787 (bitmap_tree_listify_from): New function to convert part of a splay
788 tree bitmap to a linked-list bitmap.
789 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
790 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
791 (bitmap_find_bit): Remove.
792 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
793 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
794 Handle splay tree bitmaps.
795 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
796 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
797 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
798 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
799 bitmap_intersect_compl_p, bitmap_ior_and_compl,
800 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
801 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
802 corresponding changes to use linked-list specific bitmap_element
803 manipulation functions as applicable for efficiency.
804 (bitmap_tree_to_vec): New function.
805 (debug_bitmap_elt_file): New function split out from ...
806 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
807 (bitmap_print): Likewise.
809 PR tree-optimization/63155
810 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
812 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
815 2018-10-22 Martin Liska <mliska@suse.cz>
817 PR tree-optimization/87686
819 2018-08-29 Martin Liska <mliska@suse.cz>
821 * tree-switch-conversion.c (switch_conversion::expand):
822 Strenghten assumption about gswitch statements.
824 2018-10-22 Martin Liska <mliska@suse.cz>
826 * ipa-icf.c (sem_item::compare_attributes): Remove.
827 (sem_item::compare_referenced_symbol_properties): Use
828 attribute_list_equal instead.
829 (sem_function::equals_wpa): Likewise.
830 * ipa-icf.h: Remove compare_attributes.
832 2018-10-22 Richard Biener <rguenther@suse.de>
835 * mem-stats.h (mem_usage::operator==): Fix pasto.
837 2018-10-22 Richard Biener <rguenther@suse.de>
839 PR tree-optimization/87640
840 * tree-vrp.c (set_value_range_with_overflow): Decompose
842 (extract_range_from_binary_expr_1): Adjust.
844 2018-10-22 Martin Jambor <mjambor@suse.cz>
846 * tree-eh.h (stmt_could_throw_p): Add function parameter.
847 (stmt_can_throw_external): Likewise.
848 (stmt_can_throw_internal): Likewise.
849 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
850 (lower_eh_constructs_2): Likewise.
851 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
852 (stmt_can_throw_external): Likewise.
853 (stmt_can_throw_internal): Likewise.
854 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
855 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
856 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
857 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
858 (pass_lower_eh_dispatch::execute): Pass cfun to
859 stmt_can_throw_external.
860 (cleanup_empty_eh): Likewise.
861 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
862 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
863 stmt_can_throw_external instead of pushing it to cfun.
864 (symbol_table::create_edge): Likewise.
865 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
866 stmt_can_throw_internal.
867 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
868 to stmt_could_throw_p.
869 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
870 stmt_can_throw_internal.
871 (pass_store_merging::execute): Likewise.
872 * gimple-ssa-strength-reduction.c
873 (find_candidates_dom_walker::before_dom_children): Pass cfun to
875 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
876 stmt_can_throw_internal.
877 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
878 to stmt_can_throw_external.
879 (check_stmt): Pass cfun to stmt_could_throw_p.
880 (check_stmt): Pass cfun to stmt_can_throw_external.
881 (pass_nothrow::execute): Likewise.
882 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
883 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
884 stmt_can_throw_internal.
885 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
886 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
887 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
888 * tree-complex.c (expand_complex_libcall): Pass cfun to
889 stmt_could_throw_p and to stmt_can_throw_internal.
890 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
891 * tree-inline.c (copy_edges_for_bb): Likewise.
892 (maybe_move_debug_stmts_to_successors): Likewise.
893 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
895 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
896 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
897 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
898 stmt_can_throw_internal.
899 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
900 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
902 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
903 stmt_can_throw_internal.
904 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
905 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
907 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
908 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
909 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
910 (convert_mult_to_fma_1): Likewise.
911 (convert_to_divmod): Likewise.
912 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
913 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
914 * tree-ssa-propagate.c
915 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
916 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
917 (maybe_optimize_range_tests): Likewise.
918 (linearize_expr_tree): Likewise.
919 (reassociate_bb): Likewise.
920 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
921 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
922 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
923 (handle_char_store): Likewise.
924 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
925 stmt_can_throw_internal.
926 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
928 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
929 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
930 (vectorizable_simd_clone_call): Likewise.
931 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
932 (gimple_stringop_fixed_value): Likewise.
934 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
936 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
937 literal pool references.
938 (s390_check_qrst_address): Adapt to the new behavior of
939 s390_loadrelative_operand_p ().
941 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
944 * config/i386/sse.md (*andnot<mode>3_bcst): New.
946 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
949 * config/i386/sse.md (*<code><mode>3_bcst): New.
951 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
954 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
955 V4DI, V16SI and V8DI.
956 (*sub<mode>3<mask_name>_bcst): New.
957 (*add<mode>3<mask_name>_bcst): Likewise.
959 2018-10-21 Bill Schmidt <wschmidt@linux.ibm.com>
960 Jinsong Ji <jji@us.ibm.com>
962 * config/rs6000/emmintrin.h (_mm_movemask_pd): Replace __vector
963 __m64 with __vector unsigned long long for compatibility.
964 (_mm_movemask_epi8): Likewise.
965 * config/rs6000/xmmintrin.h (_mm_cvtps_pi32): Likewise.
966 (_mm_cvttps_pi32): Likewise.
967 (_mm_cvtpi32_ps): Likewise.
968 (_mm_cvtps_pi16): Likewise.
969 (_mm_loadh_pi): Likewise.
970 (_mm_storeh_pi): Likewise.
971 (_mm_movehl_ps): Likewise.
972 (_mm_movelh_ps): Likewise.
973 (_mm_loadl_pi): Likewise.
974 (_mm_storel_pi): Likewise.
975 (_mm_movemask_ps): Likewise.
976 (_mm_shuffle_pi16): Likewise.
978 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
981 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
982 __builtin_ia32_vfnmsubpd512_mask.
983 (_mm512_mask_fnmsub_round_pd): Likewise.
984 (_mm512_fnmsub_pd): Likewise.
985 (_mm512_mask_fnmsub_pd): Likewise.
986 (_mm512_maskz_fnmsub_round_pd): Use
987 __builtin_ia32_vfnmsubpd512_maskz.
988 (_mm512_maskz_fnmsub_pd): Likewise.
989 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
990 (_mm512_mask_fnmsub_round_ps): Likewise.
991 (_mm512_fnmsub_ps): Likewise.
992 (_mm512_mask_fnmsub_ps): Likewise.
993 (_mm512_maskz_fnmsub_round_ps): Use
994 __builtin_ia32_vfnmsubps512_maskz.
995 (_mm512_maskz_fnmsub_ps): Likewise.
996 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
997 __builtin_ia32_vfnmsubpd256_mask.
998 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
999 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
1000 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
1001 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
1002 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
1003 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
1004 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
1005 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
1006 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
1007 __builtin_ia32_vfnmsubpd.
1008 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
1009 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
1010 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
1011 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
1012 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
1013 * config/i386/i386-builtin.def: Add
1014 __builtin_ia32_vfnmsubpd256_mask,
1015 __builtin_ia32_vfnmsubpd256_maskz,
1016 __builtin_ia32_vfnmsubpd128_mask,
1017 __builtin_ia32_vfnmsubpd128_maskz,
1018 __builtin_ia32_vfnmsubps256_mask,
1019 __builtin_ia32_vfnmsubps256_maskz,
1020 __builtin_ia32_vfnmsubps128_mask,
1021 __builtin_ia32_vfnmsubps128_maskz,
1022 __builtin_ia32_vfnmsubpd512_mask,
1023 __builtin_ia32_vfnmsubpd512_maskz,
1024 __builtin_ia32_vfnmsubps512_mask,
1025 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
1026 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
1027 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
1028 __builtin_ia32_vfnmsubpd256.
1029 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
1030 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
1031 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
1033 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
1035 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
1037 (fmai_vmfnmsub_<mode><round_name>): Likewise.
1039 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1042 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
1043 __builtin_ia32_vfnmaddpd512_mask.
1044 (_mm512_mask_fnmadd_round_pd): Likewise.
1045 (_mm512_fnmadd_pd): Likewise.
1046 (_mm512_mask_fnmadd_pd): Likewise.
1047 (_mm512_maskz_fnmadd_round_pd): Use
1048 __builtin_ia32_vfnmaddpd512_maskz.
1049 (_mm512_maskz_fnmadd_pd): Likewise.
1050 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
1051 (_mm512_mask_fnmadd_round_ps): Likewise.
1052 (_mm512_fnmadd_ps): Likewise.
1053 (_mm512_mask_fnmadd_ps): Likewise.
1054 (_mm512_maskz_fnmadd_round_ps): Use
1055 __builtin_ia32_vfnmaddps512_maskz.
1056 (_mm512_maskz_fnmadd_ps): Likewise.
1057 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
1058 __builtin_ia32_vfnmaddpd256_mask.
1059 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
1060 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
1061 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
1062 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
1063 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
1064 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
1065 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
1066 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
1067 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
1068 __builtin_ia32_vfnmaddpd.
1069 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
1070 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
1071 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
1072 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
1073 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
1074 * config/i386/i386-builtin.def: Add
1075 __builtin_ia32_vfnmaddpd256_mask,
1076 __builtin_ia32_vfnmaddpd256_maskz,
1077 __builtin_ia32_vfnmaddpd128_mask,
1078 __builtin_ia32_vfnmaddpd128_maskz,
1079 __builtin_ia32_vfnmaddps256_mask,
1080 __builtin_ia32_vfnmaddps256_maskz,
1081 __builtin_ia32_vfnmaddps128_mask,
1082 __builtin_ia32_vfnmaddps128_maskz,
1083 __builtin_ia32_vfnmaddpd512_mask,
1084 __builtin_ia32_vfnmaddpd512_maskz,
1085 __builtin_ia32_vfnmaddps512_mask,
1086 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
1087 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
1088 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
1089 __builtin_ia32_vfnmaddpd256.
1090 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
1091 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
1092 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
1094 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
1096 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
1098 (fmai_vmfnmadd_<mode><round_name>): Likewise.
1100 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1103 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
1104 __builtin_ia32_vfmsubpd512_mask.
1105 (_mm512_mask_fmsub_round_pd): Likewise.
1106 (_mm512_fmsub_pd): Likewise.
1107 (_mm512_mask_fmsub_pd): Likewise.
1108 (_mm512_maskz_fmsub_round_pd): Use
1109 __builtin_ia32_vfmsubpd512_maskz.
1110 (_mm512_maskz_fmsub_pd): Likewise.
1111 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
1112 (_mm512_mask_fmsub_round_ps): Likewise.
1113 (_mm512_fmsub_ps): Likewise.
1114 (_mm512_mask_fmsub_ps): Likewise.
1115 (_mm512_maskz_fmsub_round_ps): Use
1116 __builtin_ia32_vfmsubps512_maskz.
1117 (_mm512_maskz_fmsub_ps): Likewise.
1118 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
1119 __builtin_ia32_vfmsubpd256_mask.
1120 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
1121 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
1122 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
1123 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
1124 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
1125 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
1126 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
1127 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
1128 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
1129 __builtin_ia32_vfmsubpd.
1130 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
1131 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
1132 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
1133 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
1134 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
1135 * config/i386/i386-builtin.def: Add
1136 __builtin_ia32_vfmsubpd256_mask,
1137 __builtin_ia32_vfmsubpd256_maskz,
1138 __builtin_ia32_vfmsubpd128_mask,
1139 __builtin_ia32_vfmsubpd128_maskz,
1140 __builtin_ia32_vfmsubps256_mask,
1141 __builtin_ia32_vfmsubps256_maskz,
1142 __builtin_ia32_vfmsubps128_mask,
1143 __builtin_ia32_vfmsubps128_maskz,
1144 __builtin_ia32_vfmsubpd512_mask,
1145 __builtin_ia32_vfmsubpd512_maskz,
1146 __builtin_ia32_vfmsubps512_mask,
1147 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
1148 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
1149 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
1150 __builtin_ia32_vfmsubpd256.
1151 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
1152 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
1153 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
1155 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
1157 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
1159 (fmai_vmfmsub_<mode><round_name>): Likewise.
1161 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1163 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
1164 Remove plus. Renamed to ...
1165 (*sub<mode>3<mask_name>_bcst): This.
1166 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
1167 (*add<mode>3<mask_name>_bcst): This.
1169 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1172 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
1174 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1177 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
1178 (_mm_or_epi32): Likewise.
1179 (_mm256_xor_epi32): Likewise.
1180 (_mm_xor_epi32): Likewise.
1181 (_mm256_or_epi64): Likewise.
1182 (_mm_or_epi64): Likewise.
1183 (_mm256_xor_epi64): Likewise.
1184 (_mm_xor_epi64): Likewise.
1186 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
1189 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
1191 2018-10-20 Jakub Jelinek <jakub@redhat.com>
1194 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
1196 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
1198 * doc/ux.texi: Move @section directly after @node.
1200 2018-10-19 Jakub Jelinek <jakub@redhat.com>
1204 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
1205 depend closely nested inside of loop with ordered clause with
1208 2018-10-19 David Malcolm <dmalcolm@redhat.com>
1210 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
1211 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
1212 * doc/ux.texi: New file.
1214 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
1216 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
1217 be the first CR field allocated.
1219 2018-10-19 Richard Biener <rguenther@suse.de>
1222 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
1223 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
1225 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
1228 * config/i386/sse.md
1229 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
1230 (*add<mode>3<mask_name>_bcst_2): Likewise.
1232 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
1234 * config/i386/sse.md
1235 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
1236 Replace nonimmediate_operand with register_operand.
1237 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
1239 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
1242 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
1244 PR rtl-optimization/87596
1245 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
1246 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
1247 for instructions in FROM..TO range.
1249 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
1251 * cfgexpand.c (expand_one_var): Use specific wording in error message
1252 for non-local frame variables.
1253 * stor-layout.c (layout_decl): Do not issue a warning for them.
1255 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
1257 * haifa-sched.c (priority): Add force_recompute parameter.
1258 (apply_replacement): Call priority () with force_recompute = true.
1259 (restore_pattern): Likewise.
1261 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1263 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
1264 HOST_BITS_PER_WIDE_INT.
1265 (test_vector_ops_duplicate): Likewise.
1267 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1270 * config/i386/sse.md (VF_AVX512): New.
1271 (avx512bcst): Likewise.
1272 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
1274 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
1276 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
1279 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
1281 * doc/invoke.texi (-dumpversion): Improve grammar.
1282 (-dumpfullversion): Make more consistent with -dumpversion.
1284 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
1286 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
1287 Set JUMP_LABEL to the jump insn.
1288 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
1289 Predict emitted jump and add label to jump insn.
1291 2018-10-18 David Malcolm <dmalcolm@redhat.com>
1293 PR tree-optimization/87562
1294 * input.c (get_substring_ranges_for_loc): Use
1295 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
1296 getting the linemap for the endpoint. Verify that it's either
1297 in the same linemap as the start point's spelling location, or
1298 at least in the same file.
1300 2018-10-18 Richard Biener <rguenther@suse.de>
1302 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
1303 feed width-specific load/store costs through ix86_vec_cost.
1304 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
1305 (k8_cost): Likewise.
1306 (bdver_cost): Likewise.
1307 (znver1_cost): Likewise.
1308 (btver1_cost): Likewise.
1309 (btver2_cost): Likewise.
1311 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1313 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
1314 to simplify subreg of vec_merge.
1316 2018-10-18 Richard Biener <rguenther@suse.de>
1318 * config/i386/i386.c: Fix costing of vector FMA.
1320 2018-10-18 Richard Biener <rguenther@suse.de>
1322 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
1324 (ix86_builtin_vectorization_cost): For vec_construct properly
1325 cost insertion into SSE regs.
1326 (...): Adjust calls to ix86_vec_cost.
1328 2018-10-18 Richard Biener <rguenther@suse.de>
1332 2018-02-07 Richard Biener <rguenther@suse.de>
1334 PR tree-optimization/84204
1335 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
1338 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1341 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
1343 (test_vector_ops_duplicate): Add test for a scalar subreg of a
1344 VEC_MERGE of a VEC_DUPLICATE.
1346 2018-10-17 Joseph Myers <joseph@codesourcery.com>
1348 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
1349 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
1350 * doc/standards.texi (C Language): Document C2X.
1351 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
1352 (rl78_option_override): Handle "GNU C2X" language name.
1354 2018-10-17 Joseph Myers <joseph@codesourcery.com>
1356 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
1357 Document C17 as published in 2018.
1359 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
1362 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
1363 bail out if both sides do not have the same storage order.
1365 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
1367 * bitmap.c (bitmap_head::dump): New.
1368 * bitmap.h (bitmap_head): Add dump().
1369 * gimple-ssa-evrp-analyze.c
1370 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
1371 (evrp_range_analyzer::set_ssa_range_info): Same.
1372 (evrp_range_analyzer::record_ranges_from_phis): Same.
1373 (evrp_range_analyzer::record_ranges_from_stmt): Same.
1374 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
1375 * gimple-ssa-sprintf.c (get_int_range): Same.
1376 (format_integer): Same.
1377 (sprintf_dom_walker::handle_gimple_call): Same.
1378 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
1379 (ipcp_vr_lattice::top_p): Same.
1380 (ipcp_vr_lattice::bottom_p): Same.
1381 (ipcp_vr_lattice::set_to_bottom): Same.
1382 (ipa_vr_operation_and_type_effects): Same.
1383 (propagate_vr_across_jump_function): Same.
1384 (ipcp_store_vr_results): Same.
1385 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
1386 (ipa_print_node_jump_functions_for_edge): Same.
1387 (ipa_get_value_range): Same.
1388 (ipa_compute_jump_functions_for_edge): Same.
1389 (ipa_write_jump_function): Same.
1390 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
1391 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
1393 * vr-values.c (set_value_range_to_nonnegative): Same.
1394 (set_value_range_to_truthvalue): Same.
1395 (vr_values::get_value_range): Same.
1396 (vr_values::set_defs_to_varying): Same.
1397 (vr_values::update_value_range): Same.
1398 (symbolic_range_based_on_p): Same.
1399 (vr_values::op_with_boolean_value_range_p): Same.
1400 (vr_values::extract_range_for_var_from_comparison_expr): Same.
1401 (vr_values::extract_range_from_ssa_name): Same.
1402 (vr_values::extract_range_from_binary_expr): Same.
1403 (vr_values::extract_range_from_unary_expr): Same.
1404 (vr_values::extract_range_from_cond_expr): Same.
1405 (vr_values::extract_range_from_comparison): Same.
1406 (vr_values::check_for_binary_op_overflow): Same.
1407 (vr_values::extract_range_basic): Same.
1408 (vr_values::extract_range_from_assignment): Same.
1409 (compare_ranges): Same.
1410 (compare_range_with_value): Same.
1411 (vr_values::adjust_range_with_scev): Same.
1412 (vrp_valueize): Same.
1413 (vrp_valueize_1): Same.
1414 (vr_values::get_vr_for_comparison): Same.
1415 (vr_values::compare_name_with_value): Same.
1416 (vr_values::compare_names): Same.
1417 (vr_values::vrp_evaluate_conditional): Same.
1418 (find_case_label_ranges): Same.
1419 (vr_values::vrp_visit_switch_stmt): Same.
1420 (vr_values::extract_range_from_phi_node): Same.
1421 (vr_values::simplify_div_or_mod_using_ranges): Same.
1422 (vr_values::simplify_bit_ops_using_ranges): Same.
1423 (test_for_singularity): Same.
1424 (range_fits_type_p): Same.
1425 (vr_values::simplify_cond_using_ranges_1): Same.
1426 (vr_values::simplify_switch_using_ranges): Same.
1427 (vr_values::simplify_float_conversion_using_ranges): Same.
1428 (vr_values::two_valued_val_range_p): Same.
1429 (vr_values::add_equivalence): Move to value_range::equiv_add.
1430 * vr-values.h (vr_values::add_equivalence): Remove.
1431 (VR_INITIALIZER): Remove.
1432 * tree-vrp.c (value_range::set): New.
1433 (value_range::equiv_add): New.
1434 (value_range::value_range): New.
1435 (value_range::deep_copy): New.
1436 (value_range::check): New.
1437 (value_range::equal_p): New.
1438 (value_range::ignore_equivs_equal_p): New.
1439 (value_range::operator==): New.
1440 (value_range::operator!=): New.
1441 (value_range::symbolic_p): New.
1442 (value_range::numeric_p): New.
1443 (value_range::set_undefined): New.
1444 (value_range::set_varying): New.
1445 (value_range::may_contain_p): New.
1446 (value_range::equiv_clear): New.
1447 (value_range::singleton_p): New.
1448 (value_range::intersect): New.
1449 (value_range::dump): New.
1450 (value_range::set_and_canonicalize): New.
1451 (set_value_range): Adjust for value_range API.
1452 (set_value_range_to_undefined): Same.
1453 (set_value_range_to_varying): Same.
1454 (set_and_canonicalize_value_range): Same.
1455 (set_value_range_to_nonnull): Same.
1456 (set_value_range_to_null): Same.
1457 (range_is_null): Same.
1458 (range_is_nonnull): Same.
1459 (range_int_cst_p): Same.
1460 (range_int_cst_singleton_p): Same.
1461 (symbolic_range_p): Same.
1462 (range_includes_zero_p): Same.
1463 (value_range_constant_singleton): Same.
1464 (vrp_set_zero_nonzero_bits): Same.
1465 (ranges_from_anti_range): Same.
1466 (extract_range_into_wide_ints): Same.
1467 (extract_range_from_multiplicative_op): Same.
1468 (set_value_range_with_overflow): Same.
1469 (extract_range_from_binary_expr_1): Same.
1470 (extract_range_from_unary_expr): Same.
1471 (dump_value_range): Same.
1472 (debug_value_range): Same.
1473 (vrp_prop::check_array_ref): Same.
1474 (vrp_prop::check_mem_ref): Same.
1475 (vrp_prop::vrp_initialize): Same.
1476 (vrp_prop::visit_stmt): Same.
1477 (intersect_ranges): Same.
1478 (vrp_prop::visit_phi): Same.
1479 (vrp_prop::vrp_finalize): Same.
1480 (determine_value_range_1): Same.
1481 (determine_value_range): Same.
1482 (vrp_intersect_ranges_1): Rename to...
1483 (vrp_intersect_1): this.
1484 (vrp_intersect_ranges): Rename to...
1485 (value_range::intersect_helper): ...this.
1486 (vrp_meet_1): Rename to...
1487 (value_range::union_helper): ...this.
1488 (vrp_meet): Rename to...
1489 (value_range::union_): ...this.
1490 (copy_value_range): Remove.
1491 * tree-vrp.h (struct value_range): Rewrite into a proper class.
1492 (value_range::vrtype): New.
1493 (value_range::type): New.
1494 (value_range::equiv): New.
1495 (value_range::min): New.
1496 (value_range::max): New.
1497 (value_range::varying_p): New.
1498 (value_range::undefined_p): New.
1499 (value_range::null_p): New.
1500 (value_range::equiv_add): New.
1501 (copy_value_range): Remove.
1503 2018-10-17 David Malcolm <dmalcolm@redhat.com>
1505 * Makefile.in (SELFTEST_TARGETS): New.
1506 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
1507 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
1508 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
1510 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
1511 (selftest-c++-gdb, selftest-c++-valgrind): Move to
1513 * configure: Regenerate.
1514 * configure.ac (selftest_languages): New.
1516 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
1518 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
1519 overflow wraps argument.
1520 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
1521 wide_int_range_multiplicative_op.
1522 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
1523 overflow wraps argument.
1524 (wide_int_range_multiplicative_op): Same.
1525 (wide_int_range_lshift): Same.
1526 (wide_int_range_div): Same.
1527 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
1528 (wide_int_range_lshift): Same.
1529 (wide_int_range_div): Same.
1531 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
1533 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
1534 use sign as argument.
1535 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
1536 wide_int_range_shift_undefined_p.
1538 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1540 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
1542 (@despeculate_copy<ALLI_TI:mode>): ... This.
1543 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
1546 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1548 * config.gcc: Obsolete *-*-solaris2.10*.
1549 * doc/install.texi (Specific, *-*-solaris2*): Document it.
1551 2018-10-12 Jeff Law <law@redhat.com>
1553 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
1554 reg + sym +- const_int addressing modes.
1556 2018-10-15 David Malcolm <dmalcolm@redhat.com>
1558 * common.opt (fdiagnostics-minimum-margin-width=): New option.
1559 * diagnostic-show-locus.c (layout::layout): Apply the minimum
1561 (layout::start_annotation_line): Only print up to 3 of the
1562 margin character, to avoid touching the left-hand side.
1563 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
1564 minimum margin width, as set by test_diagnostic_context's ctor.
1565 (selftest::test_fixit_insert_containing_newline): Likewise.
1566 (selftest::test_fixit_insert_containing_newline_2): Likewise.
1567 (selftest::test_line_numbers_multiline_range): Clear
1568 dc.min_margin_width.
1569 * diagnostic.c (diagnostic_initialize): Initialize
1571 * diagnostic.h (struct diagnostic_context): Add field
1573 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
1574 * opts.c (common_handle_option): Handle
1575 OPT_fdiagnostics_minimum_margin_width_.
1576 * selftest-diagnostic.c
1577 (selftest::test_diagnostic_context::test_diagnostic_context):
1578 Initialize min_margin_width to 6.
1579 * toplev.c (general_init): Initialize global_dc->min_margin_width.
1581 2018-10-15 David Malcolm <dmalcolm@redhat.com>
1583 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1584 Fix usage of "error_at_rich_loc" in the comment.
1586 2018-10-15 Renlin Li <renlin.li@arm.com>
1589 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
1590 if-conversioned loop when it contains ifn with types not
1591 supported by backend.
1592 * internal-fn.c (expand_direct_optab_fn): Add an assert.
1593 (direct_internal_fn_supported_p): New helper function.
1594 * internal-fn.h (direct_internal_fn_supported_p): Declare.
1596 2018-10-15 Jakub Jelinek <jakub@redhat.com>
1599 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
1600 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
1601 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
1603 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
1605 PR tree-optimization/87022
1606 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
1607 bits in dist vector rather than the first one.
1609 2018-10-15 Richard Biener <rguenther@suse.de>
1612 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
1613 (visit_loadstore): When a used restrict tag escaped verify that
1614 the points-to solution of "other" pointers do not include
1616 (compute_dependence_clique): If a used restrict tag escaped
1617 communicated that down to visit_loadstore.
1619 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
1621 * config/s390/s390.c (s390_expand_vec_init): Force vector element
1622 into reg if it isn't a general operand.
1624 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
1627 * config/i386/sse.md (*vec_dupv2di): Add register source to
1630 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
1633 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
1634 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
1635 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
1636 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
1638 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
1640 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
1641 (notice_args_size): Set it in the current trace if no insn that can
1642 throw internally has been seen yet.
1643 (connect_traces): When connecting args_size between traces, allow the
1644 incoming values not to match if there is an insn setting it before the
1645 first insn that can throw internally; in that case, force the creation
1646 of a CFI note on this latter insn.
1648 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
1650 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
1651 the base class, because using the injected-class-name was not clearly
1652 specified until DR 176.
1654 2018-10-12 Paul Koning <ni1d@arrl.net>
1656 * config/pdp11/pdp11.md (doloop_end): New expander.
1657 (doloop_end_insn): renamed from "doloop_end".
1658 (addqi3): New pattern.
1659 (subqi3): New pattern.
1660 * config/pdp11/predicates.md (incdec_operand): New predicate.
1662 2018-10-12 Yury Gribov <tetra2005@gmail.com>
1665 * real.c (format_helper::can_represent_integral_type_p): New function
1666 * real.h (format_helper::can_represent_integral_type_p): Ditto.
1667 * match.pd: New pattern.
1669 2018-10-12 Alexandre Oliva <oliva@adacore.com>
1671 * configure.ac: Introduce --enable-large-address-aware
1672 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
1673 * doc/install.texi: Document it.
1674 * configure, config.in: Rebuilt.
1675 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
1676 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
1677 (LINK_SPEC): Insert it.
1678 * config/i386/mingw-264.h: Likewise.
1680 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
1682 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
1684 PR rtl-optimization/87600
1685 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
1687 2018-10-12 Paul Koning <ni1d@arrl.net>
1689 * doc/md.texi (doloop_end): Document that the pattern code may
1690 need to check operand mode.
1692 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
1694 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
1695 to zero-extend between int and floating-point registers.
1696 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
1697 ldp into floating-point registers. Add type and arch attributes.
1698 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
1699 Use f_loads for type attribute.
1701 2018-10-11 Martin Sebor <msebor@redhat.com>
1703 * doc/extend.texi (attribute packed): Correct typos.
1705 2018-10-11 Martin Sebor <msebor@redhat.com>
1707 * doc/extend.texi (attribute flatten): Mention interaction with
1710 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
1713 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
1714 Set new_decl virtual flag to zero.
1716 2018-10-11 Martin Sebor <msebor@redhat.com>
1719 * doc/extend.texi (attribute format_arg): Discuss using multiple
1720 attributes on a single function.
1722 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
1724 PR tree-optimization/86829
1725 * match.pd (sin (atan (x))): New simplification rules.
1726 (cos (atan (x))): Likewise.
1727 * real.c (build_sinatan_real): New function.
1728 * real.h (build_sinatan_real): Prototype.
1730 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
1732 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
1734 (fold_mergeeo_helper): New helper function.
1735 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
1736 intrinsics. Correct some whitespace indentation issues.
1738 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
1741 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
1742 Use HOST_WIDE_INT_1U for shift.
1744 2018-10-11 Doug Rupp <rupp@adacore.com>
1745 Olivier Hainque <hainque@adacore.com>
1747 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
1748 Pass --relax to the linker for RTPs.
1749 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
1751 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
1752 Jan Hubicka <jh@suse.cz>
1753 Martin Jambor <mjambor@suse.cz>
1755 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
1756 the same elements are repeated rather than printing all of them.
1757 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
1758 "repeated" elements.
1759 * read-rtl-function.c (test_loading_repeat): New function.
1760 (read_rtl_function_c_tests): Call test_loading_repeat.
1761 * rtl-tests.c (test_dumping_repeat): New function.
1762 (rtl_tests_c_tests): Call test_dumping_repeat.
1764 2018-10-11 Richard Biener <rguenther@suse.de>
1766 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
1767 bdver?_cost): Unify to ...
1768 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
1769 * config/i386/i386.c (processor_cost_table): Adjust.
1771 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
1774 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
1775 the thunk when expanding to GIMPLE.
1777 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1779 * varasm.c (mergeable_string_section): Don't try to move zero-length
1780 strings to the merge section.
1782 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
1785 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
1787 2018-10-10 Jakub Jelinek <jakub@redhat.com>
1790 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
1791 to special_args set.
1793 2018-10-10 Richard Biener <rguenther@suse.de>
1795 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
1796 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
1797 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
1798 and recursing and pattern terminating the recursion on SSE
1799 vector width using ix86_expand_reduc.
1800 (reduc_sminmax_scal_<mode>): Split into part reducing to half
1801 width and recursing and SSE2 vector variant doing the final
1802 reduction with ix86_expand_reduc.
1803 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
1804 with terminating the recursion at AVX level, splitting that
1807 2018-10-09 David Malcolm <dmalcolm@redhat.com>
1809 * genmatch.c (error_cb): Rename to...
1810 (diagnostic_cb): ...this, converting int params to enums.
1811 (fatal_at): Update for renaming.
1812 (warning_at): Likewise.
1814 * input.c (selftest::ebcdic_execution_charset::apply):
1815 Update for renaming of...
1816 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
1818 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
1819 converting level and reason to enums.
1820 (class selftest::lexer_error_sink): Rename to...
1821 (class selftest::lexer_test_options): ...this, renaming field
1822 "m_errors" to "m_diagnostics".
1823 (selftest::lexer_test_options::apply): Update for renaming of...
1824 (selftest::lexer_test_options::on_error): ...this, renaming to...
1825 (selftest::lexer_test_options::on_diagnostic): ...this
1826 converting level and reason to enums.
1827 (selftest::test_lexer_string_locations_raw_string_unterminated):
1828 Update for renamings.
1829 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
1832 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
1834 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
1835 * config/rs6000/pmmintrin.h: New file.
1837 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
1839 PR tree-optimization/86659
1840 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
1842 2018-10-09 Richard Biener <rguenther@suse.de>
1844 PR tree-optimization/63155
1845 * tree-ssa-structalias.c: Include tree-ssa.h.
1846 (get_constraint_for_ssa_var): For undefs return nothing_id.
1847 (find_func_aliases): Cleanup PHI handling.
1849 2018-10-09 Richard Biener <rguenther@suse.de>
1851 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
1854 2018-10-09 Martin Liska <mliska@suse.cz>
1856 * asan.c (asan_emit_stack_protection): If a stack variable
1857 is located in a same file as current function, then emit
1858 line info into variable definition string.
1860 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
1862 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
1865 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
1867 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
1870 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
1872 PR tree-optimization/86659
1873 * gimple-match.h (struct gimple_match_op): Add reverse field.
1874 (gimple_match_op::set_op): New overloaded method.
1875 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
1876 the REF_REVERSE_STORAGE_ORDER flag on the value.
1877 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
1878 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
1880 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
1883 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
1884 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
1886 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
1889 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
1890 Defined with __builtin_ia32_vfmaddsubpd512_mask.
1892 2018-10-08 Richard Biener <rguenther@suse.de>
1894 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
1895 cost the same as AVX128 ones.
1897 2018-10-08 Paul Koning <ni1d@arrl.net>
1899 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
1900 (expand_block_move): New function.
1901 * config/pdp11/pdp11.c (output_block_move): Remove.
1902 (expand_block_move): New function.
1903 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
1904 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
1905 (*movmemhi1): Remove.
1907 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
1909 * config/s390/2827.md: Increase latencies for some FP instructions.
1911 2018-10-08 Richard Biener <rguenther@suse.de>
1913 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
1915 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
1916 * tree-vectorizer.h (dump_stmt_cost): Adjust.
1917 (add_stmt_cost): Dump return value of the hook.
1919 2018-10-08 Richard Biener <rguenther@suse.de>
1921 PR tree-optimization/63155
1922 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
1923 (ssa_propagation_engine::ssa_propagate): Remove redundant
1924 bitmap bit clearing.
1926 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
1928 PR rtl-optimization/86939
1929 PR rtl-optimization/87479
1930 * ira.h (non_conflicting_reg_copy_p): New prototype.
1931 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
1932 (make_hard_regno_dead): Don't add conflicts for register
1933 ignore_reg_for_conflicts.
1934 (make_object_dead): Likewise.
1935 (non_conflicting_reg_copy_p): New function.
1936 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
1937 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
1938 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
1939 (make_hard_regno_dead): Don't add conflicts for register
1940 ignore_reg_for_conflicts. Remove special conflict handling of
1941 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
1942 check_pic_pseudo_p and update callers.
1943 (mark_pseudo_dead): Don't add conflicts for register
1944 ignore_reg_for_conflicts.
1945 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
1947 2018-10-05 Andrew Waterman <andrew@sifive.com>
1948 Jim Wilson <jimw@sifive.com>
1950 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
1951 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
1952 new pattern using HONOR_SNANS that emits one extra instruction.
1954 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
1956 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
1957 patterns): Merge SI and DI patterns to a GPR pattern.
1958 (unnamed define_insn and define_split for record form of that): Merge
1959 to a single define_insn_and_split pattern.
1961 2018-10-05 David Malcolm <dmalcolm@redhat.com>
1964 * input.c (expand_location_to_spelling_point): Add param "aspect"
1965 and use rather than hardcoding LOCATION_ASPECT_CARET.
1966 (get_substring_ranges_for_loc): Handle the case of a single token
1967 within a macro expansion.
1968 * input.h (expand_location_to_spelling_point): Add "aspect" param,
1969 defaulting to LOCATION_ASPECT_CARET.
1971 2018-10-05 Paul Koning <ni1d@arrl.net>
1973 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
1974 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
1975 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
1976 (pdp11_guard_type): New function.
1978 2018-10-05 Paul Koning <ni1d@arrl.net>
1980 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
1981 * config/pdp11/pdp11.opt (mfloat32): Remove.
1983 * doc/invoke.texi (pdp11 -mfloat32): Remove:
1984 (pdp11 -mfloat64): Remove.
1986 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
1988 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
1989 (*cmp<mode>_cc_i387): Ditto.
1990 (*cmpu<mode>_cc_i387): Ditto.
1991 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
1992 * config/i386/i386.c (ix86_expand_fp_compare): Remove
1994 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
1995 Emit x86_sahf_1 pattern.
1996 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
1997 (ix86_expand_carry_flag_compare): Ditto.
1999 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
2001 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
2002 to reg_or_0_operand. Add "C" constraint.
2003 (*cmpxf_cc_i387): Ditto.
2004 (*cmp<mode>_i387): Change operand 2 predicate
2005 to nonimm_or_0_operand. Add "C" constraint.
2006 (*cmp<mode>_cc_i387): Ditto.
2007 (*cmp<mode>_0_i387): Remove insn pattern.
2008 (*cmp<mode>_0_cc_i387): Ditto.
2010 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
2012 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
2013 * config/i386/predicates.md (nonimm_or_0_operand): Rename
2014 from vector_move_operand. Update all uses.
2016 2018-10-05 Martin Sebor <msebor@redhat.com>
2018 PR tree-optimization/87490
2019 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
2022 2018-10-05 Richard Biener <rguenther@suse.de>
2024 PR tree-optimization/63155
2025 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
2026 vertical space in dumpfiles.
2027 * tree-ssa-propagate.h
2028 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
2029 * tree-ssa-propagate.c (cfg_blocks_back): New global.
2030 (ssa_edge_worklist_back): Likewise.
2031 (curr_order): Likewise.
2032 (cfg_blocks_get): Remove abstraction.
2033 (cfg_blocks_add): Likewise.
2034 (cfg_blocks_empty_p): Likewise.
2035 (add_ssa_edge): Add to current or next worklist based on
2037 (add_control_edge): Likewise.
2038 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
2040 (ssa_propagation_engine::ssa_propagate): ... here. Unify
2041 iteration from CFG and SSA edge worklist so we process
2042 everything in RPO order, prioritizing forward progress
2044 (ssa_prop_init): Allocate new worklists, do not dump
2046 (ssa_prop_fini): Free new worklists.
2048 2018-10-05 Richard Biener <rguenther@suse.de>
2050 * tree-core.h (tree_block::abstract_flag): Remove.
2051 (tree_block::block_num): Make full 32bits.
2052 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
2053 * tree.h (BLOCK_ABSTRACT): Remove.
2054 * dwarf2out.c (gen_lexical_block_die): Remove dead code
2055 resulting from BLOCK_ABSTRACT being always false.
2056 (gen_inlined_subroutine_die): Likewise.
2057 (gen_block_die): Likewise.
2058 * tree.c (block_ultimate_origin): Likewise.
2059 * tree-pretty-print.c (dump_block_node): Remove code dealing
2060 with BLOCK_ABSTRACT.
2061 * tree-ssa-live.c (dump_scope_block): Likewise.
2062 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
2063 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
2065 2018-10-05 Richard Biener <rguenther@suse.de>
2067 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
2068 is asked for initialize mode to the component mode of the
2071 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
2074 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
2075 assembler for -mavx.
2076 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
2078 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2081 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
2083 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
2084 to be DImode. When using mffscrn, force the operand to a register.
2086 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
2088 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
2089 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
2090 X87MODEF mode iterator.
2091 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
2092 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
2093 X87MODEF mode iterator.
2095 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
2097 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
2098 -Wno-prio-ctor-dtor.
2100 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2102 * Makefile.in (OBJS): Add opt-problem.o.
2103 * dump-context.h: Include "selftest.h.
2104 (selftest::temp_dump_context): New forward decl.
2105 (class dump_context): Make friend of class
2106 selftest::temp_dump_context.
2107 (dump_context::dump_loc_immediate): New decl.
2108 (class dump_pretty_printer): Move here from dumpfile.c.
2109 (class temp_dump_context): Move to namespace selftest.
2110 (temp_dump_context::temp_dump_context): Add param
2111 "forcibly_enable_dumping".
2112 (selftest::verify_dumped_text):
2113 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
2114 (selftest::verify_item):
2115 (ASSERT_IS_TEXT): Move here from dumpfile.c.
2116 (ASSERT_IS_TREE): Likewise.
2117 (ASSERT_IS_GIMPLE): Likewise.
2118 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
2120 (dump_context::dump_loc_immediate): ...this new function.
2121 (class dump_pretty_printer): Move to dump-context.h.
2122 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
2123 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
2124 (temp_dump_context::temp_dump_context): Move to "selftest"
2125 namespace. Add param "forcibly_enable_dumping", and use it to
2126 conditionalize the use of m_pp;
2127 (selftest::verify_dumped_text): Make non-static.
2128 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
2129 (selftest::verify_item): Make non-static.
2130 (ASSERT_IS_TEXT): Move to dump-context.h.
2131 (ASSERT_IS_TREE): Likewise.
2132 (ASSERT_IS_GIMPLE): Likewise.
2133 (selftest::test_capture_of_dump_calls): Pass "true" for new
2134 param of temp_dump_context.
2135 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
2136 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
2138 * opt-problem.cc: New file.
2139 * opt-problem.h: New file.
2140 * optinfo-emit-json.cc
2141 (selftest::test_building_json_from_dump_calls): Pass "true" for
2142 new param of temp_dump_context.
2143 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
2144 (optinfo::emit_for_opt_problem): New function.
2145 (optinfo::emit): Clarity which emit_item is used.
2146 * optinfo.h (optinfo::get_dump_location): New accessor.
2147 (optinfo::emit_for_opt_problem): New decl.
2148 (optinfo::emit): Make const.
2149 * selftest-run-tests.c (selftest::run_tests): Call
2150 selftest::opt_problem_cc_tests.
2151 * selftest.h (selftest::opt_problem_cc_tests): New decl.
2152 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
2153 bool to opt_result, converting fprintf messages to
2154 opt_result::failure_at calls. Add "stmt" param for use by the
2156 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
2157 (runtime_alias_check_p): Convert return type from bool to
2158 opt_result, converting dump_printf calls to
2159 opt_result::failure_at, using the statement DDR_A for their
2161 (find_data_references_in_stmt): Convert return type from bool to
2162 opt_result, converting "return false" to opt_result::failure_at
2164 * tree-data-ref.h: Include "opt-problem.h".
2165 (dr_analyze_innermost): Convert return type from bool to opt_result,
2166 and add a const gimple * param.
2167 (find_data_references_in_stmt): Convert return type from bool to
2169 (runtime_alias_check_p): Likewise.
2170 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
2171 dr_analyze_innermost.
2172 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
2173 Convert return type from bool to opt_result, adding a message for
2174 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
2175 (vect_analyze_data_ref_dependence): Convert return type from bool
2176 to opt_result. Change sense of return type from "false"
2177 effectively meaning "no problems" to "false" meaning a problem,
2178 so that "return false" becomes "return opt_result::success".
2179 Convert "return true" calls to opt_result::failure_at, using
2180 the location of statement A rather than vect_location.
2181 (vect_analyze_data_ref_dependences): Convert return type from bool
2183 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
2184 calls to opt_result::failure_at, using the stmt location rather
2186 (vect_verify_datarefs_alignment): Convert return type from bool
2188 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
2189 into multiple more-tightly-scoped copies.
2190 (vect_analyze_data_refs_alignment): Convert return type from bool
2192 (vect_analyze_data_ref_accesses): Likewise, converting a
2193 "return false" to a "return opt_result::failure_at", adding a
2195 (vect_prune_runtime_alias_test_list): Convert return type from
2196 bool to opt_result, converting dump_printf_loc to
2197 opt_result::failure_at. Add a %G to show the pertinent statement,
2198 and use the stmt's location rather than vect_location.
2199 (vect_find_stmt_data_reference): Convert return type from
2200 bool to opt_result, converting dump_printf_loc to
2201 opt_result::failure_at, using stmt's location.
2202 (vect_analyze_data_refs): Convert return type from bool to
2203 opt_result. Convert "return false" to "return
2204 opt_result::failure_at", adding messages as needed.
2205 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
2206 type from bool to opt_result.
2207 (vect_determine_vf_for_stmt): Likewise.
2208 (vect_determine_vectorization_factor): Likewise, converting
2209 dump_printf_loc to opt_result::failure_at, using location of phi
2210 rather than vect_location.
2211 (vect_analyze_loop_form_1): Convert return type from bool to
2212 opt_result, converting dump_printf_loc calls, retaining the use of
2214 (vect_analyze_loop_form): Convert return type from loop_vec_info
2215 to opt_loop_vec_info.
2216 (vect_analyze_loop_operations): Convert return type from bool to
2217 opt_result, converting dump_printf_loc calls, using the location
2218 of phi/stmt rather than vect_location where available. Convert
2219 various "return false" to "return opt_result::failure_at" with
2220 "unsupported phi" messages.
2221 (vect_get_datarefs_in_loop): Convert return type from bool to
2222 opt_result. Add a message for the
2223 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
2224 (vect_analyze_loop_2): Convert return type from bool to
2225 opt_result. Ensure "ok" is set to a opt_result::failure_at before
2226 each "goto again;", adding new messages where needed.
2227 Add "unsupported grouped {store|load}" messages.
2228 (vect_analyze_loop): Convert return type from loop_vec_info to
2230 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
2232 * tree-vect-stmts.c (process_use): Likewise, converting
2233 dump_printf_loc call and using stmt location, rather than
2235 (vect_mark_stmts_to_be_vectorized): Likeise.
2236 (vect_analyze_stmt): Likewise, adding a %G.
2237 (vect_get_vector_types_for_stmt): Convert return type from bool to
2238 opt_result, converting dump_printf_loc calls and using stmt
2239 location, rather than vect_location.
2240 (vect_get_mask_type_for_stmt): Convert return type from tree to
2241 opt_tree, converting dump_printf_loc calls and using stmt location.
2242 * tree-vectorizer.c: Include "opt-problem.h.
2243 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
2244 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
2245 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
2246 enabled, use it to report at the top level "couldn't vectorize
2247 loop" followed by the problem.
2248 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
2249 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
2251 (vect_analyze_stmt): Likewise.
2252 (vect_get_vector_types_for_stmt): Likewise.
2253 (tree vect_get_mask_type_for_stmt): Likewise.
2254 (vect_analyze_data_ref_dependences): Likewise.
2255 (vect_enhance_data_refs_alignment): Likewise.
2256 (vect_analyze_data_refs_alignment): Likewise.
2257 (vect_verify_datarefs_alignment): Likewise.
2258 (vect_analyze_data_ref_accesses): Likewise.
2259 (vect_prune_runtime_alias_test_list): Likewise.
2260 (vect_find_stmt_data_reference): Likewise.
2261 (vect_analyze_data_refs): Likewise.
2262 (vect_analyze_loop): Convert return type from loop_vec_info to
2264 (vect_analyze_loop_form): Likewise.
2265 (vect_analyze_slp): Convert return type from bool to opt_result.
2267 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2269 * doc/invoke.texi (-fopt-info): Document new "internals"
2271 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
2272 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
2274 (optinfo_verbosity_options): Add "internals".
2275 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
2276 (dump_context::apply_dump_filter_p): New member function.
2277 (dump_context::dump_loc): Use apply_dump_filter_p rather than
2278 explicitly masking the dump_kind.
2279 (dump_context::begin_scope): Increment the scope depth first. Use
2280 apply_dump_filter_p rather than explicitly masking the dump_kind.
2281 (dump_context::emit_item): Use apply_dump_filter_p rather than
2282 explicitly masking the dump_kind.
2283 (dump_dec): Likewise.
2284 (dump_hex): Likewise.
2285 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
2286 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
2287 (opt_info_switch_p): Update handling of default
2288 MSG_OPTIMIZED_LOCATIONS to cope with default of
2289 MSG_PRIORITY_USER_FACING.
2290 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
2291 masking the dump_kind.
2292 (selftest::test_capture_of_dump_calls): Update test_dump_context
2293 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
2294 than MSG_ALL. Generalize scope test to be run at all four
2295 combinations of with/without MSG_PRIORITY_USER_FACING and
2296 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
2297 for each of the two values.
2298 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
2299 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
2300 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
2301 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
2302 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
2303 with MSG_PRIORITY_*.
2304 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
2305 dump messages as MSG_PRIORITY_USER_FACING.
2306 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
2307 about the interaction with MSG_PRIORITY_*.
2309 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2311 * varasm.c (output_constant): Add new parameter merge_strings.
2312 Make strings properly zero terminated in merge string sections.
2313 (mergeable_string_section): Don't fail if the last char is non-zero.
2314 (assemble_variable_contents): Handle merge string sections.
2315 (assemble_variable): Likewise.
2316 (assemble_constant_contents): Likewise.
2317 (output_constant_def_contents): Likewise.
2318 (output_constructor_array_range,
2319 output_constructor_regular_field): Adjust call to output_constant.
2320 (output_object_block): Adjust call to assemble_constant_contents
2321 and assemble_variable_contents.
2323 2018-10-04 Martin Liska <mliska@suse.cz>
2326 * cgraphunit.c (process_function_and_variable_attributes):
2327 Warn about a function with alias attribute and a body.
2329 2018-10-04 Martin Liska <mliska@suse.cz>
2332 * multiple_target.c (redirect_to_specific_clone): New function.
2333 (ipa_target_clone): Use it.
2334 * tree-inline.c: Fix comment.
2336 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2338 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
2340 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
2341 (gcc::dump_manager::register_pass): New member function, adapted
2342 from loop body in gcc::pass_manager::register_pass, adding a
2343 call to update_dfi_for_opt_info.
2344 (gcc::dump_manager::opt_info_enable_passes): Store the
2345 -fopt-info options into the new fields. Move the loop
2347 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
2349 * dumpfile.h (struct opt_pass): New forward decl.
2350 (gcc::dump_manager::register_pass): New decl.
2351 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
2352 (class gcc::dump_manager): Add fields "m_optgroup_flags",
2353 "m_optinfo_flags", and "m_optinfo_filename".
2354 * passes.c (gcc::pass_manager::register_pass): Move all of the
2355 dump-handling code to gcc::dump_manager::register_pass.
2357 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
2359 PR rtl-optimization/87466
2360 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
2361 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
2362 * doc/tm.texi: Regenerate.
2363 * ira-lives.c (process_bb_node_lives): Use the new target hook.
2364 * lra-lives.c (process_bb_lives): Likewise.
2365 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
2368 2018-10-04 Tamar Christina <tamar.christina@arm.com>
2370 * params.c (add_params): Fix initialization.
2372 2018-10-04 Martin Liska <mliska@suse.cz>
2374 PR gcov-profile/84107
2375 * tree-profile.c (init_ic_make_global_vars):
2376 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
2377 Come up with new ic_tuple* variables. Emit
2378 __gcov_indirect_call{,_topn} variables.
2379 (gimple_gen_ic_profiler): Access the variable
2381 (gimple_gen_ic_func_profiler): Access
2382 __gcov_indirect_call.callee field.
2383 (gimple_init_gcov_profiler): Use ptr_type_node.
2384 * value-prof.c (gimple_ic): Use ptr_type_node.
2386 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2388 PR tree-optimization/85787
2389 * ipa-pure-const.c (malloc_candidate_p_1): Move most of malloc_candidate_p
2390 into this function and add support for detecting multiple phis.
2391 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
2393 2018-10-04 Martin Liska <mliska@suse.cz>
2396 * ipa-inline.c (inline_to_all_callers_1):
2397 Call ultimate_alias_target for node being inlined.
2399 2018-10-03 Jeff Law <law@redhat.com>
2401 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
2403 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
2404 * tree.h (get_typenode_from_name): Prototype.
2406 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
2408 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
2409 Change operand 2 predicate to nonimmediate_operand.
2410 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2412 2018-10-03 Martin Sebor <msebor@redhat.com>
2413 Jeff Law <law@redhat.com>
2415 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
2417 (get_string_length): Detect unterminated arrays.
2418 (format_string): Same.
2419 (format_directive): Warn about unterminated arrays.
2420 (handle_gimple_call): Mark statements with no_warning as needed.
2422 2018-10-03 Jim Wilson <jimw@sifive.com>
2424 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
2425 also define __riscv_abi_rve. Delete trailing white space.
2427 2018-10-03 Paul Koning <ni1d@arrl.net>
2429 Enable LRA register allocator for PDP11.
2430 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
2433 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
2434 * config/pdp11/pdp11.opt (-mlra): New option.
2435 * doc/invoke.texi (PDP-11 Options): Document -mlra.
2437 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
2439 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
2440 (*<absneg:code>extend<mode>xf2): Ditto.
2442 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
2444 PR tree-optimization/87415
2445 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
2448 2018-10-02 Jeff Law <law@redhat.com>
2450 * gimple-fold.c (get_range_strlen): Only set *nonstr when
2451 an unterminated string is discovered. Bubble up range
2452 even for unterminated strings.
2453 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
2454 indicates the string was not terminated via NONSTR.
2456 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
2458 * tree-vrp.c (extract_range_from_unary_expr): Special case all
2459 pointer conversions.
2460 Do not do anything special for anti-ranges.
2462 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
2464 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
2465 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
2467 2018-10-03 Martin Liska <mliska@suse.cz>
2469 PR gcov-profile/86109
2470 * coverage.c (coverage_begin_function): Do not
2471 mark lambdas as artificial.
2472 * tree-core.h (struct GTY): Remove tm_clone_flag
2473 and introduce new lambda_function.
2474 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
2476 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
2479 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
2480 P8_VECTOR and VSX are enabled.
2482 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
2484 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
2485 0x3907 as CPU model number.
2487 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
2489 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
2490 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
2491 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
2492 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
2493 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
2494 * config/s390/s390.md: Likewise. Rename also the cpu attribute
2495 value from arch12 to z14.
2497 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
2499 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
2501 (isinf<mode>2): Ditto.
2503 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
2505 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
2506 before emitting fxam. Perform calculations in XFmode.
2508 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
2510 * match.pd (((X /[ex] A) +- B) * A): New transformation.
2512 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
2515 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
2516 * tree.c (signed_or_unsigned_type_for): Handle complex.
2518 2018-10-02 Jeff Law <law@redhat.com>
2520 * gimple-fold.c (get_range_strlen): Remove dead code.
2522 2018-10-02 Martin Sebor <msebor@redhat.com>
2523 Jeff Law <law@redhat.com>
2525 * builtins.c (unterminated_array): Add new arguments.
2526 If argument is not terminated, bubble up size and exact
2528 (expand_builtin_strnlen): Detect, avoid expanding
2529 and diagnose unterminated arrays.
2530 (c_strlen): Fill in offset of start of unterminated strings.
2531 * builtins.h (unterminated_array): Update prototype.
2533 2018-10-02 Richard Biener <rguenther@suse.de>
2535 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
2536 of haddv4df, first reduce to SSE width and exploit the fact
2537 that we only need element zero with the reduction result.
2538 (reduc_plus_scal_v2df): Likewise.
2540 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
2542 * dojump.h (do_jump): Delete.
2543 (do_jump_1): Likewise.
2544 (split_comparison): Move around.
2545 * dojump.c (do_jump): Make static.
2546 (do_jump_1): Likewise.
2547 (jumpifnot): Move around.
2548 (jumpifnot_1): Likewise.
2550 (jumpif_1): Likewise.
2551 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
2553 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
2555 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
2556 insns in the delay slot and add_insn_after for the jump insn.
2558 2018-10-02 Richard Biener <rguenther@suse.de>
2560 * tree-inline.c (expand_call_inline): Use the location of
2561 the callee declaration for the inline-entry marker.
2562 * final.c (notice_source_line): Remove special-casing of
2563 NOTE_INSN_INLINE_ENTRY.
2565 2018-10-01 Carl Love <cel@us.ibm.com>
2568 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
2569 (__builtin_mtfsb0): New.
2570 (__builtin_mtfsb1): New.
2571 ( __builtin_set_fpscr_rn): New.
2572 (__builtin_set_fpscr_drn): New.
2573 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
2574 (rs6000_expand_set_fpscr_rn_builtin): Add.
2575 (rs6000_expand_set_fpscr_drn_builtin): Add.
2576 (rs6000_expand_builtin): Add case statement entries for
2577 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
2578 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
2579 RS6000_BUILTIN_MFFSL.
2580 (rs6000_init_builtins): Add ftype initialization and def_builtin
2581 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
2582 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
2583 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
2584 rs6000_mffscdrn): Add define_insn.
2585 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
2586 * doc/extend.texi: Add documentation for the builtins.
2588 2018-10-01 Richard Biener <rguenther@suse.de>
2590 PR tree-optimization/87465
2591 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
2592 causing branch miscounts.
2594 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2596 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
2597 aarch64_option_default_param): New.
2598 (params.h): Include.
2599 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
2600 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
2601 stack-clash protection validation code.
2603 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2605 * params.c (validate_param): New.
2606 (add_params): Use it.
2607 (set_param_value): Refactor param validation into validate_param.
2608 (diagnostic.h): Include.
2609 * diagnostic.h (diagnostic_ready_p): New.
2611 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2613 * params.c (set_param_value):
2614 Add index of parameter being validated.
2615 * common/common-target.def (option_validate_param): New.
2616 * common/common-targhooks.h (default_option_validate_param): New.
2617 * common/common-targhooks.c (default_option_validate_param): New.
2618 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
2619 * doc/tm.texi: Regenerate.
2621 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2624 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2625 Add validation for stack-clash parameters and set defaults.
2627 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2630 * configure.ac: Add stack-clash-protection-guard-size.
2631 * doc/install.texi: Document it.
2632 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
2633 * params.def: Update comment for guard-size.
2634 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
2635 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
2636 * configure: Regenerate.
2638 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2641 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
2642 STACK_DYNAMIC_OFFSET): New.
2643 * config/aarch64/aarch64.c (aarch64_layout_frame):
2644 Update outgoing args size.
2645 (aarch64_stack_clash_protection_alloca_probe_range,
2646 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
2648 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2651 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
2653 * target.def (stack_clash_protection_alloca_probe_range): New.
2654 (stack_clash_protection_final_dynamic_probe): Remove.
2655 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
2656 (default_stack_clash_protection_final_dynamic_probe): Remove.
2657 * targhooks.c: Likewise.
2658 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
2659 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
2660 * doc/tm.texi: Regenerate.
2662 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2665 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
2666 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
2667 aarch64_clamp_to_uimm12_shift): New.
2668 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
2669 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
2671 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2674 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
2676 2018-10-01 Jeff Law <law@redhat.com>
2677 Richard Sandiford <richard.sandiford@linaro.org>
2678 Tamar Christina <tamar.christina@arm.com>
2681 * config/aarch64/aarch64.md
2682 (probe_stack_range): Add k (SP) constraint.
2683 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
2684 STACK_CLASH_MAX_UNROLL_PAGES): New.
2685 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
2686 stack probes for stack clash.
2687 (aarch64_allocate_and_probe_stack_space): New.
2688 (aarch64_expand_prologue): Use it.
2689 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
2690 (aarch64_sub_sp): Add emit_move_imm optional param.
2692 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
2694 PR tree-optimization/87261
2695 * match.pd: Remove trailing whitespace.
2696 Add (x & y) | ~(x | y) -> ~(x ^ y),
2697 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
2699 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
2701 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
2703 (ashlsi3_insn): Update instruction constraints.
2704 (ashrsi3_insn): Likewise.
2705 (rotrsi3): Likewise.
2706 (add_shift): Likewise.
2707 * config/arc/constraints.md (Csz): New 32 bit constraint. It
2708 avoids placing in the limm field small constants which, otherwise,
2709 could end into a small instruction.
2711 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
2713 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
2714 destination register is not odd-even.
2715 (umaddsidi4_split): Likewise.
2717 2018-10-01 Richard Biener <rguenther@suse.de>
2719 * tree-inline.c (expand_call_inline): Store origin of fn
2720 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
2721 * tree.c (block_ultimate_origin): Simplify and do some
2724 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
2726 * config/i386/mmx.md (EMMS): New int iterator.
2727 (emms): New int attribute.
2728 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
2729 EMMS int iterator. Explicitly declare clobbers.
2730 (mmx_emms): Remove expander.
2732 * config/i386/predicates.md (emms_operation): Remove predicate.
2733 (vzeroall_pattern): New predicate.
2734 (vzeroupper_pattern): Rename from vzeroupper_operation.
2735 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
2736 vzeroupper_pattern and vzeroall_pattern predicates.
2738 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
2740 PR rtl-optimization/86939
2741 * ira-lives.c (make_hard_regno_born): Rename from this...
2742 (make_hard_regno_live): ... to this. Remove update to conflict
2743 information. Update function comment.
2744 (make_hard_regno_dead): Add conflict information update. Update
2746 (make_object_born): Rename from this...
2747 (make_object_live): ... to this. Remove update to conflict information.
2748 Update function comment.
2749 (make_object_dead): Add conflict information update. Update function
2751 (mark_pseudo_regno_live): Call make_object_live.
2752 (mark_pseudo_regno_subword_live): Likewise.
2753 (mark_hard_reg_dead): Update function comment.
2754 (mark_hard_reg_live): Call make_hard_regno_live.
2755 (process_bb_node_lives): Likewise.
2756 * lra-lives.c (make_hard_regno_born): Rename from this...
2757 (make_hard_regno_live): ... to this. Remove update to conflict
2758 information. Remove now uneeded check_pic_pseudo_p argument.
2759 Update function comment.
2760 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
2761 to conflict information. Update function comment.
2762 (mark_pseudo_live): Remove update to conflict information. Update
2764 (mark_pseudo_dead): Add conflict information update.
2765 (mark_regno_live): Call make_hard_regno_live.
2766 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
2767 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
2769 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
2772 * config/i386/i386.c (construct_container): Use TImode for
2773 BLKmode values in 2 integer registers.
2775 2018-09-29 Jeff Law <law@redhat.com>
2777 * builtins.c (unterminated_array): Pass in c_strlen_data * to
2778 c_strlen rather than just a tree *.
2779 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
2780 Update recursive calls appropriately. If caller did not provide a
2781 suitable data pointer, create a local one. When a non-terminated
2782 string is discovered, bubble up information about the string via the
2783 c_strlen_data object.
2784 * builtins.h (c_strlen): Update prototype.
2785 (c_strlen_data): New structure.
2786 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
2787 For a type 2 call, if c_strlen indicates a non-terminated string
2788 use the length of the non-terminated string.
2789 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
2791 2018-09-29 Jakub Jelinek <jakub@redhat.com>
2794 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
2795 __m512d type for __A argument rather than __m512.
2797 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
2799 * match.pd (simple_comparison): Don't optimize if either operand is
2800 a function pointer when target needs function pointer canonicalization.
2802 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
2804 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
2805 power5 .. power9 to remove indirection.
2806 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
2807 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
2808 ASM_CPU_476_SPEC): Delete.
2809 (ASM_CPU_SPEC): Adjust.
2810 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
2811 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
2813 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
2815 * config.in: Delete HAVE_AS_DCI.
2816 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
2817 * config/rs6000/rs6000.h: Ditto.
2818 * configure.ac: Delete HAVE_AS_DCI.
2819 * configure: Regenerate.
2821 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
2823 * config.in (HAVE_AS_LWSYNC): Delete.
2824 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
2825 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
2827 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
2828 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
2830 * configure.ac: Delete HAVE_AS_LWSYNC.
2831 * configure: Regenerate.
2833 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
2834 Pierre-Marie de Rodat <derodat@adacore.com>
2836 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
2837 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
2838 (cgraph_node::create_thunk): Add indirect_offset parameter.
2839 (thunk_adjust): Likewise.
2840 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
2841 and initialize the corresponding field with it.
2842 (cgraph_node::dump): Dump indirect_offset field.
2843 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
2844 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
2845 (thunk_adjust): Add indirect_offset parameter and deal with it.
2846 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
2847 pass it to thunk_adjust. Do not call the target hook if it's non-zero
2848 or if the thunk is external or local. Fix formatting. Do not chain
2849 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
2850 if any, in the GIMPLE representation.
2851 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
2852 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
2853 (input_node): Read indirect_offset field.
2854 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
2855 call to thunk_adjust.
2856 * tree-nested.c (struct nesting_info): Add thunk_p field.
2857 (create_nesting_tree): Set it.
2858 (convert_all_function_calls): Copy static chain from targets to thunks.
2859 (finalize_nesting_tree_1): Return early for thunks.
2860 (unnest_nesting_tree_1): Do not finalize thunks.
2861 (gimplify_all_functions): Do not gimplify thunks.
2863 2018-09-28 David Malcolm <dmalcolm@redhat.com>
2865 * opt-suggestions.c (option_proposer::build_option_suggestions):
2866 Release "option_values".
2868 2018-09-28 David Malcolm <dmalcolm@redhat.com>
2870 * coverage.c (get_coverage_counts): Convert problem-reporting dump
2871 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
2872 * dumpfile.c (kind_as_string): New function.
2873 (dump_loc): Rather than a hardcoded prefix of "note: ", use
2874 kind_as_string to vary the prefix based on dump_kind.
2875 (selftest::test_capture_of_dump_calls): Update for above.
2877 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
2879 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
2880 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
2882 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
2884 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
2885 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
2886 INVALID_REGNUM instead of FPSR_REG.
2887 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
2888 * config/i386/i386.md: Update comment of FP compares.
2889 (fldenv): Do not clobber FPSR_REG.
2891 2018-09-28 Richard Biener <rguenther@suse.de>
2893 * tree.h (BLOCK_ORIGIN): New.
2894 * omp-expand.c (grid_expand_target_grid_body): Assign
2895 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
2896 * tree-inline.c (remap_block): Likewise.
2897 * auto-profile.c (get_function_decl_from_block): Simplify
2898 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
2899 * langhooks.c (lhd_print_error_function): Likewise.
2900 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
2902 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
2903 * tree.c (block_nonartificial_location): Likewise.
2904 (block_ultimate_origin): Likewise.
2905 * tree-pretty-print.c (percent_K_format): Likewise. Remove
2906 no longer needed LTO case.
2908 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
2909 Jan Hubicka <jh@suse.cz>
2910 Martin Jambor <mjambor@suse.cz>
2912 * simplify-rtx.c (simplify_merge_mask): New function.
2913 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
2914 same masks are used in op1 or op2.
2915 (test_vec_merge): New function.
2916 (test_vector_ops): Call test_vec_merge.
2918 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
2920 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
2921 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
2922 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
2923 (cypress_costs): Set it.
2924 (supersparc_costs): Likewise.
2925 (hypersparc_costs): Likewise.
2926 (leon_cost): Likewise.
2927 (leon3_costs): Likewise.
2928 (sparclet_costs): Likewise.
2929 (ultrasparc_costs): Likewise.
2930 (ultrasparc_costs): Likewise.
2931 (niagara_costs): Likewise.
2932 (niagara2_costs): Likewise.
2933 (niagara3_costs): Likewise.
2934 (niagara4_costs): Likewise.
2935 (niagara7_costs): Likewise.
2936 (m8_costs): Likewise.
2937 (TARGET_CAN_FOLLOW_JUMP): Define.
2938 (pass_work_around_errata::gate): Minor tweak.
2939 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
2940 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
2941 Automaitcally clear MASK_FSMULD mask for V7 processors.
2942 (sparc_can_follow_jump): New static function.
2943 (output_ubranch): Deal with CROSSING_JUMP_P.
2944 (sparc_use_sched_lookahead): Rewrite using switch statement.
2945 (sparc_issue_rate): Reorder.
2946 (sparc_branch_cost): New function.
2948 2018-09-27 Martin Sebor <msebor@redhat.com>
2950 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
2951 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
2952 (int_fits_type_p): Same.
2954 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
2956 * config/i386/i386.md (FPCR_REG): Remove.
2957 (UNSPEC_FLDCW): Remove.
2958 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
2959 (x86_fldcw_1): Remove insn pattern.
2960 (fnstenv): Do not clobber FPCR_REG.
2962 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
2963 (CALL_USED_REGISTERS): Ditto.
2964 (REG_ALLOC_ORDER): Ditto.
2965 (REG_CLASS_CONTENTS): Ditto.
2966 (HI_REGISTER_NAMES): Ditto.
2967 (ADDITIONAL_REGISTER_NAMES): Use defines instead
2968 of numerical constants.
2969 * config/i386/i386.c (regclass_map): Remove fpsr register.
2970 (dbx_register_map): Ditto.
2971 (dbx64_register_map): Ditto.
2972 (svr4_dbx_register_map): Ditto.
2973 (print_reg): Do not handle FPCR_REG.
2975 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
2978 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
2979 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
2980 Delete, always treat as true.
2981 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
2982 Ditto. Simplify remaining code.
2983 * config/powerpcspe/powerpcspe.h: Ditto.
2984 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
2985 Simplify remaining code.
2986 (rs6000_expand_builtin): Ditto.
2987 * config/rs6000/rs6000.h: Ditto.
2988 * configure.ac: Ditto.
2989 * configure: Regenerate.
2991 2018-09-27 Martin Liska <mliska@suse.cz>
2993 * coverage.c (get_coverage_counts): Revert the formatting
2994 of missing profile opt info.
2996 2018-09-27 Richard Biener <rguenther@suse.de>
3000 * dwarf2out.c (set_block_origin_self): Do not mark outermost
3001 block as we do not output that.
3002 (gen_inlined_subroutine_die): Elide the originally outermost
3003 block, matching what we do for concrete instances.
3004 (decls_for_scope): Add parameter specifying whether to recurse
3007 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
3008 Tom de Vries <tom@codesourcery.com>
3012 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
3013 STORE_FLAG_VALUE == 1.
3015 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
3017 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
3018 constant definitions.
3019 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
3020 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
3021 ("speculation_barrier"): New expander definition.
3023 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
3025 PR gcov-profile/86957
3026 * common.opt: New warning option -Wmissing-profile.
3027 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
3028 * doc/invoke.texi: Document -Wmissing-profile.
3030 2018-09-26 Jim Wilson <jimw@sifive.com>
3032 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
3033 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
3035 2018-09-26 Martin Sebor <msebor@redhat.com>
3037 * tree.c (zerop): Change return type to bool.
3038 (integer_zerop, integer_onep, integer_each_onep): Same.
3039 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
3040 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
3041 (real_onep, real_minus_onep, chain_index): Same.
3042 (print_type_hash_statistics, type_list_equal): Same.
3043 * tree.h (zerop): Same.
3044 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
3045 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
3046 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
3047 (real_onep, real_minus_onep, chain_index): Same.
3048 (print_type_hash_statistics, type_list_equal): Same.
3050 2018-09-26 Jim Wilson <jimw@sifive.com>
3052 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
3054 2018-09-26 Jakub Jelinek <jakub@redhat.com>
3057 * config/i386/i386.c: Include debug.h and dwarf2out.h.
3058 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
3061 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
3063 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
3065 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
3067 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
3068 and FP_SECOND_SSE_REGS.
3069 (REG_CLASS_NAMES): Ditto.
3070 (REG_CLASS_CONTENTS): Ditto.
3071 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
3072 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
3073 (ix86_preferred_output_reload_class): Ditto.
3074 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
3075 clobber constraint to "=&f".
3076 (fix_truncdi_i387): Ditto.
3077 (lrintxfdi2): Ditto.
3078 (fistdi2_<rounding>): Ditto.
3079 (fpremxf4_i387): Change "=u" constraint to "=f".
3080 (fprem1xf4_i387): Ditto.
3082 (fptanxf4_i387): Ditto.
3083 (fxtractxf3_i387): Ditto.
3084 (fscalexf4_i387): Ditto.
3085 (atan2xf3): Change "u" constraint to "f".
3086 (fyl2xxf3_i387): Ditto.
3087 (fyl2xp1xf3_i387): Ditto.
3089 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
3092 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
3093 for removed I387_MASK_PM entity.
3096 2018-09-26 Jeff Law <law@redhat.com>
3098 2018-09-26 Alexey Neyman <stilor@att.net>
3100 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
3101 headers are no longer pulled in by <isl/val.h>.
3103 2018-09-26 Richard Biener <rguenther@suse.de>
3106 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
3107 or concrete instance DIE to the tree. Create abstract origin
3108 attributes also for concrete instances.
3110 2018-09-26 Alexey Neyman <stilor@att.net>
3112 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
3113 headers are no longer pulled in by <isl/val.h>.
3115 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
3117 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
3118 Use new helper functions.
3119 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
3120 Use new helper functions.
3121 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
3122 aarch_mm_needs_release): New declarations.
3123 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
3124 aarch_mm_needs_release): New.
3126 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
3128 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
3129 (arm32_output_mi_thunk): Deal with long calls.
3131 2018-09-26 Richard Biener <rguenther@suse.de>
3135 * tree-inline.c (expand_call_inline): When the location
3136 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
3137 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
3138 the inserted BLOCK to make inlined_function_outer_scope_p
3140 * dwarf2out.c (add_call_src_coords_attributes): Do not add
3141 coords for reserved locations.
3143 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3145 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
3146 (*call_indirect_nonlocal_sysv<mode>): Ditto.
3147 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
3148 (*sibcall_nonlocal_sysv<mode>): Ditto.
3149 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
3150 (<bd>_<mode>): Ditto.
3151 (<bd>tf_<mode>): Ditto.
3153 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3155 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
3156 control string as a list of templates instead of as C code.
3157 (*altivec_movti): Ditto.
3158 * config/rs6000/darwin.md (movdf_low_di): Ditto.
3160 2018-09-25 Jim Wilson <jimw@sifive.com>
3162 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
3163 when target symbol is weak.
3165 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
3168 * builtins.c (unterminated_array): Simplify.
3169 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
3170 where pointer arithmetic is safe.
3172 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3175 * config/rs6000/altivec.md (altivec_vspltb): Use
3176 const_0_to_15_operand instead of u5bit_cint_operand.
3177 (*altivec_vspltb_internal): Ditto.
3178 (altivec_vspltb_direct): Ditto.
3179 (altivec_vsplth): Use const_0_to_7_operand instead of
3181 (*altivec_vsplth_internal): Ditto.
3182 (altivec_vsplth_direct): Ditto.
3183 (altivec_vspltw): Use const_0_to_3_operand instead of
3185 (*altivec_vspltw_internal): Ditto.
3186 (altivec_vspltw_direct): Ditto.
3187 (altivec_vspltsf): Ditto.
3188 (*altivec_vspltsf_internal): Ditto.
3189 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
3190 various splats with the proper size immediate. Reorder the various
3191 cases by ascending size of immediate, and put all such together.
3193 2018-09-25 Richard Biener <rguenther@suse.de>
3196 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
3198 (maybe_create_die_with_external_ref): Do not create
3199 DW_TAG_imported_unit here.
3200 (add_abstract_origin_attribute): Handle external BLOCK refs.
3201 (dwarf2out_abstract_function): Simplify LTO case.
3202 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
3203 rather than using maybe_create_die_with_external_ref.
3205 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
3208 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
3209 (frndintxf2_mask_pm_i387): Ditto.
3210 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
3211 Enable for !flag_trapping_math.
3212 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
3213 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
3214 Change operand 1 predicate to nonimmediate_operand.
3215 (attr "i387_cw"): Remove mask_pm.
3216 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
3217 (enum ix86_entity): Remove I387_MASK_PM.
3218 * config/i386/i386.c (ix86_i387_mode_needed): Do not
3219 handle I387_MASK_PM.
3220 (ix86_mode_needed): Ditto.
3221 (ix86_mode_after): Ditto.
3222 (ix86_mode_entry): Ditto.
3223 (ix86_mode_exit): Ditto.
3224 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
3226 2018-09-25 Jakub Jelinek <jakub@redhat.com>
3228 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
3229 to_update_switch_stmts to vNULL instead of calling create on them
3232 2018-09-25 Richard Biener <rguenther@suse.de>
3234 PR tree-optimization/87402
3235 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
3236 (visit_phi): Re-instantiate handling of supposed to be VARYING
3237 but non-VARYING backedge value.
3239 2018-09-25 Richard Biener <rguenther@suse.de>
3242 * dwarf2out.c (struct sym_off_pair): New.
3243 (external_die_map): New global.
3244 (lookup_decl_die): When in LTO create DIEs lazily from the
3246 (lookup_block_die): New function, create DIEs lazily in LTO.
3247 (equate_block_to_die): New function.
3248 (dwarf2out_die_ref_for_decl): During WPA get the association
3249 from the external DIE map.
3250 (dwarf2out_register_external_die): Record mapping into the
3252 (maybe_create_die_with_external_ref): New function split out from
3253 DIE generation part of old dwarf2out_register_external_die.
3254 (add_abstract_origin_attribute): Do not return the DIE. When
3255 in LTO reference externals directly.
3256 (dwarf2out_abstract_function): When in LTO ignore calls for
3257 decls with external DIEs (already present abstract instances).
3258 (gen_call_site_die): Adjust.
3259 (add_high_low_attributes): Likewise.
3260 (gen_lexical_block_die): Likewise.
3261 (gen_inlined_subroutine_die): Likewie.
3262 (gen_block_die): Likewise.
3263 (dwarf2out_inline_entry): Likewise.
3264 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
3267 2018-09-25 Martin Liska <mliska@suse.cz>
3269 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
3270 integers and not by a float value.
3272 2018-09-25 Martin Liska <mliska@suse.cz>
3275 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
3277 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
3278 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
3279 * godump.c (go_format_type): Remove extra parenthesis.
3281 2018-09-25 Martin Liska <mliska@suse.cz>
3283 * alias.c (set_dest_equal_p): Remove unused function.
3284 * config/i386/i386.c (def_builtin_pure2): Likewise.
3285 * diagnostic-show-locus.c (class layout): Remove
3287 (layout::layout): Likewise here.
3288 * dump-context.h (class temp_dump_context): Likewise.
3289 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
3290 (add_AT_loclistsptr): Likewise.
3291 (add_AT_offset): Likewise.
3292 (get_AT_hi_pc): Likewise.
3293 (is_comdat_die): Likewise.
3294 (type_is_enum): Likewise.
3295 (ceiling): Likewise.
3296 (add_AT_vms_delta): Likewise.
3297 (is_class_die): Likewise.
3298 * edit-context.c (class line_event): Remove unused field.
3299 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
3301 * ipa-cp.c (ipa_get_vr_lat): Likewise.
3302 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
3303 (ok_for_base_p_nonstrict): Likewise.
3304 * tree-chrec.c (is_not_constant_evolution): Likewise.
3305 (chrec_fold_poly_cst): Likewise.
3306 * tree-if-conv.c (has_pred_critical_p): Likewise.
3307 * tree-ssa-coalesce.c (print_exprs): Likewise.
3308 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
3309 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
3310 * tree-vrp.c (value_ranges_intersect_p): Likewise.
3311 (value_range_nonnegative_p): Likewise.
3313 2018-09-25 Martin Liska <mliska@suse.cz>
3315 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
3316 Do not handle "GNU Pascal".
3317 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
3319 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
3320 from documentation. Likewise.
3321 * dbxout.c (dbxout_range_type): Likewise.
3322 * doc/cpp.texi: Likewise.
3323 * doc/extend.texi: Likewise.
3324 * doc/frontends.texi: Likewise.
3325 * doc/invoke.texi: Remove Pascal entry.
3326 * tree.def (CLEANUP_POINT_EXPR): Likewise.
3327 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
3329 2018-09-25 Martin Liska <mliska@suse.cz>
3332 * doc/invoke.texi: Document all parameters and remove default
3335 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
3338 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
3339 contains HOST_WIDE_INTs when computing its size.
3341 2018-09-24 Jim Wilson <jimw@sifive.com>
3344 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
3346 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
3348 2018-09-24 Andrew Pinski <apinski@marvell.com>
3350 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
3351 access prev before checking it for NULLness in the
3352 AARCH64_FUSE_CMP_BRANCH case.
3354 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
3357 * config/i386/i386.c (rest_of_insert_endbranch): Set
3358 endbr_queued_at_entrance to true and don't insert ENDBR if
3359 x86_function_profiler will be called.
3360 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
3362 * config/i386/i386.h (machine_function): Add
3363 endbr_queued_at_entrance.
3365 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
3367 * genattrtab.c (mk_attr_alt): Use alternative_mask.
3368 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
3370 (check_attr_test): Use alternative_mask.
3371 (get_attr_value): Likewise.
3372 (compute_alternative_mask): Use alternative_mask and XWINT.
3373 (make_alternative_compare): Use alternative_mask.
3374 (attr_alt_subset_p): Use XWINT.
3375 (attr_alt_subset_of_compl_p): Likewise.
3376 (attr_alt_intersection): Use alternative_mask and XWINT.
3377 (attr_alt_union): Likewise.
3378 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
3379 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
3380 (simplify_test_exp): Use alternative_mask and XWINT.
3381 (write_test_expr): Use alternative_mask and XWINT, adjust bit
3382 number calculation to support 64 bits. Generate code that
3383 checks 64-bit masks.
3384 (main): Use alternative_mask.
3385 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
3387 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
3390 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
3391 RETURN+USE when returning via %r14.
3393 2018-09-24 Martin Liska <mliska@suse.cz>
3395 * gcov.c (output_lines): Print colorization legend
3396 for both flag_use_colors and flag_use_hotness_colors.
3399 2018-09-24 Martin Liska <mliska@suse.cz>
3401 * coverage.c (get_coverage_counts): Use warning_at
3402 with current_function_decl location. Use %qD in warning
3405 2018-09-24 Martin Liska <mliska@suse.cz>
3407 * memory-block.h (memory_block_pool::release): Annotate with
3408 valgrind that the memory is not accessible.
3410 2018-09-24 Martin Liska <mliska@suse.cz>
3413 * asan.c: Make asan_handled_variables extern.
3415 * cfgexpand.c (expand_stack_vars): Make sure
3416 a representative is unpoison if another
3417 variable in the partition is handled by
3418 use-after-scope sanitization.
3420 2018-09-24 Richard Biener <rguenther@suse.de>
3422 PR tree-optimization/63155
3423 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
3424 the worklist when the edge of the respective argument isn't
3427 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3429 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
3430 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
3431 (MASK_CLASS_P): Update for rename.
3432 (MAYBE_MASK_CLASS_P): Ditto.
3433 (REG_CLASS_NAMES): Update.
3434 (REG_CLASS_CONTENT): Update.
3435 * config/i386/i386.c (regclass_map): Update for MASK_REG
3436 and ALL_MASK_REGS rename.
3437 * config/i386/constraints.md (Yk): Update for rename.
3440 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3442 * config/i386/i386.h (enum reg_class): Remove
3443 EVEX_SSE_REGS and MOD4_SSE_REGS.
3444 (REG_CLASS_NAMES): Update.
3445 (REG_CLASS_CONTENT): Update.
3446 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
3447 registers as ALL_SSE_REGS.
3448 (ix86_additional_allocno_class_p): Remove.
3449 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
3450 (ix86_register_priority): Lower priority of EVEX SSE registers.
3451 Use IN_RANGE macro where appropriate.
3452 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
3453 AVX-5124VNNIW checks.
3454 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
3455 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
3456 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
3457 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
3458 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
3459 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
3460 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
3461 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
3462 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
3463 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
3464 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
3465 * config/i386/constraints.md (Yh): Remove.
3467 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3469 * config/i386/i386.c (regclass_map): Declare integer REX registers
3472 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
3474 * doc/service.texi (Service): Switch the fsf.org link to https.
3476 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
3479 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3480 Define to speculation_safe_value_not_needed.
3482 2018-09-21 Florian Weimer <fweimer@redhat.com>
3485 * doc/extend.texi (Common Function Attributes): Mention that
3486 noreturn suppresses tail call optimization.
3488 2018-09-21 Jeff Law <law@redhat.com>
3490 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
3491 vr_values::cleanup_edges_and_switches.
3492 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
3494 (identify_jump_threads): Remove EDGE_IGNORE handling.
3495 (execute_vrp): Move handling of to_remove_edges and
3496 to_update_switch_stmts into vr_values class member functions.
3497 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
3498 (to_update_switch_stmts): Likewise.
3499 * vr-values.c: Include cfghooks.h.
3500 (vr_values::vr_values): Initialize to_remove_edges and
3501 to_update_switch_stmts.
3502 (vr_values::~vr_values): Verify to_remove_edges and
3503 to_update_switch_stmts are empty.
3504 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
3505 (vr_values::cleanup_edges_and_switches): New member function.
3506 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
3507 function. Add new data members.
3509 2018-09-21 David Malcolm <dmalcolm@redhat.com>
3511 PR tree-optimization/87309
3512 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
3513 calls with pflags and alt_flags.
3514 (selftest::test_capture_of_dump_calls): Add test of interaction of
3515 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
3517 2018-09-21 Olivier Hainque <hainque@adacore.com>
3519 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
3521 2018-09-21 Olivier Hainque <hainque@adacore.com>
3523 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
3525 2018-09-21 Olivier Hainque <hainque@adacore.com>
3527 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
3528 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
3529 (VXWORKS_LIBS_RTP): Minor reordering.
3531 2018-09-21 Olivier Hainque <hainque@adacore.com>
3533 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
3534 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
3536 2018-09-21 Olivier Hainque <hainque@adacore.com>
3538 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
3539 (PTRDIFF_TYPE): Likewise.
3541 2018-09-21 Olivier Hainque <hainque@adacore.com>
3543 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
3544 triplet, similar to support for VxWorks7.
3545 * config/vxworks-dummy.h: Provide a default definition
3546 of TARGET_VXWORKS64 to 0.
3548 2018-09-21 Olivier Hainque <hainque@adacore.com>
3550 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
3551 * config/vxworks-dummy.h: here.
3553 2018-09-21 Olivier Hainque <hainque@adacore.com>
3555 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
3557 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
3558 Bo Zhou <zbo.zhou@hisilicon.com>
3560 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
3561 * config/aarch64/aarch64-tune.md: Regenerated.
3562 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
3563 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
3564 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
3566 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
3567 Julian Brown <julian@codesourcery.com>
3569 * builtins.c (get_builtin_sync_mem): Handle address spaces.
3571 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
3573 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
3574 if the call takes a static chain.
3576 2018-09-21 Martin Liska <mliska@suse.cz>
3578 * auto-profile.c (autofdo_source_profile::read): Do not
3580 (read_profile): Do not add working sets.
3581 (read_autofdo_file): Remove sum_all.
3582 (afdo_callsite_hot_enough_for_early_inline): Remove const
3584 * coverage.c (struct counts_entry): Remove gcov_summary.
3585 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
3586 do not support GCOV_TAG_PROGRAM_SUMMARY.
3587 (get_coverage_counts): Remove summary and expected
3589 * coverage.h (get_coverage_counts): Likewise.
3590 * doc/gcov-dump.texi: Remove -w option.
3591 * gcov-dump.c (dump_working_sets): Remove.
3592 (main): Do not support '-w' option.
3593 (print_usage): Likewise.
3594 (tag_summary): Likewise.
3595 * gcov-io.c (gcov_write_summary): Do not dump
3597 (gcov_read_summary): Likewise.
3598 (gcov_histo_index): Remove.
3599 (gcov_histogram_merge): Likewise.
3600 (compute_working_sets): Likewise.
3601 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
3603 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
3604 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
3605 (GCOV_HISTOGRAM_SIZE): Remove.
3606 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
3607 (struct gcov_summary): Simplify rapidly just
3608 to runs and sum_max fields.
3609 (gcov_histo_index): Remove.
3610 (NUM_GCOV_WORKING_SETS): Likewise.
3611 (compute_working_sets): Likewise.
3612 * gcov-tool.c (print_overlap_usage_message): Remove
3613 trailing empty line.
3614 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
3615 (output_lines): Remove program related line.
3616 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
3617 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
3619 (input_profile_summary): Do not read it.
3620 (merge_profile_summaries): And do not merge it.
3621 (input_symtab): Do not call removed function.
3622 * modulo-sched.c (sms_schedule): Do not print sum_max.
3623 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
3624 removed when histogram method was invented.
3625 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
3627 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
3629 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
3630 and dump selected value.
3631 * profile.c (add_working_set): Remove.
3632 (get_working_sets): Likewise.
3633 (find_working_set): Likewise.
3634 (get_exec_counts): Do not work with working sets.
3635 (read_profile_edge_counts): Do not inform as sum_max is removed.
3636 (compute_branch_probabilities): Likewise.
3637 (compute_value_histograms): Remove argument for call of
3638 get_coverage_counts.
3639 * profile.h: Do not make gcov_summary const.
3641 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
3643 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
3645 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
3647 PR tree-optimization/86990
3648 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
3649 Check that the entire merged store group is made of constants only for
3652 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
3654 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
3655 (VTABLE_VERIFICATION_SPEC): Likewise.
3656 (SANITIZER_EARLY_SPEC): Likewise.
3657 (SANITIZER_SPEC): Likewise.
3658 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
3659 * doc/invoke.texi (Link Options): Document -r.
3661 2018-09-20 Richard Biener <rguenther@suse.de>
3664 * gimplify.c (gimplify_expr): Retain alignment of
3665 addressable lvalue in dereference.
3667 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
3670 * configure.ac: Check for .loc is_stmt support.
3671 * configure, config.in: Rebuilt.
3672 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
3675 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3677 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
3678 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
3681 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3683 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
3685 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
3688 2018-09-20 Marek Polacek <polacek@redhat.com>
3690 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
3692 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
3694 PR tree-optimization/87288
3695 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
3696 into account when determining PEELING_FOR_NITERS.
3698 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
3700 PR tree-optimization/86877
3701 * tree-vect-loop.c (vect_analyze_loop_2): Call
3702 vect_verify_datarefs_alignment.
3704 2018-09-19 Marek Polacek <polacek@redhat.com>
3706 * doc/invoke.texi: Document -Wclass-conversion.
3708 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
3710 * config/pa/pa.c (pa_adjust_priority): Delete.
3711 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
3713 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
3714 (atomic_storehi): Likewise.
3715 (atomic_storesi): Likewise.
3716 (atomic_loaddi): Restore compare and swap exchange loop code.
3718 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
3720 PR rtl-optimization/86902
3721 * combine.c (try_combine): When changing the CC mode used, don't change
3722 an unrelated mode in other_insn to that new CC mode.
3724 2018-09-19 David Malcolm <dmalcolm@redhat.com>
3726 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
3727 with %T in place of calls to dump_generic_expr.
3728 (prune_runtime_alias_test_list): Likewise.
3729 (create_runtime_alias_checks): Likewise.
3730 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
3731 (vect_analyze_data_ref_dependence): Likewise.
3732 (vect_slp_analyze_data_ref_dependence): Likewise.
3733 (vect_record_base_alignment): Likewise. Use %G in place of call
3734 to dump_gimple_stmt.
3735 (vect_compute_data_ref_alignment): Likewise.
3736 (verify_data_ref_alignment): Likewise.
3737 (vect_find_same_alignment_drs): Likewise.
3738 (vect_analyze_group_access_1): Likewise.
3739 (vect_analyze_data_ref_accesses): Likewise.
3740 (dependence_distance_ge_vf): Likewise.
3741 (dump_lower_bound): Likewise.
3742 (vect_prune_runtime_alias_test_list): Likewise.
3743 (vect_find_stmt_data_reference): Likewise.
3744 (vect_analyze_data_refs): Likewise.
3745 (vect_create_addr_base_for_vector_ref): Likewise.
3746 (vect_create_data_ref_ptr): Likewise.
3747 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
3748 (vect_can_advance_ivs_p): Likewise.
3749 (vect_update_ivs_after_vectorizer): Likewise.
3750 (vect_gen_prolog_loop_niters): Likewise.
3751 (vect_prepare_for_masked_peels): Likewise.
3752 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
3753 (vect_determine_vectorization_factor): Likewise.
3754 (vect_is_simple_iv_evolution): Likewise.
3755 (vect_analyze_scalar_cycles_1): Likewise.
3756 (vect_analyze_loop_operations): Likewise.
3757 (report_vect_op): Likewise.
3758 (vect_is_slp_reduction): Likewise.
3759 (check_reduction_path): Likewise.
3760 (vect_is_simple_reduction): Likewise.
3761 (vect_create_epilog_for_reduction): Likewise.
3762 (vect_finalize_reduction:): Likewise.
3763 (vectorizable_induction): Likewise.
3764 (vect_transform_loop_stmt): Likewise.
3765 (vect_transform_loop): Likewise.
3766 (optimize_mask_stores): Likewise.
3767 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
3768 (vect_split_statement): Likewise.
3769 (vect_recog_over_widening_pattern): Likewise.
3770 (vect_recog_average_pattern): Likewise.
3771 (vect_determine_min_output_precision_1): Likewise.
3772 (vect_determine_precisions_from_range): Likewise.
3773 (vect_determine_precisions_from_users): Likewise.
3774 (vect_mark_pattern_stmts): Likewise.
3775 (vect_pattern_recog_1): Likewise.
3776 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
3777 (vect_record_max_nunits): Likewise.
3778 (vect_build_slp_tree_1): Likewise.
3779 (vect_build_slp_tree_2): Likewise.
3780 (vect_print_slp_tree): Likewise.
3781 (vect_analyze_slp_instance): Likewise.
3782 (vect_detect_hybrid_slp_stmts): Likewise.
3783 (vect_detect_hybrid_slp_1): Likewise.
3784 (vect_slp_analyze_operations): Likewise.
3785 (vect_slp_analyze_bb_1): Likewise.
3786 (vect_transform_slp_perm_load): Likewise.
3787 (vect_schedule_slp_instance): Likewise.
3788 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
3789 (vect_mark_stmts_to_be_vectorized): Likewise.
3790 (vect_init_vector_1): Likewise.
3791 (vect_get_vec_def_for_operand): Likewise.
3792 (vect_finish_stmt_generation_1): Likewise.
3793 (vect_check_load_store_mask): Likewise.
3794 (vectorizable_call): Likewise.
3795 (vectorizable_conversion): Likewise.
3796 (vectorizable_operation): Likewise.
3797 (vectorizable_load): Likewise.
3798 (vect_analyze_stmt): Likewise.
3799 (vect_is_simple_use): Likewise.
3800 (vect_get_vector_types_for_stmt): Likewise.
3801 (vect_get_mask_type_for_stmt): Likewise.
3802 * tree-vectorizer.c (increase_alignment): Likewise.
3804 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
3806 * doc/rtl.texi: Adjust vec_select description.
3807 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
3808 non-constant selectors.
3810 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
3812 * config/aarch64/aarch64-protos.h
3813 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
3814 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
3815 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
3816 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
3817 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
3818 (AARCH64_FL_PROFILE): Move index so flags are ordered.
3819 (AARCH64_ISA_RCPC8_4): New flag.
3820 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
3821 to aarch64_offset_9bit_signed_unscaled_p.
3822 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
3824 * config/aarch64/constraints.md (Ust): New constraint.
3825 * config/aarch64/predicates.md.
3826 (aarch64_9bit_offset_memory_operand): New predicate.
3827 (aarch64_rcpc_memory_operand): New predicate.
3829 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
3831 PR rtl-optimization/87361
3832 * rtlanal.c (nonzero_bits1): Revert accidental change.
3834 2018-09-19 Richard Biener <rguenther@suse.de>
3836 PR tree-optimization/87349
3837 PR tree-optimization/87342
3838 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
3840 2018-09-18 Marek Polacek <polacek@redhat.com>
3842 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
3843 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
3845 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
3847 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
3849 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
3851 PR rtl-optimization/86882
3852 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
3854 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
3856 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
3857 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
3859 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
3862 * doc/invoke.texi (Link Options): Fix formatting and grammar.
3864 2018-09-18 Richard Biener <rguenther@suse.de>
3867 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
3868 (compute_samebase_partition_bases): Likewise.
3869 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
3870 (gimple_can_coalesce_p): Simplify.
3872 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
3874 Handle a library implementation of ffs calling __builtin_ffs.
3875 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
3876 (mmix_init_libfuncs): New function: make __builtin_ffs expand
3879 2018-09-17 David Malcolm <dmalcolm@redhat.com>
3881 * diagnostic-show-locus.c (class layout_range): Add field
3883 (layout_range::layout_range): Add "original_idx" param and use it
3884 to initialize new field.
3885 (make_range): Use 0 for original_idx.
3886 (layout::layout): Pass in index to calls to
3887 maybe_add_location_range.
3888 (layout::maybe_add_location_range): Add param "original_idx" and
3889 pass it on to layout_range.
3890 (layout::print_any_labels): Pass on range->m_original_idx to
3892 (gcc_rich_location::add_location_if_nearby): Use 0 for
3894 * gcc-rich-location.h (text_range_label::get_text): Update for new
3896 (range_label_for_type_mismatch::get_text): Likewise.
3898 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
3900 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
3902 2018-09-17 David Malcolm <dmalcolm@redhat.com>
3904 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
3905 format_string_diagnostic_t.
3906 (fmtwarn_n): Likewise.
3907 * substring-locations.c
3908 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
3909 (format_warning_n_va): Convert to...
3910 (format_string_diagnostic_t::emit_warning_n_va): ...this.
3911 (format_warning_va): Convert to...
3912 (format_string_diagnostic_t::emit_warning_va): ...this.
3913 (format_warning_at_substring): Convert to...
3914 (format_string_diagnostic_t::emit_warning): ...this.
3915 (format_warning_at_substring_n): Convert to...
3916 (format_string_diagnostic_t::emit_warning_n): ...this.
3917 * substring-locations.h (class format_string_diagnostic_t): New
3919 (format_warning_va): Convert to
3920 format_string_diagnostic_t::emit_warning_va.
3921 (format_warning_n_va): Convert to
3922 format_string_diagnostic_t::emit_warning_n_va.
3923 (format_warning_at_substring): Convert to
3924 format_string_diagnostic_t::emit_warning.
3925 (format_warning_at_substring_n): Convert to
3926 format_string_diagnostic_t::emit_warning_n.
3928 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
3929 Bernd Schmidt <bernds_cb1@t-online.de>
3931 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
3934 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
3936 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
3937 operand 0 predicate to nonimmediate operand.
3938 (rint<mode>2_frndint): Remove insn pattern.
3939 (rint<mode>2): Change operand 1 predicate to general_operand.
3940 Extend operand 1 to XFmode and generate rintxf2 insn.
3941 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
3942 Do not use X87MODEF mode macro.
3943 (frndintxf2_<rounding>_i387): Rename from
3944 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
3945 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
3946 to XFmode and generate significandxf3 insn.
3948 2018-09-17 Richard Biener <rguenther@suse.de>
3950 PR tree-optimization/87328
3951 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
3952 visiting unexecutable backedges when not iterating.
3953 (do_rpo_vn): Mark all edges not executable even when not
3956 2018-09-17 Martin Jambor <mjambor@suse.cz>
3959 * doc/invoke.texi (Warning Options): Likewise.
3961 2018-09-17 Richard Biener <rguenther@suse.de>
3963 PR tree-optimization/87301
3964 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
3965 clean EH info from leftover copy assignments.
3967 2018-09-17 Martin Liska <mliska@suse.cz>
3969 PR gcov-profile/85871
3970 * gcov.c (output_intermediate_file): Fix out of bounds
3973 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
3975 * config/arc/arc.c: Object attributes for core4 not reflected
3977 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
3980 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
3982 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
3984 2018-09-17 Martin Liska <mliska@suse.cz>
3986 * doc/gcov.texi: Document new option --use-hotness-colors.
3987 * gcov.c (struct source_info): Declare new field.
3988 (source_info::source_info): Set default for maximum_count.
3989 (print_usage): Add new -q option.
3990 (process_args): Process it.
3991 (accumulate_line_info): Save src->maximum_count.
3992 (output_line_beginning): Make color line number if
3993 flag_use_hotness_colors is set.
3994 (output_line_details): Pass default argument value.
3995 (output_lines): Pass src->maximum_count.
3997 2018-09-17 Martin Liska <mliska@suse.cz>
3999 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
4000 Use processor_names table.
4001 * config/i386/i386.c (ix86_default_align): Use
4002 processor_cost_table for alignment values.
4003 (ix86_option_override_internal): Use processor_names.
4004 (ix86_function_specific_print): Likewise.
4005 * config/i386/i386.h (struct processor_costs):
4006 Add alignment values.
4007 (struct ptt): Remove and replace with const char *.
4008 * config/i386/x86-tune-costs.h (struct processor_costs):
4009 Declare default alignments for all costs.
4011 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
4013 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
4014 symbolics or VR_VARYING ranges for ABS_EXPR.
4015 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
4016 when range will wrap.
4018 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
4021 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
4022 before and after a JUMP_TABLE_DATA.
4024 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
4027 * dojump.c (do_compare_and_jump): Canonicalize function pointers
4028 when one operand is a function pointer. Use POINTER_TYPE_P and
4029 FUNC_OR_METHOD_TYPE_P.
4030 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
4031 * fold-const.c (build_range_check): Likewise.
4032 * match.pd (simple_comparison): Likewise.
4034 2018-09-14 David Malcolm <dmalcolm@redhat.com>
4037 * spellcheck.c (get_edit_distance_cutoff): New function.
4038 (selftest::test_edit_distance_unit_test_oneway): Rename to...
4039 (selftest::test_get_edit_distance_one_way): ...this.
4040 (selftest::test_get_edit_distance_unit): Rename to...
4041 (selftest::test_get_edit_distance_both_ways): ...this.
4042 (selftest::test_edit_distances): Move tests to this new function,
4043 and test some more pairs of strings. Update for above renaming.
4044 (selftest::get_old_cutoff): New function.
4045 (selftest::test_get_edit_distance_cutoff): New function.
4046 (selftest::assert_suggested_for): New function.
4047 (ASSERT_SUGGESTED_FOR): New macro.
4048 (selftest::assert_not_suggested_for): New function.
4049 (ASSERT_NOT_SUGGESTED_FOR): New macro.
4050 (selftest::test_suggestions): New function.
4051 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
4052 tests to selftest::test_edit_distances and call it. Add calls to
4053 selftest::test_get_edit_distance_cutoff and
4054 selftest::test_suggestions.
4055 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
4056 (best_match::consider): Replace hard-coded cutoff calculation with
4058 (best_match::get_cutoff): New declaration.
4059 (best_match::get_best_meaningful_candidate): Likewise.
4061 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
4063 * builtins.c (fold_builtin_strlen): Remove TODO comment.
4065 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
4068 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
4070 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
4071 terminated string literal.
4073 2018-09-14 Martin Sebor <msebor@redhat.com>
4075 * builtins.c (unterminated_array): Handle ARRAY_REF.
4076 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
4077 * builtins.h (unterminated_array): Declare extern.
4078 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
4080 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
4083 2018-09-14 Martin Sebor <msebor@redhat.com>
4084 Jeff Law <law@redhat.com>
4086 * builtins.c (unterminated_array): New.
4087 (expand_builtin_strcpy): Adjust.
4088 (expand_builtin_strcpy_args): Detect unterminated arrays.
4089 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
4090 unterminated arrays.
4091 * gimple-fold.h (get_maxval_strlen): Add argument.
4092 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
4094 * gimple-fold.c (get_range_strlen): Add argument.
4095 (get_maxval_strlen): Adjust.
4096 * gimple-fold.h (get_range_strlen): Add argument.
4098 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
4100 * config/i386/movdirintrin.h: Fix copyright year.
4102 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
4104 * reg-stack.c: Include regs.h.
4105 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
4106 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
4107 FIRST_STACK_REG, not DFmode.
4108 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
4109 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
4110 (change stack): Default register mode to the reg_raw_mode of
4111 FIRST_STACK_REG, not DFmode.
4112 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
4113 (*swapxf): Rename from swapxf.
4115 2018-09-14 Carl Love <cel@us.ibm.com>
4117 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
4118 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
4120 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
4123 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
4126 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
4129 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
4131 2018-09-14 Jason Merrill <jason@redhat.com>
4133 Fix --enable-gather-detailed-mem-stats.
4134 * hash-table.c (hash_table_usage): Change from variable to function.
4135 * hash-table.h: Adjust.
4136 * Makefile.in: Add missing dependencies on hash-table.h.
4138 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4140 PR tree-optimization/87259
4142 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
4143 execute_cse_reciprocals_1 has tried transforming.
4145 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
4147 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
4148 VR_VARYING for PLUS/MINUS_EXPR.
4150 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
4152 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
4155 2018-09-14 Richard Biener <rguenther@suse.de>
4158 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
4159 bits for the merged partition.
4161 2018-09-13 Martin Sebor <msebor@redhat.com>
4162 Bernd Edlinger <bernd.edlinger@hotmail.de>
4164 * builtins.h (c_srlen): Add argument.
4165 * builtins.c (warn_string_no_nul): New function.
4166 (c_strlen): Add argument and use it. Update recursive calls.
4167 Pass DECL argument to string_constant to get info on non
4168 terminated strings. Update *NONSTR as needed.
4169 (fold_builtin_strlen): Add argument to calls to c_strlen.
4170 Warn for unterminated arrays.
4171 (warn_string_no_null): Add prototype.
4172 * expr.c (string_constant): Update arguments. Update recursive
4173 calls appropriately. Detect missing NUL terminator and outermost
4174 declaration its missing in.
4175 Improve checks for arrays with nonzero lower bound or elements
4176 that are not a single byte. Simplify offset computation.
4177 Simplify checks for non-NUL terminated strings.
4178 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
4179 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
4181 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
4183 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
4185 * fold-const.c (c_getstr): Fix function comment. Remove unused third
4186 argument. Fix range checks.
4187 * fold-const.h (c_getstr): Adjust protoype.
4188 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
4189 string is constant but contains no NUL byte.
4191 * expr.c (string_constant): Adjust function comment.
4192 Remove bogus check for zero termination.
4194 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
4196 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
4197 (get_constant_size): Don't make STRING_CSTs larger than they are.
4198 (check_string_literal): New check function for STRING_CSTs.
4199 (output_constant): Use it.
4201 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
4204 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
4206 2018-09-13 Richard Biener <rguenther@suse.de>
4208 PR tree-optimization/87263
4209 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
4210 (struct unwind_state): Add max_rpo field.
4211 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
4212 Compute max_rpo, the max RPO number a block can be backwards reached
4213 from. Re-write non-iterating mode to a RPO ordered worklist approach,
4214 separating it from the iterating mode.
4216 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
4218 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
4219 (rfs_decision): New scheduling decision.
4221 2018-09-13 Richard Biener <rguenther@suse.de>
4224 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
4225 (vn_nary_op_insert_pieces_predicated): Do not write useless
4226 valid_dominated_by_p entry outside of the allocated storage.
4228 2018-09-13 Omar Sandoval <osandov@osandov.com>
4229 Tom de Vries <tdevries@suse.de>
4232 * dwarf2out.c (is_c): New function.
4233 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
4235 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
4238 * config/aarch64/aarch64.md (*aarch64_bfxil):
4240 * config/aarch64/constraints.md (Ulc): Define.
4241 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
4243 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
4246 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
4248 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
4249 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
4250 aarch64_layout_frame call.
4251 (aarch64_expand_epilogue): Likewise.
4252 (aarch64_initial_elimination_offset): Likewise.
4253 (aarch64_get_separate_components): Likewise.
4254 (aarch64_use_return_insn_p): Likewise.
4255 (aarch64_layout_frame): Remove unneeded check.
4257 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4259 * configure.ac: Only append
4260 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
4261 gcc_config_arguments if it was never reconfigured or last reconfigure
4262 was with different arguments.
4263 * configure: Regenerated.
4265 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4266 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4269 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
4270 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
4272 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4274 PR tree-optimization/87287
4275 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
4276 X % C == 0 to X % (unsigned) C == 0 optimization to ...
4277 * match.pd (X % C == 0): ... here. New optimization.
4279 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4282 * expr.h (maybe_optimize_mod_cmp): Declare.
4283 * expr.c (mod_inv): New function.
4284 (maybe_optimize_mod_cmp): New function.
4285 (do_store_flag): Use it.
4286 * cfgexpand.c (expand_gimple_cond): Likewise.
4288 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
4289 Julian Brown <julian@codesourcery.com>
4292 * gimplify.c (gimplify_scan_omp_clauses): Set
4293 target_firstprivatize_array_bases in OpenACC parallel and kernels
4294 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
4295 OpenACC data regions.
4297 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
4299 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
4300 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
4302 2018-09-12 Richard Biener <rguenther@suse.de>
4304 PR tree-optimization/87280
4305 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
4306 edge but unreachable target.
4307 (do_rpo_vn): For conservatively handling a PHI only mark
4308 the backedge executable but not the block reachable.
4310 2018-09-12 Richard Biener <rguenther@suse.de>
4312 PR tree-optimization/87266
4313 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
4316 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
4318 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
4320 ("trunc<BFP:mode><DFP_ALL:mode>2")
4321 ("trunc<DFP_ALL:mode><BFP:mode>2")
4322 ("extend<BFP:mode><DFP_ALL:mode>2")
4323 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
4324 according to the target operand type.
4326 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4327 Andreas Krebbel <krebbel@linux.ibm.com>
4329 PR tree-optimization/86844
4330 * gimple-ssa-store-merging.c
4331 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
4332 there are any overlapping stores in between them, make sure they are
4333 also coalesced or we give up completely.
4335 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4338 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
4339 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
4341 2018-09-12 Tom de Vries <tdevries@suse.de>
4343 * common.opt (gdescribe-dies): Add option.
4344 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
4345 attribute for artifical and nameless decls.
4346 (dwarf2out_register_external_die): Add description attribute to
4347 external reference die.
4348 (add_desc_attribute): New functions.
4349 (gen_subprogram_die): Add description attribute to
4350 DW_TAG_call_site_parameter.
4351 * tree-pretty-print.c (print_generic_expr_to_str): New function.
4352 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
4353 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
4355 (@item -gdescribe-dies): Add.
4357 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
4359 * tree-vrp.c (vrp_shift_undefined_p): Remove.
4360 (extract_range_from_binary_expr_1: Call
4361 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
4362 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
4365 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
4367 * gimple-ssa-warn-alloca.c
4368 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
4369 field for ALLOCA_BOUND_*_LARGE.
4371 2018-09-11 Nathan Sidwell <nathan@acm.org>
4373 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
4375 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4377 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
4378 for clobbers. Remove obsolete comment.
4380 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4382 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
4383 mpxchk, mpxld and mpxst types.
4384 (define_attr length_immediate): Remove all processing of mpx types.
4385 (define_attr prefix_0f): Ditto.
4386 (define_attr memory): Ditto.
4388 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4390 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
4391 (log<mode>2): Change operand 1 predicate to general_operand.
4392 Extend operand 1 to XFmode and generate logxf3 insn.
4393 (log10<mode>2): Change operand 1 predicate to general_operand.
4394 Extend operand 1 to XFmode and generate log10xf3 insn.
4395 (log2<mode>2): Change operand 1 predicate to general_operand.
4396 Extend operand 1 to XFmode and generate log2xf3 insn.
4397 (fyl2xp1_extend<mode>xf3_i387): Remove.
4398 (log1p<mode>2): Change operand 1 predicate to general_operand.
4399 Extend operand 1 to XFmode and generate log1pxf3 insn.
4400 (fxtract_extend<mode>xf3_i387): Remove.
4401 (logb<mode>2): Change operand 1 predicate to general_operand.
4402 Extend operand 1 to XFmode and generate logbxf3 insn.
4403 (ilogb<mode>2): Change operand 1 predicate to general_operand.
4404 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
4405 (significand<mode>2): Change operand 1 predicate to general_operand.
4406 Extend operand 1 to XFmode and generate significandxf3 insn.
4408 2018-09-11 Nathan Sidwell <nathan@acm.org>
4410 * gcc.c (perror_with_name, pfatal_with_name): Delete.
4411 (load_specs): Use fatal_error.
4412 (DELETE_IF_ORDINARY, process_command): Use error.
4413 (execute, run_attempt): Use fatal_error.
4415 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
4417 * diagnostic-core.h (sorry_at): New prototype.
4418 * diagnostic.c (sorry_at): New function.
4420 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
4422 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
4423 by zero as VR_UNDEFINED.
4425 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
4427 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
4428 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
4429 (<sincos>mode2): New expander.
4430 (sincos_extend<mode>xf3_i387): Remove insn pattern.
4431 (sincos -> sin, cos splitters): Remove splitter patterns.
4432 (sincos<mode>3): Change operand 2 predicate to general_operand.
4433 Extend operand 2 to XFmode and generate sincosxf3 insn.
4434 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
4435 Change operand 3 predicate to const1_operand.
4436 (fptan_extend<mode>xf4_i387): Remove insn pattern.
4437 (tanxf2): Update operands in the call to fptanxf4_i387.
4438 (tan<mode>2): Change operand 1 predicate to general_operand.
4439 Extend operand 1 to XFmode and generate tanxf3 insn.
4440 (atan2xf3): Rename from *fpatanxf3_i387.
4441 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
4442 (atan2xf3): Remove expander.
4443 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
4444 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
4445 (atan<mode>2): Change operand 1 predicate to general_operand.
4446 Extend operand 1 to XFmode and generate atanxf3 insn.
4448 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
4450 * config/i386/i386.md (x87/SSE constant load splitter): Use
4451 memory_operand instead of nonimmediate_operand for input operand
4454 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
4456 * config/i386/i386.md (float partial SSE register stall splitter): Move
4457 splitter near its instruction pattern.
4458 (float_extend partial SSE register stall splitter): Ditto.
4459 (float_truncate partial SSE register stall splitter): Ditto.
4461 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
4464 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
4465 to speculation_safe_value_not_needed.
4468 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
4469 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
4470 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
4471 leaf_function_p, instead use has_hard_reg_initial_val.
4473 2018-09-09 Nathan Sidwell <nathan@acm.org>
4475 * gcc.h (pfatal_with_name): Don't declare here.
4476 * gcc.c (pfatal_with_name): Make static.
4478 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
4480 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
4483 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
4485 PR rtl-optimization/85458
4486 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
4487 priority hook to reduce the priority of EXPR.
4489 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
4491 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
4492 DImode for x87 on 32bit targets. Conditionally disable x87 modes
4493 with X87_ENABLE_FLOAT. Remove preparation code.
4494 (*float<SWI48:mode><MODEF:mode>2): Rename from
4495 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
4496 math using "enabled" attribute.
4497 (*floatdi<MODEF:mode>2_i387): Rename from
4498 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
4499 enable for 32bit targets only.
4500 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
4502 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
4503 as operand 1 predicate. Rewrite as define_insn_and_split.
4504 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
4506 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
4508 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
4509 to fallthru to FLOAT case.
4511 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
4514 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
4515 around folding of vec_sl to handle out of range shift values.
4517 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
4519 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
4520 Update callers to gen_fix_trunc<mode>_i387_fisttp
4521 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
4522 nonimmediate_operand.
4523 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
4524 and corresponding splitters.
4525 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
4526 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
4527 (fix_truncdi_i387_with_temp): Remove insn pattern
4528 and corresponding splitters.
4529 (fix_trunc<mode>_i387): Change operand 0 predicate to
4530 nonimmediate_operand.
4531 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
4532 and corresponding splitters.
4533 (*fistdi2_1): Remove.
4535 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
4536 (lrintxfdi2): Remove expander. Reimplement as define_insn.
4537 (*fist<mode>2_1): Remove.
4538 (fist<mode>2): Ditto.
4539 (fist<mode>2_with_temp): Remove insn pattern and corresponding
4541 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
4542 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
4543 (fistdi2_<rounding>): Change operand 0 predicate to
4544 nonimmediate_operand.
4545 (fistdi2_<rounding>_with_temp): Remove insn pattern
4546 and corresponding splitters.
4547 (fist<mode>2_<rounding>): Change operand 0 predicate to
4548 nonimmediate_operand.
4549 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
4550 and corresponding splitters.
4552 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
4554 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
4556 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
4559 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
4561 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4562 early gimple folding of vec_splat().
4563 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
4564 * gimple-fold.h: Add an extern define for tree_vec_extract().
4566 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
4568 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
4569 wrappers around TREE_TYPE comparisons.
4571 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
4574 * config/s390/predicates.md: Add nonsym_memory_operand.
4575 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
4576 contains a SYMBOL_REF, load it into an intermediate pseudo.
4577 (s390_emit_compare_and_swap): Legitimize operand.
4578 * config/s390/s390.md: Use the new nonsym_memory_operand
4579 with UNSPECV_CAS patterns.
4581 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
4584 * config/s390/s390-passes.def: New file.
4585 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
4587 (make_pass_s390_early_mach): Add declaration.
4588 * config/s390/s390.c (make_pass_s390_early_mach):
4589 (s390_option_override): Remove dynamic registration.
4590 * config/s390/t-s390: Add s390-passes.def.
4592 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
4594 * config/s390/s390.c (s390_decompose_constant_pool_ref):
4595 Remove UNSPEC_LTREL_BASE check.
4596 (annotate_constant_pool_refs): Likewise.
4597 (find_constant_pool_ref): Likewise.
4598 (find_ltrel_base): Removed.
4599 (replace_ltrel_base): Removed.
4600 (s390_mainpool_finish): Remove replace_ltrel_base call.
4601 (s390_chunkify_start): Remove pending LTREL_BASE logic.
4602 (s390_chunkify_finish): Remove replace_ltrel_base call.
4603 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
4605 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
4608 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
4609 to speculation_safe_value_not_needed.
4611 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
4612 Bernd Schmidt <bernds_cb1@t-online.de>
4614 * config/nvptx/nvptx-opts.h: New file.
4615 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
4616 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
4618 (TARGET_SM35): New macro.
4619 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
4621 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
4623 (misa=): New option.
4624 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
4626 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
4628 * config/i386/i386.md (truncdfsf2): Remove expander.
4629 (truncdfsf2_with_temp): Ditto.
4630 (truncxf<mode>2): Ditto.
4631 (*truncdfsf_fast_mixed): Remove insn pattern.
4632 (*truncdfsf_fast_i387): Ditto.
4633 (*truncdfsf_mixed): Ditto.
4634 (*truncdfsf_i387): Ditto.
4635 (*truncdfsf2_i387_1): Ditto.
4636 (*truncxfsf2_mixed): Ditto.
4637 (*truncxfdf2_mixed): Ditto.
4638 (*truncxf<mode>2_i387_noop): Ditto. Update callers
4639 to call gen_truncxf<mode>2 instead.
4640 (*truncxf<mode>2_i387): Remove.
4641 (reg->reg splitters): Remove splitter pattern.
4642 (reg->mem splitters): Ditto.
4644 (truncdfsf2): New insn pattern.
4645 (truncxf<mode>2): Ditto.
4647 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4649 * tree-ssa-math-opts.c (is_mult_by): New function.
4650 (is_square_of): Use the above.
4651 (optimize_recip_sqrt): New function.
4652 (pass_cse_reciprocals::execute): Use the above.
4654 2018-09-05 Richard Biener <rguenther@suse.de>
4657 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
4658 to zero-init the emplaced vec.
4660 2018-09-05 Martin Liska <mliska@suse.cz>
4662 PR tree-optimization/87205
4663 * tree-switch-conversion.c (pass_lower_switch::execute):
4664 Group cases for switch statements.
4666 2018-09-05 Richard Biener <rguenther@suse.de>
4668 PR tree-optimization/87217
4669 * tree-ssa-sccvn.c (vuse_valueize): New.
4670 (vn_reference_lookup_pieces): Use it.
4671 (vn_reference_lookup): Likewise.
4673 2018-09-05 Nathan Sidwell <nathan@acm.org>
4676 * stor-layout.c (place_field): Scan forwards to check last
4677 bitfield when ms_bitfield_placement is in effect.
4679 2018-09-05 Richard Biener <rguenther@suse.de>
4682 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
4685 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
4686 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
4688 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
4689 * config.gcc (extra_objs): Build it.
4690 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
4692 * config/aarch64/aarch64-passes.def
4693 (pass_tag_collision_avoidance): New pass.
4694 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
4695 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
4696 (aarch64_classify_address): Remove static qualifier.
4697 (aarch64_address_info, aarch64_address_type): Move to...
4698 * config/aarch64/aarch64-protos.h: ... here.
4699 (make_pass_tag_collision_avoidance): New function.
4700 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
4703 2018-09-05 Martin Liska <mliska@suse.cz>
4705 * doc/gcov.texi: Update documentation of humar
4707 * gcov.c (format_count): Print one decimal place, it provides
4708 more fine number of situations like '1G' vs. '1.4G'.
4710 2018-09-05 Martin Liska <mliska@suse.cz>
4713 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
4714 * optc-gen.awk: Allow 'Var' for Deprecated options in order
4715 to generate a MASK value.
4717 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
4720 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
4721 if frame pointer isn't used.
4722 (compute_frame_pointer_to_fb_displacement): Likewise.
4724 2018-09-04 Jakub Jelinek <jakub@redhat.com>
4727 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
4728 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
4729 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
4730 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
4731 and OPTION_MASK_ISA_XSAVEC_UNSET.
4733 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
4735 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
4736 XOR operations in NAND case.
4738 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
4740 * wide-int-range.cc (wide_int_range_convert): New.
4741 * wide-int-range.h (wide_int_range_convert): New.
4742 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
4743 code into wide_int_range_convert.
4744 (extract_range_into_wide_ints): Do not munge anti range constants
4745 into the entire domain. Just return the range back.
4747 2018-09-04 Martin Liska <mliska@suse.cz>
4749 * genmatch.c (output_line_directive): Add new argument
4751 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
4753 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
4755 * doc/invoke.texi (Option Summary): Add whitespace.
4757 * doc/invoke.texi (Option Summary): Add -Waligned-new.
4759 2018-09-04 Richard Biener <rguenther@suse.de>
4761 PR tree-optimization/87211
4762 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
4763 backedge value we're supposed to treat as VARYING also number
4764 the PHI to VARYING in case it got a different value-number already.
4766 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
4768 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
4769 (extract_range_from_binary_expr_1): Do not call
4770 vrp_can_optimize_bit_op.
4771 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
4773 (wide_int_range_get_mask_and_bounds): New.
4774 (wide_int_range_optimize_bit_op): New.
4775 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
4776 (wide_int_range_bit_and): Same.
4777 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
4778 (wide_int_range_optimize_bit_op): New.
4779 (wide_int_range_get_mask_and_bounds): New.
4781 2018-09-04 Richard Biener <rguenther@suse.de>
4783 PR tree-optimization/87176
4784 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
4785 variable. When value-numbering a virtual PHI node make sure
4786 to not value-number to the backedge value.
4788 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
4790 * doc/extend.texi (Long Long, Hex Floats): Document support for
4791 long long and hex floats in more recent versions of ISO C++.
4793 2018-09-03 Richard Biener <rguenther@suse.de>
4795 PR tree-optimization/87177
4796 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
4799 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4801 * bb-reorder.c (edge_order): Convert to C-qsort-style
4802 tri-state comparator.
4803 (reorder_basic_blocks_simple): Change std::stable_sort to
4806 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4808 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
4809 tri-state comparator.
4810 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
4812 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4814 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
4815 (mergesort): ... here as maximum count for using netsort.
4816 (gcc_qsort): Set nlim to 3 if stable sort is requested.
4817 (gcc_stablesort): New.
4818 * system.h (gcc_stablesort): Declare.
4820 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4822 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
4823 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
4824 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
4826 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
4828 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
4829 lxsdx and stxsdx alternatives.
4830 (*mov<mode>_hardfloat64): Ditto.
4831 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
4833 2018-09-03 Richard Biener <rguenther@suse.de>
4835 PR tree-optimization/87200
4836 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
4839 2018-09-03 Martin Liska <mliska@suse.cz>
4841 PR tree-optimization/87201
4842 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
4843 Fix parenthesis in an expression.
4845 2018-09-03 Richard Biener <rguenther@suse.de>
4847 PR tree-optimization/87197
4848 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
4849 visited. CSE the VN_INFO hashtable lookup.
4851 PR tree-optimization/87169
4852 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
4853 iterating make sure there's no extra backedges from irreducible
4854 regions feeding the header. Mark the destination block
4857 2018-09-03 Martin Liska <mliska@suse.cz>
4860 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
4861 * common/common-targhooks.c (default_get_valid_option_values):
4863 * common/common-targhooks.h (default_get_valid_option_values):
4865 * common/config/i386/i386-common.c: Move processor_target_table
4867 (ix86_get_valid_option_values): New function.
4868 (TARGET_GET_VALID_OPTION_VALUES): New macro.
4869 * config/i386/i386.c (struct ptt): Move to i386-common.c.
4870 (PTA_*): Move all defined masks into i386-common.c.
4871 (ix86_function_specific_restore): Use new processor_cost_table.
4872 * config/i386/i386.h (struct ptt): Moved from i386.c.
4873 (struct pta): Likewise.
4874 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
4875 * doc/tm.texi.in: Likewise.
4876 * opt-suggestions.c (option_proposer::suggest_option):
4877 Pass prefix to build_option_suggestions.
4878 (option_proposer::get_completions): Likewise.
4879 (option_proposer::build_option_suggestions): Use the new target
4881 * opts.c (struct option_help_tuple): New struct.
4882 (print_filtered_help): Use the new target hook.
4884 2018-09-03 Martin Liska <mliska@suse.cz>
4887 * predict.c (set_even_probabilities): Add likely_edges
4888 argument and handle cases where we have precisely one
4890 (combine_predictions_for_bb): Catch also likely_edges.
4891 (tree_predict_by_opcode): Handle gswitch statements.
4892 * tree-cfg.h (find_case_label_for_value): New declaration.
4893 (find_taken_edge_switch_expr): Likewise.
4894 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
4895 Find pivot in decision tree based on probabily, not by number of
4898 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
4900 * doc/standards.texi (Standards): Update Objective-C reference.
4902 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
4904 * doc/install.texi (Prerequisites): Update link for MPC.
4906 2018-09-01 Michael Matz <matz@suse.de>
4908 PR tree-optimization/87074
4909 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
4910 PHIs for outer-loop uses.
4912 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
4914 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
4915 * doc/invoke.texi (C Dialect Options): Ditto.
4917 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
4919 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
4921 2018-08-31 Richard Biener <rguenther@suse.de>
4923 PR tree-optimization/87168
4924 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
4925 (rpo_elim::eliminate_avail): When OP was not visited it must
4928 2018-08-31 David Malcolm <dmalcolm@redhat.com>
4930 * tree-vrp.c (copy_value_range): Convert param "from" from
4931 "value_range *" to "const value_range *".
4932 (range_is_null): Likewise for param "vr".
4933 (range_int_cst_p): Likewise.
4934 (range_int_cst_singleton_p): Likewise.
4935 (symbolic_range_p): Likewise.
4936 (value_ranges_intersect_p): Likewise for both params.
4937 (value_range_nonnegative_p): Likewise for param "vr".
4938 (value_range_constant_singleton): Likewise.
4939 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
4940 (extract_range_into_wide_ints): Likewise for param "vr".
4941 (extract_range_from_multiplicative_op): Likewise for params "vr0"
4943 (vrp_can_optimize_bit_op): Likewise.
4944 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
4946 (extract_range_from_unary_expr): Likewise.
4947 (debug_value_range): Likewise for param "vr".
4948 (value_range::dump): Add "const" qualifier.
4949 (vrp_prop::check_array_ref): Convert local "vr" from
4950 "value_range *" to "const value_range *".
4951 (vrp_prop::check_mem_ref): Likewise.
4952 (vrp_prop::visit_stmt): Likewise for local "old_vr".
4953 (vrp_intersect_ranges_1): Likewise for param "vr_1".
4954 (vrp_intersect_ranges): Likewise.
4955 (simplify_stmt_for_jump_threading): Likewise for local "vr".
4956 (vrp_prop::vrp_finalize): Likewise.
4957 * tree-vrp.h (value_range::dump): Add "const" qualifier.
4958 (vrp_intersect_ranges): Add "const" qualifier to params as above.
4959 (extract_range_from_unary_expr): Likewise.
4960 (value_range_constant_singleton): Likewise.
4961 (symbolic_range_p): Likewise.
4962 (copy_value_range): Likewise.
4963 (extract_range_from_binary_expr_1): Likewise.
4964 (range_int_cst_p): Likewise.
4965 (vrp_set_zero_nonzero_bits): Likewise.
4966 (range_int_cst_singleton_p): Likewise.
4968 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
4970 * config/aarch64/arm_neon.h (vabsd_s64): New.
4971 (vnegd_s64): Likewise.
4973 2018-08-31 Martin Jambor <mjambor@suse.cz>
4975 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
4977 2018-08-31 Martin Liska <mliska@suse.cz>
4979 * ipa-icf.c (sem_item::add_type): Use
4980 sem_item::m_type_hash_cache.
4981 * ipa-icf.h: Move the cache from sem_item_optimizer
4984 2018-08-31 Nathan Sidwell <nathan@acm.org>
4986 * doc/extend.texi (Backwards Compatibility): Remove implicit
4987 extern C leeway of () being (...).
4989 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4991 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
4993 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
4997 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
4999 2018-08-31 Jakub Jelinek <jakub@redhat.com>
5002 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
5003 gen_int_mode. Formatting fixes.
5005 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
5007 * target.def (custom_function_descriptors): Improve documentation.
5008 * doc/tm.texi.in (Trampolines): Expand discussion of function
5009 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
5010 beginning of the section.
5011 * doc/tm.texi: Regenerated.
5013 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
5015 * cfg.h (class auto_edge_flag): Spell out the template-id of the
5016 base class in the initializer list. This is a workaround for
5017 building with older GCC.
5018 (class auto_bb_flag): Likewise.
5020 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
5022 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
5023 (altivec_vcmpequ<VI_char>_p): Remove star.
5024 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
5026 (expand_strncmp_vec_sequence): New function.
5027 (emit_final_str_compare_vec): New function.
5028 (expand_strn_compare): Add support for vector strncmp.
5029 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
5030 length specification to bytes.
5031 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
5032 (vcmpnezb_p): New pattern.
5033 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
5034 for option -mstring-compare-inline-limit.
5036 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
5038 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
5039 (PTA_SKYLAKE): Add PTA_AES.
5040 (PTA_GOLDMONT): Likewise.
5042 2018-08-29 Jan Hubicka <jh@suse.cz>
5045 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
5046 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
5048 2018-08-29 Jan Hubicka <jh@suse.cz>
5050 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
5052 (hash_tree): Do not visit TYPE_STUB_DECL.
5053 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
5054 stream TYPE_STUB_DECL.
5055 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
5056 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
5057 after free_lang_data.
5058 (type_in_anonymous_namespace_p): Likewise.
5060 2018-08-29 Jan Hubicka <jh@suse.cz>
5062 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
5063 comment that it has to be even number.
5064 (class sreal): Change m_sig type to int32_t.
5065 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
5066 int64_t for temporary calculations.
5067 (sreal_verify_basics): Drop one bit from minimum and maximum.
5069 2018-08-30 Richard Biener <rguenther@suse.de>
5071 PR tree-optimization/87147
5072 * tree-ssa-sccvn.c (SSA_VISITED): New function.
5073 (visit_phi): When the degenerate result is from the backedge and
5074 we didn't visit its definition yet drop to VARYING.
5075 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
5077 2018-08-29 Jan Hubicka <jh@suse.cz>
5079 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
5081 (hash_tree): Likewise.
5083 2018-08-29 Jan Hubicka <jh@suse.cz>
5085 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
5086 and TYPE_NEXT_REF_TO.
5088 2018-08-29 Jan Hubicka <jh@suse.cz>
5090 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
5091 comment that it has to be even number.
5092 (class sreal): Change m_sig type to int32_t.
5093 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
5094 int64_t for temporary calculations.
5095 (sreal_verify_basics): Drop one bit from minimum and maximum.
5097 2018-08-30 Tamar Christina <tamar.christina@arm.com>
5099 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
5101 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
5104 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
5105 if to_add is negative.
5107 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
5110 * builtins.c (c_strlen): Improve range checks.
5112 2018-08-29 Martin Sebor <msebor@redhat.com>
5113 Jeff Law <law@redhat.com>
5115 PR tree-optimization/86714
5116 PR tree-optimization/86711
5117 * builtins.c (c_strlen): Add arguments to call to string_constant.
5118 * expr.c (string_constant): Add argument. Detect missing nul
5119 terminator and outermost declaration it's missing in.
5120 * expr.h (string_constant): Add argument.
5121 * fold-const.c (read_from_constant_string): Add arguments to call to
5123 (c_getstr): Likewise.
5124 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
5126 * tree-ssa-strlen.c (get_stridx): Likewise.
5128 2018-08-29 Jan Hubicka <jh@suse.cz>
5130 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
5131 Do not stream DECL_VINDEX.
5132 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
5133 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
5134 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
5136 2018-08-29 Richard Biener <rguenther@suse.de>
5138 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
5139 virtual operands that are not default defs to honor region
5141 (rpo_vn_valueize): Remove ineffective code here.
5143 2018-08-29 Richard Biener <rguenther@suse.de>
5145 PR tree-optimization/87132
5146 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
5147 when skipping defs reachable over backedges.
5149 2018-08-29 Richard Biener <rguenther@suse.de>
5151 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
5152 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
5153 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
5154 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
5155 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
5156 (vn_reference_lookup_3): Remove use of const_parms.
5157 (free_rpo_vn): Do not free const_parms.
5158 (do_rpo_vn): Do not call init_const_parms.
5159 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
5160 SSA_NAME_POINTS_TO_READONLY_MEMORY.
5161 (call_may_clobber_ref_p_1): Likewise.
5163 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
5166 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
5168 (-ftree-scev-cprop): Document.
5170 2018-08-29 Jan Hubicka <jh@suse.cz>
5172 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
5174 (sreal constructor): Update.
5175 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
5176 sreal:operator/): Update.
5178 2018-08-29 Martin Liska <mliska@suse.cz>
5180 * tree-switch-conversion.c (switch_conversion::expand):
5181 Strenghten assumption about gswitch statements.
5183 2018-08-29 Richard Biener <rguenther@suse.de>
5185 PR tree-optimization/87117
5186 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
5187 re-value-number released SSA VDEFs.
5189 2018-08-29 Richard Biener <rguenther@suse.de>
5191 PR tree-optimization/87126
5192 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
5194 2018-08-28 Jim Wilson <jimw@sifive.com>
5196 * config/riscv/pic.md: Rewrite.
5197 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
5199 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
5200 (SOFTF, default_load, softload, softstore): New.
5202 2018-08-28 Jeff Law <law@redhat.com>
5204 * fold-const.c (fold_binary_loc): Remove recently added assert.
5206 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
5208 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
5209 to OP parmeter of generated function.
5211 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
5213 PR tree-optimization/87009
5214 * match.pd: Add boolean optimizations.
5216 2018-08-28 Martin Sebor <msebor@redhat.com>
5219 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
5220 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
5221 (pass_walloca::gate): Use it.
5222 (alloca_call_type): Same.
5223 (pass_walloca::execute): Same.
5224 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
5226 2018-08-28 David Malcolm <dmalcolm@redhat.com>
5228 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
5229 GCC_VERSION for usage of "__gcc_dump_printf__" format from
5232 2018-08-28 Richard Biener <rguenther@suse.de>
5234 PR tree-optimization/87124
5235 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
5236 constants before looking up avail.
5238 2018-08-28 Jakub Jelinek <jakub@redhat.com>
5241 * calls.c (maybe_warn_nonstring_arg): Punt early if
5242 warn_stringop_overflow is zero. Don't call get_range_strlen
5243 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
5244 Swap comparison operands to have constants on rhs. Only use
5245 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
5246 increment lenrng[0].
5248 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
5250 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
5251 use of tree_to_shwi. Remove duplicated test for the size being
5252 a whole number of bytes.
5254 2018-08-28 Richard Biener <rguenther@suse.de>
5256 PR tree-optimization/87117
5257 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
5258 Handle removed stmt without LHS (GIMPLE_NOP).
5260 2018-08-28 Richard Biener <rguenther@suse.de>
5262 PR tree-optimization/87117
5263 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
5264 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
5266 2018-08-28 Richard Biener <rguenther@suse.de>
5268 PR tree-optimization/87117
5269 * tree-ssa-pre.c (compute_avail): Do not make expressions
5270 with predicated values available.
5271 (get_expr_value_id): Assert we do not run into predicated value
5274 2018-08-28 Richard Biener <rguenther@suse.de>
5276 PR tree-optimization/87117
5277 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
5278 get virtual operands.
5279 (get_expr_operands): Handle STRING_CST like other decls.
5281 2018-08-28 Martin Liska <mliska@suse.cz>
5283 * tree.h: Update documentation of fndecl_built_in_p
5287 2018-08-27 Jeff Law <law@redhat.com>
5288 PR tree-optimization/87110
5289 * tree-ssa-dse.c (compute_trims): Handle non-constant
5292 2018-08-27 Martin Sebor <msebor@redhat.com>
5294 PR tree-optimization/86914
5295 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
5297 2018-08-27 Martin Sebor <msebor@redhat.com>
5299 PR tree-optimization/87112
5300 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
5301 the type of the bound argument.
5303 2018-08-27 Jeff Law <law@redhat.com>
5305 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
5306 type does not have a TYPE_SIZE_UNIT.
5308 2018-08-27 Steve Ellcey <sellcey@cavium.com>
5310 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
5311 with include of backend.h.
5313 2018-08-27 Richard Biener <rguenther@suse.de>
5315 PR tree-optimization/86927
5316 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
5317 use const cond reduction code.
5319 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
5321 PR tree-optimization/85758
5322 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
5324 2018-08-27 David Malcolm <dmalcolm@redhat.com>
5327 * diagnostic-show-locus.c (class layout_range): Update for
5328 conversion of show_caret_p to a tri-state.
5329 (layout_range::layout_range): Likewise.
5330 (make_range): Likewise.
5331 (layout::maybe_add_location_range): Likewise.
5332 (layout::should_print_annotation_line_p): Don't show annotation
5333 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
5334 (layout::get_state_at_point): Update for conversion of
5335 show_caret_p to a tri-state. Bail out early for
5336 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
5337 underlining or source colorization.
5338 (gcc_rich_location::add_location_if_nearby): Update for conversion
5339 of show_caret_p to a tri-state.
5340 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
5341 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
5343 (selftest::test_one_liner_labels): Likewise.
5344 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
5345 conversion of show_caret_p to a tri-state.
5346 * pretty-print.c (text_info::set_location): Likewise.
5347 * pretty-print.h (text_info::set_location): Likewise.
5348 * substring-locations.c (format_warning_n_va): Likewise.
5349 * tree-diagnostic.c (default_tree_printer): Likewise.
5350 * tree-pretty-print.c (newline_and_indent): Likewise.
5352 2018-08-27 David Malcolm <dmalcolm@redhat.com>
5355 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
5356 line above for line-insertion fix-it hints.
5357 (selftest::test_fixit_insert_containing_newline): Update the
5358 expected results, and add a test with line-numbering enabled.
5360 2018-08-27 Martin Liska <mliska@suse.cz>
5363 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
5364 params with DECL_HAS_VALUE_EXPR_P.
5366 2018-08-27 Martin Liska <mliska@suse.cz>
5368 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
5369 selected expansion strategy.
5371 2018-08-27 Martin Liska <mliska@suse.cz>
5373 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
5374 * builtins.c (is_builtin_fn): Likewise.
5375 * attribs.c (diag_attr_exclusions): Use new function
5376 fndecl_built_in_p and remove check for FUNCTION_DECL if
5378 (builtin_mathfn_code): Likewise.
5379 (fold_builtin_expect): Likewise.
5380 (fold_call_expr): Likewise.
5381 (fold_builtin_call_array): Likewise.
5382 (fold_call_stmt): Likewise.
5383 (set_builtin_user_assembler_name): Likewise.
5384 (is_simple_builtin): Likewise.
5385 * calls.c (gimple_alloca_call_p): Likewise.
5386 (maybe_warn_nonstring_arg): Likewise.
5387 * cfgexpand.c (expand_call_stmt): Likewise.
5388 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
5389 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
5390 (cgraph_node::verify_node): Likewise.
5391 * cgraphclones.c (build_function_decl_skip_args): Likewise.
5392 (cgraph_node::create_clone): Likewise.
5393 * config/arm/arm.c (arm_insert_attributes): Likewise.
5394 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
5395 * dse.c (scan_insn): Likewise.
5396 * expr.c (expand_expr_real_1): Likewise.
5397 * fold-const.c (operand_equal_p): Likewise.
5398 (fold_binary_loc): Likewise.
5399 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
5400 * gimple-low.c (lower_stmt): Likewise.
5401 * gimple-pretty-print.c (dump_gimple_call): Likewise.
5402 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
5404 * gimple.c (gimple_build_call_from_tree): Likewise.
5405 (gimple_call_builtin_p): Likewise.
5406 (gimple_call_combined_fn): Likewise.
5407 * gimplify.c (gimplify_call_expr): Likewise.
5408 (gimple_boolify): Likewise.
5409 (gimplify_modify_expr): Likewise.
5410 (gimplify_addr_expr): Likewise.
5411 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
5412 * ipa-cp.c (determine_versionability): Likewise.
5413 * ipa-fnsummary.c (compute_fn_summary): Likewise.
5414 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
5415 * ipa-split.c (visit_bb): Likewise.
5416 (split_function): Likewise.
5417 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
5418 * lto-cgraph.c (input_node): Likewise.
5419 * lto-streamer-out.c (write_symbol): Likewise.
5420 * omp-low.c (setjmp_or_longjmp_p): Likewise.
5421 (lower_omp_1): Likewise.
5422 * predict.c (strip_predict_hints): Likewise.
5423 * print-tree.c (print_node): Likewise.
5424 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
5425 * trans-mem.c (is_tm_irrevocable): Likewise.
5426 (is_tm_load): Likewise.
5427 (is_tm_simple_load): Likewise.
5428 (is_tm_store): Likewise.
5429 (is_tm_simple_store): Likewise.
5430 (is_tm_abort): Likewise.
5431 (tm_region_init_1): Likewise.
5432 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
5433 * tree-cfg.c (verify_gimple_call): Likewise.
5434 (move_stmt_r): Likewise.
5435 (stmt_can_terminate_bb_p): Likewise.
5436 * tree-eh.c (lower_eh_constructs_2): Likewise.
5437 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
5438 * tree-inline.c (remap_gimple_stmt): Likewise.
5439 (copy_bb): Likewise.
5440 (estimate_num_insns): Likewise.
5441 (fold_marked_statements): Likewise.
5442 * tree-sra.c (scan_function): Likewise.
5443 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
5444 (optimize_stack_restore): Likewise.
5445 (pass_fold_builtins::execute): Likewise.
5446 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5447 (mark_all_reaching_defs_necessary_1): Likewise.
5448 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
5449 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
5450 (pass_forwprop::execute): Likewise.
5451 * tree-ssa-loop-im.c (stmt_cost): Likewise.
5452 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5453 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
5454 * tree-ssa-strlen.c (get_string_length): Likewise.
5455 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
5456 (find_func_aliases_for_call): Likewise.
5457 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
5458 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
5459 * tree-tailcall.c (find_tail_calls): Likewise.
5460 * tree.c (need_assembler_name_p): Likewise.
5461 (free_lang_data_in_decl): Likewise.
5462 (get_call_combined_fn): Likewise.
5463 * ubsan.c (is_ubsan_builtin_p): Likewise.
5464 * varasm.c (incorporeal_function_p): Likewise.
5465 * tree.h (DECL_BUILT_IN): Remove and replace with
5467 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
5468 (fndecl_built_in_p): New.
5470 2018-08-27 Martin Liska <mliska@suse.cz>
5472 PR tree-optimization/86847
5473 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
5474 Dump also subtree probability.
5475 (switch_decision_tree::do_jump_if_equal): New function.
5476 (switch_decision_tree::emit_case_nodes): Handle special
5477 situations in balanced tree that can be emitted much simpler.
5478 Fix calculation of probabilities that happen in tree expansion.
5479 * tree-switch-conversion.h (struct cluster): Add
5481 (struct simple_cluster): Likewise.
5482 (struct case_tree_node): Add new function has_child.
5483 (do_jump_if_equal): New.
5485 2018-08-27 Martin Liska <mliska@suse.cz>
5487 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
5488 Add new argument to bit_test_cluster constructor.
5489 (bit_test_cluster::emit): Set bits really number of values
5491 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
5492 probability argument.
5493 * tree-switch-conversion.h (struct bit_test_cluster):
5494 Add m_handles_entire_switch.
5496 2018-08-27 Martin Liska <mliska@suse.cz>
5498 PR tree-optimization/86702
5499 * tree-switch-conversion.c (jump_table_cluster::emit):
5500 Make probabilities even for values in jump table
5501 according to number of cases handled.
5502 (switch_decision_tree::compute_cases_per_edge): Pass
5503 argument to reset_out_edges_aux function.
5504 (switch_decision_tree::analyze_switch_statement): Likewise.
5505 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
5508 2018-08-27 Martin Liska <mliska@suse.cz>
5510 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
5511 cfun argument explicitly.
5512 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
5513 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
5514 function gimple_switch_default_bb.
5515 (convert_switch_statements):
5517 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
5518 * stmt.c (label_to_block_fn): Use label_to_block and pass
5519 cfun argument explicitly and use gimple_switch_label_bb.
5520 (expand_case): Likewise.
5521 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
5522 cfun argument explicitly. Likewise.
5523 (make_edges_bb): Likewise.
5524 (make_cond_expr_edges): Likewise.
5525 (get_cases_for_edge): Likewise.
5526 (make_gimple_switch_edges): Likewise.
5527 (label_to_block_fn): Likewise.
5528 (label_to_block): Likewise.
5529 (make_goto_expr_edges): Likewise.
5530 (make_gimple_asm_edges): Likewise.
5531 (main_block_label): Likewise.
5532 (group_case_labels_stmt): Likewise.
5533 (find_taken_edge_computed_goto): Likewise.
5534 (find_taken_edge_switch_expr): Likewise.
5535 (gimple_verify_flow_info): Likewise.
5536 (gimple_redirect_edge_and_branch): Likewise.
5537 (gimple_switch_label_bb): New function.
5538 (gimple_switch_default_bb): Likewise.
5539 (gimple_switch_edge): Likewise.
5540 (gimple_switch_default_edge): Likewise.
5541 * tree-cfg.h (label_to_block_fn): Remove and replace ...
5542 (label_to_block): ... with this.
5543 (gimple_switch_label_bb): New.
5544 (gimple_switch_default_bb): Likewise.
5545 (gimple_switch_edge): Likewise.
5546 (gimple_switch_default_edge): Likewise.
5547 * tree-cfgcleanup.c (convert_single_case_switch): Use
5548 new gimple functions and pass new argument to label_to_block.
5549 (cleanup_control_flow_bb):
5550 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
5551 cfun argument explicitly.
5552 (make_eh_edges): Likewise.
5553 (redirect_eh_dispatch_edge): Likewise.
5554 (lower_resx): Likewise.
5555 (lower_eh_dispatch): Likewise.
5556 (maybe_remove_unreachable_handlers): Likewise.
5557 (unsplit_eh): Likewise.
5558 (cleanup_empty_eh): Likewise.
5559 (verify_eh_edges): Likewise.
5560 (verify_eh_dispatch_edge): Likewise.
5561 * tree-ssa-dom.c (record_edge_info): Likewise.
5562 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
5563 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
5564 (thread_through_normal_block): Likewise.
5565 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
5566 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
5567 * tree-switch-conversion.c (switch_conversion::collect): Use new
5569 (switch_conversion::check_final_bb): Likewise.
5570 (switch_conversion::gather_default_values): Pass new argument
5572 (switch_conversion::build_constructors): Likewise.
5573 (switch_decision_tree::compute_cases_per_edge): Use new
5574 gimple_switch_edge function.
5575 (switch_decision_tree::analyze_switch_statement): Pass new argument
5577 (switch_decision_tree::try_switch_expansion): Use
5578 gimple_switch_default_edge.
5579 * tree-vrp.c (find_switch_asserts): Pass new argument
5581 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
5582 (vr_values::simplify_switch_using_ranges): Likewise.
5584 2018-08-27 Richard Biener <rguenther@suse.de>
5586 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
5587 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
5589 * tree-ssa-sccvn.h (struct vn_pval): New structure.
5590 (struct vn_nary_op_s): Add unwind_to member. Add
5591 predicated_values flag and put result into a union together
5592 with a linked list of vn_pval.
5593 (struct vn_ssa_aux): Add name member to make maintaining
5594 a map of SSA name to vn_ssa_aux possible. Remove no longer
5595 needed info, dfsnum, low, visited, on_sccstack, use_processed
5596 and range_info_anti_range_p members.
5597 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
5598 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
5600 (vn_valueize): New global.
5601 (vn_context_bb): Likewise.
5602 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
5603 VN_INFO_PTR_INFO): Remove.
5604 * tree-ssa-sccvn.c: ... (rewrite)
5605 (pass_fre::execute): For -O2+ initialize loops and run
5606 RPO VN in optimistic mode (iterating). For -O1 and -Og
5607 run RPO VN in non-optimistic mode.
5608 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
5609 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
5610 * doc/invoke.texi (sccvn-max-scc-size): Remove.
5611 (rpo-vn-max-loop-depth): Document.
5612 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
5613 when valuezing the VUSE signals we walked out of the region.
5614 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
5615 (phi_translate): Set VN context block to use for availability
5617 (compute_avail): Likewise.
5618 (pre_valueize): New function.
5619 (pass_pre::execute): Adjust to the RPO VN API.
5621 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
5622 (propagate_constants_for_unrolling): Remove.
5623 (tree_unroll_loops_completely): Perform value-numbering
5624 on the unrolled bodies loop parent.
5626 2018-08-27 Richard Biener <rguenther@suse.de>
5628 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
5629 for partial antic compute.
5631 2018-08-27 Jakub Jelinek <jakub@redhat.com>
5633 PR rtl-optimization/87065
5634 * combine.c (simplify_if_then_else): Formatting fix.
5635 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
5637 (known_cond): Don't return const_true_rtx for vector modes. Use
5638 CONST0_RTX instead of const0_rtx. Formatting fixes.
5640 2018-08-27 Martin Liska <mliska@suse.cz>
5642 PR gcov-profile/87069
5643 * gcov.c (process_file): Record files already processed
5644 and warn about a file being processed multiple times.
5646 2018-08-27 Martin Liska <mliska@suse.cz>
5649 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5650 Set default values for x_aarch64_*_string strings.
5651 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
5652 prefix. For -mabi do not print '=ABI' in help and use
5653 <option_value> format for -msve-vector-bits and -moverride
5656 2018-08-26 Jeff Law <law@redhat.com>
5658 * config/mips/frame-header-opt.c: Include "backend.h" rather than
5661 2018-08-26 Marek Polacek <polacek@redhat.com>
5663 PR c++/87029, Implement -Wredundant-move.
5664 * doc/invoke.texi: Document -Wredundant-move.
5666 2018-08-25 Martin Sebor <msebor@redhat.com>
5668 PR tree-optimization/87059
5669 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
5670 to the same type as the other.
5671 * fold-const.c (fold_binary_loc): Assert expectation.
5673 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
5675 * config/darwin.c (machopic_legitimize_pic_address): Clean up
5676 extraneous parentheses, dead code section and formatting.
5678 2018-08-24 David Malcolm <dmalcolm@redhat.com>
5681 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
5682 wide enough for jumps in the line-numbering to be visible.
5683 (layout::print_gap_in_line_numbering): New member function.
5684 (layout::calculate_line_spans): When using line numbering, merge
5685 line spans that are only 1 line apart.
5686 (diagnostic_show_locus): When printing line numbers, show gaps in
5687 line numbering directly, rather than printing headers.
5688 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
5689 line-numbering with multiple line spans.
5690 (selftest::test_fixit_insert_containing_newline_2): Add test of
5691 line-numbering, in which the spans are close enough to be merged.
5693 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
5695 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
5696 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
5697 * tree-vrp.c (range_is_nonnull): Remove.
5698 (range_includes_zero_p): Accept value_range instead of min/max.
5699 (extract_range_from_binary_expr_1): Do not early bail on
5701 Use range_includes_zero_p instead of range_is_nonnull.
5702 (extract_range_from_unary_expr): Use range_includes_zero_p instead
5703 of range_is_nonnull.
5704 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
5705 special case VR_ANTI_RANGE.
5706 (vrp_finalize): Same.
5707 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
5709 (range_is_nonnull): Remove.
5710 * vr-values.c (vrp_stmt_computes_nonzero): Use
5711 range_includes_zero_p instead of range_is_nonnull.
5712 (extract_range_basic): Pass value_range to range_includes_zero_p
5713 instead of range_is_nonnull.
5715 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
5717 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
5718 * emit-rtl.h (rtl_data): Remove return_bnd.
5719 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
5720 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
5721 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
5722 (POINTER_BOUNDS_MODE): Remove definition.
5723 (make_pointer_bounds_mode): Remove.
5724 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
5725 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
5726 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
5727 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
5728 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
5729 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
5730 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
5732 * config/i386/i386-modes.def (BND32, BND64): Remove.
5733 * config/i386/i386.c (dbx_register_map): Remove bound registers.
5734 (dbx64_register_map): Ditto.
5735 (svr4_dbx_register_map): Ditto.
5736 (indirect_thunk_bnd_needed): Remove.
5737 (indirect_thunks_bnd_used): Ditto.
5738 (indirect_return_bnd_needed): Ditto.
5739 (indirect_return_via_cx_bnd): Ditto.
5740 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
5741 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
5742 (output_indirect_thunk): Ditto. Remove need_prefix argument.
5743 (output_indirect_thunk_function): Remove handling of
5744 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
5745 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
5746 (ix86_save_reg): Remove handling of crtl->return_bnd.
5747 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
5748 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
5749 and UNSPEC_BNDLX_ADDR.
5750 (ix86_output_indirect_branch_via_reg): Remove handling of
5751 indirect_thunk_prefix_bnd.
5752 (ix86_output_indirect_branch_via_push): Ditto.
5753 (ix86_output_function_return): Ditto.
5754 (ix86_output_indirect_function_return): Ditto.
5755 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
5756 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
5757 (CALL_USED_REGISTERS): Ditto.
5758 (REG_ALLOC_ORDER): Update for removal of bound registers.
5759 (HI_REGISTER_NAMES): Ditto.
5760 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
5761 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
5762 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
5763 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
5764 (FIRST_PSEUDO_REG): Update.
5765 (BND): Remove mode iterator.
5766 * config/i386/predicates.md (bnd_mem_operator): Remove.
5768 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
5770 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
5773 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
5775 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
5776 the case in which the permute needs only a single element and
5777 repeats for every vector of the result. Extend that case to
5778 handle variable-length vectors.
5779 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
5781 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
5784 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
5785 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
5787 2018-08-24 Richard Biener <rguenther@suse.de>
5789 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
5790 bb_flags_allocated members.
5791 (auto_flag): New RAII class for allocating flags.
5792 (auto_edge_flag): New RAII class for allocating edge flags.
5793 (auto_bb_flag): New RAII class for allocating bb flags.
5794 * cfgloop.c (verify_loop_structure): Allocate temporary edge
5796 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
5797 in favor of temporarily allocated BB flag.
5798 * hsa-brig.c: Re-order includes.
5799 * hsa-dump.c: Likewise.
5800 * hsa-regalloc.c: Likewise.
5801 * print-rtl.c: Likewise.
5802 * profile-count.c: Likewise.
5804 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
5807 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
5810 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
5813 * wide-int-range.cc (wide_int_range_div): Do not ignore result
5814 from wide_int_range_multiplicative_op.
5816 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5818 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
5821 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
5823 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
5826 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
5828 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
5829 full GENERIC expression used for replacement.
5831 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
5833 * tree-vrp.c (abs_extent_range): Remove.
5834 (extract_range_into_wide_ints): Pass wide ints by reference.
5835 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
5836 Pass wide ints by reference in all calls to
5837 extract_range_into_wide_ints.
5838 * wide-int-range.cc (wide_int_range_div): New.
5839 * wide-int-range.h (wide_int_range_div): New.
5840 (wide_int_range_includes_zero_p): New.
5841 (wide_int_range_zero_p): New.
5843 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
5845 * config/aarch64/aarch64.md (arches): New enum.
5846 (arch): New enum attr.
5847 (arch_enabled): New attr.
5848 (enabled): Now uses arch_enabled only.
5849 (simd, sve, fp16): Removed attribute.
5850 (fp): Attr now defined in terms of 'arch'.
5851 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
5852 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
5853 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
5854 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
5855 attributes into 'arch'.
5856 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
5857 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
5858 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
5859 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
5860 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
5861 'simd' attribute into 'arch'.
5862 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
5863 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
5864 Convert use of 'fp' attribute to 'arch'.
5865 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
5866 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
5868 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
5869 (different modes) Merge 'fp' and 'simd' into 'arch'.
5870 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
5873 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
5875 PR rtl-optimization/87026
5876 * expmed.c (canonicalize_comparison): If we can no longer create
5877 pseudoregisters, don't.
5879 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
5882 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
5884 * config/arm/arm.c (speculation_barrier_libfunc): New static
5886 (arm_init_libfuncs): Initialize it.
5887 (arm_emit_speculation_barrier): New function.
5888 * config/arm/arm.md (speculation_barrier): Call
5889 arm_emit_speculation_barrier for architectures that do not have
5891 (speculation_barrier_insn): Only match on Armv7 or later.
5893 2018-08-23 Richard Biener <rguenther@suse.de>
5896 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
5899 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
5901 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
5902 of single-vector TBLs.
5903 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
5906 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
5909 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
5910 aarch64_evpc_tbl guard.
5912 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
5914 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
5917 2018-08-22 Martin Sebor <msebor@redhat.com>
5920 * tree-pretty-print.c (pretty_print_string): Add argument.
5921 (dump_generic_node): Call to pretty_print_string with string size.
5923 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
5925 PR rtl-optimization/86771
5926 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
5927 of two SETs into those two SETs, one to be placed at i2, if that SETs
5928 destination is modified between i2 and i3.
5930 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
5932 PR tree-optimization/86725
5933 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
5935 (vect_analyze_scalar_cycles_1): Check it.
5937 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
5939 PR tree-optimization/86725
5940 * tree-vect-loop.c (vect_is_simple_reduction): When treating
5941 an outer loop phi as a double reduction, make sure that the
5942 single user of the phi result is an inner loop phi.
5944 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
5946 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
5947 grouped stores with gaps to a strided group.
5949 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
5951 * tree-vect-stmts.c (get_group_load_store_type)
5952 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
5953 first statement in a group.
5955 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5957 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
5958 the sequence used in gcc/gcc.c.
5960 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5963 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
5966 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5968 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
5969 Darwin10-specific unwinder-shim.
5970 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
5971 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
5972 New to cater for Darwin10 Rosetta.
5974 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5976 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
5977 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
5979 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5984 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
5985 (dwarf2out_switch_text_section): Generate a local label for the
5986 second function sub-section and apply it as the second FDE start
5988 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
5989 second sub-section start.
5991 2018-08-22 Richard Biener <rguenther@suse.de>
5993 PR tree-optimization/86988
5994 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
5996 2018-08-22 Richard Biener <rguenther@suse.de>
5998 PR tree-optimization/86945
5999 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
6001 2018-08-22 Alexandre Oliva <oliva@adacore.com>
6003 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
6004 a comment about how uses of r2 for .sdata2 come about.
6006 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
6008 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
6010 2018-08-21 Marek Polacek <polacek@redhat.com>
6012 PR c++/86981, Implement -Wpessimizing-move.
6013 * doc/invoke.texi: Document -Wpessimizing-move.
6015 2018-08-21 Jan Hubicka <jh@suse.cz>
6017 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
6018 * tree.h (is_redundant_typedef): Remove.
6019 * dwarf2out.c (is_redundant_typedef): Turn into static function.
6021 2018-08-21 Jan Hubicka <jh@suse.cz>
6023 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
6026 2018-08-21 Tamar Christina <tamar.christina@arm.com>
6028 * expmed.c (extract_low_bits): Reject invalid subregs early.
6030 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
6033 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
6036 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6038 * config/vxworks.h: Guard vxworks_asm_out_constructor and
6039 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
6040 * config/vxworks.c: Likewise.
6042 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6044 * config/vxworks.c: Set targetm.have_ctors_dtors
6045 if HAVE_INITFINI_ARRAY_SUPPORT.
6046 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
6047 if HAVE_INITFINI_ARRAY_SUPPORT.
6049 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6051 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
6052 default search path for VxWorks < 7.
6054 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
6056 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
6057 (get_format_string): Refer to c_getstr.
6059 2018-08-21 Tom de Vries <tdevries@suse.de>
6061 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
6062 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
6063 (debuginfo_early_stop): Declare.
6064 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
6065 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
6066 (debuginfo_early_stop): New function.
6067 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
6068 and debuginfo_early_stop.
6069 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
6070 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
6071 (general_init): Call debuginfo_early_init.
6072 (finalize): Call debuginfo_fini.
6073 (do_compile): Call debuginfo_init.
6074 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
6076 (@item -fdump-debug, @item -fdump-earlydebug): Add.
6078 2018-08-21 Tom de Vries <tdevries@suse.de>
6080 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
6081 flag_dump_noaddr and flag_dump_unnumbered.
6083 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
6085 * wide-int-range.cc (wide_int_range_abs): New.
6086 (wide_int_range_order_set): Rename from wide_int_range_min_max.
6087 * wide-int-range.h (wide_int_range_abs): New.
6088 (wide_int_range_min_max): New.
6089 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
6090 case to call wide_int_range_abs.
6091 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
6092 (extract_range_from_abs_expr): Delete.
6094 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
6097 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
6098 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
6101 2018-08-20 Nathan Sidwell <nathan@acm.org>
6102 Jeff Law <law@redhat.com>
6104 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
6105 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
6107 2018-08-20 David Malcolm <dmalcolm@redhat.com>
6110 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
6111 (decl_attributes): Likewise.
6112 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
6114 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
6115 * diagnostic-core.h (class auto_diagnostic_group): New class.
6116 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
6117 (diagnostic_report_diagnostic): Handle the first diagnostics within
6119 (emit_diagnostic): Add auto_diagnostic_group instance.
6121 (inform_n): Likewise.
6122 (warning): Likewise.
6123 (warning_at): Likewise.
6124 (warning_n): Likewise.
6125 (pedwarn): Likewise.
6126 (permerror): Likewise.
6128 (error_n): Likewise.
6129 (error_at): Likewise.
6131 (fatal_error): Likewise.
6132 (internal_error): Likewise.
6133 (internal_error_no_backtrace): Likewise.
6134 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
6135 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
6136 * diagnostic.h (struct diagnostic_context): Add fields
6137 "diagnostic_group_nesting_depth",
6138 "diagnostic_group_emission_count", "begin_group_cb",
6140 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
6141 Add auto_diagnostic_group instance(s).
6142 (find_explicit_erroneous_behavior): Likewise.
6143 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
6144 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
6145 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
6146 (gimplify_va_arg_expr): Likewise.
6147 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
6148 (HSA_SORRY_AT): Likewise.
6149 * ipa-devirt.c (compare_virtual_tables): Likewise.
6150 (warn_odr): Likewise.
6151 * multiple_target.c (expand_target_clones): Likewise.
6152 * opts-common.c (cmdline_handle_error): Likewise.
6153 * reginfo.c (globalize_reg): Likewise.
6154 * substring-locations.c (format_warning_n_va): Likewise.
6155 * tree-inline.c (expand_call_inline): Likewise.
6156 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
6157 * tree-ssa-loop-niter.c
6158 (do_warn_aggressive_loop_optimizations): Likewise.
6159 * tree-ssa-uninit.c (warn_uninit): Likewise.
6160 * tree.c (warn_deprecated_use): Likewise.
6162 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
6165 * config/i386/i386.md (eh_return): Always update EH return
6166 address in word_mode.
6168 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
6170 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
6171 TARGET_SPLIT_COMPLEX_ARG is defined.
6173 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
6175 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
6177 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
6180 * expr.c (expand_assignment): Assert that bitpos is positive.
6181 (store_field): Likewise
6182 (expand_expr_real_1): Make sure that bitpos is positive.
6183 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
6186 2018-08-20 Nathan Sidwell <nathan@acm.org>
6188 * Makefile.in (CPP_ID_DATA_H): Delete.
6189 (CPP_INTERNAL_H): Don't add it.
6190 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
6191 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
6193 2018-08-20 Richard Biener <rguenther@suse.de>
6195 PR tree-optimization/78655
6196 * tree-vrp.c (extract_range_from_binary_expr_1): Make
6197 pointer + offset nonnull if either operand is nonnull work.
6199 2018-08-20 Tom de Vries <tdevries@suse.de>
6201 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
6202 unless the referenced die describes the added property using
6203 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
6204 Otherwise, add a DW_AT_location to the referenced die.
6206 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
6209 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
6210 register_operand when calling ix86_set_reg_reg_cost.
6211 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
6212 Set *total to 0 for operands that satisfy x86_64_immediate_operand
6213 predicate and to 1 otherwise.
6215 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
6217 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
6218 emit a diagnostic that it is not supported and reset the option.
6219 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
6220 supported and consume the option. (ASM_FINAL_SPEC): New.
6222 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
6224 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
6227 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
6229 C-SKY port: Documentation
6231 * doc/extend.texi (C-SKY Function Attributes): New section.
6232 * doc/invoke.texi (Option Summary): Add C-SKY options.
6233 (C-SKY Options): New section.
6234 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
6236 2018-08-17 Jojo <jijie_rong@c-sky.com>
6237 Huibin Wang <huibin_wang@c-sky.com>
6238 Sandra Loosemore <sandra@codesourcery.com>
6239 Chung-Lin Tang <cltang@codesourcery.com>
6241 C-SKY port: Backend implementation
6243 * config/csky/*: New.
6244 * common/config/csky/*: New.
6246 2018-08-17 Jojo <jijie_rong@c-sky.com>
6247 Huibin Wang <huibin_wang@c-sky.com>
6248 Sandra Loosemore <sandra@codesourcery.com>
6249 Chung-Lin Tang <cltang@codesourcery.com>
6250 Andrew Jenner <andrew@codesourcery.com>
6252 C-SKY port: Configury
6254 * config.gcc (csky-*-*): New.
6255 * configure.ac: Add csky to targets for dwarf2 debug_line support.
6256 * configure: Regenerated.
6258 2018-08-17 David Malcolm <dmalcolm@redhat.com>
6260 * dump-context.h: Include "dumpfile.h".
6261 (dump_context::dump_printf_va): Convert final param from va_list
6262 to va_list *. Convert from ATTRIBUTE_PRINTF to
6263 ATTRIBUTE_GCC_DUMP_PRINTF.
6264 (dump_context::dump_printf_loc_va): Likewise.
6265 * dumpfile.c: Include "stringpool.h".
6266 (make_item_for_dump_printf_va): Delete.
6267 (make_item_for_dump_printf): Delete.
6268 (class dump_pretty_printer): New class.
6269 (dump_pretty_printer::dump_pretty_printer): New ctor.
6270 (dump_pretty_printer::emit_items): New member function.
6271 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
6273 (dump_pretty_printer::emit_item): New member function.
6274 (dump_pretty_printer::stash_item): New member function.
6275 (dump_pretty_printer::format_decoder_cb): New member function.
6276 (dump_pretty_printer::decode_format): New member function.
6277 (dump_context::dump_printf_va): Reimplement in terms of
6278 dump_pretty_printer.
6279 (dump_context::dump_printf_loc_va): Convert final param from va_list
6281 (dump_context::begin_scope): Reimplement call to
6282 make_item_for_dump_printf.
6283 (dump_printf): Update for change to dump_printf_va.
6284 (dump_printf_loc): Likewise.
6285 (selftest::test_capture_of_dump_calls): Convert "stmt" from
6286 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
6287 with %T, %E, and %G.
6288 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
6289 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
6290 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
6291 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
6292 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
6293 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
6294 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
6295 within a dump_printf_loc call to "%wu".
6296 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
6297 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
6298 missing space after "=".
6299 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
6300 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
6301 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
6302 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
6303 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
6304 duplicate "vectorized" from message.
6306 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
6308 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
6309 polyNxK_t element's TYPE_STRING_FLAG.
6311 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
6313 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
6314 (they were unnamed before). Fix comments.
6316 2018-08-17 Nathan Sidwell <nathan@acm.org>
6318 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
6320 2018-08-17 Richard Biener <rguenther@suse.de>
6322 PR tree-optimization/86841
6323 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
6325 2018-08-17 Martin Liska <mliska@suse.cz>
6327 * common.opt: Remove Warn, Init and Report for options with
6328 Ignore/Deprecated flag. Warning is done automatically for
6330 * config/i386/i386.opt: Likewise.
6331 * config/ia64/ia64.opt: Likewise.
6332 * config/rs6000/rs6000.opt: Likewise.
6333 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
6334 Remove usage of flag_check_pointer_bounds.
6335 * lto-wrapper.c (merge_and_complain): Do not handle
6336 OPT_fcheck_pointer_bounds.
6337 (append_compiler_options): Likewise.
6338 * opt-functions.awk: Do not handle Deprecated.
6339 * optc-gen.awk: Check that Var, Report and Init are not
6340 used for an option with Ignore/Deprecated flag.
6341 * opts-common.c (decode_cmdline_option): Do not report
6343 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
6345 * opts.h (struct cl_option): Remove cl_deprecated flag.
6346 (CL_ERR_DEPRECATED): Remove error enum value.
6348 2018-08-17 Richard Biener <rguenther@suse.de>
6351 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
6352 across a va-arg-pack using call adjust its return value accordingly.
6354 2018-08-16 Martin Sebor <msebor@redhat.com>
6356 PR tree-optimization/86853
6357 * gimple-ssa-sprintf.c (struct format_result): Rename member.
6358 (struct fmtresult): Add member and initialize it in ctors.
6359 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
6360 (format_string): Handle %S the same as %ls. Set MAYFAIL.
6361 (format_directive): Set POSUNDER4K when MAYFAIL is set.
6362 (parse_directive): Handle %C same as %c.
6363 (sprintf_dom_walker::compute_format_length): Adjust.
6364 (is_call_safe): Adjust.
6366 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
6368 * builtins.c (c_strlen): Add new parameter eltsize. Use it
6369 for determining how to count the elements.
6370 * builtins.h (c_strlen): Adjust prototype.
6371 * expr.c (string_constant): Add new parameter mem_size.
6372 Set *mem_size appropriately.
6373 * expr.h (string_constant): Adjust protoype.
6374 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
6375 * gimple-fold.h (get_range_strlen): Adjust prototype.
6376 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
6377 (format_string): Call get_string_length with eltsize.
6379 2018-08-16 David Malcolm <dmalcolm@redhat.com>
6381 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
6382 to emit the span, rather than setting it as the prefix.
6384 2018-08-16 David Malcolm <dmalcolm@redhat.com>
6386 * diagnostic-show-locus.c (layout::start_annotation_line): Add
6387 "margin_char" parameter, defaulting to space. Use it in place
6388 of pp_space for the initial part of the margin.
6389 (layout::print_leading_fixits): Use '+' when filling the margin
6390 of line-insertion fix-it hints.
6392 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6394 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
6397 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6399 * config/rs6000/altivec.md: Don't set length attribute to the default
6401 * config/rs6000/darwin.md: Ditto.
6402 * config/rs6000/dfp.md: Ditto.
6403 * config/rs6000/htm.md: Ditto.
6404 * config/rs6000/rs6000.md: Ditto.
6405 * config/rs6000/sync.md: Ditto.
6406 * config/rs6000/vsx.md: Ditto.
6408 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6410 * config/rs6000/altivec.md: Don't set length attribute to the default
6411 value, for branch instructions.
6412 * config/rs6000/darwin.md: Ditto.
6413 * config/rs6000/rs6000.md: Ditto.
6415 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6417 * config/rs6000/rs6000.md (length): Always define as const_int 4.
6418 (unnamed conditional branch define_insn): Set length to 4 or 8
6419 depending on offset.
6420 (<bd>_<mode>): Similar, for alternative 0.
6421 (<bd>tf_<mode>): Ditto.
6423 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6425 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
6427 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
6429 * doc/rtl.texi: Replace old RTX class names with new names.
6432 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
6434 * expmed.h (canonicalize_comparison): New declaration.
6435 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
6436 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
6437 * optabs.c (prepare_cmp_insn): Likewise.
6438 * rtl.h (unsigned_condition_p): New function which checks if a
6439 comparison operator is unsigned.
6441 2018-08-16 Nathan Sidwell <nathan@acm.org>
6443 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
6444 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
6446 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6449 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
6450 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
6451 (mov<mov>): ..this and enable unconditionally.
6453 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6455 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
6457 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
6459 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
6460 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
6461 "Common" with "Target".
6463 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
6465 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
6466 * doc/invoke.texi (mmitigate-rop): Remove.
6467 * config/i386/i386.c: Do not include "regrename.h".
6468 (ix86_rop_should_change_byte_p, reg_encoded_number)
6469 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
6471 (ix86_reorg): Remove call to ix86_mitigate_rop.
6472 * config/i386/i386.md (attr "modrm_class"): Remove.
6473 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
6474 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
6475 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
6477 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6479 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
6480 allow folding of mergeh() and mergel() for the float and double types.
6481 (fold_mergehl_helper): Rework to handle building a permute tree
6484 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
6486 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
6489 2018-08-15 David Malcolm <dmalcolm@redhat.com>
6491 * common.opt (fdiagnostics-show-labels): New option.
6492 * diagnostic-show-locus.c (class layout_range): Add field
6494 (class layout): Add field "m_show_labels_p".
6495 (layout_range::layout_range): Add param "label" and use it to
6497 (make_range): Pass in NULL for new "label" param of layout_range's
6499 (layout::layout): Initialize m_show_labels_p.
6500 (layout::maybe_add_location_range): Pass in loc_range->m_label
6501 when constructing layout_range instances.
6502 (struct line_label): New struct.
6503 (layout::print_any_labels): New member function.
6504 (layout::print_line): Call it if label-printing is enabled.
6505 (selftest::test_one_liner_labels): New test.
6506 (selftest::test_diagnostic_show_locus_one_liner): Call it.
6507 * diagnostic.c (diagnostic_initialize): Initialize
6508 context->show_labels_p.
6509 * diagnostic.h (struct diagnostic_context): Add field
6511 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
6512 -fno-diagnostics-show-labels.
6513 * dwarf2out.c (gen_producer_string): Add
6514 OPT_fdiagnostics_show_labels to the ignored options.
6515 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
6517 (gcc_rich_location::maybe_add_expr): Likewise.
6518 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
6519 label" param, defaulting to NULL.
6520 (gcc_rich_location::add_expr): Add "label" param.
6521 (gcc_rich_location::maybe_add_expr): Likewise.
6522 (class text_range_label): New class.
6523 (class range_label_for_type_mismatch): New class.
6524 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
6525 of format_warning_va.
6526 (fmtwarn_n): Likewise for new params of format_warning_n_va.
6527 * lto-wrapper.c (merge_and_complain): Add
6528 OPT_fdiagnostics_show_labels to the "pick one setting" options.
6529 (append_compiler_options): Likewise to the dropped options.
6530 (append_diag_options): Likewise to the passed-on options.
6531 * opts.c (common_handle_option): Handle the new option.
6532 * selftest-diagnostic.c
6533 (test_diagnostic_context::test_diagnostic_context): Enable
6535 * substring-locations.c: Include "gcc-rich-location.h".
6536 (format_warning_n_va): Add "fmt_label" and "param_label" params
6537 and use them as appropriate.
6538 (format_warning_va): Add "fmt_label" and "param_label" params,
6539 passing them on to format_warning_n_va.
6540 (format_warning_at_substring): Likewise.
6541 (format_warning_at_substring_n): Likewise.
6542 * substring-locations.h (format_warning_va): Add "fmt_label" and
6543 "param_label" params.
6544 (format_warning_n_va): Likewise.
6545 (format_warning_at_substring): Likewise.
6546 (format_warning_at_substring_n): Likewise.
6547 * toplev.c (general_init): Initialize global_dc->show_labels_p.
6549 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
6552 * builtins.c (expand_builtin_memcmp): Do not expand the call
6553 when overflow is detected.
6555 2018-08-15 Martin Sebor <msebor@redhat.com>
6557 PR tree-optimization/71625
6558 * config/aarch64/aarch64-builtins.c
6559 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
6561 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
6563 * config/s390/s390.c (s390_reorg): Remove loop.
6565 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
6568 (darwin_function_switched_text_sections): Delete.
6569 * gcc/config/darwin.h
6570 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
6572 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
6575 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
6576 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
6577 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
6579 2018-08-15 Martin Liska <mliska@suse.cz>
6581 PR tree-optimization/86925
6582 * predict.c (expr_expected_value_1): When taking
6583 later predictor, assign also probability.
6584 Use fold_build2_initializer_loc in order to fold
6585 the expression in -frounding-math.
6587 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
6589 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
6591 (expand_vec_perm_1): Try the new method.
6593 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
6596 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
6597 Check whether lra_live_max_point is 0 before dividing.
6599 2018-08-14 Martin Sebor <msebor@redhat.com>
6601 PR tree-optimization/86650
6602 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
6603 (vrp_prop::check_mem_ref): Same.
6605 2018-08-13 Liu Hao <lh_mouse@126.com>
6607 * pretty-print.c (eat_esc_sequence): Swap the foreground and
6608 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
6609 and clear it thereafter, as it only works for DBCS.
6611 2018-08-13 Liu Hao <lh_mouse@126.com>
6613 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
6614 handle returned by _get_osf_handle().
6616 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6618 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
6619 for folding vec_perm.
6621 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6623 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
6624 Add support for gimple-folding of vec_pack() and vec_unpack()
6627 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6629 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
6630 vec_xst variants to the list.
6631 (rs6000_gimple_fold_builtin): Add support for folding unaligned
6632 vector loads and stores.
6634 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
6636 * config.gcc (rs6000-ibm-aix4.x): Delete.
6637 (rs6000-ibm-aix5.1): Delete.
6638 (rs6000-ibm-aix5.2): Delete.
6639 (rs6000-ibm-aix5.3): Delete.
6640 * config/rs6000/aix43.h: Delete.
6641 * config/rs6000/aix51.h: Delete.
6642 * config/rs6000/aix52.h: Delete.
6643 * config/rs6000/t-aix43: Delete.
6645 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
6647 * config/s390/s390.c (s390_decompose_constant_pool_ref):
6649 (s390_decompose_address): Factor out constant pool ref
6652 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6654 * config/nds32/nds32-predicates.c
6655 (nds32_can_use_bclr_p): Change return type as bool.
6656 (nds32_can_use_bset_p): Ditto.
6657 (nds32_can_use_btgl_p): Ditto.
6658 (nds32_can_use_bitci_p): Ditto.
6659 * config/nds32/nds32-protos.h
6660 (nds32_can_use_bclr_p): Change declaration.
6661 (nds32_can_use_bset_p): Ditto.
6662 (nds32_can_use_btgl_p): Ditto.
6663 (nds32_can_use_bitci_p): Ditto.
6665 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6667 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
6668 Support -msched-prolog-epilog option.
6669 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
6671 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6673 * common/config/nds32/nds32-common.c
6674 (nds32_option_optimization_table): Enalbe -malways-align.
6676 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6678 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
6680 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
6681 OPT_misr_secure_ case.
6682 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
6683 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
6684 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
6686 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
6687 (nds32_isr_info): New field security_level.
6688 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
6689 * config/nds32/nds32.md (return_internal): Consider critical attribute.
6690 * config/nds32/nds32.opt (misr-secure): New option.
6691 * config/nds32/nds32_init.inc: New file.
6692 * config/nds32/nds32_isr.h: New file.
6694 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
6696 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
6697 Update comment for atomic instructions.
6698 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
6699 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
6701 (atomic_loaddi): Revise fence expansion to only emit fence prior to
6702 load for __ATOMIC_SEQ_CST model.
6703 (atomic_loaddi_1): Remove float register target.
6704 (atomic_storedi): Handle CONST_INT values.
6705 (atomic_storedi_1): Remove float register source. Add special case
6707 (memory_barrier): New expander and insn.
6709 2018-08-11 Jakub Jelinek <jakub@redhat.com>
6711 PR tree-optimization/86835
6712 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
6713 new_stmt after def_gsi, make sure to insert new_square_stmt after
6714 that stmt, not 2 stmts before it.
6716 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
6719 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
6722 2018-08-10 Martin Liska <mliska@suse.cz>
6725 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
6727 * builtins.c (expand_builtin_expect_with_probability):
6729 (expand_builtin_expect_with_probability): New function.
6730 (build_builtin_expect_predicate): Add new argumnet probability
6731 for BUILT_IN_EXPECT_WITH_PROBABILITY.
6732 (fold_builtin_expect):
6735 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
6736 * builtins.h (fold_builtin_expect): Set new argument.
6737 * doc/extend.texi: Document __builtin_expect_with_probability.
6738 * doc/invoke.texi: Likewise.
6739 * gimple-fold.c (gimple_fold_call): Pass new argument.
6740 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
6741 also BUILT_IN_EXPECT_WITH_PROBABILITY.
6742 * predict.c (get_predictor_value): New function.
6743 (expr_expected_value): Add new argument probability. Assume
6744 that predictor and probability are always non-null.
6745 (expr_expected_value_1): Likewise. For __builtin_expect and
6746 __builtin_expect_with_probability set probability. Handle
6747 combination in binary expressions.
6748 (tree_predict_by_opcode): Simplify code by simply calling
6749 get_predictor_value.
6750 (pass_strip_predict_hints::execute): Add handling of
6751 BUILT_IN_EXPECT_WITH_PROBABILITY.
6752 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
6754 * tree.h (DECL_BUILT_IN_P): New function.
6756 2018-08-10 Martin Liska <mliska@suse.cz>
6758 PR tree-optimization/85799
6759 * passes.def: Add argument for pass_strip_predict_hints.
6760 * predict.c (class pass_strip_predict_hints): Add new argument
6762 (strip_predictor_early): New function.
6763 (pass_strip_predict_hints::execute): Call the function to
6765 (strip_predict_hints): New function.
6766 * predict.def: Fix comment.
6768 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
6770 * Makefile.in: Clarify which tm.texi to copy over to assert the
6771 right to grant a GFDL license for all.
6773 2018-08-09 Jeff Law <law@redhat.com>
6775 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
6778 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
6780 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
6783 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
6785 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
6786 reductions for variable-length vectors.
6788 2018-08-09 David Malcolm <dmalcolm@redhat.com>
6791 * common.opt (fdiagnostics-show-line-numbers): New option.
6792 * diagnostic-show-locus.c (class layout): Add fields
6793 "m_show_line_numbers_p" and "m_linenum_width";
6794 (num_digits): New function.
6795 (test_num_digits): New function.
6796 (layout::layout): Initialize new fields. Update m_x_offset
6797 logic to handle any left margin.
6798 (layout::print_source_line): Print line number when requested.
6799 (layout::start_annotation_line): New member function.
6800 (layout::print_annotation_line): Call it.
6801 (layout::print_leading_fixits): Likewise.
6802 (layout::print_trailing_fixits): Likewise. Update calls to
6803 move_to_column for new parameter.
6804 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
6805 it to potentially call start_annotation_line.
6806 (layout::show_ruler): Call start_annotation_line.
6807 (selftest::test_line_numbers_multiline_range): New selftest.
6808 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
6809 and selftest::test_line_numbers_multiline_range.
6810 * diagnostic.c (diagnostic_initialize): Initialize
6811 show_line_numbers_p.
6812 * diagnostic.h (struct diagnostic_context): Add field
6813 "show_line_numbers_p".
6814 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
6815 -fno-diagnostics-show-line-numbers.
6816 * dwarf2out.c (gen_producer_string): Add
6817 OPT_fdiagnostics_show_line_numbers to the ignored options.
6818 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
6819 one setting" options.
6820 (append_compiler_options): Likewise to the dropped options.
6821 (append_diag_options): Likewise to the passed-on options.
6822 * opts.c (common_handle_option): Handle the new option.
6823 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
6825 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
6827 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
6828 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
6829 third argument of type "const signed char" to descriptions of
6830 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
6831 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
6832 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
6833 __builtin_bcdsub_ov functions.
6835 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
6837 PR tree-optimization/86858
6838 * tree-vect-loop.c (vect_is_simple_reduction): Restore
6839 flow_bb_inside_loop_p calls.
6841 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
6843 PR tree-optimization/86871
6844 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
6845 instead of gimple_assign_lhs.
6847 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
6850 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
6851 register constraint to operand 0.
6852 (add<mode>3_carryinC): Likewise.
6853 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
6855 2018-08-09 Martin Liska <mliska@suse.cz>
6858 * common.opt: Remove extra line.
6860 2018-08-09 Martin Liska <mliska@suse.cz>
6862 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
6863 at the end of a line, make first letter capital and end up
6864 a sentence with a dot.
6865 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
6866 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
6867 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
6868 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
6869 (PARAM_MAX_ISL_OPERATIONS): Likewise.
6870 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
6871 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
6872 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
6873 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
6874 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
6875 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
6876 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
6877 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
6878 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
6879 (PARAM_TREE_REASSOC_WIDTH): Likewise.
6880 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
6881 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
6882 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
6884 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
6887 * config/s390/s390.c (s390_option_override_internal): Reduce the
6888 stack-clash-protection-probe-interval param if it would be too big
6891 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
6894 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
6895 * config/m68k/m68k.c (handle_move_double): Don't call
6896 m68k_final_prescan_insn.
6897 (m68k_adjust_decorated_operand): Renamed from
6898 m68k_final_prescan_insn, remove first and third operand and
6900 (print_operand): Call it.
6901 (print_operand_address): Call it.
6903 2018-08-08 Nathan Sidwell <nathan@acm.org>
6905 * diagnostic.c (diagnostic_report_current_module): Use
6906 linemap_included_from & linemap_included_from_linemap.
6908 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
6910 * config/aarch64/aarch64-cores.def: Add phecda core.
6911 * config/aarch64/aarch64-tune.md: Regenerate.
6912 * doc/invoke.texi: Add phecda core.
6914 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
6917 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
6919 * config/s390/s390.md ("movti"): Add more alternatives for
6920 constant to GPR copies.
6922 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
6924 * config/s390/s390.c: Fix whitespace damage throughout the file.
6925 * config/s390/s390.h: Likewise.
6926 * config/s390/tpf.h: Likewise.
6928 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
6930 * config/s390/s390.c (s390_loadrelative_operand_p):
6931 Remove TARGET_CPU_ZARCH usages.
6932 (s390_rtx_costs): Likewise.
6933 (s390_legitimate_constant_p): Likewise.
6934 (s390_cannot_force_const_mem): Likewise.
6935 (legitimate_reload_constant_p): Likewise.
6936 (s390_preferred_reload_class): Likewise.
6937 (legitimize_pic_address): Likewise.
6938 (legitimize_tls_address): Likewise.
6939 (s390_split_branches): Removed.
6940 (s390_add_execute): Removed.
6941 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
6942 (s390_mainpool_start): Likewise.
6943 (s390_mainpool_finish): Likewise.
6944 (s390_mainpool_cancel): Removed.
6945 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
6946 (s390_chunkify_cancel): Likewise.
6947 (s390_return_addr_rtx): Likewise.
6948 (s390_register_info): Remove split_branches_pending_p uages.
6949 (s390_optimize_register_info): Likewise.
6950 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
6951 split_branches_pending_p usages.
6952 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
6953 (s390_load_got): Likewise.
6954 (s390_expand_split_stack_prologue): Likewise.
6955 (output_asm_nops): Likewise.
6956 (s390_function_profiler): Likewise.
6957 (s390_emit_call): Likewise.
6958 (s390_conditional_register_usage): Likewise.
6959 (s390_optimize_prologue): Likewise.
6960 (s390_reorg): Remove TARGET_CPU_ZARCH and
6961 split_branches_pending_p usages.
6962 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
6964 (s390_output_indirect_thunk_function): Likewise.
6965 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
6966 (TARGET_CPU_ZARCH_P): Removed.
6967 (struct machine_function): Remove split_branches_pending_p.
6968 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
6970 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
6972 * common/config/s390/s390-common.c (processor_flags_table):
6974 * config.gcc: Remove with_arch/with_tune support.
6975 * config/s390/2064.md: Remove cpu attribute comparisons.
6976 * config/s390/driver-native.c (s390_host_detect_local_cpu):
6978 * config/s390/linux.h (ASM_SPEC):
6979 Remove -march support.
6980 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
6981 Use a table to get an arch level.
6982 * config/s390/s390-opts.h (enum processor_type):
6984 * config/s390/s390.c
6985 (processor_table): Remove entries, add arch_level values.
6986 (s390_issue_rate): Remove cases.
6987 (s390_option_override): Adjust
6988 s390_option_override_internal() call.
6989 (s390_option_override_internal): Remove deprecation warning.
6990 (s390_valid_target_attribute_tree): Adjust
6991 s390_option_override_internal() call.
6992 * config/s390/s390.h (struct s390_processor):
6993 Share with s390-c.c, add arch_level field.
6994 * config/s390/s390.md:
6995 Remove occurrences in cpu attribute.
6996 * config/s390/s390.opt: Remove -march/-mtune support.
6997 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
6998 * doc/invoke.texi: Remove deprecation warning.
7000 2018-08-08 Luis Machado <luis.machado@linaro.org>
7002 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
7004 (qdf24xx_tunings): Set vector cost structure to
7005 qdf24xx_vector_cost.
7007 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
7008 <register_sextend>: Set to 3.
7010 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
7013 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
7014 * config/aarch64/aarch64-simd.md
7015 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
7016 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
7017 (aarch64_frecpx<mode>): ...this new pattern.
7018 * config/aarch64/aarch64-simd-builtins.def: Remove comment
7019 about aarch64_frecp<FRECP:frecp_suffix><mode>.
7021 2018-08-07 Martin Liska <mliska@suse.cz>
7024 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
7025 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
7026 * predict.def (PRED_MALLOC_NONNULL): New predictor.
7027 * doc/extend.texi: Document that malloc attribute adds
7030 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
7033 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7034 Define to speculation_safe_value_not_needed.
7036 2018-08-06 Jeff Law <law@redhat.com>
7038 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
7039 the vr_values instance to cprop_into_stmt.
7040 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
7041 (cprop_operand): Also query EVRP to determine if OP is a constant.
7043 2018-08-06 Nathan Sidwell <nathan@acm.org>
7045 * diagnostic.c (diagnostic_report_current_module): Reroll
7046 included-at loop. Translate text.
7048 2018-08-06 David Malcolm <dmalcolm@redhat.com>
7050 * function-tests.c (selftest::test_expansion_to_rtl): Call
7051 free_after_compilation.
7053 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7055 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
7057 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
7059 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
7060 loops with memory block operations from getting unrolled.
7062 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
7065 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7066 Define to speculation_safe_value_not_needed.
7068 2018-08-06 Jeff Law <law@redhat.com>
7070 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
7073 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
7076 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
7077 with all enabled __intN types.
7079 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
7081 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7083 * alias.c (record_set): Check for clobber high.
7084 * cfgexpand.c (expand_gimple_stmt): Likewise.
7085 * combine-stack-adj.c (single_set_for_csa): Likewise.
7086 * combine.c (find_single_use_1): Likewise.
7087 (set_nonzero_bits_and_sign_copies): Likewise.
7088 (get_combine_src_dest): Likewise.
7089 (is_parallel_of_n_reg_sets): Likewise.
7090 (try_combine): Likewise.
7091 (record_dead_and_set_regs_1): Likewise.
7092 (reg_dead_at_p_1): Likewise.
7093 (reg_dead_at_p): Likewise.
7094 * dce.c (deletable_insn_p): Likewise.
7095 (mark_nonreg_stores_1): Likewise.
7096 (mark_nonreg_stores_2): Likewise.
7097 * df-scan.c (df_find_hard_reg_defs): Likewise.
7098 (df_uses_record): Likewise.
7099 (df_get_call_refs): Likewise.
7100 * dwarf2out.c (mem_loc_descriptor): Likewise.
7101 * haifa-sched.c (haifa_classify_rtx): Likewise.
7102 * ira-build.c (create_insn_allocnos): Likewise.
7103 * ira-costs.c (scan_one_insn): Likewise.
7104 * ira.c (equiv_init_movable_p): Likewise.
7105 (rtx_moveable_p): Likewise.
7106 (interesting_dest_for_shprep): Likewise.
7107 * jump.c (mark_jump_label_1): Likewise.
7108 * postreload-gcse.c (record_opr_changes): Likewise.
7109 * postreload.c (reload_cse_simplify): Likewise.
7110 (struct reg_use): Add source expr.
7111 (reload_combine): Check for clobber high.
7112 (reload_combine_note_use): Likewise.
7113 (reload_cse_move2add): Likewise.
7114 (move2add_note_store): Likewise.
7115 * print-rtl.c (print_pattern): Likewise.
7116 * recog.c (decode_asm_operands): Likewise.
7117 (store_data_bypass_p): Likewise.
7118 (if_test_bypass_p): Likewise.
7119 * regcprop.c (kill_clobbered_value): Likewise.
7120 (kill_set_value): Likewise.
7121 * reginfo.c (reg_scan_mark_refs): Likewise.
7122 * reload1.c (maybe_fix_stack_asms): Likewise.
7123 (eliminate_regs_1): Likewise.
7124 (elimination_effects): Likewise.
7125 (mark_not_eliminable): Likewise.
7126 (scan_paradoxical_subregs): Likewise.
7127 (forget_old_reloads_1): Likewise.
7128 * reorg.c (find_end_label): Likewise.
7129 (try_merge_delay_insns): Likewise.
7130 (redundant_insn): Likewise.
7131 (own_thread_p): Likewise.
7132 (fill_simple_delay_slots): Likewise.
7133 (fill_slots_from_thread): Likewise.
7134 (dbr_schedule): Likewise.
7135 * resource.c (update_live_status): Likewise.
7136 (mark_referenced_resources): Likewise.
7137 (mark_set_resources): Likewise.
7138 * rtl.c (copy_rtx): Likewise.
7139 * rtlanal.c (reg_referenced_p): Likewise.
7140 (single_set_2): Likewise.
7141 (noop_move_p): Likewise.
7142 (note_stores): Likewise.
7143 * sched-deps.c (sched_analyze_reg): Likewise.
7144 (sched_analyze_insn): Likewise.
7146 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7148 * cse.c (invalidate_reg): New function extracted from...
7149 (invalidate): ...here.
7150 (canonicalize_insn): Check for clobber high.
7151 (invalidate_from_clobbers): invalidate clobber highs.
7152 (invalidate_from_sets_and_clobbers): Likewise.
7153 (count_reg_usage): Check for clobber high.
7154 (insn_live_p): Likewise.
7155 * cselib.c (cselib_expand_value_rtx_1):Likewise.
7156 (cselib_invalidate_regno): Check for clobber in setter.
7157 (cselib_invalidate_rtx): Pass through setter.
7158 (cselib_invalidate_rtx_note_stores):
7159 (cselib_process_insn): Check for clobber high.
7160 * cselib.h (cselib_invalidate_rtx): Add operand.
7162 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7164 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
7165 (mark_not_eliminable): Likewise.
7166 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
7167 * lra-lives.c (process_bb_lives): Check for clobber high.
7168 * lra.c (new_insn_reg): Remember clobber highs.
7169 (collect_non_operand_hard_regs): Check for clobber high.
7170 (lra_set_insn_recog_data): Likewise.
7171 (add_regs_to_insn_regno_info): Likewise.
7172 (lra_update_insn_regno_info): Likewise.
7174 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7176 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
7177 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
7179 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7181 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
7182 (copy_insn_1): Likewise.
7183 (gen_hard_reg_clobber_high): New gen function.
7184 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
7185 * genemit.c (gen_exp): Likewise.
7186 (gen_emit_seq): Pass through info.
7187 (gen_insn): Check for CLOBBER_HIGH.
7188 (gen_expand): Pass through info.
7189 (gen_split): Likewise.
7190 (output_add_clobbers): Likewise.
7191 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
7192 (remove_clobbers): Likewise.
7193 * rtl.h (gen_hard_reg_clobber_high): New declaration.
7195 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7197 * doc/rtl.texi (clobber_high): Add.
7198 (parallel): Add in clobber high
7199 * rtl.c (rtl_check_failed_code3): Add function.
7200 * rtl.def (CLOBBER_HIGH): Add expression.
7201 * rtl.h (RTL_CHECKC3): Add macro.
7202 (rtl_check_failed_code3): Add declaration.
7203 (XC3EXP): Add macro.
7205 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
7208 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
7209 cfun->machine->max_used_stack_alignment if needed.
7211 2018-08-04 Martin Sebor <msebor@redhat.com>
7213 PR tree-optimization/86571
7214 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
7217 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
7220 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7223 2018-08-03 Jeff Law <law@redhat.com>
7226 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7227 Define to speculation_safe_value_not_needed.
7229 2018-08-03 David Malcolm <dmalcolm@redhat.com>
7231 * doc/gcov.texi (-x): Remove duplicate "to".
7232 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
7233 (-Wif-not-aligned): Remove duplicate "is".
7234 (-flto): Remove duplicate "the".
7235 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
7236 duplicate "v5.00.b".
7237 (MSP430 Options): Remove duplicate "and" from the description
7238 of "-mgprel-sec=regexp".
7239 (x86 Options): Remove duplicate copies of "vmldLog102" and
7240 vmlsLog104 from description of "-mveclibabi=type".
7242 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
7244 * internal-fn.h (first_commutative_argument): Declare.
7245 * internal-fn.c (first_commutative_argument): New function.
7246 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
7247 restrictions for pattern statements. Use first_commutative_argument
7248 to look for commutative operands in calls to internal functions.
7250 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
7252 * Makefile.in (wide-int-range.o): New.
7253 * tree-vrp.c: Move all the wide_int_* functions to...
7254 * wide-int-range.cc: ...here.
7255 * tree-vrp.h: Move all the wide_int_* prototypes to...
7256 * wide-int-range.h: ...here.
7258 2018-08-03 Tom de Vries <tdevries@suse.de>
7260 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
7262 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
7263 * except.c (output_function_exception_table): Do early exit if
7264 targetm_common.except_unwind_info (&global_options) == UI_NONE.
7266 2018-08-03 Martin Liska <mliska@suse.cz>
7268 * predict.c (dump_prediction): Change to 2 digits
7271 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
7273 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
7275 (falkor_am_1_gtov_gtov): ... a new insn reservation.
7277 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
7279 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
7280 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
7281 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
7282 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
7283 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
7285 2018-08-02 David Malcolm <dmalcolm@redhat.com>
7287 * diagnostic-show-locus.c (diagnostic_show_locus): Use
7288 pp_take_prefix when saving the existing prefix.
7289 * diagnostic.c (diagnostic_append_note): Likewise.
7290 * langhooks.c (lhd_print_error_function): Likewise.
7291 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
7292 param's type. Free the existing prefix.
7293 (pp_take_prefix): New function.
7294 (pretty_printer::pretty_printer): Drop the prefix parameter.
7295 Rename the length parameter to match the comment.
7296 (pretty_printer::~pretty_printer): Free the prefix.
7297 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
7299 (struct pretty_printer): Drop the "const" from "prefix" field's
7300 type and clarify memory management.
7301 (pp_set_prefix): Drop the "const" from the 2nd param.
7302 (pp_take_prefix): New decl.
7304 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
7306 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
7307 for word_mode_ok here instead of passing as argument.
7308 (expand_block_compare): Change select_block_compare_mode() call.
7309 (expand_strncmp_gpr_sequence): New function.
7310 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
7312 2018-08-02 Jeff Law <law@redhat.com>
7315 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7316 Define to speculation_safe_value_not_needed.
7319 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7320 Define to speculation_safe_value_not_needed.
7322 2018-08-02 Tom de Vries <tdevries@suse.de>
7325 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
7326 function. Return UI_TARGET unconditionally.
7327 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
7328 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
7330 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7332 * genemit.c (print_overload_test): Fix typo.
7334 2018-08-02 Richard Biener <rguenther@suse.de>
7336 PR tree-optimization/86816
7337 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
7338 which checks for value availability before querying it.
7339 (gvn_uses_equal): Use it.
7340 (same_succ_hash): Likewise.
7341 (gimple_equal_p): Likewise.
7343 2018-08-02 Nick Clifton <nickc@redhat.com>
7346 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7347 Define to speculation_safe_value_not_needed.
7350 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7351 Define to speculation_safe_value_not_needed.
7354 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7355 Define to speculation_safe_value_not_needed.
7358 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7359 Define to speculation_safe_value_not_needed.
7362 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7363 Define to speculation_safe_value_not_needed.
7366 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7367 Define to speculation_safe_value_not_needed.
7370 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7371 Define to speculation_safe_value_not_needed.
7374 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7375 Define to speculation_safe_value_not_needed.
7378 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
7379 speculation_safe_value_not_needed.
7382 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
7383 to speculation_safe_value_not_needed.
7385 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7387 * doc/md.texi: Expand the documentation of instruction names
7388 to mention port-local uses. Document '@' in pattern names.
7389 * read-md.h (overloaded_instance, overloaded_name): New structs.
7391 (md_reader::handle_overloaded_name): New member function.
7392 (md_reader::get_overloads): Likewise.
7393 (md_reader::m_first_overload): New member variable.
7394 (md_reader::m_next_overload_ptr): Likewise.
7395 (md_reader::m_overloads_htab): Likewise.
7396 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
7397 m_next_overload_ptr and m_overloads_htab.
7398 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
7399 (get_mode_token, get_code_token, get_int_token): New functions.
7400 (map_attr_string): Add an optional argument that passes back
7401 the associated iterator.
7402 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
7403 (md_reader::handle_overloaded_name, add_overload_instance): New
7405 (apply_iterators): Handle '@' names. Report an error if '@'
7406 is used without iterators.
7407 (initialize_iterators): Initialize the new iterator_group fields.
7408 * genopinit.c (handle_overloaded_code_for)
7409 (handle_overloaded_gen): New functions.
7410 (main): Use them to print declarations of maybe_code_for_* and
7411 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
7412 * genemit.c (print_overload_arguments, print_overload_test)
7413 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
7414 (main): Use it to print definitions of maybe_code_for_* and
7415 maybe_gen_* functions.
7416 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
7417 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
7418 instead of explicit mode checks.
7419 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
7420 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
7421 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
7422 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
7423 (aarch64_expand_compare_and_swap): Likewise
7424 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
7425 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
7426 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
7427 (aarch64_constant_pool_reload_icode): Delete.
7428 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
7429 instead of aarch64_constant_pool_reload_icode. Use
7430 code_for_aarch64_reload_mov instead of explicit mode checks.
7431 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
7432 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
7433 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
7434 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
7435 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
7436 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
7437 (aarch64_atomic_load_op_code): Delete.
7438 (aarch64_emit_atomic_load_op): Likewise.
7439 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
7440 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
7441 instead of aarch64_emit_atomic_load_op.
7442 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
7443 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
7444 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
7445 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
7446 character before the pattern name.
7447 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7448 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
7449 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
7450 (aarch64_frecps<mode>): Likewise.
7451 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
7452 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
7453 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
7454 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
7455 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
7457 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7459 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
7460 Allow HFmode constants if TARGET_FP_F16INST.
7462 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
7465 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
7466 No longer check last store for clobber of address register.
7468 2018-08-02 Martin Liska <mliska@suse.cz>
7470 PR gcov-profile/86817
7471 * gcov.c (process_all_functions): New function.
7473 (process_file): Move functions processing to
7474 process_all_functions.
7476 2018-08-02 David Malcolm <dmalcolm@redhat.com>
7478 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
7479 "const" to the "gimple *" and "rtx_insn *" parameters.
7480 * dumpfile.h (dump_user_location_t::dump_user_location_t):
7482 (dump_location_t::dump_location_t): Likewise.
7484 2018-08-01 Martin Sebor <msebor@redhat.com>
7486 PR tree-optimization/86650
7487 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
7488 rather than a "gcall *". Directly pass the data of interest
7489 to percent_K_format, rather than building a temporary CALL_EXPR
7491 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
7492 (gimple_fold_builtin_strncat): Adjust.
7493 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
7494 gcall* argument with gimple*.
7495 * gimple-ssa-warn-restrict.c (check_call): Same.
7496 (wrestrict_dom_walker::before_dom_children): Same.
7497 (builtin_access::builtin_access): Same.
7498 (check_bounds_or_overlap): Same
7499 (maybe_diag_overlap): Same.
7500 (maybe_diag_offset_bounds): Same.
7501 * tree-diagnostic.c (default_tree_printer): Move usage of
7502 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
7504 * tree-pretty-print.c (percent_K_format): Add argument.
7505 * tree-pretty-print.h: Add argument.
7506 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
7507 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
7508 (maybe_diag_stxncpy_trunc): Same.
7509 (handle_builtin_stxncpy): Same.
7510 (handle_builtin_strcat): Same.
7512 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7514 * match.pd: Optimise pointer range checks.
7516 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7518 PR tree-optimization/86758
7519 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
7520 to remove pattern statements.
7522 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7524 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
7525 result of dfs_enumerate_from when constructing stmt_vec_infos,
7526 instead of additionally calling get_loop_body.
7528 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7530 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
7532 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
7533 When creating an iv, assert that the step is not known to be zero.
7534 (vect_setup_realignment): Update call accordingly.
7535 * tree-vect-stmts.c (vectorizable_store): Likewise.
7536 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
7538 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7540 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
7541 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
7542 (vectorizable_reduction): Likewise.
7543 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
7544 (vect_detect_hybrid_slp_stmts): Likewise.
7545 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
7547 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
7549 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
7550 (wide_int_set_zero_nonzero_bits): ...this.
7551 (zero_nonzero_bits_from_vr): Rename to...
7552 (vrp_set_zero_nonzero_bits): ...this.
7553 (extract_range_from_multiplicative_op_1): Abstract wide int
7555 (wide_int_range_multiplicative_op): ...here.
7556 (extract_range_from_binary_expr_1): Extract wide int binary
7557 operations into their own functions.
7558 (wide_int_range_lshift): New.
7559 (wide_int_range_can_optimize_bit_op): New.
7560 (wide_int_range_shift_undefined_p): New.
7561 (wide_int_range_bit_xor): New.
7562 (wide_int_range_bit_ior): New.
7563 (wide_int_range_bit_and): New.
7564 (wide_int_range_trunc_mod): New.
7565 (extract_range_into_wide_ints): New.
7566 (vrp_shift_undefined_p): New.
7567 (extract_range_from_multiplicative_op): New.
7568 (vrp_can_optimize_bit_op): New.
7569 * tree-vrp.h (value_range::dump): New.
7570 (wide_int_range_multiplicative_op): New.
7571 (wide_int_range_lshift):New.
7572 (wide_int_range_shift_undefined_p): New.
7573 (wide_int_range_bit_xor): New.
7574 (wide_int_range_bit_ior): New.
7575 (wide_int_range_bit_and): New.
7576 (wide_int_range_trunc_mod): New.
7577 (zero_nonzero_bits_from_bounds): Rename to...
7578 (wide_int_set_zero_nonzero_bits): ...this.
7579 (zero_nonzero_bits_from_vr): Rename to...
7580 (vrp_set_zero_nonzero_bits): ...this.
7581 (range_easy_mask_min_max): Rename to...
7582 (wide_int_range_can_optimize_bit_op): this.
7583 * vr-values.c (simplify_bit_ops_using_ranges): Rename
7584 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
7586 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7588 * tree-vectorizer.h (vect_orig_stmt): New function.
7589 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
7590 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
7591 (vect_create_epilog_for_reduction): Likewise.
7592 (vectorizable_live_operation): Likewise.
7593 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
7594 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
7595 * tree-vect-stmts.c (vectorizable_call): Likewise.
7596 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
7598 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7600 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
7602 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
7603 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
7604 (vect_transform_loop): Likewise.
7605 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7607 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7609 * tree-vectorizer.h (vect_schedule_slp): Return void.
7610 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7611 (vect_schedule_slp): Likewise.
7613 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7615 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
7617 (vect_transform_loop): Update calls accordingly. Schedule SLP
7618 instances before the main loop, if any exist.
7620 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7622 PR tree-optimization/86749
7623 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
7624 If the lhs is used in a COND_EXPR, check that it is being used
7625 as the "then" or "else" value.
7627 2018-08-01 Tom de Vries <tdevries@suse.de>
7630 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
7631 speculation_safe_value_not_needed.
7633 2018-08-01 Richard Biener <rguenther@suse.de>
7635 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
7638 2018-08-01 Martin Liska <mliska@suse.cz>
7640 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
7641 format how successful transformation is dumped.
7642 (gimple_mod_pow2_value_transform): Likewise.
7643 (gimple_mod_subtract_transform): Likewise.
7644 (gimple_stringops_transform): Likewise.
7646 2018-08-01 Martin Liska <mliska@suse.cz>
7649 * value-prof.c (interesting_stringop_to_profile_p):
7650 Simplify the code and add BUILT_IN_MEMMOVE.
7651 (gimple_stringops_transform): Likewise.
7653 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
7655 * config/aarch64/aarch64-simd.md
7656 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
7657 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
7658 use GPI iterator instead of SI mode.
7660 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
7662 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
7663 rs6000_speculation_barrier.
7664 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
7665 new barrier pattern name.
7667 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
7669 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
7670 (speculation_barrier): New insn.
7672 2018-08-01 Richard Biener <rguenther@suse.de>
7675 * graphite.h: Include isl/id.h and isl/space.h to allow build
7678 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
7681 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
7682 mode for COFF targets.
7683 * defaults.h (TARGET_COFF): Define.
7684 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
7685 TARGET_COFF): Define.
7686 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
7687 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
7688 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
7690 2018-07-31 Alexandre Oliva <oliva@adacore.com>
7692 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
7693 * gimple-streamer-out.c (output_bb): Save it.
7694 * lto-streamer-in.c (input_struct_function_base): Restore
7695 instance discriminator if available. Create map on demand.
7696 * lto-streamer-out.c (output_struct_function_base): Save it if
7698 * final.c (decl_to_instance_map): Document LTO strategy.
7700 2018-07-31 Alexandre Oliva <oliva@adacore.com>
7701 Olivier Hainque <hainque@adacore.com>
7703 * debug.h (decl_to_instance_map_t): New type.
7704 (decl_to_instance_map): Declare.
7705 (maybe_create_decl_to_instance_map): New inline function.
7706 * final.c (bb_discriminator, last_bb_discriminator): New statics,
7707 to track basic block discriminators.
7708 (final_start_function_1): Initialize them.
7709 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
7711 (decl_to_instance_map): New variable.
7712 (map_decl_to_instance, maybe_set_discriminator): New functions.
7713 (notice_source_line): Set discriminator.
7715 2018-07-31 Ian Lance Taylor <iant@golang.org>
7717 * targhooks.c (default_have_speculation_safe_value): Add
7720 2018-07-31 David Malcolm <dmalcolm@redhat.com>
7722 * dump-context.h: Include "pretty-print.h".
7723 (dump_context::refresh_dumps_are_enabled): New decl.
7724 (dump_context::emit_item): New decl.
7725 (class dump_context): Add fields "m_test_pp" and
7727 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
7728 (temp_dump_context::get_dumped_text): New decl.
7729 (class temp_dump_context): Add field "m_pp".
7730 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
7731 (dump_context::refresh_dumps_are_enabled): ...and add a test for
7733 (set_dump_file): Update for above change.
7734 (set_alt_dump_file): Likewise.
7735 (dump_loc): New overload, taking a pretty_printer *.
7736 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
7737 to any test pretty-printer.
7738 (make_item_for_dump_gimple_stmt): New function, adapted from
7739 optinfo::add_gimple_stmt.
7740 (dump_context::dump_gimple_stmt): Call it, and use the result,
7741 eliminating the direct usage of dump_file and alt_dump_file in
7742 favor of indirectly using them via emit_item.
7743 (make_item_for_dump_gimple_expr): New function, adapted from
7744 optinfo::add_gimple_expr.
7745 (dump_context::dump_gimple_expr): Call it, and use the result,
7746 eliminating the direct usage of dump_file and alt_dump_file in
7747 favor of indirectly using them via emit_item.
7748 (make_item_for_dump_generic_expr): New function, adapted from
7750 (dump_context::dump_generic_expr): Call it, and use the result,
7751 eliminating the direct usage of dump_file and alt_dump_file in
7752 favor of indirectly using them via emit_item.
7753 (make_item_for_dump_printf_va): New function, adapted from
7754 optinfo::add_printf_va.
7755 (make_item_for_dump_printf): New function.
7756 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
7757 and use the result, eliminating the direct usage of dump_file and
7758 alt_dump_file in favor of indirectly using them via emit_item.
7759 (make_item_for_dump_dec): New function.
7760 (dump_context::dump_dec): Call it, and use the result,
7761 eliminating the direct usage of dump_file and alt_dump_file in
7762 favor of indirectly using them via emit_item.
7763 (make_item_for_dump_symtab_node): New function, adapted from
7764 optinfo::add_symtab_node.
7765 (dump_context::dump_symtab_node): Call it, and use the result,
7766 eliminating the direct usage of dump_file and alt_dump_file in
7767 favor of indirectly using them via emit_item.
7768 (dump_context::begin_scope): Reimplement, avoiding direct usage
7769 of dump_file and alt_dump_file in favor of indirectly using them
7771 (dump_context::emit_item): New member function.
7772 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
7773 Set up test pretty-printer on the underlying context. Call
7774 refresh_dumps_are_enabled.
7775 (temp_dump_context::~temp_dump_context): Call
7776 refresh_dumps_are_enabled.
7777 (temp_dump_context::get_dumped_text): New member function.
7778 (selftest::verify_dumped_text): New function.
7779 (ASSERT_DUMPED_TEXT_EQ): New macro.
7780 (selftest::test_capture_of_dump_calls): Run all tests twice, with
7781 and then without optinfo enabled. Add uses of
7782 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
7783 * dumpfile.h: Update comment for the dump_* API.
7784 * optinfo-emit-json.cc
7785 (selftest::test_building_json_from_dump_calls): Update for new
7786 param for temp_dump_context ctor.
7787 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
7788 and "m_owned" field.
7789 (optinfo_item::~optinfo_item): Likewise.
7790 (optinfo::add_item): New member function.
7791 (optinfo::emit): Update comment.
7792 (optinfo::add_string): Delete.
7793 (optinfo::add_printf): Delete.
7794 (optinfo::add_printf_va): Delete.
7795 (optinfo::add_gimple_stmt): Delete.
7796 (optinfo::add_gimple_expr): Delete.
7797 (optinfo::add_tree): Delete.
7798 (optinfo::add_symtab_node): Delete.
7799 (optinfo::add_dec): Delete.
7800 * optinfo.h (class dump_context): New forward decl.
7801 (optinfo::add_item): New decl.
7802 (optinfo::add_string): Delete.
7803 (optinfo::add_printf): Delete.
7804 (optinfo::add_printf_va): Delete.
7805 (optinfo::add_gimple_stmt): Delete.
7806 (optinfo::add_gimple_expr): Delete.
7807 (optinfo::add_tree): Delete.
7808 (optinfo::add_symtab_node): Delete.
7809 (optinfo::add_dec): Delete.
7810 (optinfo::add_poly_int): Delete.
7811 (optinfo_item::optinfo_item): Remove "owned" param.
7812 (class optinfo_item): Remove field "m_owned".
7814 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
7817 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
7818 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
7819 requested variable alignment.
7820 (expand_one_ssa_partition): Likewise.
7821 (expand_one_var): Likewise.
7823 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7825 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
7826 to speculation_safe_value_not_needed.
7828 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7830 * targhooks.h (speculation_safe_value_not_needed): New prototype.
7831 * targhooks.c (speculation_safe_value_not_needed): New function.
7832 * target.def (have_speculation_safe_value): Update documentation.
7833 * doc/tm.texi: Regenerated.
7835 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7837 * config/aarch64/iterators.md (ALLI_TI): New iterator.
7838 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
7840 (despeculate_copy<ALLI:mode>_insn): New insn.
7841 (despeculate_copyti_insn): New insn.
7842 (despeculate_simple<ALLI:mode>): New insn
7843 (despeculate_simpleti): New insn.
7844 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
7846 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
7847 aarch64_speculation_safe_value.
7848 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
7850 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7852 * config/aarch64/aarch64-speculation.cc: New file.
7853 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
7854 before pass_reorder_blocks.
7855 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
7857 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
7858 X14 and X15 when tracking speculation.
7859 * config/aarch64/aarch64.md (register name constants): Add
7860 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
7861 (unspec): Add UNSPEC_SPECULATION_TRACKER.
7862 (speculation_barrier): New insn attribute.
7863 (cmp<mode>): Allow SP in comparisons.
7864 (speculation_tracker): New insn.
7865 (speculation_barrier): Add speculation_barrier attribute.
7866 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
7867 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
7868 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
7870 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7872 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
7873 aarch64_track_speculation is true.
7874 (tb<optab><mode>1): Likewise.
7875 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
7876 generate CB[N]Z when tracking speculation.
7877 (aarch64_split_compare_and_swap): Likewise.
7878 (aarch64_split_atomic_op): Likewise.
7880 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7882 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
7884 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7886 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
7887 (speculation_barrier): New insn.
7889 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7891 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
7892 * config/arm/arm.md (speculation_barrier): New expand.
7893 (speculation_barrier_insn): New pattern.
7895 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7897 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
7898 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
7899 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
7900 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
7902 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
7903 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
7904 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
7905 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
7906 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
7907 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
7908 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
7909 * builtins.c (expand_speculation_safe_value): New function.
7910 (expand_builtin): Call it.
7911 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
7912 * doc/extend.texi: Document __builtin_speculation_safe_value.
7913 * doc/md.texi: Document "speculation_barrier" pattern.
7914 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
7915 TARGET_HAVE_SPECULATION_SAFE_VALUE.
7916 * doc/tm.texi: Regenerated.
7917 * target.def (have_speculation_safe_value, speculation_safe_value): New
7919 * targhooks.c (default_have_speculation_safe_value): New function.
7920 (default_speculation_safe_value): New function.
7921 * targhooks.h (default_have_speculation_safe_value): Add prototype.
7922 (default_speculation_safe_value): Add prototype.
7924 2018-07-31 David Malcolm <dmalcolm@redhat.com>
7926 * dump-context.h (dump_context::dump_loc): New decl.
7927 * dumpfile.c (dump_context::dump_loc): New member function.
7928 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
7929 and dump_gimple_stmt.
7930 (dump_context::dump_gimple_expr_loc): Likewise, using
7932 (dump_context::dump_generic_expr_loc): Likewise, using
7934 (dump_context::dump_printf_loc_va): Likewise, using
7936 (dump_context::begin_scope): Explicitly using the global function
7937 "dump_loc", rather than the member function.
7939 2018-07-31 Martin Sebor <msebor@redhat.com>
7941 PR tree-optimization/86741
7942 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
7944 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
7946 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
7947 depend on whether prefetch instructions will be emitted or not.
7948 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
7949 will be emitted or not.
7950 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
7951 (TARGET_SETMEM_PFD): New macros.
7953 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7955 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
7956 (NULL_STMT_VEC_INFO): Delete.
7957 (stmt_vec_info::operator*): Likewise.
7958 (stmt_vec_info::operator gimple *): Likewise.
7959 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
7960 of NULL_STMT_VEC_INFO.
7961 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
7962 (vect_reassociating_reduction_p): Likewise.
7963 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
7964 (vectorizable_store): Likewise.
7965 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
7966 (vec_info::free_stmt_vec_infos): Likewise.
7968 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7970 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
7971 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
7973 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7975 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
7976 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
7977 (vec_info::free_stmt_vec_info): New private member functions.
7978 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
7979 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
7980 * tree-parloops.c (gather_scalar_reductions): Remove calls to
7981 set_stmt_vec_info_vec and free_stmt_vec_infos.
7982 * tree-vect-loop.c (_loop_vec_info): Remove call to
7983 set_stmt_vec_info_vec.
7984 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
7985 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
7986 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
7987 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
7988 (vec_info::free_stmt_vec_info): ...these new functions. Remove
7989 assignments in {vec_info::,}new_stmt_vec_info that are redundant
7990 with the clearing in the xcalloc.
7991 (stmt_vec_info_vec): Delete.
7992 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
7993 (vectorize_loops): Likewise.
7994 (vec_info::~vec_info): Remove argument from call to
7995 free_stmt_vec_infos.
7996 (vec_info::add_stmt): Remove vinfo argument from call to
7999 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8001 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
8002 rather than a gimple stmt.
8003 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
8004 information for pattern statements when passed the original
8005 statement; instead wait to be passed the pattern statement itself.
8006 Don't call set_vinfo_for_stmt here.
8007 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
8008 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
8009 stmt_vec_infos here.
8010 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
8011 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
8012 stmt_vec_infos entry.
8014 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8016 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
8017 * tree-vectorizer.c (vec_info::replace_stmt): New function.
8018 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
8019 * tree-vect-stmts.c (vectorizable_call): Likewise.
8020 (vectorizable_simd_clone_call): Likewise.
8022 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8024 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
8025 * tree-vectorizer.c (vec_info::remove_stmt): New function.
8026 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
8027 * tree-vect-loop.c (vect_transform_loop): Likewise.
8028 * tree-vect-slp.c (vect_schedule_slp): Likewise.
8029 * tree-vect-stmts.c (vect_remove_stores): Likewise.
8031 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8033 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
8034 (vect_dr_stmt): Delete.
8035 * tree-vectorizer.c (vec_info::lookup_dr): New function.
8036 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
8038 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
8039 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
8040 (vect_verify_datarefs_alignment, vect_peeling_supportable)
8041 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
8042 (vect_analyze_data_refs): Likewise.
8043 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
8045 (vect_find_same_alignment_drs): Likewise.
8046 (vect_slp_analyze_node_dependences): Update calls accordingly.
8047 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
8048 instead of DR_VECT_AUX.
8049 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
8050 of a vector data references. Use vec_info::lookup_dr instead of
8052 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
8053 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
8054 instead of DR_VECT_AUX.
8056 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8058 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
8060 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
8062 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8063 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
8064 (vect_gen_prolog_loop_niters): Likewise.
8066 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8068 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
8069 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
8070 (vect_known_alignment_in_bytes, vect_dr_behavior)
8071 (vect_get_scalar_dr_size): Take references as dr_vec_infos
8072 instead of data_references. Update calls to other routines for
8073 which the same change has been made.
8074 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
8075 dr_vec_infos instead of stmt_vec_infos.
8076 (vect_analyze_data_ref_dependence): Update call accordingly.
8077 (vect_slp_analyze_data_ref_dependence)
8078 (vect_record_base_alignments): Use DR_VECT_AUX.
8079 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
8080 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
8081 (vector_alignment_reachable_p, vect_get_data_access_cost)
8082 (vect_peeling_supportable, vect_analyze_group_access_1)
8083 (vect_analyze_group_access, vect_analyze_data_ref_access)
8084 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
8085 (vect_compile_time_alias, vect_small_gap_p)
8086 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
8087 (vect_supportable_dr_alignment): Take references as dr_vec_infos
8088 instead of data_references. Update calls to other routines for
8089 which the same change has been made.
8090 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
8091 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
8092 (vect_slp_analyze_and_verify_node_alignment)
8093 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
8094 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
8095 (vect_setup_realignment): Use dr_vec_infos. Update calls after
8097 (_vect_peel_info::dr): Replace with...
8098 (_vect_peel_info::dr_info): ...this new field.
8099 (vect_peeling_hash_get_most_frequent)
8100 (vect_peeling_hash_choose_best_peeling): Update accordingly.
8101 (vect_peeling_hash_get_lowest_cost):
8102 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
8103 routines for which the same change has been made.
8104 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
8106 * tree-vect-loop-manip.c (get_misalign_in_elems)
8107 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
8109 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8110 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
8111 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
8112 (get_group_load_store_type, get_negative_load_store_type)
8113 (vect_get_data_ptr_increment, vectorizable_store)
8114 (vectorizable_load): Likewise.
8115 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
8116 Update calls to other routines for which the same change has been made.
8118 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8120 * tree-vectorizer.h (vec_info::move_dr): New member function.
8121 (dataref_aux): Rename to...
8122 (dr_vec_info): ...this and add "dr" and "stmt" fields.
8123 (_stmt_vec_info::dr_aux): Update accordingly.
8124 (_stmt_vec_info::data_ref_info): Delete.
8125 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
8126 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
8127 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
8128 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
8130 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
8131 (STMT_VINFO_DR_INFO): New macro.
8132 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
8133 (set_dr_misalignment): Update after rename of dataref_aux.
8134 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
8135 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
8136 initialization of STMT_VINFO_DATA_REF.
8137 * tree-vectorizer.c (vec_info::move_dr): New function.
8138 * tree-vect-patterns.c (vect_recog_bool_pattern)
8139 (vect_recog_mask_conversion_pattern)
8140 (vect_recog_gather_scatter_pattern): Use it.
8141 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
8142 the "dr" and "stmt" fields of dr_vec_info instead of
8143 STMT_VINFO_DATA_REF.
8145 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8147 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
8148 (is_pattern_stmt_p): Use it.
8149 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
8150 on pattern statements.
8152 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8154 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
8155 original stmt as a stmt_vec_info rather than a gimple stmt.
8156 (vect_pattern_recog_1): Take the statement directly as a
8157 stmt_vec_info, rather than via a gimple_stmt_iterator.
8158 Update call to vect_mark_pattern_stmts.
8159 (vect_pattern_recog): Update calls accordingly.
8161 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8163 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
8164 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
8165 a vect_def_type for the first argument.
8166 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
8167 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
8168 operand if it isn't defined by a vectorized statement.
8169 (vect_build_gather_load_calls): Remove the mask_dt argument and
8170 update calls to vect_get_vec_def_for_stmt_copy.
8171 (vectorizable_bswap): Likewise the dt argument.
8172 (vectorizable_call): Update calls to vectorizable_bswap and
8173 vect_get_vec_def_for_stmt_copy.
8174 (vectorizable_simd_clone_call, vectorizable_assignment)
8175 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
8176 (vectorizable_comparison): Update calls to
8177 vect_get_vec_def_for_stmt_copy.
8178 (vectorizable_store): Likewise. Remove now-unnecessary calls to
8180 (vect_get_loop_based_defs): Remove dt argument and update call
8181 to vect_get_vec_def_for_stmt_copy.
8182 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
8183 and vect_get_vec_def_for_stmt_copy.
8184 (vectorizable_load): Update calls to vect_build_gather_load_calls
8185 and vect_get_vec_def_for_stmt_copy.
8186 * tree-vect-loop.c (vect_create_epilog_for_reduction)
8187 (vectorizable_reduction, vectorizable_live_operation): Update calls
8188 to vect_get_vec_def_for_stmt_copy.
8190 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8192 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
8193 and gimple stmt arguments with a stmt_vec_info.
8194 (vect_record_base_alignments): Update calls accordingly.
8195 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
8196 and gimple stmt arguments with a stmt_vec_info.
8197 (vect_build_slp_tree_1): Remove vinfo argument and update call
8198 to vect_record_max_nunits.
8199 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
8200 and vect_record_max_nunits.
8202 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8204 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
8205 file and take a stmt_vec_info instead of a gimple stmt.
8206 (supportable_widening_operation, vect_finish_replace_stmt)
8207 (vect_finish_stmt_generation, vect_get_store_rhs)
8208 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
8209 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
8210 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
8211 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
8212 (vect_create_data_ref_ptr, bump_vector_ptr)
8213 (vect_permute_store_chain, vect_setup_realignment)
8214 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
8215 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
8216 (vectorizable_reduction, vectorizable_induction)
8217 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
8218 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
8219 than gimple stmts as arguments.
8220 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
8221 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
8222 (can_group_stmts_p, vect_check_gather_scatter)
8223 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
8224 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
8225 (vect_permute_load_chain, vect_shift_permute_load_chain)
8226 (vect_transform_grouped_load)
8227 (vect_record_grouped_load_vectors): Likewise.
8228 * tree-vect-loop.c (vect_fixup_reduc_chain)
8229 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
8230 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
8231 (vectorizable_reduction, vectorizable_induction)
8232 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
8233 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
8234 (vect_get_load_store_mask): Likewise.
8235 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
8236 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
8237 * tree-vect-stmts.c (vect_mark_relevant)
8238 (is_simple_and_all_uses_invariant)
8239 (exist_non_indexing_operands_for_use_p, process_use)
8240 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
8241 (vect_get_vec_def_for_operand, vect_get_vec_defs)
8242 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
8243 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
8244 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
8245 (get_negative_load_store_type, get_load_store_type)
8246 (vect_check_load_store_mask, vect_check_store_rhs)
8247 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
8248 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
8249 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
8250 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8251 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
8252 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
8253 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
8254 (supportable_widening_operation): Likewise.
8256 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8258 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
8259 a stmt_vec_info instead of a gcall.
8260 (vect_check_gather_scatter): Update call accordingly.
8261 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
8263 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
8264 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
8265 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
8266 instead of a gimple stmt.
8267 (vect_transform_loop): Update calls accordingly.
8268 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
8269 stmt_vec_infos instead of gimple stmts.
8270 (vect_analyze_slp_instance): Update use accordingly.
8271 * tree-vect-stmts.c (read_vector_array, write_vector_array)
8272 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
8273 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
8274 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
8275 (vect_gen_widened_results_half, vect_get_loop_based_defs)
8276 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
8277 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
8278 down to subroutines.
8280 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8282 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
8283 of the worklist from a vector of gimple stmts to a vector of
8285 * tree-vect-stmts.c (vect_mark_relevant, process_use)
8286 (vect_mark_stmts_to_be_vectorized): Likewise
8288 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8290 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
8291 statement before passing it to vect_analyze_stmt.
8292 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
8293 the chain of phi vector definitions. Track the exit phi via its
8295 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
8296 STMT_VINFO_REDUC_DEF.
8297 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
8298 stmt_vec_infos to handle the statement chains.
8299 (vect_get_slp_defs): Record the first statement in the node
8300 using a stmt_vec_info.
8301 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
8302 statements here and pass their stmt_vec_info down to subroutines.
8303 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
8304 down to vect_finish_stmt_generation.
8305 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
8306 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
8307 stmt_vec_infos to subroutines.
8308 (vect_remove_stores): Use stmt_vec_infos to handle the statement
8311 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8313 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
8314 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
8315 (vect_permute_store_chain, vect_setup_realignment)
8316 (vect_permute_load_chain, vect_shift_permute_load_chain)
8317 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
8318 stmts internally, and when passing values to other vectorizer routines.
8319 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
8320 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
8321 (vect_analyze_loop_operations, get_initial_def_for_reduction)
8322 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
8323 (vectorizable_reduction, vectorizable_induction)
8324 (vectorizable_live_operation, vect_transform_loop_stmt)
8325 (vect_transform_loop): Likewise.
8326 * tree-vect-patterns.c (vect_reassociating_reduction_p)
8327 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
8328 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
8329 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8330 (vect_slp_analyze_node_operations_1): Likewise.
8331 * tree-vect-stmts.c (vect_mark_relevant, process_use)
8332 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
8333 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
8334 (vect_finish_stmt_generation_1, get_group_load_store_type)
8335 (get_load_store_type, vect_build_gather_load_calls)
8336 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
8337 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
8338 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8339 (vectorizable_store, vectorizable_load, vectorizable_condition)
8340 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
8341 (supportable_widening_operation): Likewise.
8342 (vect_get_vector_types_for_stmt): Likewise.
8343 * tree-vectorizer.h (vect_dr_behavior): Likewise.
8345 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8347 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
8348 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
8349 (vect_permute_store_chain, vect_permute_load_chain)
8350 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
8351 repeated stmt_vec_info lookups.
8352 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
8353 (vect_update_ivs_after_vectorizer): Likewise.
8354 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
8355 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
8356 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
8357 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8358 (vect_bb_slp_scalar_cost): Likewise.
8359 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
8361 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8363 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
8364 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
8365 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
8366 to get gassigns and gcalls, rather than operating on generc gimple
8368 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
8369 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
8370 (vectorizable_load, vect_analyze_stmt): Likewise.
8371 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
8373 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8375 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
8376 return stmt_vec_infos rather than gimple stmts. Do not accept
8378 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
8380 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
8381 Update use of get_later_stmt.
8382 (vect_get_constant_vectors): Update call accordingly.
8383 (vect_schedule_slp_instance): Likewise
8384 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
8385 (vect_slp_analyze_instance_dependence): Likewise.
8386 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
8388 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8390 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
8391 (stmt_info_for_cost::stmt_info): ...this new field.
8392 (add_stmt_costs): Update accordingly.
8393 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
8394 (vect_get_known_peeling_cost): Likewise.
8395 (vect_estimate_min_profitable_iters): Likewise.
8396 * tree-vect-stmts.c (record_stmt_cost): Likewise.
8398 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8400 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
8401 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
8402 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
8404 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
8406 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8408 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
8409 a gimple stmt to a stmt_vec_info.
8410 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
8412 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8414 * tree-vectorizer.h (vec_info::grouped_stores): Change from
8415 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
8416 (_loop_vec_info::reduction_chains): Likewise.
8417 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
8419 * tree-vect-slp.c (vect_analyze_slp): Likewise.
8421 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8423 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
8424 a gimple stmt to a stmt_vec_info.
8425 (_stmt_vec_info::next_element): Likewise.
8426 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
8427 (vect_slp_analyze_and_verify_node_alignment)
8428 (vect_analyze_group_access_1, vect_analyze_group_access)
8429 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
8430 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
8431 (vect_supportable_dr_alignment): Update accordingly.
8432 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
8433 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
8434 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
8435 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
8436 * tree-vect-slp.c (vect_build_slp_tree_1)
8437 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
8438 (vect_split_slp_store_group, vect_analyze_slp_instance)
8439 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
8440 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
8441 (get_group_load_store_type, get_load_store_type)
8442 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
8443 (vect_transform_stmt, vect_remove_stores): Likewise.
8445 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8447 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
8449 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
8450 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
8451 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
8452 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
8453 (vector_alignment_reachable_p, vect_get_data_access_cost)
8454 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
8455 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
8456 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
8457 (vect_analyze_group_access_1, vect_analyze_group_access)
8458 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
8459 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
8460 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
8461 result of vect_dr_stmt and use the stmt_vec_info instead of
8462 the associated gimple stmt.
8463 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
8464 (vect_gen_prolog_loop_niters): Likewise.
8465 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8467 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8469 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
8470 to a vec<stmt_vec_info>.
8471 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
8472 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
8474 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
8475 to a vec<stmt_vec_info>.
8476 (bst_traits::value_type, bst_traits::value_type): Likewise.
8477 (bst_traits::hash): Update accordingly.
8478 (vect_get_and_check_slp_defs): Change the stmts parameter from
8479 a vec<gimple *> to a vec<stmt_vec_info>.
8480 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
8481 (vect_build_slp_tree): Likewise.
8482 (vect_build_slp_tree_2): Likewise. Update uses of
8483 SLP_TREE_SCALAR_STMTS.
8484 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
8485 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
8486 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
8487 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
8488 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
8489 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
8490 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
8491 (vect_get_constant_vectors, vect_get_slp_defs)
8492 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
8493 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
8494 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
8495 instead of gimple stmts.
8496 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
8497 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
8498 (vect_slp_analyze_instance_dependence): Update uses of
8499 SLP_TREE_SCALAR_STMTS.
8500 (vect_slp_analyze_and_verify_node_alignment): Likewise.
8501 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
8502 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
8503 (get_initial_defs_for_reduction): Likewise.
8504 (vect_create_epilog_for_reduction): Likewise.
8505 (vectorize_fold_left_reduction): Likewise.
8506 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
8507 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
8508 (can_vectorize_live_stmts): Likewise.
8510 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8512 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
8513 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
8514 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
8516 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
8518 (gather_scalar_reductions): Update after above interface changes.
8519 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
8520 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
8522 (vect_force_simple_reduction): Likewise.
8523 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
8524 LOOP_VINFO_REDUCTIONS.
8525 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8527 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8529 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
8530 a gimple stmt to a stmt_vec_info.
8531 * tree-vect-loop.c (vect_active_double_reduction_p)
8532 (vect_force_simple_reduction, vectorizable_reduction): Update
8535 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8537 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
8538 vec<gimple *> to a vec<stmt_vec_info>.
8539 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
8540 the reduction_phis argument from a vec<gimple *> to a
8542 (vectorizable_reduction): Likewise the phis local variable that
8543 is passed to vect_create_epilog_for_reduction. Update for new type
8544 of SLP_TREE_VEC_STMTS.
8545 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
8546 (vectorizable_live_operation): Likewise.
8547 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
8548 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
8550 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8552 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
8553 a gimple stmt to a stmt_vec_info.
8554 (vectorizable_condition, vectorizable_live_operation)
8555 (vectorizable_reduction, vectorizable_induction): Pass back the
8556 vectorized statement as a stmt_vec_info.
8557 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
8558 use of STMT_VINFO_VEC_STMT.
8559 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
8560 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
8561 as stmt_vec_infos rather than gimple stmts.
8562 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
8564 (vectorizable_live_operation): Likewise.
8565 (vectorizable_reduction, vectorizable_induction): Likewise,
8566 updating use of STMT_VINFO_VEC_STMT.
8567 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
8568 of STMT_VINFO_VEC_STMT.
8569 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
8570 (vectorizable_simd_clone_call, vectorizable_conversion)
8571 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8572 (vectorizable_store, vectorizable_load, vectorizable_condition)
8573 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
8574 from a gimple stmt to a stmt_vec_info.
8575 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
8576 pointer to a stmt_vec_info to the vectorizable_* routines.
8578 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8580 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
8581 a gimple stmt to a stmt_vec_info.
8582 (is_pattern_stmt_p): Update accordingly.
8583 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
8584 (vect_record_grouped_load_vectors): Likewise.
8585 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
8586 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
8587 (vect_model_reduction_cost): Likewise.
8588 (vect_create_epilog_for_reduction): Likewise.
8589 (vectorizable_reduction, vectorizable_induction): Likewise.
8590 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8591 Return the stmt_vec_info for the pattern statement.
8592 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
8593 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
8594 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
8595 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
8596 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
8597 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
8598 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
8599 (free_stmt_vec_info, vect_is_simple_use): Likewise.
8601 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8603 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
8604 (vect_finish_stmt_generation): Likewise.
8605 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
8606 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
8607 (vect_build_gather_load_calls): Use the return value of the above
8608 functions instead of a separate call to vinfo_for_stmt. Use narrow
8609 scopes for the input gimple stmt and wider scopes for the associated
8610 stmt_vec_info. Use vec_info::lookup_def when setting these
8611 stmt_vec_infos from an SSA_NAME definition.
8612 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
8613 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
8614 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8615 (vectorizable_store, vectorizable_load, vectorizable_condition)
8616 (vectorizable_comparison): Likewise.
8617 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
8618 (vectorizable_reduction): Likewise.
8620 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8622 * tree-vectorizer.h (vect_is_simple_use): Add an optional
8623 stmt_vec_info * parameter before the optional gimple **.
8624 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
8625 (process_use, vect_get_vec_def_for_operand_1): Update callers.
8626 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
8627 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8628 (vectorizable_live_operation): Likewise.
8629 * tree-vect-patterns.c (type_conversion_p): Likewise.
8630 (vect_look_through_possible_promotion): Likewise.
8631 (vect_recog_rotate_pattern): Likewise.
8632 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
8634 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8636 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
8637 a typedef to a wrapper class.
8638 (NULL_STMT_VEC_INFO): New macro.
8639 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
8640 (stmt_vec_info::operator*): New function.
8641 (stmt_vec_info::operator gimple *): Likewise.
8642 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
8643 (add_stmt_costs): Likewise.
8644 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
8645 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
8646 (vect_get_known_peeling_cost): Likewise.
8647 (vect_estimate_min_profitable_iters): Likewise.
8648 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8649 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
8650 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
8651 (vectorizable_store, free_stmt_vec_infos): Likewise.
8652 (new_stmt_vec_info): Change return type of xcalloc to
8655 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8657 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
8658 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
8659 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
8660 a single_imm_use-based sequence.
8661 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
8663 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8665 * tree-vectorizer.h (vec_info::lookup_def): Declare.
8666 * tree-vectorizer.c (vec_info::lookup_def): New function.
8667 * tree-vect-patterns.c (vect_get_internal_def): Use it.
8668 (vect_widened_op_tree): Likewise.
8669 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
8670 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
8671 (vectorizable_reduction): Likewise.
8672 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
8674 (vect_is_slp_reduction): Update calls accordingly. Use
8675 vec_info::lookup_def.
8676 (vect_is_simple_reduction): Likewise
8677 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
8679 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8681 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
8682 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
8683 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
8685 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
8686 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
8687 (vect_update_vf_for_slp, vect_analyze_loop_operations)
8688 (vect_is_slp_reduction, vectorizable_induction)
8689 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
8690 * tree-vect-patterns.c (vect_init_pattern_stmt):
8691 (vect_determine_min_output_precision_1, vect_determine_precisions)
8692 (vect_pattern_recog): Likewise.
8693 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
8694 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
8695 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
8696 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
8697 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
8698 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
8699 info field from a loop to a loop_vec_info.
8701 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8703 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
8704 (vec_info::add_stmt): Declare.
8705 * tree-vectorizer.c (vec_info::add_stmt): New function.
8706 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
8707 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
8708 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
8709 (vectorizable_induction): Likewise.
8710 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
8711 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
8712 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
8713 (vectorizable_load): Likewise.
8714 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8715 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
8716 (vect_recog_gather_scatter_pattern): Likewise.
8717 (append_pattern_def_seq): Likewise. Remove a check that is
8718 performed by add_stmt itself.
8720 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8722 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
8723 which make_ssa_name was called with new_stmt before new_stmt
8726 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8728 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
8730 (vect_is_slp_reduction): ...here...
8731 (vect_is_simple_reduction): ...and here. Remove repetition of tests
8732 that are already known to be false.
8734 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8736 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
8737 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
8738 STMT_VINFO_NUM_SLP_USES when it's true.
8739 (vect_free_slp_instance): Add a final_p parameter and pass it to
8741 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
8742 (vect_analyze_slp_instance): Likewise.
8743 (vect_slp_analyze_operations): Likewise.
8744 (vect_slp_analyze_bb_1): Likewise.
8745 * tree-vectorizer.c (vec_info): Likewise.
8746 * tree-vect-loop.c (vect_transform_loop): Likewise.
8748 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8750 * tree-vect-loop.c (vectorizable_reduction): Assert that the
8751 function is not called for second and subsequent members of
8754 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8756 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
8757 cases for nested loops from here to ...
8758 (vect_create_epilog_for_reduction): ...here. Only call
8759 vect_is_simple_use for inner-loop reductions.
8761 2018-07-31 Martin Liska <mliska@suse.cz>
8763 PR gcov-profile/85338
8764 PR gcov-profile/85350
8765 PR gcov-profile/85372
8766 * profile.c (struct location_triplet): New.
8767 (struct location_triplet_hash): Likewise.
8768 (output_location): Do not output a BB that
8769 is already recorded for a line.
8770 (branch_prob): Use streamed_locations.
8772 2018-07-31 Martin Liska <mliska@suse.cz>
8774 PR gcov-profile/85370
8775 * coverage.c (coverage_begin_function): Do not mark target
8776 clones as artificial functions.
8778 2018-07-31 Martin Liska <mliska@suse.cz>
8780 PR gcov-profile/83813
8781 PR gcov-profile/84758
8782 PR gcov-profile/85217
8783 PR gcov-profile/85332
8784 * profile.c (branch_prob): Do not record GOTO expressions
8785 for GIMPLE statements which locations are already streamed.
8787 2018-07-31 Olivier Hainque <hainque@adacore.com>
8789 * gcc.c (handle_spec_function): Accept a soft_matched_part
8790 argument, as do_spec_1. Pass it down to ...
8791 (eval_spec_function): Accept a soft_matched_part argument,
8792 and pass it down to ...
8793 (do_spec_2): Accept a soft_matched_part argument, and pass
8794 it down to do_spec_1.
8795 (do_spec_1): Pass soft_matched_part to handle_spec_function.
8796 (handle_braces): Update call to handle_spec_function.
8797 (driver::set_up_specs): Update calls to do_spec_2.
8798 (compare_debug_dump_opt_spec_function): Likewise.
8799 (compare_debug_self_opt_spec_function): Likewise.
8801 2018-07-31 Olivier Hainque <hainque@adacore.com>
8803 * common.opt (nolibc): New option.
8804 * doc/invoke.texi (Link Options): Document it.
8805 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
8806 * config/alpha/linux.h: Likewise.
8807 * config/arc/elf.h: Likewise.
8808 * config/arm/uclinux-elf.h: Likewise.
8809 * config/arm/unknown-elf.h: Likewise.
8810 * config/avr/avrlibc.h: Likewise.
8811 * config/bfin/bfin.h: Likewise.
8812 * config/bfin/linux.h: Likewise.
8813 * config/bfin/uclinux.h: Likewise.
8814 * config/darwin.h: Likewise.
8815 * config/darwin10.h: Likewise.
8816 * config/darwin12.h: Likewise.
8817 * config/gnu-user.h: Likewise.
8818 * config/lm32/uclinux-elf.h: Likewise.
8819 * config/pa/pa-hpux11.h: Likewise.
8820 * config/pa/pa64-hpux.h: Likewise.
8821 * config/sparc/sparc.h: Likewise.
8823 2018-07-31 Olivier Hainque <hainque@adacore.com>
8825 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
8826 undefined variables.
8828 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
8831 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
8834 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
8836 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
8837 terminated string literal.
8839 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
8841 PR rtl-optimization/85160
8842 * combine.c (is_just_move): New function.
8843 (try_combine): Allow combining two instructions into two if neither of
8844 the original instructions was a move.
8846 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
8849 * doc/extend.texi (Global Register Variables): Discourage use of type
8851 (Local Register Variables): Likewise.
8853 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
8855 PR tree-optimization/86506
8856 * hwint.h (ceil_log2): Resync with hwint.c implementation.
8858 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
8861 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
8862 hard_regno, make sure no insn between `from` and `to` clobbers it.
8864 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
8865 Tom de Vries <tdevries@suse.de>
8867 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
8868 (PTX_DEFAULT_RUNTIME_DIM): ... this.
8869 (nvptx_goacc_validate_dims): Set default worker and gang dims to
8870 PTX_DEFAULT_RUNTIME_DIM.
8871 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
8873 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
8875 * config/pa/pa.c (pa_output_addr_vec): Align address table.
8876 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
8877 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
8879 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
8881 * config/rs6000/constraints.md (wG constraint): Delete, no longer
8883 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
8884 predicate to reflect toc fusion has been deleted.
8885 (toc_fusion_mem_raw): Delete, no longer used.
8886 (toc_fusion_mem_wrapped): Likewise.
8887 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
8889 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
8890 Delete, no longer used.
8891 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
8892 meant to be used for toc fusion.
8893 (rs6000_debug_print_mode): Delete toc fusion debugging.
8894 (rs6000_debug_reg_global): Likewise.
8895 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
8896 fusion and secondary reload support that were never used.
8897 (rs6000_option_override_internal): Delete TOC fusion, that was only
8898 partially defined, and it did not work unless you also used the
8900 (rs6000_legitimate_address_p): Delete TOC fusion support.
8901 (rs6000_opt_masks): Likewise.
8902 (fusion_wrap_memory_address): Delete function, no longer used.
8903 (fusion_split_address); Delete TOC fusion support.
8904 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
8905 longer used with toc fusion being deleted.
8906 (TARGET_TOC_FUSION_FP): Likewise.
8907 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
8909 (toc fusion spliter): Delete TOC fusion support.
8910 (toc_fusionload_<mode>): Likewise.
8911 (toc_fusionload_di): Likewise.
8912 (fusion_gpr_load_<mode>): Delete generator function, this insn no
8913 longer needs to be named. Rename predicate to delete TOC fusion.
8914 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
8915 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
8916 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
8917 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
8918 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
8920 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
8922 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
8923 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
8924 __int128 in built-in function prototypes.
8925 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
8926 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
8928 2018-07-27 Martin Sebor <msebor@redhat.com>
8930 PR tree-optimization/86696
8931 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
8932 types, including enums.
8933 (handle_char_store): Be prepared for the above function to fail.
8935 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
8937 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
8938 when optimization level is lower than 2 or optimize for size.
8940 2018-07-26 Martin Sebor <msebor@redhat.com>
8942 PR tree-optimization/86043
8943 PR tree-optimization/86042
8944 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
8945 (get_string_cst_length): Rename...
8946 (get_min_string_length): ...to this. Add argument.
8947 (handle_char_store): Extend to handle multi-character stores by
8949 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
8950 * tree.h (initializer_zerop): Add argument.
8952 2018-07-26 Jakub Jelinek <jakub@redhat.com>
8955 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
8956 declare target to variables if they have always,{to,from,tofrom} map
8959 2018-07-26 Martin Liska <mliska@suse.cz>
8962 * lto-wrapper.c: Add linker_output as prefix
8963 for ltrans_output_file.
8965 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
8967 PR rtl-optimization/85805
8968 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
8969 value for hard registers if that was written in the same mode.
8971 2018-07-26 Martin Liska <mliska@suse.cz>
8973 PR gcov-profile/86536
8974 * gcov.c (format_gcov): Use printf format %.*f directly
8975 and do not handle special values.
8977 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
8979 * common/config/arc/arc-common.c (arc_option_optimization_table):
8980 Update default optimizations for size.
8982 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
8984 * config/arc/arc.md (movsf_insn): Add short instruction selection.
8985 * config/arc/constraints.md (CfZ): New constraint.
8986 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
8987 (subsf3_fpu): Likewise.
8988 (cmpsf_fpu): Likewise.
8989 (cmpsf_fpu_uneq): Likewise.
8991 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
8993 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
8995 (arc_is_uncached_mem_p): uncached applies to both the variable and
8998 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9000 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
9003 2018-07-25 David Malcolm <dmalcolm@redhat.com>
9005 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
9006 field "m_scopes" from vec to auto_vec.
9008 2018-07-25 Martin Liska <mliska@suse.cz>
9010 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
9013 2018-07-25 Richard Biener <rguenther@suse.de>
9016 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
9017 special wrt context_die late.
9018 (gen_subprogram_die): Re-use DIEs in local scope.
9020 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
9022 PR tree-optimization/86644
9023 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
9025 2018-07-25 Martin Liska <mliska@suse.cz>
9028 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
9029 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
9031 2018-07-25 Martin Liska <mliska@suse.cz>
9034 * params.def: Explain ASan abbreviation and provide
9035 a documentation link.
9037 2018-07-24 Martin Sebor <msebor@redhat.com>
9039 PR tree-optimization/86622
9040 PR tree-optimization/86532
9041 * builtins.h (string_length): Declare.
9042 * builtins.c (c_strlen): Correct handling of non-constant offsets.
9043 (check_access): Be prepared for non-constant length ranges.
9044 (string_length): Make extern.
9045 * expr.c (string_constant): Only handle the minor non-constant
9046 array index. Use string_constant to compute the length of
9047 a generic string constant.
9049 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
9051 PR tree-optimization/86618
9052 * tree-vect-stmts.c (vectorizable_call): Don't take the address
9053 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
9055 2018-07-24 David Malcolm <dmalcolm@redhat.com>
9057 PR tree-optimization/86636
9058 * json.cc (json::object::set): Fix comment. Add assertions.
9059 (json::array::append): Move here from json.h. Add comment and an
9061 (json::string::string): Likewise.
9062 * json.h (json::array::append): Move to json.cc.
9063 (json::string::string): Likewise.
9064 * optinfo-emit-json.cc
9065 (optrecord_json_writer::impl_location_to_json): Assert that we
9066 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
9067 wrapper around it. Expand the location once, rather than three
9069 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
9070 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
9072 (optrecord_json_writer::optinfo_to_json): Likewise, in four
9073 places. Fix some overlong lines.
9075 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
9077 * config/aarch64/aarch64-simd.md
9078 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
9079 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
9080 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
9081 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
9082 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
9083 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
9084 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
9085 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
9086 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
9088 2018-07-24 Jakub Jelinek <jakub@redhat.com>
9091 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
9092 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
9093 and abs_d == d, do the power of two handling if profitable.
9095 2018-07-24 Richard Biener <rguenther@suse.de>
9097 * match.pd: Add BIT_FIELD_REF canonicalizations.
9099 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9102 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
9104 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9106 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
9107 terminated STRING_CST object.
9109 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9111 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
9113 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
9115 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
9116 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
9117 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
9118 the elements into a register.
9119 (rs6000_split_v4si_init_di_reg): Delete.
9120 (rs6000_split_v4si_init): Delete.
9121 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
9122 (vsx_init_v4si): Rewrite as a define_expand.
9124 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
9126 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
9127 zero_extend argument from memory): New.
9129 2018-07-22 Martin Sebor <msebor@redhat.com>
9132 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
9133 diagnosing calls with unknown arguments unless -Walloca-larger-than
9134 is restricted to less than PTRDIFF_MAX bytes.
9136 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
9138 * doc/gcov.texi (Invoking Gcov): Editorial changes.
9140 2018-07-20 David Malcolm <dmalcolm@redhat.com>
9142 * pretty-print.c (text_info::set_location): Remove redundant
9143 "line_table" parameter from call to rich_location::set_range.
9145 2018-07-20 Martin Sebor <msebor@redhat.com>
9148 * builtins.c (expand_builtin_alloca): Adjust.
9149 * calls.c (alloc_max_size): Simplify.
9150 * cgraphunit.c (cgraph_node::expand): Adjust.
9151 * common.opt (larger_than_size, warn_frame_larger_than): Remove
9153 (frame_larger_than_size): Same.
9154 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
9155 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
9157 * doc/invoke.texi (GCC Command Options): Document option arguments.
9158 Explain byte-size arguments and suffixes.
9159 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
9160 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
9161 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
9162 * doc/options.texi (UInteger): Expand.
9163 (Host_Wide_Int, ByteSize): Document new properties.
9164 * final.c (final_start_function_1): Include sizes in an error message.
9165 * function.c (frame_offset_overflow): Same.
9166 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
9167 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
9168 Diagnose unbounded alloca calls only for limits of less than
9170 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
9171 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
9173 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
9174 only for limits of less than PTRDIFF_MAX.
9175 * langhooks-def.h (lhd_handle_option): Change function argument
9177 * langhooks.c (lhd_handle_option): Same.
9178 * langhooks.h (handle_option): Same.
9179 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
9181 (var_type, var_type_struct): Same.
9182 (var_set): Handle ByteSize flag.
9183 * optc-gen.awk: Add comments to output to ease debugging. Make
9184 use of HOST_WIDE_INT where appropriate.
9185 * opts-gen-save.awk: Use %lx to format unsigned long.
9186 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
9187 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
9188 arguments. Parse bytes-size suffixes.
9189 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
9190 (enum_value_to_arg): Same.
9191 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
9192 (handle_option): Adjust.
9193 (generate_option): Change function argument to HOST_WIDE_INT.
9194 (cmdline_handle_error): Adjust.
9195 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
9196 (set_option): Change function argument to HOST_WIDE_INT.
9197 (option_enabled): Handle cl_host_wide_int.
9198 (get_option_state): Handle CLVC_SIZE.
9199 (control_warning_option): Same.
9200 * opts.c (common_handle_option): Change function argument to
9201 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
9202 OPT_Wvla_larger_than_.
9203 * opts.h (enum cl_var_type): Add an enumerator.
9204 * stor-layout.c (layout_decl): Print a more meaningful warning.
9205 * toplev.c (output_stack_usage): Adjust.
9207 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
9209 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
9210 call to inline_expand_builtin_string_cmp.
9211 (expand_builtin_strcmp): Likewise.
9212 (expand_builtin_strncmp): Likewise.
9213 (inline_string_cmp): Delete the last parameter, change char_type_node
9214 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
9216 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
9217 the inlining expansion on target where the type of the call has same or
9218 narrower precision than unsigned char.
9220 2018-07-20 David Malcolm <dmalcolm@redhat.com>
9222 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
9223 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
9224 * common.opt (fsave-optimization-record): New option.
9225 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
9226 * doc/invoke.texi (-fsave-optimization-record): New option.
9227 * dumpfile.c: Include "optinfo-emit-json.h".
9228 (struct kv_pair): Move to coretypes.h.
9229 (optgroup_options): Make non-static.
9230 (dump_context::end_scope): Call
9231 optimization_records_maybe_pop_dump_scope.
9232 * dumpfile.h (optgroup_options): New decl.
9233 * json.cc: New file.
9235 * optinfo-emit-json.cc: New file.
9236 * optinfo-emit-json.h: New file.
9237 * optinfo.cc: Include "optinfo-emit-json.h".
9238 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
9239 (optinfo_enabled_p): Check optimization_records_enabled_p.
9240 (optinfo_wants_inlining_info_p): Likewise.
9241 * optinfo.h: Update comment.
9242 * profile-count.c (profile_quality_as_string): New function.
9243 * profile-count.h (profile_quality_as_string): New decl.
9244 (profile_count::quality): New accessor.
9245 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
9246 and optinfo_emit_json_cc_tests.
9247 * selftest.h (selftest::json_cc_tests): New decl.
9248 (selftest::optinfo_emit_json_cc_tests): New decl.
9249 * toplev.c: Include "optinfo-emit-json.h".
9250 (compile_file): Call optimization_records_finish.
9251 (do_compile): Call optimization_records_start.
9252 * tree-ssa-live.c: Include optinfo.h.
9253 (remove_unused_scope_block_p): Retain inlining information if
9254 optinfo_wants_inlining_info_p returns true.
9256 2018-07-20 Richard Biener <rguenther@suse.de>
9259 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
9260 to cover -flto-partition=none.
9262 2018-07-20 Martin Liska <mliska@suse.cz>
9264 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
9265 (get_decl_source_range): Remove unused function.
9267 2018-07-20 Richard Biener <rguenther@suse.de>
9269 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
9270 (struct vn_phi_s): Likewise.
9271 (struct vn_reference_s): Likewise.
9272 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
9273 for searching the slot of an entry known to be in the hash itself.
9274 (vn_phi_hasher::equal): Likewise.
9275 (vn_reference_hasher::equal): Likewise.
9276 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
9278 (optimistic_info, current_info): Remove, keeping only valid_info.
9279 (vn_reference_lookup_1): Remove fallback lookup.
9280 (vn_reference_lookup_2): Likewise.
9281 (vn_nary_op_lookup_1): Likewise.
9282 (vn_phi_lookup): Likewise.
9283 (vn_nary_build_or_lookup_1): Make sure to not chain the built
9285 (vn_reference_insert): Adjust, chain the inserted hash element
9286 at last_inserted_ref.
9287 (vn_reference_insert_pieces): Likewise.
9288 (visit_reference_op_call): Likewise.
9289 (vn_nary_op_insert_into): Chain the inserted hash element at
9291 (vn_nary_op_insert_pieces): Adjust.
9292 (vn_nary_op_insert): Likewise.
9293 (vn_nary_op_insert_stmt): Likewise.
9294 (vn_phi_insert): Adjust, chain the inserted hash element at
9296 (process_scc): Remove clearing and copying the optimistic
9297 table. Instead remove elements inserted during an optimistic
9298 iteration from the single table we maintain.
9299 (init_scc_vn): Adjust.
9300 (free_scc_vn): Likewise.
9301 (sccvn_dom_walker::record_cond): Likewise.
9302 (sccvn_dom_walker::after_dom_children): Likewise.
9304 2018-07-19 Martin Sebor <msebor@redhat.com>
9306 PR tree-optimization/84047
9307 PR tree-optimization/83776
9308 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
9309 (check_array_bounds): Call it.
9311 2018-07-19 Martin Sebor <msebor@redhat.com>
9313 * align.h (align_flags): Use member initialization.
9315 2018-07-19 David Malcolm <dmalcolm@redhat.com>
9317 * Makefile.in (OBJS): Add optinfo.o.
9318 * coretypes.h (class symtab_node): New forward decl.
9319 (struct cgraph_node): New forward decl.
9320 (class varpool_node): New forward decl.
9321 * dump-context.h: New file.
9322 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
9324 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
9325 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
9326 (set_alt_dump_file): Likewise.
9327 (dump_context::~dump_context): New dtor.
9328 (dump_gimple_stmt): Move implementation to...
9329 (dump_context::dump_gimple_stmt): ...this new member function.
9330 Add the stmt to any pending optinfo, creating one if need be.
9331 (dump_gimple_stmt_loc): Move implementation to...
9332 (dump_context::dump_gimple_stmt_loc): ...this new member function.
9333 Start a new optinfo and add the stmt to it.
9334 (dump_gimple_expr): Move implementation to...
9335 (dump_context::dump_gimple_expr): ...this new member function.
9336 Add the stmt to any pending optinfo, creating one if need be.
9337 (dump_gimple_expr_loc): Move implementation to...
9338 (dump_context::dump_gimple_expr_loc): ...this new member function.
9339 Start a new optinfo and add the stmt to it.
9340 (dump_generic_expr): Move implementation to...
9341 (dump_context::dump_generic_expr): ...this new member function.
9342 Add the tree to any pending optinfo, creating one if need be.
9343 (dump_generic_expr_loc): Move implementation to...
9344 (dump_context::dump_generic_expr_loc): ...this new member
9345 function. Add the tree to any pending optinfo, creating one if
9347 (dump_printf): Move implementation to...
9348 (dump_context::dump_printf_va): ...this new member function. Add
9349 the text to any pending optinfo, creating one if need be.
9350 (dump_printf_loc): Move implementation to...
9351 (dump_context::dump_printf_loc_va): ...this new member function.
9352 Start a new optinfo and add the stmt to it.
9353 (dump_dec): Move implementation to...
9354 (dump_context::dump_dec): ...this new member function. Add the
9355 value to any pending optinfo, creating one if need be.
9356 (dump_context::dump_symtab_node): New member function.
9357 (dump_context::get_scope_depth): New member function.
9358 (dump_context::begin_scope): New member function.
9359 (dump_context::end_scope): New member function.
9360 (dump_context::ensure_pending_optinfo): New member function.
9361 (dump_context::begin_next_optinfo): New member function.
9362 (dump_context::end_any_optinfo): New member function.
9363 (dump_context::s_current): New global.
9364 (dump_context::s_default): New global.
9365 (dump_scope_depth): Delete global.
9366 (dumpfile_ensure_any_optinfo_are_flushed): New function.
9367 (dump_symtab_node): New function.
9368 (get_dump_scope_depth): Reimplement in terms of dump_context.
9369 (dump_begin_scope): Likewise.
9370 (dump_end_scope): Likewise.
9371 (selftest::temp_dump_context::temp_dump_context): New ctor.
9372 (selftest::temp_dump_context::~temp_dump_context): New dtor.
9373 (selftest::verify_item): New function.
9374 (ASSERT_IS_TEXT): New macro.
9375 (ASSERT_IS_TREE): New macro.
9376 (ASSERT_IS_GIMPLE): New macro.
9377 (selftest::test_capture_of_dump_calls): New test.
9378 (selftest::dumpfile_c_tests): Call it.
9379 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
9380 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
9381 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
9382 descriptive comment.
9383 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
9384 (dump_node, dump_bb): Move these unrelated decls.
9385 (class dump_manager): Add leading comment.
9386 * optinfo.cc: New file.
9387 * optinfo.h: New file.
9389 2018-07-19 Michael Collison <michael.collison@arm.com>
9390 Richard Henderson <rth@redhat.com>
9392 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
9393 (subti): Handle op1 zero.
9394 (subvti4, usub4ti4): New.
9395 (*sub<GPI>3_compare1_imm): New.
9396 (sub<GPI>3_carryinCV): New.
9397 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
9398 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
9400 2018-07-19 Michael Collison <michael.collison@arm.com>
9401 Richard Henderson <rth@redhat.com>
9403 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
9404 (addti3): Create simpler code if low part is already known to be 0.
9405 (addvti4, uaddvti4): New.
9406 (*add<GPI>3_compareC_cconly_imm): New.
9407 (*add<GPI>3_compareC_cconly): New.
9408 (*add<GPI>3_compareC_imm): New.
9409 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
9410 handle constants within this pattern..
9411 (*add<GPI>3_compareV_cconly_imm): New.
9412 (*add<GPI>3_compareV_cconly): New.
9413 (*add<GPI>3_compareV_imm): New.
9414 (add<GPI>3_compareV): New.
9415 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
9416 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
9417 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
9418 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
9419 with 'comparison' operator.
9420 (*add<GPI>3_compareV_cconly_imm): Ditto.
9421 (*add<GPI>3_compareV_cconly): Ditto.
9422 (*add<GPI>3_compareV_imm): Ditto.
9423 (add<GPI>3_compareV): Ditto.
9424 (add<mode>3_carryinC): Ditto.
9425 (*add<mode>3_carryinC_zero): Ditto.
9426 (*add<mode>3_carryinC): Ditto.
9427 (add<mode>3_carryinV): Ditto.
9428 (*add<mode>3_carryinV_zero): Ditto.
9429 (*add<mode>3_carryinV): Ditto.
9431 2018-07-19 Michael Collison <michael.collison@arm.com>
9432 Richard Henderson <rth@redhat.com>
9434 * config/aarch64/aarch64-modes.def (CC_V): New.
9435 * config/aarch64/aarch64-protos.h
9436 (aarch64_addti_scratch_regs): Declare
9437 (aarch64_subvti_scratch_regs): Declare.
9438 (aarch64_expand_subvti): Declare.
9439 (aarch64_gen_unlikely_cbranch): Declare
9440 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
9441 for signed overflow using CC_Vmode.
9442 (aarch64_get_condition_code_1): Handle CC_Vmode.
9443 (aarch64_gen_unlikely_cbranch): New function.
9444 (aarch64_addti_scratch_regs): New function.
9445 (aarch64_subvti_scratch_regs): New function.
9446 (aarch64_expand_subvti): New function.
9448 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9450 * config/aarch64/aarch64-option-extensions.def: New entry for profile
9452 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
9453 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
9456 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9459 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
9460 address check not strict.
9462 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9464 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
9466 (store_pair_lanes<mode>): Likewise.
9467 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
9468 enum value 'ADDR_QUERY_LDP_STP_N'.
9469 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
9470 (aarch64_print_address_internal): Add declaration.
9471 (aarch64_print_ldpstp_address): Remove.
9472 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
9473 (aarch64_print_operand): Change printing of 'y'.
9474 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
9475 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
9476 'true' rather than '1'.
9477 * config/aarch64/constraints.md (Uml): Likewise.
9478 (Uml): Rename to Umn.
9481 2018-07-19 Richard Biener <rguenther@suse.de>
9483 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
9485 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
9486 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
9487 (vn_reference_hasher): Likewise.
9488 (struct vn_tables_s): Remove obstack and alloc-pool members.
9489 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
9490 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
9491 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
9492 (vn_reference_insert_pieces): Likewise.
9493 (alloc_vn_nary_op_noinit): Adjust.
9494 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
9495 (vn_phi_eq): Adjust.
9496 (shared_lookup_phiargs): Remove.
9497 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
9498 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
9499 (visit_reference_op_call): Likewise.
9500 (copy_nary, copy_phi, copy_reference): Remove.
9501 (process_scc): Rewind the obstack when iterating. Do not
9502 copy the elements to valid_info but just move them from one
9503 hashtable to the other.
9504 (allocate_vn_table): Adjust.
9505 (free_vn_table): Likewise.
9506 (init_scc_vn): Likewise.
9507 (free_scc_vn): Likewise.
9509 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
9512 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
9513 indirect_return as function type attribute.
9514 (ix86_attribute_table): Change indirect_return to function
9516 * doc/extend.texi: Update indirect_return attribute.
9518 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
9520 * wide-int.h (widest2_int): New.
9521 * gimple-fold.c (arith_overflowed_p): Use it.
9522 * tree.h (widest2_int_cst): New.
9523 * tree-vrp.c (wide_int_binop_overflow): Rename from
9524 vrp_int_const_binop.
9525 Rewrite to work on trees.
9526 (extract_range_from_multiplicative_op_1): Abstract code to...
9527 (wide_int_range_min_max): ...here.
9528 (wide_int_range_cross_product): ...and here.
9529 (extract_range_from_binary_expr_1): Abstract overflow code to...
9530 (wide_int_range_mult_wrapping): ...here.
9531 * tree-vrp.h (wide_int_range_cross_product): New.
9532 (wide_int_range_mult_wrapping): New.
9534 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
9535 Julia Koval <julia.koval@intel.com>
9537 * config/i386/x86-tune-costs.h (skylake_memcpy,
9538 skylake_memset): Replace rep_prefix with unrolling for size 512.
9540 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
9543 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
9544 comparision with EQ_EXPR in last stmt.
9546 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
9548 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
9549 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
9550 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
9551 previously known as "PowerPC AltiVec Built-in Functions". Move
9552 some material to new subsubsections "PowerPC AltiVec Built-in
9553 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
9555 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
9556 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
9557 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
9558 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
9560 2018-07-18 Richard Biener <rguenther@suse.de>
9562 PR tree-optimization/86557
9563 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
9566 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
9568 * config/s390/s390.c (s390_function_profiler): Generate CFI.
9570 2018-07-17 Jeff Law <law@redhat.com>
9572 * config/arm/arm.c (get_label_padding): Update for recent
9573 changes to label_to_alignment.
9575 PR tree-optimization/86010
9576 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
9578 * config/mips/mips.c (vr4130_align_insns): Update for recent
9579 changes to label_to_alignment.
9581 * config/frv/frv.c (frv_label_align): Update for recent changes
9582 to label_to_alignment.
9584 * config/nios2/nios2.c (nios2_label_align): Update for recent
9585 changes which dropped ALIGN_LABELS_LOG.
9587 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
9589 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
9590 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
9592 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9594 * config/arc/arc.c (arc_label_align): Use align_labels instead of
9595 deprecated align_labels_log.
9597 2018-07-17 Richard Biener <rguenther@suse.de>
9600 * dwarf2out.c (init_sections_and_labels): Always generate
9601 a debug_line_str_section for early LTO debug.
9602 (dwarf2out_finish): Reset debug_line_str_hash output early.
9603 Bump counter for extra dwarf5 .debug_loc labels to not conflict
9605 (dwarf2out_early_finish): Output debug_line_str.
9607 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
9609 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
9610 index register on z196 or later.
9612 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
9614 * config/s390/s390.c (s390_default_align): Set default function
9616 (s390_override_options_after_change): Call s390_default align.
9617 (s390_option_override_internal): Call s390_default_align.
9618 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
9620 2018-07-17 Jakub Jelinek <jakub@redhat.com>
9623 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
9624 to _looptemp_ clauses, other than the first two.
9626 2018-07-17 Martin Liska <mliska@suse.cz>
9628 * opts.c: Do not enable OPT_falign_* for -Os.
9630 2018-07-17 Martin Liska <mliska@suse.cz>
9632 * align.h (MAX_CODE_ALIGN): New.
9633 (MAX_CODE_ALIGN_VALUE): New.
9634 * common/config/i386/i386-common.c (ix86_handle_option):
9635 (MAX_CODE_ALIGN): Moved to align.h.
9636 * final.c (MAX_CODE_ALIGN): Likewise.
9637 * opts.c (parse_and_check_align_values):
9638 (MAX_CODE_ALIGN): Likewise.
9639 (MAX_CODE_ALIGN_VALUE): Likewise.
9641 2018-07-17 Martin Liska <mliska@suse.cz>
9643 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
9644 in order to fulfil coding style.
9645 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
9646 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9647 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
9648 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
9649 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
9650 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
9651 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
9652 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
9653 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9655 2018-07-17 Martin Liska <mliska@suse.cz>
9657 * align.h: New file.
9658 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
9660 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
9661 align_flags of label_to_alignment.
9662 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
9664 * config/m68k/m68k.c: Do not use removed align_labels_value and
9666 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
9667 (LOOP_ALIGN): Likewise.
9668 (LABEL_ALIGN): Likewise.
9669 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
9670 Remove not used macro.
9671 (rs6000_loop_align): Change return type to align_flags.
9672 (rs6000_loop_align_max_skip): Remove.
9673 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
9674 Change return type to align_flags.
9675 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
9676 Remove not used macro.
9677 (rs6000_loop_align): Change return type to align_flags.
9678 (rs6000_loop_align_max_skip): Remove.
9679 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
9680 * config/rx/rx-protos.h (rx_align_for_label): Make it
9682 * config/rx/rx.c (rx_align_for_label): Change return type
9684 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
9686 into align_flags class.
9687 (LABEL_ALIGN): Likewise.
9688 (LOOP_ALIGN): Likewise.
9689 * config/s390/s390.c (s390_label_align): Use align_flags
9691 (s390_asm_output_function_label): Likewise.
9692 * config/sh/sh.c (sh_override_options_after_change):
9693 Use align_flags class directly without macros.
9694 (find_barrier): Likewise.
9695 (barrier_align): Likewise.
9696 (sh_loop_align): Likewise.
9697 * config/spu/spu.c (spu_option_override):
9698 Use align_flags_tuple::get_value instead of removed macros.
9699 (spu_sched_init): Likewise.
9700 * config/spu/spu.h (GTY): Likewise.
9701 * config/visium/visium.c (visium_option_override):
9702 Set "8" as default secondary alignment.
9703 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
9704 in order to guarantee secondary alignment of 8.
9705 * coretypes.h: Include align.h header file.
9706 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
9707 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
9708 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
9709 * doc/tm.texi.in: Likewise.
9710 * final.c (struct label_alignment): Remove not used structure.
9711 (LABEL_ALIGN): Change type to align_flags.
9712 (LOOP_ALIGN): Likewise.
9713 (JUMP_ALIGN): Likewise.
9714 (default_loop_align_max_skip): Remove.
9715 (default_label_align_max_skip): Likewise.
9716 (default_jump_align_max_skip): Likewise.
9717 (default_label_align_after_barrier_max_skip):
9718 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
9719 (LABEL_TO_MAX_SKIP): Remove.
9720 (label_to_alignment): Return align_flags type instead of integer.
9721 (label_to_max_skip): Remove.
9722 (align_fuzz): Use align_flags type.
9723 (compute_alignments): Use align_flags type and use align_flags::max
9724 to combine multiple alignments.
9725 (grow_label_align): Grow vec instead of C array.
9726 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
9727 (shorten_branches): Use align_flags type and use align_flags::max
9728 to combine multiple alignments.
9729 (final_scan_insn_1): Remove usage of secondary alignment that comes
9730 from label alignment, but instead use proper secondary alignment
9731 which is computed in grow_label_align.
9732 * flags.h (struct align_flags_tuple): Move to align.h.
9733 (struct align_flags): Likewise.
9734 (state_align_loops): Rename to align_loops.
9735 (state_align_jumps): Rename to align_jumps.
9736 (state_align_labels): Rename to align_labels.
9737 (state_align_functions): Rename to align_functions.
9738 (align_loops_log): Remove.
9739 (align_jumps_log): Remove.
9740 (align_labels_log): Remove.
9741 (align_functions_log): Remove.
9742 (align_loops_max_skip): Remove.
9743 (align_jumps_max_skip): Remove.
9744 (align_labels_max_skip): Remove.
9745 (align_functions_max_skip): Remove.
9746 (align_loops_value): Remove.
9747 (align_jumps_value): Remove.
9748 (align_labels_value): Remove.
9749 (align_functions_value): Remove.
9750 * output.h (label_to_alignment): Change return type to align_flags.
9751 (label_to_max_skip): Remove.
9752 * target.def: Remove loop_align_max_skip, label_align_max_skip,
9753 jump_align_max_skip macros.
9754 * targhooks.h (default_loop_align_max_skip): Remove.
9755 (default_label_align_max_skip): Likewise.
9756 (default_jump_align_max_skip): Likewise.
9757 (default_label_align_after_barrier_max_skip): Remove.
9758 * toplev.c (read_log_maxskip): Use ::normalize function.
9759 (parse_N_M): Remove not used argument and also call ::normalize.
9760 (parse_alignment_opts): Do not pass unused arguments.
9761 * varasm.c (assemble_start_function): Use directly align_functions
9762 instead of removed macros.
9763 * system.h: Do not poison removed macros.
9765 2018-07-17 Jakub Jelinek <jakub@redhat.com>
9768 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
9769 and cond temporaries don't have reference type if iterator has
9770 pointer type. For init use &for_pre_body instead of pre_p if
9771 for_pre_body is non-empty.
9773 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
9775 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
9776 double-double modes to SFmode directly directly.
9777 (trunc<mode>sf2_fprs): Delete.
9779 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
9781 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
9782 for conversions between IFmode and the decimal floating point modes.
9783 (init_float128_ieee): Use the correct names for conversions between
9784 KFmode and the decimal floating point modes.
9786 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
9788 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
9789 for the conversions between TDmode and IFmode.
9790 (init_float128_ieee): Use more correct names for the conversions
9791 between TDmode and KFmode.
9793 2018-07-16 Jakub Jelinek <jakub@redhat.com>
9795 PR tree-optimization/86526
9796 * builtins.c (expand_builtin_memcmp): Formatting fixes.
9797 (inline_expand_builtin_string_cmp): Likewise.
9798 (inline_string_cmp): Likewise. Use c_readstr instead of
9799 builtin_memcpy_read_str. Add unit_mode temporary.
9801 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
9804 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
9805 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
9807 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
9809 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
9810 Alphabetize prototypes of built-in functions, separating out
9811 built-in functions that are listed in this section but should be
9812 described elsewhere.
9814 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
9817 * expmed.c (emit_store_flag): Do not emit setcc followed by a
9818 conditional move when trapping comparison was split to a
9819 non-trapping one (and vice versa).
9821 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
9823 * config/s390/s390.c (s390_function_profiler): Generate nops
9824 instead of profiler call sequences.
9825 * config/s390/s390.opt: Add the new option.
9827 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
9829 * config/s390/s390.c (s390_function_profiler): Generate
9830 __mcount_loc section.
9831 * config/s390/s390.opt: Add the new option.
9833 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
9835 * common.opt: Add the new warning.
9836 * config/s390/s390.c (s390_function_profiler): Emit "brasl
9837 %r0,__fentry__" when -mfentry is specified.
9838 (s390_option_override_internal): Disallow -mfentry for 31-bit
9840 * config/s390/s390.opt: Add the new option.
9842 2018-07-16 Richard Biener <rguenther@suse.de>
9845 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
9846 for function-local FUNCTION_DECL and RESULT_DECL immediately.
9848 2018-07-16 Martin Liska <mliska@suse.cz>
9851 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
9854 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
9856 * config/arc/arcHS.md: Update ARCHS scheduling rules.
9858 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
9860 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
9862 * config/arc/arc-cpus.def (hs4x): New cpu.
9864 * config/arc/arc-tables.opt: Regenerate.
9865 * config/arc/arc.c (arc_sched_issue_rate): New function.
9866 (TARGET_SCHED_ISSUE_RATE): Define.
9867 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
9868 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
9870 (attr tune): Add ARCHS4x tune values.
9871 (attr tune_dspmpy): Define.
9872 (*tst): Correct instruction type.
9873 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
9874 * config/arc/arcHS4x.md: New file.
9875 * config/arc/fpu.md: Update instruction type attributes.
9876 * config/arc/t-multilib: Regenerate.
9878 2018-07-16 Tom de Vries <tdevries@suse.de>
9881 * var-tracking.c (vt_initialize): Fix pre_dec handling.
9883 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9885 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
9888 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
9890 PR tree-optimization/86514
9891 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
9892 conversion to a boolean type from a type with greater precision.
9894 2018-07-16 Tom de Vries <tdevries@suse.de>
9896 * var-tracking.c (vt_initialize): Print adjusted insn slim if
9897 dump_flags request TDF_SLIM.
9899 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
9901 * fold-const.c (int_const_binop_1): Abstract...
9902 (wide_int_binop): ...wide int code here.
9903 (poly_int_binop): ...poly int code here.
9904 Abstract the rest of int_const_binop_1 into int_const_binop.
9905 * fold-const.h (wide_int_binop): New.
9906 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
9907 Remove useless PLUS/MINUS_EXPR case.
9908 (zero_nonzero_bits_from_vr): Move wide int code...
9909 (zero_nonzero_bits_from_bounds): ...here.
9910 (extract_range_from_binary_expr_1): Move mask optimization code...
9911 (range_easy_mask_min_max): ...here.
9912 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
9913 (range_easy_mask_min_max): New.
9915 2018-07-15 Jeff Law <law@redhat.com>
9918 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
9921 2018-07-14 Jim Wilson <jimw@sifive.com>
9923 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
9925 2018-07-14 Paul Koning <ni1d@arrl.net>
9927 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
9929 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
9931 * lto-streamer-out.c (copy_function_or_variable): Dump info about
9934 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
9935 Steve Munroe <munroesj52@gmail.com>
9937 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
9938 (_mm_andnot_si128): Likewise.
9939 (_mm_or_si128): Likewise.
9940 (_mm_xor_si128): Likewise.
9942 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
9945 * builtins.c (expand_builtin_memcmp): Inline the calls first
9946 when result_eq is false.
9947 (expand_builtin_strcmp): Inline the calls first.
9948 (expand_builtin_strncmp): Likewise.
9949 (inline_string_cmp): New routine. Expand a string compare
9950 call by using a sequence of char comparison.
9951 (inline_expand_builtin_string_cmp): New routine. Inline expansion
9952 a call to str(n)cmp/memcmp.
9953 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
9955 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
9957 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
9959 * config/arm/driver-arm.c: Include arm-native.h.
9960 (host_detect_local_cpu): Use auto-generated data tables.
9961 (vendors, arm_cpu_table): Delete. Move part information to ...
9962 * config/arm/arm-cpus.in: ... here.
9963 * config/arm/parsecpu.awk (gen_native): New function.
9964 (vendor, part): New CPU fields.
9965 (END): Add support for building the native CPU detection tables.
9966 * config/arm/t-arm (arm-native.h): Add build rule.
9967 (driver-arm.o): Add dependency on arm-native.h.
9969 2018-07-13 Richard Biener <rguenther@suse.de>
9972 * match.pd (addr1 - addr2): Allow either of the operand to
9975 2018-07-13 Tom de Vries <tdevries@suse.de>
9977 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
9980 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
9982 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
9983 arrays instead of numbered variables.
9985 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
9987 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
9988 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
9990 2018-07-13 Richard Biener <rguenther@suse.de>
9993 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
9994 instead of get_context_die.
9996 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9997 Richard Biener <rguenther@suse.de>
10000 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
10001 that the loop latch destination where phi is defined.
10003 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
10005 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
10006 (riscv_expand_epilogue): Add assertion to check interrupt mode.
10007 (riscv_set_current_function): Extract getting interrupt type to new
10009 (riscv_get_interrupt_type): New function.
10010 (riscv_merge_decl_attributes): New function, checking interrupt type is
10012 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
10014 2018-07-12 Paul Koning <ni1d@arrl.net>
10016 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
10019 2018-07-12 Paul Koning <ni1d@arrl.net>
10021 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
10022 zero reference, add doloop_end instead.
10023 * doc/md.texi (decrement_and_branch_until_zero): Remove.
10024 (Looping patterns): Remove decrement_and_branch_until_zero. Add
10025 detail for doloop_end.
10027 2018-07-12 Martin Sebor <msebor@redhat.com>
10030 * attribs.c (decl_attributes): Reject conflicting attributes before
10031 calling attribute handlers.
10033 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
10035 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
10037 (gcc::dump_manager::get_dump_file_name): likewise.
10038 (dump_begin): Likewise.
10039 * dumpfile.h (dump_begin): Update prototype.
10040 (gcc::dump_manager::get_dump_file_name,
10041 gcc::dump_manager::get_dump_file_name): Update prototype.
10043 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10045 * internal-fn.h (vectorizable_internal_fn_p): New function.
10046 * tree-vect-slp.c (compatible_calls_p): Likewise.
10047 (vect_build_slp_tree_1): Remove nops argument. Handle calls
10048 to internal functions.
10049 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
10051 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10053 * fold-const.h (inverse_conditions_p): Declare.
10054 * fold-const.c (inverse_conditions_p): New function.
10055 * match.pd: Use inverse_conditions_p. Add folds of view_converts
10056 that test the inverse condition of a conditional internal function.
10057 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
10058 * internal-fn.c (internal_fn_mask_index): Handle conditional
10059 internal functions.
10060 (vectorized_internal_fn_supported_p): New function.
10061 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
10062 (any_pred_load_store): Replace with...
10063 (need_to_predicate): ...this new variable.
10064 (redundant_ssa_names): New variable.
10065 (ifcvt_can_use_mask_load_store): Move initial checks to...
10066 (ifcvt_can_predicate): ...this new function. Handle tree codes
10067 for which a conditional internal function exists.
10068 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
10069 instead of ifcvt_can_use_mask_load_store. Update after variable
10071 (predicate_load_or_store): New function, split out from
10072 predicate_mem_writes.
10073 (check_redundant_cond_expr): New function.
10074 (value_available_p): Likewise.
10075 (predicate_rhs_code): Likewise.
10076 (predicate_mem_writes): Rename to...
10077 (predicate_statements): ...this. Use predicate_load_or_store
10078 and predicate_rhs_code.
10079 (combine_blocks, tree_if_conversion): Update after above name changes.
10080 (ifcvt_local_dce): Handle redundant_ssa_names.
10081 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
10082 general conditional functions.
10083 * tree-vect-stmts.c (vectorizable_call): Likewise.
10085 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10086 Alan Hayward <alan.hayward@arm.com>
10087 David Sherwood <david.sherwood@arm.com>
10089 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
10090 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
10091 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
10092 plus and minus and convert them into IFN_COND_FMA-based sequences.
10093 (convert_mult_to_fma): Handle conditional plus and minus.
10095 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10097 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
10098 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
10099 (cond_fnms_optab): New optabs.
10100 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
10101 internal functions.
10102 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
10103 * internal-fn.h (get_conditional_internal_fn): Declare.
10104 (get_unconditional_internal_fn): Likewise.
10105 * internal-fn.c (cond_ternary_direct): New macro.
10106 (expand_cond_ternary_optab_fn): Likewise.
10107 (direct_cond_ternary_optab_supported_p): Likewise.
10108 (FOR_EACH_COND_FN_PAIR): Likewise.
10109 (get_conditional_internal_fn): New function.
10110 (get_unconditional_internal_fn): Likewise.
10111 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
10112 (gimple_match_op::gimple_match_op): Add a new overload for 5
10114 (gimple_match_op::set_op): Likewise.
10115 (gimple_resimplify5): Declare.
10116 * genmatch.c (decision_tree::gen): Generate simplifications for
10118 * gimple-match-head.c (gimple_simplify): Define an overload for
10119 5 operands. Handle calls with 5 arguments in the top-level overload.
10120 (convert_conditional_op): Handle conversions from unconditional
10121 internal functions to conditional ones.
10122 (gimple_resimplify5): New function.
10123 (build_call_internal): Pass a fifth operand.
10124 (maybe_push_res_to_seq): Likewise.
10125 (try_conditional_simplification): Try converting conditional
10126 internal functions to unconditional internal functions.
10127 Handle 3-operand unconditional forms.
10128 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
10129 Define ternary equivalents of the current rules for binary conditional
10130 internal functions.
10131 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
10132 ternary operations.
10133 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
10134 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
10135 (optab): Handle them.
10136 (SVE_COND_FP_TERNARY): New int iterator.
10137 (sve_fmla_op, sve_fmad_op): New int attributes.
10138 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
10139 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
10140 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
10142 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10144 * target.def (preferred_else_value): New target hook.
10145 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
10146 * doc/tm.texi: Regenerate.
10147 * targhooks.h (default_preferred_else_value): Declare.
10148 * targhooks.c (default_preferred_else_value): New function.
10149 * internal-fn.h (conditional_internal_fn_code): Declare.
10150 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
10151 (get_conditional_internal_fn): Use it.
10152 (conditional_internal_fn_code): New function.
10153 * gimple-match.h (gimple_match_cond): New struct.
10154 (gimple_match_op): Add a cond member function.
10155 (gimple_match_op::gimple_match_op): Update all forms to take a
10157 * genmatch.c (expr::gen_transform): Use the same condition as res_op
10158 for the suboperation, but don't specify a particular else_value.
10159 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
10160 (visit_nary_op, visit_reference_op_load): Pass
10161 gimple_match_cond::UNCOND to the gimple_match_op constructor.
10162 * gimple-match-head.c: Include tree-eh.h
10163 (convert_conditional_op): New function.
10164 (maybe_resimplify_conditional_op): Likewise.
10165 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
10166 (gimple_resimplify2): Likewise.
10167 (gimple_resimplify3): Likewise.
10168 (gimple_resimplify4): Likewise.
10169 (maybe_push_res_to_seq): Return null for conditional operations.
10170 (try_conditional_simplification): New function.
10171 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
10173 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
10175 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
10177 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
10179 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
10181 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
10183 (hash_tree): Do not hash DECL_FCONTEXT
10184 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
10185 Do not stream DECL_FCONTEXT.
10186 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
10187 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
10189 2018-07-12 Richard Biener <rguenther@suse.de>
10192 * dwarf2out.c (gen_block_die): Only output blocks when they have
10193 at least one !DECL_IGNORED_P variable.
10195 2018-07-12 Richard Biener <rguenther@suse.de>
10198 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
10199 Remove -mieee-fp handling.
10201 2018-07-12 Richard Biener <rguenther@suse.de>
10203 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
10204 left-over from last patch.
10206 2018-07-12 Jakub Jelinek <jakub@redhat.com>
10208 PR tree-optimization/86492
10209 * gimple-ssa-store-merging.c
10210 (imm_store_chain_info::coalesce_immediate_stores): Call
10211 check_no_overlap even for the merge_overlapping case. Formatting fix.
10213 2018-07-12 Richard Biener <rguenther@suse.de>
10215 PR middle-end/86479
10216 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
10217 move possibly trapping operations into the conditional.
10219 2018-07-12 Richard Biener <rguenther@suse.de>
10221 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
10222 (vn_lookup_simplify_result): Remove recursion limit applied
10224 (vn_nary_build_or_lookup_1): Adjust.
10225 (try_to_simplify): Likewise.
10226 * gimple-match-head.c (gimple_resimplify1): Instead apply one
10228 (gimple_resimplify2): Likewise.
10229 (gimple_resimplify3): Likewise.
10230 (gimple_resimplify4): Likewise.
10232 2018-07-11 Jakub Jelinek <jakub@redhat.com>
10234 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
10235 Use __mmask64 type instead of __mmask8 for __M argument.
10236 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
10237 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
10239 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
10240 __mmask16 for __M argument.
10241 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
10242 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
10243 to __mmask16 instead of __mmask8.
10244 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
10245 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
10246 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
10247 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
10248 instead of __mmask16 for __U argument.
10249 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
10250 __mmask16 instead of __mmask8 for __U argument.
10251 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
10253 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
10255 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
10257 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
10259 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
10261 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
10263 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
10265 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
10267 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
10269 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
10271 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
10272 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
10273 return type as well as __M argument type and all casts from __mmask8
10275 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
10276 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
10277 return type as well as __M argument type and all casts from __mmask8
10279 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
10280 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
10281 return type as well as __M argument type and all casts from __mmask8
10283 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
10284 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
10285 return type as well as __M argument type and all casts from __mmask8
10287 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
10288 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
10291 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
10293 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
10294 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
10297 2018-07-11 Paul Koning <ni1d@arrl.net>
10299 * doc/md.texi (define_subst): Document how multiple occurrences of
10300 the same argument in the replacement pattern are handled.
10302 2018-07-11 Paul Koning <ni1d@arrl.net>
10304 * doc/extend.texi (Common Variable Attributes): Move "mode" into
10305 alphabetical order.
10306 (Common Type Attributes): Add "mode" attribute.
10308 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
10310 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
10311 stream DECL_ORIGINAL_TYPE.
10312 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
10313 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
10314 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
10315 Do not walk original type.
10316 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
10318 (write_ts_decl_non_common_tree_pointers): Do not stream
10320 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
10321 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
10323 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
10325 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
10326 thread twice from the same starting edge.
10328 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
10330 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
10331 * gimple.c (gimple_call_nonnull_result_p): ...here...
10332 (gimple_call_nonnull_arg): ...and here.
10333 * gimple.h (gimple_call_nonnull_result_p): New.
10334 (gimple_call_nonnull_arg): New.
10336 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
10338 * config/arm/arm-cpus.in: Move information from fpu field of each
10339 cpu definition to the isa field.
10340 * config/arm/parsecpu.awk (fpu): Delete match rule.
10341 (gen_comm_data): Don't add bits from the CPU's FPU entry.
10343 2018-07-11 Richard Biener <rguenther@suse.de>
10346 * dwarf2out.c (init_sections_and_labels): Use
10347 output_asm_line_debug_info consistently.
10348 (dwarf2out_early_finish): Likewise.
10349 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
10352 2018-07-11 Richard Biener <rguenther@suse.de>
10354 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
10355 Rework father_bb setting in a way to avoid propagating constants
10356 multiple times on a loop body.
10358 2018-07-10 Mark Wielaard <mark@klomp.org>
10361 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
10363 2018-07-10 Richard Biener <rguenther@suse.de>
10365 * hash-map.h (hash_map::iterator::operator*): Return
10366 references to key and value.
10368 2018-07-10 Jakub Jelinek <jakub@redhat.com>
10371 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
10372 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
10373 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
10374 (gimplify_omp_for): For composite loops, move outer
10375 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
10376 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
10377 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
10378 TREE_LIST for both the original class iterator and the "last" helper
10379 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
10380 loop, remember has_decl_expr from outer composite loops for the
10381 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
10383 2018-07-09 Martin Sebor <msebor@redhat.com>
10385 PR middle-end/77357
10386 PR middle-end/86428
10387 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
10388 accessing implicitly initialized array elements.
10389 * expr.c (string_constant): Handle string initializers of
10390 character arrays within aggregates.
10391 * gimple-fold.c (fold_array_ctor_reference): Add argument.
10392 Store element offset. As a special case, handle zero size.
10393 (fold_nonarray_ctor_reference): Same.
10394 (fold_ctor_reference): Add argument. Store subobject offset.
10395 * gimple-fold.h (fold_ctor_reference): Add argument.
10397 2018-07-09 Paul Koning <ni1d@arrl.net>
10399 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
10400 (pdp11_insn_cost): New function.
10401 (pdp11_md_asm_adjust): New function.
10402 (TARGET_INVALID_WITHIN_DOLOOP): Define.
10403 (pdp11_rtx_costs): Update to match machine better.
10404 (output_addr_const_pdp11): Correct format mismatch warnings.
10405 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
10406 * config/pdp11/pdp11.md: General change to add base_cost and/or
10407 length attributes for use by new pdp11_insn_cost function.
10408 (MIN_BRANCH): Correct definition.
10410 (doloop_end): Use standard pattern name for looping pattern.
10411 (doloop_end_nocc): New.
10412 (movsf): Add another constraint alternative.
10413 (zero_extendqihi2): Add constraint alternatives for not in place
10415 (zero_extendhisi2): Remove.
10416 (shift patterns): Add CC handling variants.
10420 (define_peephole2): New peephole to recognize mov that sets CC for
10423 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10425 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
10426 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
10429 2018-07-09 Jakub Jelinek <jakub@redhat.com>
10432 * real.c (real_nextafter): Return true if result is denormal.
10434 2018-07-09 Martin Liska <mliska@suse.cz>
10436 * common.opt: Add back wrongly removed attribute.
10438 2018-07-09 Richard Biener <rguenther@suse.de>
10441 * dwarf2out.c (gen_block_die): For an early generated DIE
10442 always output high/low PC attributes.
10444 2018-07-09 Tom de Vries <tdevries@suse.de>
10446 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
10447 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
10448 onto VAR_DECL with abstract origin.
10450 2018-07-07 Jim Wilson <jimw@sifive.com>
10452 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
10454 2018-07-07 Tom de Vries <tdevries@suse.de>
10456 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
10458 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
10460 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
10462 (combine_bound): Use wide-int overflow calculation instead of
10464 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
10466 * fold-const.c (int_const_binop_2): Same.
10467 (extract_muldiv_1): Same.
10468 (fold_div_compare): Same.
10469 (fold_abs_const): Same.
10471 * poly-int.h (add): Same.
10475 * predict.c (predict_iv_comparison): Same.
10476 * profile-count.c (slow_safe_scale_64bit): Same.
10477 * simplify-rtx.c (simplify_const_binary_operation): Same.
10478 * tree-chrec.c (tree_fold_binomial): Same.
10479 * tree-data-ref.c (split_constant_offset_1): Same.
10480 * tree-if-conv.c (idx_within_array_bound): Same.
10481 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
10482 * tree-ssa-phiopt.c (minmax_replacement): Same.
10483 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
10484 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
10485 * vr-values.c (vr_values::adjust_range_with_scev): Same.
10486 * wide-int.cc (wi::add_large): Same.
10487 (wi::mul_internal): Same.
10488 (wi::sub_large): Same.
10489 (wi::divmod_internal): Same.
10490 * wide-int.h: Change overflow type to overflow_type for neg, add,
10491 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
10492 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
10493 mul_internal, divmod_internal.
10494 (overflow_type): New enum.
10495 (accumulate_overflow): New.
10497 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
10499 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
10500 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
10502 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
10504 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
10505 argument is checked for zero before entering loop, avoid checking again.
10507 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
10509 * gimplify.h (generic_expr_could_trap_p): Set as global function.
10510 * gimplify.h (generic_expr_could_trap_p): Likwise.
10511 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
10513 2018-07-06 Jakub Jelinek <jakub@redhat.com>
10515 PR tree-optimization/86401
10516 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
10517 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
10518 (fold_bit_and_mask): ... here. New helper function for match.pd.
10519 * fold-const.h (fold_bit_and_mask): Declare.
10520 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
10522 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
10525 * target.def (translate_mode_attribute): New hook.
10526 * targhooks.h (default_translate_mode_attribute): Declare.
10527 * targhooks.c (default_translate_mode_attribute): New function.
10528 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
10529 * doc/tm.texi: Regenerate.
10530 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
10531 (rs6000_translate_mode_attribute): New function.
10533 2018-07-06 Paul Koning <ni1d@arrl.net>
10535 * doc/md.texi (define_split): Document DONE and FAIL.
10536 (define_peephole2): Ditto.
10538 2018-07-05 Jeff Law <law@redhat.com>
10540 PR tree-optimization/86010
10541 * tree-ssa-dse.c (compute_trims): More aggressively trim at
10542 both the head and tail of mem* and str* calls.
10544 2018-07-05 Jim Wilson <jimw@sifive.com>
10546 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
10548 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
10550 * config/aarch64/aarch64-simd.md: correct flags text for
10551 MIN_EXPR replacement.
10553 2018-07-05 James Clarke <jrtc27@jrtc27.com>
10555 * configure: Regenerated.
10557 2018-07-05 Carl Love <cel@us.ibm.com>
10559 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
10560 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
10561 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
10562 VSX_BUILTIN_DOUBLEL_V4SF.
10564 2018-07-05 Martin Sebor <msebor@redhat.com>
10567 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
10568 than its domain to compute its the upper bound of a char array.
10570 2018-07-05 Nathan Sidwell <nathan@acm.org>
10572 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
10573 * doc/cpp.texi: Update comment.
10574 * doc/tm.texi: Rebuilt.
10575 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
10576 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
10577 * doc/extend.texi (Backwards Compatibility): Clarify it is system
10578 headers affected by extern "C".
10579 * system.h: Poison NO_IMPLICIT_EXTERN_C.
10580 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
10581 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
10582 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
10583 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
10584 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
10585 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
10586 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
10587 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
10588 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
10589 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
10590 config/sparc/sp64-elf.h, config/spu/spu.h,
10591 config/stormy16/stormy16.h, config/v850/v850.h,
10592 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
10593 define NO_IMPLICIT_EXTERN_C.
10594 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
10596 2018-07-05 Tamar Christina <tamar.christina@arm.com>
10599 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
10600 instead of GET_MODE_SIZE when comparing Units.
10602 2018-07-05 Tamar Christina <tamar.christina@arm.com>
10605 * rtlanal.c (set_noop_p): Constrain on mode change,
10606 include hard-reg-set.h
10608 2018-07-05 Tamar Christina <tamar.christina@arm.com>
10610 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
10612 2018-07-05 Jakub Jelinek <jakub@redhat.com>
10615 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
10618 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
10620 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
10622 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
10625 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
10627 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
10629 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
10631 PR middle-end/86380
10632 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
10634 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
10636 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
10637 neg_*_op* variables.
10639 2018-07-04 Martin Liska <mliska@suse.cz>
10641 * tree-switch-conversion.c: Define
10642 max_ratio_for_speed and max_ratio_for_size constants.
10644 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
10645 Martin Liska <mliska@suse.cz>
10647 PR middle-end/66240
10650 * common.opt: Rename align options with 'str_' prefix.
10651 * common/config/i386/i386-common.c (set_malign_value): New
10653 (ix86_handle_option): Use it to set -falign-* options/
10654 * config/aarch64/aarch64-protos.h (struct tune_params): Change
10655 type from int to string.
10656 * config/aarch64/aarch64.c: Update default values from int
10658 * config/alpha/alpha.c (alpha_override_options_after_change):
10660 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
10661 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10662 max skip conditionally.
10663 * config/i386/freebsd.h (SUBALIGN_LOG): New.
10664 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10665 max skip conditionally.
10666 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10667 max skip conditionally.
10668 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
10669 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10670 max skip conditionally.
10671 * config/i386/i386.c (struct ptt): Change type from int to
10673 (ix86_default_align): Set default values.
10674 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
10675 max skip conditionally.
10676 * config/i386/iamcu.h (SUBALIGN_LOG): New.
10677 (ASM_OUTPUT_MAX_SKIP_ALIGN):
10678 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
10679 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10680 max skip conditionally.
10681 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
10682 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
10683 * config/i386/x86-64.h (SUBALIGN_LOG): New.
10684 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10685 max skip conditionally.
10686 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
10687 * config/ia64/ia64.c (ia64_option_override): Set default values
10688 for alignment options.
10689 * config/m68k/m68k.c: Handle new str_align_* options.
10690 * config/mips/mips.c (mips_set_compression_mode): Change
10692 (mips_option_override): Set default values for options.
10693 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
10695 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10697 * config/rx/rx.c (rx_option_override): Likewise.
10698 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
10699 (LABEL_ALIGN): Use align_labels_log.
10700 (LOOP_ALIGN): Use align_loops_align.
10701 * config/s390/s390.c (s390_asm_output_function_label): Use new
10703 * config/sh/sh.c (sh_override_options_after_change):
10704 Change type of constants.
10705 * config/spu/spu.c (spu_sched_init): Likewise.
10706 * config/sparc/sparc.c (sparc_option_override): Set default
10707 values for options.
10708 * config/visium/visium.c (visium_option_override): Likewise.
10709 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
10710 emit p2align format with last argument if it's not needed.
10711 * doc/invoke.texi: Document extended format of -falign-*.
10712 * final.c: Use align_labels alignment.
10713 * flags.h (struct target_flag_state): Change type to use
10715 (struct align_flags_tuple): New.
10716 (struct align_flags): Likewise.
10717 (align_loops_log): Redefine macro to use new types.
10718 (align_loops_max_skip): Redefine macro to use new types.
10719 (align_jumps_log): Redefine macro to use new types.
10720 (align_jumps_max_skip): Redefine macro to use new types.
10721 (align_labels_log): Redefine macro to use new types.
10722 (align_labels_max_skip): Redefine macro to use new types.
10723 (align_functions_log): Redefine macro to use new types.
10724 (align_loops): Redefine macro to use new types.
10725 (align_jumps): Redefine macro to use new types.
10726 (align_labels): Redefine macro to use new types.
10727 (align_functions): Redefine macro to use new types.
10728 (align_functions_max_skip): Redefine macro to use new types.
10729 (align_loops_value): New macro.
10730 (align_jumps_value): New macro.
10731 (align_labels_value): New macro.
10732 (align_functions_value): New macro.
10733 * function.c (invoke_set_current_function_hook): Propagate
10734 alignment values from flags to global variables default in
10736 * ipa-icf.c (sem_function::equals_wpa): Use
10737 cl_optimization_option_eq instead of memcmp.
10738 * lto-streamer.h (cl_optimization_stream_out): Support streaming
10740 (cl_optimization_stream_in): Likewise.
10741 * optc-save-gen.awk: Support strings in cl_optimization.
10742 * opth-gen.awk: Likewise.
10743 * opts.c (finish_options): Remove error checking of invalid
10745 (MAX_CODE_ALIGN): Remove.
10746 (MAX_CODE_ALIGN_VALUE): Likewise.
10747 (parse_and_check_align_values): New function.
10748 (check_alignment_argument): Likewise.
10749 (common_handle_option): Use check_alignment_argument.
10750 * opts.h (parse_and_check_align_values): Declare.
10751 * toplev.c (init_alignments): Remove.
10752 (read_log_maxskip): New.
10753 (parse_N_M): Likewise.
10754 (parse_alignment_opts): Likewise.
10755 (backend_init_target): Remove usage of init_alignments.
10756 * toplev.h (parse_alignment_opts): Declare.
10757 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
10759 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
10760 * tree.c (cl_option_hasher::equal): New.
10761 * varasm.c: Use new global macros.
10763 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
10765 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
10766 Use a simpler align directive also if MAXSKIP = ALIGN-1.
10767 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10768 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10769 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10770 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
10771 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
10772 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
10773 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
10774 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10775 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10776 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10777 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10779 2018-07-04 Martin Liska <mliska@suse.cz>
10780 Jonathan Wakely <jwakely@redhat.com>
10782 * coverage.c: Use correct type.
10783 * doc/invoke.texi: Language correction.
10785 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
10788 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
10789 ENDBRANCH for non-tail call which may return via indirect branch.
10790 * doc/extend.texi: Document indirect_return attribute.
10792 2018-07-03 Martin Sebor <msebor@redhat.com>
10794 PR tree-optimization/86274
10795 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
10797 (format_floating): Correct handling of infinities and NaNs.
10799 2018-07-03 Martin Sebor <msebor@redhat.com>
10801 * print-tree.c (print_real_cst): New function.
10802 (print_node_brief): Call it.
10803 (print_node): Ditto.
10805 2018-07-03 Jeff Law <law@redhat.com>
10807 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
10808 into a single pattern.
10810 * config/h8300/h8300.md (ors code_iterator): New.
10811 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
10812 a single pattern and single splitter.
10813 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
10814 (iorqi3_1, xorqi3_1): Likewise.
10815 (iorqi3, xorqi3 expanders): Similarly.
10817 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
10818 (movmd_internal) into a single pattern using the P mode iterator.
10819 (movmd splitters): Similarly.
10820 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
10821 (movsd splitters): Similarly.
10823 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
10824 ADDB, ADDW and ADDL into a single ADD attribute which selects the
10825 right table based on the size of the operand.
10826 * config/h8300/h8300.md (length_table): Corresponding changes. All
10827 references to "addb", "addw" and "addl" changed to "add".
10828 (btst patterns): Merge two variants into a single pattern.
10829 (tstqi, tsthi): Likewise.
10830 (addhi3_incdec, addsi3_incdec): Likewise.
10831 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
10832 (mulhi3, mulsi3): Likewise.
10833 (udivhi3, udivsi3): Likewise.
10834 (divhi3, divsi3): Likewise.
10835 (andorqi3, andorhi3, andorsi3): Likewise.
10837 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
10840 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
10841 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
10843 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10845 PR tree-optimization/85694
10846 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
10847 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
10849 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
10850 (<u>avg<mode>3_ceil): New patterns.
10852 2018-07-03 David Malcolm <dmalcolm@redhat.com>
10854 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
10855 scan-tree-dump directive.
10856 * gcc.dg/vect/slp-perm-2.c: Likewise.
10857 * gcc.dg/vect/slp-perm-3.c: Likewise.
10858 * gcc.dg/vect/slp-perm-5.c: Likewise.
10859 * gcc.dg/vect/slp-perm-6.c: Likewise.
10860 * gcc.dg/vect/slp-perm-7.c: Likewise.
10861 * gcc.dg/vect/slp-perm-8.c: Likewise.
10863 2018-07-03 Marek Polacek <polacek@redhat.com>
10865 PR middle-end/86202
10866 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
10868 2018-07-03 Richard Biener <rguenther@suse.de>
10871 * tree-ssa-structalias.c (find_func_clobbers): Properly
10872 handle indirect calls.
10874 2018-07-03 Jeff Law <law@redhat.com>
10876 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
10877 (shifts): New code iterator.
10878 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
10879 expander. Fix HImode handling on H8/SX.
10880 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
10881 (subqi3, subhi3, subsi3 expanders): Likewise.
10882 (andqi3, andhi3, andsi3 expanders): Likewise.
10883 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
10884 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
10885 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
10886 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
10887 (zero_extendqihi2, zero_extendqisi2): Likewise.
10888 (extendqihi2, extendqisi2): Likewise.
10889 (rotlqi3, rotlhi3, rotlsi3): Likewise.
10890 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
10891 (rotlqi3_1, rotlhi3_1): Likewise.
10892 (logicalhi3_sn, logicalsi3_sn): Likewise.
10893 (logicalhi3, logicalsi3): Likewise.
10895 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10897 * tree-vect-patterns.c (vect_recog_rotate_pattern)
10898 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
10899 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
10900 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
10901 type to append_pattern_def_seq instead of creating a stmt_vec_info
10903 (build_mask_conversion): Likewise. Remove vinfo argument.
10904 (vect_add_conversion_to_patterm): Likewise, renaming to...
10905 (vect_add_conversion_to_pattern): ...this.
10906 (vect_recog_mask_conversion_pattern): Update call to
10907 build_mask_conversion. Pass the vector type to
10908 append_pattern_def_seq here too.
10909 (vect_recog_gather_scatter_pattern): Update call to
10910 vect_add_conversion_to_pattern.
10912 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10914 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
10915 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
10916 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
10917 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
10918 STMT_VINFO_PATTERN_DEF_SEQ to null here.
10919 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
10920 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
10921 append_pattern_def_seq instead of new_pattern_def_seq.
10922 (vect_recog_divmod_pattern): Do both of the above.
10923 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
10926 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10928 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
10929 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
10930 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
10931 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
10932 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
10933 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
10934 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
10935 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
10936 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
10937 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
10938 parameter with a single stmt_vec_info.
10939 (vect_recog_func_ptr): Likewise.
10940 (vect_recog_gather_scatter_pattern): Likewise, folding in...
10941 (vect_try_gather_scatter_pattern): ...this.
10942 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
10943 the stmt_vec_info of the statement to be matched. Don't clear
10944 STMT_VINFO_RELATED_STMT.
10945 (vect_pattern_recog): Update call accordingly.
10947 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10949 PR tree-optimization/85694
10950 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
10951 (uavgM3_ceil): Document new optabs.
10952 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
10953 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
10955 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
10956 (savg_ceil_optab): New optabs.
10957 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
10958 (vect_vect_recog_func_ptrs): Add it.
10959 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
10960 constant directly from the associated lhs.
10962 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10964 * tree-vect-patterns.c (vect_split_statement): New function.
10965 (vect_convert_input): Use it to try to split an existing cast.
10967 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10969 * poly-int.h (print_hex): New function.
10970 * dumpfile.h (dump_dec, dump_hex): Declare.
10971 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
10972 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
10973 min_input_precision, operation_precision and operation_sign.
10974 * tree-vect-patterns.c (vect_get_range_info): New function.
10975 (vect_same_loop_or_bb_p, vect_single_imm_use)
10976 (vect_operation_fits_smaller_type): Delete.
10977 (vect_look_through_possible_promotion): Add an optional
10978 single_use_p parameter.
10979 (vect_recog_over_widening_pattern): Rewrite to use new
10980 stmt_vec_info infomration. Handle one operation at a time.
10981 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
10982 (vect_truncatable_operation_p, vect_set_operation_type)
10983 (vect_set_min_input_precision): New functions.
10984 (vect_determine_min_output_precision_1): Likewise.
10985 (vect_determine_min_output_precision): Likewise.
10986 (vect_determine_precisions_from_range): Likewise.
10987 (vect_determine_precisions_from_users): Likewise.
10988 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
10989 (vect_vect_recog_func_ptrs): Put over_widening first.
10991 (vect_pattern_recog): Call vect_determine_precisions.
10993 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10995 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
10996 statements that have been replaced by further pattern statements.
10997 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
10999 2018-07-03 Richard Biener <rguenther@suse.de>
11001 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
11002 always set *dt. Dump vectype in vectype overload.
11003 * dumpfile.h (dump_gimple_expr): New function.
11004 (dump_gimple_expr_loc): Likewise.
11005 * dumpfile.c (dump_gimple_expr): New function.
11006 (dump_gimple_expr_loc): Likewise.
11008 2018-07-02 Jeff Law <law@redhat.com>
11010 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
11011 the H8/300, H8/300H and H8/S variants into a single pattern.
11012 (movhi_h8300, movqi_h8300hs): Similarly.
11013 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
11014 (QHI mode iterator): New.
11016 * config/h8300/h8300.md: Remove trailing whitespace.
11018 2018-07-02 Jim Wilson <jimw@sifive.com>
11020 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
11021 instead of emit_insn for interrupt returns.
11022 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
11023 (riscv_sret, riscv_uret): Likewise.
11025 2018-07-02 David Malcolm <dmalcolm@redhat.com>
11027 * pretty-print.c (selftest::test_pp_format): Move save and restore
11028 of quotes to class auto_fix_quotes, and add an instance.
11029 * selftest.c: Include "intl.h".
11030 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
11031 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
11032 * selftest.h (selftest::auto_fix_quotes): New class.
11034 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
11036 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
11037 (aarch64_sve_prepare_conditional_op): Remove.
11038 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
11039 Allow aarch64_simd_reg_or_zero as select operand; remove
11040 the aarch64_sve_prepare_conditional_op call.
11041 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
11042 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
11043 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
11044 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
11045 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
11046 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
11047 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
11048 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
11049 and a splitters to match all of the *_any patterns.
11050 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
11052 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
11053 (SVE_COND_FP_BINARY_REV): Remove.
11054 (sve_int_op_rev, sve_fp_op_rev): New.
11055 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
11056 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
11057 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
11058 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
11059 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
11060 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
11061 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
11062 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
11063 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
11065 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
11066 Remove match_dup 1 from the inner unspec.
11067 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
11069 * config/aarch64/aarch64.md (movprfx): New attr.
11070 (length): Default movprfx to 8.
11071 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
11072 (*madd<SVE_I>, *msub<SVE_I): Likewise.
11073 (*<su>mul<SVE_I>3_highpart): Likewise.
11074 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
11075 (*v<ASHIFT><SVE_I>3): Likewise.
11076 (*<su><MAXMIN><SVE_I>3): Likewise.
11077 (*<su><MAXMIN><SVE_F>3): Likewise.
11078 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
11079 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
11080 (*div<SVE_F>4): Likewise.
11082 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
11084 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
11087 2018-07-02 Richard Biener <rguenther@suse.de>
11089 PR tree-optimization/86363
11090 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
11091 memset argument refers to a non-variable address.
11093 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
11095 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
11096 {PLUS,MINUS}_EXPR code to...
11097 (adjust_symbolic_bound): ...here,
11098 (combine_bound): ...here,
11099 (set_value_range_with_overflow): ...and here.
11101 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
11103 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
11105 (extract_range_from_abs_expr): ...here.
11107 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
11109 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
11110 -fno-omit-frame-pointer when not optimizing.
11112 2018-07-02 Martin Liska <mliska@suse.cz>
11115 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
11116 (propagate_nothrow): Likewise.
11118 2018-07-02 Martin Liska <mliska@suse.cz>
11121 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
11123 2018-07-02 David Malcolm <dmalcolm@redhat.com>
11125 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
11126 function in r262149, changing "loc" param from source_location to
11127 const dump_location_t &.
11128 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
11129 declaration, as above.
11131 2018-07-01 Paul Koning <ni1d@arrl.net>
11133 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
11134 -munit-asm, -mgnu-asm, -mdec-asm.
11135 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
11136 (pdp11_output_labelref): New.
11137 (pdp11_output_def): New.
11138 (pdp11_output_addr_vec_elt): New.
11139 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
11140 %# and %@ format codes.
11141 (pdp11_option_override): New.
11142 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
11143 (pdp11_output_ident): New.
11144 (pdp11_asm_named_section): New.
11145 (pdp11_asm_init_sections): New.
11146 (pdp11_file_start): New.
11147 (pdp11_file_end): New.
11148 (output_ascii): Use .ascii/.asciz for -mdec-asm.
11149 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
11150 %o, like %c but octal.
11151 (pdp11_option_override): New.
11152 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
11154 (DATA_SECTION_ASM_OP): Ditto.
11155 (READONLY_DATA_SECTION_ASM_OP): New.
11156 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
11157 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
11158 (ASM_OUTPUT_LABELREF): Ditto.
11159 (ASM_OUTPUT_DEF): Ditto.
11160 (ASM_OUTPUT_EXTERNAL): New.
11161 (ASM_OUTPUT_SOURCE_FILENAME): New.
11162 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
11163 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
11164 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
11165 %# and %@ format codes.
11166 * config/pdp11/pdp11.opt (mgnu-asm): New.
11167 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
11168 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
11169 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
11171 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
11173 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
11174 dereferencing path[] beyond its length.
11176 (debug_all_paths): New.
11177 (rewire_first_differing_edge): New.
11178 (adjust_paths_after_duplication): New.
11179 (duplicate_thread_path): Call adjust_paths_after_duplication.
11181 (thread_through_all_blocks): Add new argument to
11182 duplicate_thread_path.
11184 2018-06-30 Jim Wilson <jimw@sifive.com>
11186 * config/riscv/predicates.md (p2m1_shift_operand): New.
11187 (high_mask_shift_operand): New.
11188 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
11189 pattern using p2m1_shift_operand.
11190 (lshsi3_zero_extend_3+2): New combiner pattern using
11191 high_mask_shift_operand.
11193 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11195 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
11197 (vect_recog_rotate_pattern): ...here.
11198 (vect_convert_input): Try to insert casts of invariants in the
11200 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
11201 preheader to be empty.
11203 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11205 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
11206 vector type. If given, install it in the new statement's
11207 STMT_VINFO_VECTYPE.
11208 (vect_element_precision): New function.
11209 (vect_unpromoted_value): New struct.
11210 (vect_unpromoted_value::vect_unpromoted_value): New function.
11211 (vect_unpromoted_value::set_op): Likewise.
11212 (vect_look_through_possible_promotion): Likewise.
11213 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
11214 (vect_widened_op_tree, vect_convert_input): Likewise.
11215 (vect_convert_inputs, vect_convert_output): Likewise.
11216 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
11217 to handle the optional cast of the multiplication result and
11218 vect_widened_op_tree to detect the widened multiplication itself.
11219 Do not require the input and output of promotion casts to have
11220 the same sign, but base the signedness of the operation on the
11221 input rather than the result. If the pattern includes two
11222 promotions, check that those promotions have the same sign.
11223 Do not restrict the MULT_EXPR handling to a double-width result;
11224 handle quadruple-width results and wider. Use vect_convert_inputs
11225 to convert the inputs to the common type.
11226 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
11227 to handle the optional cast of the ABS result. Also allow a sign
11228 change or a sign extension between the ABS and MINUS.
11229 Use vect_widened_op_tree to detect the widened subtraction and use
11230 vect_convert_inputs to convert the inputs to the common type.
11231 (vect_handle_widen_op_by_const): Delete.
11232 (vect_recog_widen_op_pattern): New function.
11233 (vect_recog_widen_mult_pattern): Use it.
11234 (vect_recog_widen_shift_pattern): Likewise.
11235 (vect_recog_widen_sum_pattern): Use
11236 vect_look_through_possible_promotion to handle the promoted
11239 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11241 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
11242 the containing gimple_seq *.
11243 * gimple-iterator.h (gsi_for_stmt): Declare it.
11244 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
11245 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
11246 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
11247 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
11248 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
11250 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
11252 (vect_mark_pattern_stmts): ...here. Handle cases in which the
11253 statement being replaced is part of an existing pattern
11254 definition sequence, inserting the new pattern statements before
11256 (vect_pattern_recog_1): Don't return a bool. If the statement
11257 is already part of a pattern, instead apply pattern matching
11258 to the pattern definition statements. Don't clear the
11259 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
11260 (vect_pattern_recog): Don't break after the first match;
11261 continue processing the pattern definition statements instead.
11262 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
11264 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11266 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
11267 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
11268 (vect_recog_widen_sum_pattern): Use it.
11270 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11272 * tree-vect-loop.c (vectorizable_reduction): Assert that the
11273 phi is not a pattern statement and has not been replaced by
11274 a pattern statement.
11275 * tree-vect-patterns.c (type_conversion_p): Don't check
11276 STMT_VINFO_IN_PATTERN_P.
11277 (vect_recog_vector_vector_shift_pattern): Likewise.
11278 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
11279 the pattern statement rather than the original statement; check
11280 directly for a WIDEN_MULT_EXPR here.
11281 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
11282 vect_is_simple_use to return the pattern statement rather
11283 than the original statement; use is_pattern_stmt_p to check
11284 for such a pattern statement.
11285 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
11286 to return the pattern statement rather than the original statement;
11287 don't do the same transformation here.
11288 (vect_is_simple_use): If the defining statement has been replaced
11289 by a pattern statement, return the pattern statement instead.
11290 Remove the corresponding (local) transformation from the vectype
11293 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11295 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
11296 end and default to null.
11297 * tree-vect-loop.c (vect_create_epilog_for_reduction)
11298 (vectorizable_reduction): Update calls accordingly, dropping the
11299 gimple ** argument if the passed-back statement isn't needed.
11300 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
11301 (vect_recog_rotate_pattern): Likewise.
11302 (vect_recog_mask_conversion_pattern): Likewise.
11303 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
11304 (vect_mask_constant_operand_p): Likewise.
11305 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
11306 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
11307 (get_group_load_store_type, get_load_store_type): Likewise.
11308 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
11309 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
11310 (vectorizable_conversion, vectorizable_assignment): Likewise.
11311 (vectorizable_shift, vectorizable_operation): Likewise.
11312 (vectorizable_store, vect_is_simple_cond): Likewise.
11313 (vectorizable_condition, vectorizable_comparison): Likewise.
11314 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
11315 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
11316 and move it to the end. Cope with null def_stmt_outs.
11318 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
11320 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
11322 2018-06-29 Jeff Law <law@redhat.com>
11324 * config/v850/v850.c (v850_legitimate_address_p): Handle large
11325 displacements for TARGET_V850E2V3 and newer.
11326 (TARGET_LRA_P): Remove. Defaults to LRA now.
11327 * config/v850/v850.md (sign23byte_load): Remove.
11328 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
11329 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
11331 2018-06-29 Martin Liska <mliska@suse.cz>
11334 * coverage.c (coverage_init): Mangle full path name.
11335 * doc/invoke.texi: Document the change.
11336 * gcov-io.c (mangle_path): New.
11337 * gcov-io.h (mangle_path): Likewise.
11338 * gcov.c (mangle_name): Use mangle_path for path mangling.
11340 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11342 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
11343 if starting source register is not even.
11345 2018-06-29 Martin Liska <mliska@suse.cz>
11347 PR tree-optimization/86263
11348 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
11349 Make edge redirection.
11351 2018-06-29 David Malcolm <dmalcolm@redhat.com>
11353 * dumpfile.c (dump_loc): Add indentation based on scope depth.
11354 (dump_scope_depth): New variable.
11355 (get_dump_scope_depth): New function.
11356 (dump_begin_scope): New function.
11357 (dump_end_scope): New function.
11358 * dumpfile.h (get_dump_scope_depth): New declaration.
11359 (dump_begin_scope): New declaration.
11360 (dump_end_scope): New declaration.
11361 (class auto_dump_scope): New class.
11362 (AUTO_DUMP_SCOPE): New macro.
11363 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
11366 2018-06-29 Richard Biener <rguenther@suse.de>
11368 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
11369 compute_all_dependences succeeds.
11370 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
11371 exceed --param loop-max-datarefs-for-datadeps.
11373 2018-06-29 Jakub Jelinek <jakub@redhat.com>
11375 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
11377 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
11380 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
11381 alternative 0 in preferred_for_speed attribute.
11383 2018-06-28 Paul Koning <ni1d@arrl.net>
11385 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
11386 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
11387 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
11388 * config/pdp11/pdp11.md: Correct "length" attribute calculation
11389 for shift insn patterns.
11391 2018-06-28 David Malcolm <dmalcolm@redhat.com>
11393 * cgraph.c (cgraph_node::get_body): Replace assignments to
11394 "dump_file" with calls to set_dump_file.
11395 * dumpfile.c (alt_dump_file): Make static, and group with...
11396 (alt_flags): ...this definition.
11397 (dumps_are_enabled): New variable.
11398 (refresh_dumps_are_enabled): New function.
11399 (set_dump_file): New function.
11400 (set_alt_dump_file): New function.
11401 (gcc::dump_manager::dump_start): Replace assignments to
11402 "dump_file" and "alt_dump_file" with calls to set_dump_file and
11404 (gcc::dump_manager::dump_finish): Likewise.
11405 * dumpfile.h (alt_dump_file): Delete decl.
11406 (dumps_are_enabled): New variable decl.
11407 (set_dump_file): New function decl.
11408 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
11410 * tree-nested.c (lower_nested_functions): Replace assignments to
11411 "dump_file" with calls to set_dump_file.
11413 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
11415 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
11416 goto_locus of each outgoing edge of each basic block.
11418 2018-06-28 Richard Biener <rguenther@suse.de>
11420 * dwarf2out.c (decl_scope_table): Remove.
11421 (push_decl_scope): Likewise.
11422 (pop_decl_scope): Likewise.
11423 (gen_type_die_for_member): Do not call push/pop_decl_scope.
11424 (gen_struct_or_union_type_die): Likewise.
11425 (gen_tagged_type_die): Likewise.
11426 (dwarf2out_init): Do not initialize decl_scope_table.
11427 (dwarf2out_c_finalize): Do not free it.
11429 2018-06-28 Richard Biener <rguenther@suse.de>
11431 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
11432 deciding whether to not re-use a DIE.
11434 2018-06-28 Richard Biener <rguenther@suse.de>
11436 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
11437 DW_AT_abstract_origin attribute.
11439 2018-06-28 Martin Liska <mliska@suse.cz>
11441 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
11442 Use newly introduced constants.
11443 * tree-switch-conversion.h (struct jump_table_cluster):
11444 Define max_ratio_for_size and max_ratio_for_speed.
11446 2018-06-28 Martin Liska <mliska@suse.cz>
11448 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
11449 Add new checking assert to catch invalid state.
11450 (jump_table_cluster::can_be_handled): Handle single case
11452 (jump_table_cluster::is_beneficial): Bail out for such case.
11453 (bit_test_cluster::find_bit_tests):
11454 Add new checking assert to catch invalid state.
11455 (bit_test_cluster::can_be_handled): Handle single case
11457 (bit_test_cluster::is_beneficial): Bail out for such case.
11458 (switch_decision_tree::analyze_switch_statement):
11461 2018-06-28 Martin Liska <mliska@suse.cz>
11463 * common.opt: Introduce -completion option.
11464 * gcc.c (driver_handle_option): Handle it.
11465 (driver::main): Print completions if completion
11467 * opt-suggestions.c (option_proposer::get_completions):
11469 (option_proposer::suggest_completion): Likewise.
11470 (option_proposer::find_param_completions): Likewise.
11471 (verify_autocompletions): Likewise.
11472 (test_completion_valid_options): Likewise.
11473 (test_completion_valid_params): Likewise.
11474 (in_completion_p): Likewise.
11475 (empty_completion_p): Likewise.
11476 (test_completion_partial_match): Likewise.
11477 (test_completion_garbage): Likewise.
11478 (opt_proposer_c_tests): Likewise.
11479 * opt-suggestions.h: Declare new functions.
11480 * opts.c (common_handle_option): Handle OPT__completion_.
11481 * selftest-run-tests.c (selftest::run_tests): Add
11482 opt_proposer_c_tests.
11483 * selftest.c (assert_str_startswith): New.
11484 * selftest.h (assert_str_startswith): Likewise.
11485 (opt_proposer_c_tests): New.
11486 (ASSERT_STR_STARTSWITH): Likewise.
11488 2018-06-28 Martin Liska <mliska@suse.cz>
11490 * Makefile.in: Add opt-suggestions.o.
11491 * gcc-main.c: Include opt-suggestions.h.
11492 * gcc.c (driver::driver): Likewise.
11493 (driver::~driver): Remove m_option_suggestions.
11494 (driver::build_option_suggestions): Moved to option_proposer.
11495 (driver::suggest_option): Likewise.
11496 (driver::handle_unrecognized_options): Use option_proposer.
11497 * gcc.h (class driver): Add new memver m_option_proposer.
11498 * opt-suggestions.c: New file.
11499 * opt-suggestions.h: New file.
11501 2018-06-28 Martin Liska <mliska@suse.cz>
11503 * vec.h (class auto_string_vec): New (moved from auto_argvec).
11504 (auto_string_vec::~auto_string_vec): Likewise.
11506 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
11508 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
11509 prevent_decl_creation_for_types fields up and add reset_location field.
11510 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
11511 statement if id->reset_location is true.
11512 (copy_edges_for_bb): Do not set goto_locus on the new edges if
11513 id->reset_location is true.
11514 (copy_phis_for_bb): Force input_location on the arguments if
11515 id->reset_location is true.
11516 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
11517 is set on the function to be inlined.
11519 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
11521 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
11523 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
11525 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
11526 registers for Pmode.
11527 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
11528 hard registers for the clobbered pseudo.
11530 2018-06-27 Paul Koning <ni1d@arrl.net>
11532 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
11533 mutually exclusive options.
11534 * config/pdp11/constraints.md (h): New constraint.
11535 (O): Update definition to match shift code generation.
11536 (D): New constraint.
11537 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
11539 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
11541 (output_jump): Change arguments.
11542 (pdp11_fixed_cc_regs): New function.
11543 (pdp11_cc_mode): Ditto.
11544 (pdp11_expand_shift): Ditto.
11545 (pdp11_assemble_shift): Ditto.
11546 (pdp11_small_shift): Ditto.
11547 (pdp11_branch_cost): Remove.
11548 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
11550 (pdp11_register_move_cost): Update for CC registers.
11551 (pdp11_rtx_costs): Add case for LSHIFTRT.
11552 (pdp11_output_jump): Add CCNZ mode conditional branches.
11553 (notice_update_cc_on_set): Remove.
11554 (pdp11_cc_mode): New function.
11555 (simple_memory_operand): Correct pre/post decrement case.
11556 (no_side_effect_operand): New function.
11557 (pdp11_regno_reg_class): Add CC_REGS class.
11558 (pdp11_fixed_cc_regs): New function.
11559 (pdp11_small_shift): New function.
11560 (pdp11_expand_shift): New function to expand shift insns.
11561 (pdp11_assemble_shift): New function to output shifts.
11562 (pdp11_branch_cost): Remove.
11563 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
11564 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
11565 (WCHAR_TYPE): Ditto.
11566 (PTRDIFF_TYPE): Ditto.
11567 (ADJUST_INSN_LENGTH): New macro.
11568 (FIXED_REGISTERS): Add CC registers.
11569 (CALL_USED_REGISTERS): Ditto.
11570 (reg_class): Ditto.
11571 (REG_CLASS_NAMES): Ditto.
11572 (REG_CLASS_CONTENTS): Ditto.
11573 (SELECT_CC_MODE): Use new function.
11574 (TARGET_FLAGS_REGNUM): New macro.
11575 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
11576 (cc0_reg_rtx): Remove.
11577 (CC_STATUS_MDEP): Remove.
11578 (CC_STATUS_MDEFP_INIT): Remove.
11579 (CC_IN_FPU): Remove.
11580 (NOTICE_UPDATE_CC): Remove.
11581 (REGISTER_NAMES): Add CC registers.
11582 (BRANCH_COST): Change to constant 1.
11583 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
11585 * config/pdp11/pdp11.opt (mbcopy): Remove.
11586 (mbcopy-builtin): Remove.
11587 (mbranch-cheap): Remove.
11588 (mbranch-expensive): Remove.
11589 * config/pdp11/predicates.md (expand_shift_operand): Update to
11590 match shift code generation.
11591 (ccnz_operator): New predicate.
11592 * doc/invoke.texi (PDP-11 Options): Remove deleted options
11593 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
11594 Remove non-existent option -mabshi, -mno-abshi. Document mutually
11596 * doc/md.texi (PDP-11): Document new D and h constraints. Update
11597 description of O constraint.
11599 2018-06-27 Jeff Law <law@redhat.com>
11600 Austin Law <austinklaw@gmail.com>
11602 * config/v850/v850.md (addsi3_set_flags): New pattern.
11603 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
11604 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
11605 (zero_extendhisi2_v850_set_flags): Likewise.
11606 (zero_extendqisi2_v850_set_flags): Likewise.
11607 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
11608 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
11609 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
11611 * config/v850/v850-protos.h (notice_update_cc): Remove.
11612 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
11613 (v850_print_operand): Handle 'D' and "d".
11614 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
11615 Add handling of arithmetic/logical operations compared against zero.
11616 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
11617 Do not look at v850_compare_op, instead get mode from last argument.
11618 (v850_gen_compare): Remove
11619 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
11620 after reload for prologue insns.
11621 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
11623 (construct_save_jarl): Likewise.
11624 (TARGET_FLAGS_REGNUM): Define.
11625 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
11626 (NOTICE_UPDATE_CC): Remove.
11627 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
11628 than cc0. Conditionalize on reload_completed.
11629 (cmpsi_insn, setfcc_insn): Likewise.
11630 (tst1 splitter): Turn into define_and_split which sets the flags
11632 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
11633 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
11634 (cstoresf4, cstoredf4): Clobber the flags.
11635 (cmpsi, cmpsf, cmpdf): Remove expanders.
11636 (setf_insn): Remove pattern.
11637 (addsi3): Turn into define_and_split which clobbers the flags after
11638 reload and a suitable pattern (addsi3_clobber_flags) for use after
11640 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
11641 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
11642 (ashrsi3, ashrsi3_v850e2): Likewise.
11643 (bins): Clobber the flags.
11644 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
11645 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
11646 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
11647 (call_value_internal_short, call_value_internal_long): Likewise.
11648 (callt_save_interrupt, callt_return_interrupt): Likewise.
11649 (save_interrupt, return_interrupt): Likewise.
11650 (callt_save_all_interrupt, save_all_interrupt): Likewise.
11651 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
11652 (restore_all_interrupt, _restore_all_interrupt): Likewise.
11653 (All FP comparisons): Only allow after reload has completed.
11655 (divh, divhu): Tweak output template.
11656 (branch_z_normal, branch_z_invert): Remove
11657 (branch_nz_normal, branch_nz_invert): Likewise.
11658 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
11660 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
11661 * config/v850/v850.c (notice_update_cc): Remove.
11662 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
11663 (CC_NO_CARRY): Likewise.
11664 (NOTICE_UPDATE_CC): Define to nothing.
11665 * config/v850/v850.md: Remove block comment on cc0 handling
11666 Remove "cc" attribute from all patterns. Remove cc_status handling
11667 from all patterns. Minor formatting fixes.
11669 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11671 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
11672 (cortex-a76.cortex-a55): Likewise.
11673 * config/aarch64/aarch64-tune.md: Regenerate.
11674 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
11675 cortex-a76.cortex-a55.
11677 2018-06-27 Jeff Law <law@redhat.com>
11679 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
11680 (MULTILIB_DIRNAMES): Similarly.
11682 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
11684 * gimple.h (gimple_return_retbnd): Delete.
11685 (gimple_return_set_retbnd): Likewise.
11686 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
11687 gimple_return_set_retbnd.
11688 * gimple-pretty-print.c (dump_gimple_return): Remove call to
11689 gimple_return_retbnd and adjust.
11690 * tree-inline.h (struct copy_body_data): Remove retbnd field.
11691 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
11692 Explicitly return NULL in a couple more cases. Move assertion
11693 on debug statements and remove unreachable code.
11694 (reset_debug_binding): Do not test id->retbnd.
11695 (expand_call_inline): Do not set it.
11697 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
11699 * configure.ac: Add --disable-gcov option.
11700 * configure: Regenerate.
11701 * Makefile.in: Honour @enable_gcov@.
11702 * doc/install.texi: Document --disable-gcov.
11704 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11706 * config/arm/arm-cpus.in (cortex-a76): New entry.
11707 (cortex-a76.cortex-a55): Likewise.
11708 * config/arm/arm-tables.opt: Regenerate.
11709 * config/arm/arm-tune.md: Likewise.
11710 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
11711 * doc/invoke.texi (ARM Options): Document cortex-a76 and
11712 cortex-a76.cortex-a55.
11714 2018-06-27 Tamar Christina <tamar.christina@arm.com>
11717 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
11719 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
11721 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
11723 (EPILOGUE_USES): Likewise.
11725 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
11727 * tree-inline.c (remap_location): New function extracted from...
11728 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
11729 (copy_phis_for_bb): ...here. Call remap_location.
11730 (copy_cfg_body): Adjust call to copy_edges_for_bb.
11732 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
11734 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
11735 unaligned vsx for 16B memset.
11737 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
11740 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
11741 ieee128_float_type_node to long_double_type_node unless
11742 TARGET_LONG_DOUBLE_128 is set.
11744 2018-06-26 David Malcolm <dmalcolm@redhat.com>
11746 * cfgloop.c (get_loop_location): Convert return type from
11747 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
11748 by implicit construction from rtx_insn *, and using
11749 dump_user_location_t::from_function_decl for the fallback case.
11750 * cfgloop.h (get_loop_location): Convert return type from
11751 location_t to dump_user_location_t.
11752 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
11753 dump_printf_loc to pass in a dump_location_t rather than a
11754 location_t, via the gimple stmt.
11755 * coverage.c (get_coverage_counts): Update calls to
11756 dump_printf_loc to pass in dump_location_t rather than a
11758 * doc/optinfo.texi (Dump types): Convert example of
11759 dump_printf_loc from taking "locus" to taking "insn". Update
11760 description of the "_loc" calls to cover dump_location_t.
11761 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
11763 (dump_user_location_t::dump_user_location_t): New constructors,
11764 from gimple *stmt and rtx_insn *.
11765 (dump_user_location_t::from_function_decl): New function.
11766 (dump_loc): Make static.
11767 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
11768 const dump_location_t &.
11769 (dump_generic_expr_loc): Delete.
11770 (dump_printf_loc): Convert param "loc" from location_t to
11771 const dump_location_t &.
11772 (selftest::test_impl_location): New function.
11773 (selftest::dumpfile_c_tests): New function.
11774 * dumpfile.h: Include "profile-count.h".
11775 (class dump_user_location_t): New class.
11776 (struct dump_impl_location_t): New struct.
11777 (class dump_location_t): New class.
11778 (dump_printf_loc): Convert 2nd param from source_location to
11779 const dump_location_t &.
11780 (dump_generic_expr_loc): Delete.
11781 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
11782 const dump_location_t &.
11783 * gimple-fold.c (fold_gimple_assign): Update call to
11784 dump_printf_loc to pass in a dump_location_t rather than a
11785 location_t, via the gimple stmt.
11786 (gimple_fold_call): Likewise.
11787 * gimple-loop-interchange.cc
11788 (loop_cand::analyze_iloop_reduction_var): Update for change to
11789 check_reduction_path.
11790 (tree_loop_interchange::interchange): Update for change to
11791 find_loop_location.
11792 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
11793 change in return-type of find_loop_location.
11794 (graphite_regenerate_ast_isl): Likewise.
11795 * graphite-optimize-isl.c (optimize_isl): Likewise.
11796 * graphite.c (graphite_transform_loops): Likewise.
11797 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
11798 pass in a dump_location_t rather than a location_t, via the
11800 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
11801 * ipa.c (walk_polymorphic_call_targets): Likewise.
11802 * loop-unroll.c (report_unroll): Convert "locus" param from
11803 location_t to dump_location_t.
11804 (decide_unrolling): Update for change to get_loop_location's
11806 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
11807 location_t to dump_user_location_t.
11808 (grid_find_single_omp_among_assignments_1): Updates calls to
11809 dump_printf_loc to pass in a dump_location_t rather than a
11810 location_t, via the gimple stmt.
11811 (grid_parallel_clauses_gridifiable): Convert "tloc" from
11812 location_t to dump_location_t. Updates calls to dump_printf_loc
11813 to pass in a dump_location_t rather than a location_t, via the
11815 (grid_inner_loop_gridifiable_p): Likewise.
11816 (grid_dist_follows_simple_pattern): Likewise.
11817 (grid_gfor_follows_tiling_pattern): Likewise.
11818 (grid_target_follows_gridifiable_pattern): Likewise.
11819 (grid_attempt_target_gridification): Convert initialization
11820 of local "grid" from memset to zero-initialization; FIXME: does
11821 this require C++11? Update call to dump_printf_loc to pass in a
11822 optinfo_location rather than a location_t, via the gimple stmt.
11823 * profile.c (read_profile_edge_counts): Updates call to
11824 dump_printf_loc to pass in a dump_location_t rather than a
11826 (compute_branch_probabilities): Likewise.
11827 * selftest-run-tests.c (selftest::run_tests): Call
11829 * selftest.h (dumpfile_c_tests): New decl.
11830 * tree-loop-distribution.c (pass_loop_distribution::execute):
11831 Update for change in return type of find_loop_location.
11832 * tree-parloops.c (parallelize_loops): Likewise.
11833 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
11834 "locus" from location_t to dump_user_location_t.
11835 (canonicalize_loop_induction_variables): Likewise.
11836 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
11837 for change in return type of find_loop_location.
11838 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
11839 to dump_printf_loc to pass in a dump_location_t rather than a
11840 location_t, via the stmt.
11841 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
11843 * tree-vect-loop-manip.c (find_loop_location): Convert return
11844 type from source_location to dump_user_location_t.
11845 (vect_do_peeling): Update for above change.
11846 (vect_loop_versioning): Update for change in type of
11848 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
11849 from location_t to dump_user_location_t.
11850 (vect_estimate_min_profitable_iters): Update for change in type
11852 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
11853 location_t to dump_location_t.
11854 (vect_slp_bb): Update for change in type of vect_location.
11855 * tree-vectorizer.c (vect_location): Convert from source_location
11856 to dump_user_location_t.
11857 (try_vectorize_loop_1): Update for change in vect_location's type.
11858 (vectorize_loops): Likewise.
11859 (increase_alignment): Likewise.
11860 * tree-vectorizer.h (vect_location): Convert from source_location
11861 to dump_user_location_t.
11862 (find_loop_location): Convert return type from source_location to
11863 dump_user_location_t.
11864 (check_reduction_path): Convert 1st param from location_t to
11865 dump_user_location_t.
11866 * value-prof.c (check_counter): Update call to dump_printf_loc to
11867 pass in a dump_user_location_t rather than a location_t; update
11868 call to error_at for change in type of "locus".
11869 (check_ic_target): Update call to dump_printf_loc to
11870 pass in a dump_user_location_t rather than a location_t, via the
11873 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
11875 * config/s390/s390.h (enum processor_flags): Do not use
11876 default tune parameter when -march was specified.
11878 2018-06-26 Jakub Jelinek <jakub@redhat.com>
11881 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
11882 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
11885 2018-06-26 Richard Biener <rguenther@suse.de>
11887 PR tree-optimization/86287
11889 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
11890 (vect_analyze_loop): Initialize n_stmts.
11892 2018-06-26 Richard Biener <rguenther@suse.de>
11894 PR middle-end/86271
11895 * fold-const.c (fold_convertible_p): Pointer extension
11898 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
11901 * dwarf2out.c (loc_list_has_views): Adjust comments.
11902 (dw_loc_list): Split single cross-partition range with
11905 2018-06-25 Jeff Law <law@redhat.com>
11907 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
11908 on -mbig-switch by default.
11910 * config/v850/predicates.md (const_float_1_operand): Fix match_code
11912 (const_float_0_operand): Remove unused predicate.
11913 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
11914 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
11915 (recipsf2): New expander. Original pattern now called
11917 (recipdf2, recipdf2_insn): Similarly.
11918 (rsqrtsf2, rsqrtsf2_insn): Similarly
11919 (rsqrtdf2, rsqrtdf2_insn): Similarly
11921 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
11923 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
11924 Simplify logic for FreeBSD (twice).
11926 2018-06-25 Martin Sebor <msebor@redhat.com>
11928 PR tree-optimization/86204
11929 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
11930 a strnlen result if it's less than the length of the string.
11932 2018-06-25 Martin Sebor <msebor@redhat.com>
11934 PR tree-optimization/85700
11935 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
11936 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
11937 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
11939 2018-06-25 Martin Sebor <msebor@redhat.com>
11941 * doc/extend.texi (Zero-length arrays): Update and clarify.
11943 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
11945 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
11946 added IEEE/IBM long double multilib support on PowerPC little
11947 endian Linux systems.
11948 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
11949 (MULTILIB_DEFAULTS): Likewise.
11950 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11952 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
11953 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
11954 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
11956 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
11958 PR middle-end/86311
11959 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
11960 (REORDER_45): Likewise.
11962 2018-06-25 Jeff Law <law@redhat.com>
11964 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
11965 dividend to 32 bits. Adjust length.
11966 (udivmodhi4): Cleanup output template. Fix length.
11968 2018-06-25 Carl Love <cel@us.ibm.com>
11970 * config/rs6000/vsx.md: Change word selector to prefered location.
11972 2018-06-25 Richard Biener <rguenther@suse.de>
11974 PR tree-optimization/86304
11975 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
11976 epilogue-if-converted loops as well.
11978 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
11980 * lto-section-out.c (lto_begin_section): Do not print section
11981 name for noaddr and unnumbered dumps.
11983 2018-06-25 Richard Biener <rguenther@suse.de>
11985 * tree-vectorizer.h (struct vec_info_shared): New structure
11986 with parts split out from struct vec_info and loop_nest from
11987 struct _loop_vec_info.
11988 (struct vec_info): Adjust accordingly.
11989 (struct _loop_vec_info): Likewise.
11990 (LOOP_VINFO_LOOP_NEST): Adjust.
11991 (LOOP_VINFO_DATAREFS): Likewise.
11992 (LOOP_VINFO_DDRS): Likewise.
11993 (struct _bb_vec_info): Likewise.
11994 (BB_VINFO_DATAREFS): Likewise.
11995 (BB_VINFO_DDRS): Likewise.
11996 (struct _stmt_vec_info): Add dr_aux member.
11997 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
11998 (DR_MISALIGNMENT_UNINITIALIZED): New.
11999 (set_dr_misalignment): Adjust.
12000 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
12001 (vect_analyze_loop): Adjust prototype.
12002 (vect_analyze_loop_form): Likewise.
12003 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
12004 Compute dependences lazily.
12005 (vect_record_base_alignments): Use shared datarefs/ddrs.
12006 (vect_verify_datarefs_alignment): Likewise.
12007 (vect_analyze_data_refs_alignment): Likewise.
12008 (vect_analyze_data_ref_accesses): Likewise.
12009 (vect_analyze_data_refs): Likewise.
12010 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
12011 constructor parameter for shared part.
12012 (vect_analyze_loop_form): Pass in shared part and adjust.
12013 (vect_analyze_loop_2): Pass in storage for the number of
12014 stmts. Move loop nest finding to the caller. Compute
12016 (vect_analyze_loop): Pass in shared part.
12017 (vect_transform_loop): Verify shared datarefs are unchanged.
12018 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
12019 constructor parameter for shared part.
12020 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
12021 (vect_slp_bb): Verify shared datarefs are unchanged before
12023 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
12025 (new_stmt_vec_info): Initialize DR_AUX misalignment to
12026 DR_MISALIGNMENT_UNINITIALIZED.
12027 * tree-vectorizer.c (vec_info::vec_info): Add constructor
12028 parameter for shared part.
12029 (vec_info::~vec_info): Adjust.
12030 (vec_info_shared::vec_info_shared): New.
12031 (vec_info_shared::~vec_info_shared): Likewise.
12032 (vec_info_shared::save_datarefs): Likewise.
12033 (vec_info_shared::check_datarefs): Likewise.
12034 (try_vectorize_loop_1): Construct shared part live for analyses
12035 of a single loop for multiple vector sizes.
12036 * tree-parloops.c (gather_scalar_reductions): Adjust.
12038 2018-06-25 Richard Biener <rguenther@suse.de>
12040 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
12041 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
12042 (vect_analyze_data_refs): Remove similar code from here and
12043 simplify accordingly.
12045 2018-06-25 Richard Biener <rguenther@suse.de>
12047 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
12048 for reverse storage order accesses rather than asserting
12049 they cannot happen here.
12051 2018-06-25 Tom de Vries <tdevries@suse.de>
12054 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
12055 Use data16 instead of .byte for insn prefix.
12057 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
12060 * parser.c (make_char_string_pack): Pass this literal chars
12061 through cpp_interpret_string.
12062 (cp_parser_userdef_numeric_literal): Check the result of
12063 make_char_string_pack.
12065 2018-06-24 Maya Rashish <coypu@sdf.org>
12067 * ginclude/stddef.h: Simplify conditions around avoiding
12068 re-definition of __size_t.
12070 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12072 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
12073 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
12075 2018-06-22 Maya Rashish <coypu@sdf.org>
12077 * doc/invoke.texi (mno-fancy-math-387): Update for changes
12078 made to OpenBSD and NetBSD through the years.
12080 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
12082 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
12083 behavior of vec_pack (vector double, vector double) to match
12084 behavior of vec_float2 (vector double, vector double).
12086 2018-06-22 Olivier Hainque <hainque@adacore.com>
12088 * gimplify.c (gimplify_function_tree): Prevent creation
12089 of a trampoline for the address of the current function
12090 passed to entry/exit instrumentation hooks.
12092 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
12095 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
12098 2018-06-22 Martin Liska <mliska@suse.cz>
12100 PR tree-optimization/86263
12101 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
12102 Bail out if is_enabled is false.
12103 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
12105 (jump_table_cluster::is_enabled): New function.
12107 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12109 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
12110 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
12111 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
12112 (lto_input_ts_binfo_tree_pointers): Likewise.
12113 * tree-streamer-out.c (streamer_write_tree_bitfields,
12114 write_ts_binfo_tree_pointers): Likewise.
12115 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
12117 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12119 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
12121 2018-06-22 Martin Liska <mliska@suse.cz>
12123 * symbol-summary.h (get): Make it pure and inline move
12124 functionality from ::get function.
12125 (get): Remove and inline into ::get and ::get_create.
12126 (get_create): Move code from ::get function.
12128 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12131 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
12132 -x assembler-with-cpp.
12134 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12136 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
12137 _FILE_OFFSET_BITS=64 for C++.
12139 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
12141 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
12142 conversion insn that shows up when pr85657-3.c is compiled using
12143 IEEE 128-bit long double.
12144 (neg<mode>2_internal): Use the correct mode to check whether the
12145 mode is IBM extended.
12146 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
12147 multiply and divide external functions from being created more
12150 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12152 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
12154 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
12155 the edge can be forwarded.
12156 (cfg_layout_merge_blocks): Likewise.
12158 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12160 * except.c (finish_eh_generation): Commit edge insertions only after
12161 the EH edges have been redirected from post-landing to landing pads.
12163 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12165 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
12166 create_tmp_var_for to create the FRAME decl.
12167 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
12169 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12171 * tree-inline.c (copy_edges_for_bb): Minor tweak.
12172 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
12173 debug statement when resetting its value.
12174 (expand_call_inline): Copy the locus of the call onto the assignment
12175 of the return value, if any. Use local variable in more cases.
12177 2018-06-21 Martin Liska <mliska@suse.cz>
12179 * ipa-pure-const.c (propagate_nothrow): Use
12180 funct_state_summaries->get.
12181 (dump_malloc_lattice): Likewise.
12182 (propagate_malloc): Likewise.
12184 2018-06-21 Richard Biener <rguenther@suse.de>
12186 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
12187 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
12188 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
12190 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
12191 BLOCK_ABSTRACT_ORIGIN unconditionally.
12193 2018-06-21 David Malcolm <dmalcolm@redhat.com>
12195 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
12197 * ipa-reference.c (ipa_reference_c_finalize): Delete
12198 ipa_ref_opt_sum_summaries and set it to NULL.
12200 2018-06-21 Tom de Vries <tdevries@suse.de>
12202 PR tree-optimization/85859
12203 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
12204 test with comment from bb_no_side_effects_p.
12206 2018-06-21 Richard Biener <rguenther@suse.de>
12208 PR tree-optimization/86232
12209 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
12210 max for constant niter.
12212 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
12214 * config/aarch64/aarch64-simd.md
12215 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
12217 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
12219 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
12220 Make opernads of the unspec commutative.
12222 2018-06-21 Richard Biener <rguenther@suse.de>
12224 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
12225 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
12226 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
12227 (vect_analyze_data_ref_dependence): Re-order checks to deal with
12229 (vect_record_base_alignments): Do not record base alignment
12230 for gathers or scatters.
12231 (vect_compute_data_ref_alignment): Drop return value that is always
12232 true. Bail out early for gathers or scatters.
12233 (vect_enhance_data_refs_alignment): Bail out early for gathers
12235 (vect_find_same_alignment_drs): Likewise.
12236 (vect_analyze_data_refs_alignment): Remove dead code.
12237 (vect_slp_analyze_and_verify_node_alignment): Likewise.
12238 (vect_analyze_data_refs): For possible gathers or scatters do
12239 not create an alternate DR, just check their possible validity
12240 and mark them. Adjust DECL_NONALIASED handling to not rely
12241 on DR_BASE_ADDRESS.
12242 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
12243 update inits of gathers or scatters.
12244 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
12245 Also copy gather/scatter flag to pattern vinfo.
12247 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
12249 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
12250 behavior of vec_packsu (vector unsigned long long, vector unsigned
12251 long long) to match behavior of vec_packs with same signature.
12253 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
12254 Thomas Schwinge <thomas@codesourcery.com>
12255 Cesar Philippidis <cesar@codesourcery.com>
12257 * gimplify.c (gimplify_scan_omp_clauses): Add support for
12258 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
12259 (gimplify_adjust_omp_clauses): Likewise.
12260 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
12261 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
12262 (gimplify_omp_target_update): Update handling of acc update and
12264 * omp-low.c (install_var_field): Remove unused parameter
12265 base_pointers_restrict.
12266 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
12267 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
12269 (omp_target_base_pointers_restrict_p): Delete.
12270 (scan_omp_target): Update call to scan_sharing_clauses.
12271 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
12273 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
12274 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
12275 (convert_local_omp_clauses): Likewise.
12276 * tree-pretty-print.c (dump_omp_clause): Likewise.
12277 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
12279 (omp_clause_code_name): Likewise.
12281 2018-06-20 Jakub Jelinek <jakub@redhat.com>
12284 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
12287 PR tree-optimization/86231
12288 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
12289 anti-range don't overwrite *vr0min before using it to compute *vr0max.
12291 2018-06-20 Tom de Vries <tdevries@suse.de>
12293 PR tree-optimization/86097
12294 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
12295 iv type if signedness of iv type is not the same as that of *nit.
12297 2018-06-20 Jakub Jelinek <jakub@redhat.com>
12299 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
12300 EDGE_EH edges, verify they are all EDGE_EH.
12302 2018-06-20 Maya Rashish <coypu@sdf.org>
12304 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
12306 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12308 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
12309 * config/aarch64/aarch64.c (xgene1_tunings): Add
12310 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
12311 (aarch64_mode_valid_for_sched_fusion_p):
12312 Allow 16-byte modes.
12313 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
12314 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
12316 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
12318 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
12319 * config/aarch64/iterators.md (VQ2): New mode iterator.
12321 2018-06-20 Martin Liska <mliska@suse.cz>
12323 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
12324 Change default ratio from 10 to 8.
12326 2018-06-20 Martin Liska <mliska@suse.cz>
12328 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
12330 (bit_test_cluster::find_bit_tests): Likewise.
12331 (switch_decision_tree::analyze_switch_statement): Find clusters.
12332 * tree-switch-conversion.h (struct jump_table_cluster): Document
12335 2018-06-20 Martin Liska <mliska@suse.cz>
12337 * tree-switch-conversion.c (switch_conversion::collect):
12338 Record m_uniq property.
12339 (switch_conversion::expand): Bail out for special conditions.
12340 (group_cluster::~group_cluster): New.
12341 (group_cluster::group_cluster): Likewise.
12342 (group_cluster::dump): Likewise.
12343 (jump_table_cluster::emit): New.
12344 (switch_decision_tree::fix_phi_operands_for_edges): New.
12345 (struct case_node): Remove struct.
12346 (jump_table_cluster::can_be_handled): New.
12347 (case_values_threshold): Moved to header.
12348 (reset_out_edges_aux): Likewise.
12349 (jump_table_cluster::is_beneficial): New.
12350 (bit_test_cluster::can_be_handled): Likewise.
12351 (add_case_node): Remove.
12352 (bit_test_cluster::is_beneficial): New.
12353 (case_bit_test::cmp): New.
12354 (bit_test_cluster::emit): New.
12355 (expand_switch_as_decision_tree_p): Remove.
12356 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
12357 (fix_phi_operands_for_edge): Likewise.
12358 (switch_decision_tree::analyze_switch_statement): New.
12359 (compute_cases_per_edge): Move ...
12360 (switch_decision_tree::compute_cases_per_edge): ... here.
12361 (try_switch_expansion): Likewise.
12362 (switch_decision_tree::try_switch_expansion): Likewise.
12363 (record_phi_operand_mapping): Likewise.
12364 (switch_decision_tree::record_phi_operand_mapping): Likewise.
12365 (emit_case_decision_tree): Likewise.
12366 (switch_decision_tree::emit): Likewise.
12367 (balance_case_nodes): Likewise.
12368 (switch_decision_tree::balance_case_nodes): Likewise.
12369 (dump_case_nodes): Likewise.
12370 (switch_decision_tree::dump_case_nodes): Likewise.
12371 (emit_jump): Likewise.
12372 (switch_decision_tree::emit_jump): Likewise.
12373 (emit_cmp_and_jump_insns): Likewise.
12374 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
12375 (emit_case_nodes): Likewise.
12376 (switch_decision_tree::emit_case_nodes): Likewise.
12377 (conditional_probability): Remove.
12378 * tree-switch-conversion.h (enum cluster_type): New.
12380 (struct cluster): Likewise.
12381 (cluster::cluster): Likewise.
12382 (struct simple_cluster): Likewise.
12383 (simple_cluster::simple_cluster): Likewise.
12384 (struct group_cluster): Likewise.
12385 (struct jump_table_cluster): Likewise.
12386 (struct bit_test_cluster): Likewise.
12387 (struct min_cluster_item): Likewise.
12388 (struct case_tree_node): Likewise.
12389 (case_tree_node::case_tree_node): Likewise.
12390 (jump_table_cluster::case_values_threshold): Likewise.
12391 (struct case_bit_test): Likewise.
12392 (struct switch_decision_tree): Likewise.
12393 (struct switch_conversion): Likewise.
12394 (switch_decision_tree::reset_out_edges_aux): Likewise.
12396 2018-06-20 Martin Liska <mliska@suse.cz>
12398 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
12399 (hoist_edge_and_branch_if_true): Likewise.
12400 (expand_switch_using_bit_tests_p): Likewise.
12401 (struct case_bit_test): Likewise.
12402 (case_bit_test_cmp): Likewise.
12403 (emit_case_bit_tests): Likewise.
12404 (switch_conversion::switch_conversion): New class.
12405 (struct switch_conv_info): Remove old struct.
12406 (collect_switch_conv_info): More to ...
12407 (switch_conversion::collect): ... this.
12408 (check_range): Likewise.
12409 (switch_conversion::check_range): Likewise.
12410 (check_all_empty_except_final): Likewise.
12411 (switch_conversion::check_all_empty_except_final): Likewise.
12412 (check_final_bb): Likewise.
12413 (switch_conversion::check_final_bb): Likewise.
12414 (create_temp_arrays): Likewise.
12415 (switch_conversion::create_temp_arrays): Likewise.
12416 (free_temp_arrays): Likewise.
12417 (gather_default_values): Likewise.
12418 (switch_conversion::gather_default_values): Likewise.
12419 (build_constructors): Likewise.
12420 (switch_conversion::build_constructors): Likewise.
12421 (constructor_contains_same_values_p): Likewise.
12422 (switch_conversion::contains_same_values_p): Likewise.
12423 (array_value_type): Likewise.
12424 (switch_conversion::array_value_type): Likewise.
12425 (build_one_array): Likewise.
12426 (switch_conversion::build_one_array): Likewise.
12427 (build_arrays): Likewise.
12428 (switch_conversion::build_arrays): Likewise.
12429 (gen_def_assigns): Likewise.
12430 (switch_conversion::gen_def_assigns): Likewise.
12431 (prune_bbs): Likewise.
12432 (switch_conversion::prune_bbs): Likewise.
12433 (fix_phi_nodes): Likewise.
12434 (switch_conversion::fix_phi_nodes): Likewise.
12435 (gen_inbound_check): Likewise.
12436 (switch_conversion::gen_inbound_check): Likewise.
12437 (process_switch): Use the newly created class.
12438 (switch_conversion::expand): New.
12439 (switch_conversion::~switch_conversion): New.
12440 * tree-switch-conversion.h: New file.
12442 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12444 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
12445 tree-vect-patterns.c.
12446 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
12447 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
12448 (vect_recog_sad_pattern): Likewise.
12449 (vect_recog_widen_sum_pattern): Likewise.
12450 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
12451 (vect_recog_widen_shift_pattern): Remove the type_in argument.
12452 (vect_recog_rotate_pattern): Likewise.
12453 (vect_recog_mult_pattern): Likewise.
12454 (vect_recog_vector_vector_shift_pattern): Likewise.
12455 (vect_recog_divmod_pattern): Likewise.
12456 (vect_recog_mixed_size_cond_pattern): Likewise.
12457 (vect_recog_bool_pattern): Likewise.
12458 (vect_recog_mask_conversion_pattern): Likewise.
12459 (vect_try_gather_scatter_pattern): Likewise.
12460 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
12461 (vect_recog_over_widening_pattern): Likewise.
12462 (vect_recog_gather_scatter_pattern): Likewise.
12463 (vect_recog_func_ptr): Move from tree-vectorizer.h
12464 (vect_vect_recog_func_ptrs): Move further down the file.
12465 (vect_recog_func): Likewise. Remove the third argument.
12466 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
12467 (vect_pattern_recog_1): Expect the pattern function to do any
12468 necessary target tests. Also expect it to provide a vector type.
12469 Remove the type_in handling.
12471 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12473 * tree-vect-patterns.c (vect_pattern_detected): New function.
12474 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
12475 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
12476 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
12477 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
12478 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
12479 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
12480 (vect_recog_mask_conversion_pattern)
12481 (vect_try_gather_scatter_pattern): Likewise.
12483 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12485 * tree-vect-patterns.c (vect_get_internal_def): New function.
12486 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
12487 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
12488 (search_type_for_mask_1): Use it.
12490 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12492 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
12493 redundant WIDEN_SUM_EXPR handling.
12494 (vect_recog_sad_pattern): Likewise.
12496 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12498 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
12499 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
12500 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
12501 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
12502 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
12504 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12506 * tree-vect-stmts.c (vectorizable_call): Make sure that we
12507 use the stmt_vec_info of the original bb statement for the
12508 new zero assignment, even if the call is part of a pattern.
12510 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12512 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
12513 that the sequence is attached to the original statement rather
12514 than the pattern statement.
12515 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
12516 PATTERN_DEF_SEQ from the original statement rather than
12517 the main pattern statement.
12518 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
12519 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
12520 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
12522 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12524 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
12525 definition statements before the early exit for statements that aren't
12527 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
12529 (vect_transform_loop): ...here. Process pattern definition
12530 statements without first checking whether the main pattern
12531 statement is live or relevant.
12533 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
12535 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
12536 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
12538 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
12540 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
12541 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
12542 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
12543 (expand_block_compare): Change select_block_compare_mode call.
12544 (expand_strncmp_align_check): Use new functions, fix comment.
12545 (emit_final_str_compare_gpr): New function.
12546 (expand_strn_compare): Refactor and clean up code.
12547 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
12549 2018-06-19 Tony Reix <tony.reix@atos.com>
12550 Damien Bergamini <damien.bergamini@atos.com>
12551 David Edelsohn <dje.gcc@gmail.com>
12553 * collect2.c (static_obj): New variable.
12554 (static_libs): New variable.
12555 (is_in_list): Uncomment declaration.
12556 (main): Track AIX libraries linked statically.
12557 (is_in_list): Uncomment definition.
12558 (scan_prog_file): Don't add AIX shared libraries initializer
12559 to constructor list if linking statically.
12561 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
12563 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
12565 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
12567 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
12569 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
12572 2018-06-19 Martin Liska <mliska@suse.cz>
12574 * config/i386/i386.c (ix86_can_inline_p): Do not use
12575 ipa_fn_summaries::get_create.
12576 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
12578 (devirtualization_time_bonus): Likewise.
12579 (ipcp_propagate_stage): Likewise.
12580 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
12581 (edge_set_predicate): Likewise.
12582 (evaluate_conditions_for_known_args): Likewise.
12583 (evaluate_properties_for_edge): Likewise.
12584 (ipa_call_summary::reset): Tranform to ...
12585 (ipa_call_summary::~ipa_call_summary): ... this.
12586 (ipa_fn_summary::reset): Transform to ...
12587 (ipa_fn_summary::~ipa_fn_summary): ... this.
12588 (ipa_fn_summary_t::remove): Rename to ...
12589 (ipa_fn_summary_t::remove_callees): ... this.
12590 (ipa_fn_summary_t::duplicate): Use placement new
12591 instead of memory copy.
12592 (ipa_call_summary_t::duplicate): Likewise.
12593 (ipa_call_summary_t::remove): Remove.
12594 (dump_ipa_call_summary): Change get_create to get.
12595 (ipa_dump_fn_summary): Dump only when summary exists.
12596 (analyze_function_body): Use symbol_summary::get instead
12598 (compute_fn_summary): Likewise.
12599 (estimate_edge_devirt_benefit): Likewise.
12600 (estimate_edge_size_and_time): Likewise.
12601 (inline_update_callee_summaries): Likewise.
12602 (remap_edge_change_prob): Likewise.
12603 (remap_edge_summaries): Likewise.
12604 (ipa_merge_fn_summary_after_inlining): Likewise.
12605 (write_ipa_call_summary): Likewise.
12606 (ipa_fn_summary_write): Likewise.
12607 (ipa_free_fn_summary): Likewise.
12608 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
12609 (struct ipa_call_summary): Likewise.
12610 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
12612 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
12613 (estimate_size_after_inlining): Likewise.
12614 (estimate_growth): Likewise.
12615 (growth_likely_positive): Likewise.
12616 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
12617 (inline_call): Likewise.
12618 * ipa-inline.c (caller_growth_limits): Likewise.
12619 (can_inline_edge_p): Likewise.
12620 (can_inline_edge_by_limits_p): Likewise.
12621 (compute_uninlined_call_time): Likewise.
12622 (compute_inlined_call_time): Likewise.
12623 (want_inline_small_function_p): Likewise.
12624 (edge_badness): Likewise.
12625 (update_caller_keys): Likewise.
12626 (update_callee_keys): Likewise.
12627 (inline_small_functions): Likewise.
12628 (inline_to_all_callers_1): Likewise.
12629 (dump_overall_stats): Likewise.
12630 (early_inline_small_functions): Likewise.
12631 (early_inliner): Likewise.
12632 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12633 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
12634 * ipa-pure-const.c (malloc_candidate_p): Likewise.
12635 * ipa-split.c (execute_split_functions): Likewise.
12636 * symbol-summary.h: Likewise.
12637 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
12639 2018-06-19 Richard Biener <rguenther@suse.de>
12641 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
12642 (vectorize_loops): ... here. Fix dbgcnt handling.
12643 (try_vectorize_loop): Wrap try_vectorize_loop_1.
12645 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
12648 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
12649 ieee128 argument takes up only one (vector) register, not two (floating
12652 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
12654 * gimplify.c (gimplify_init_constructor): Really never clear for an
12655 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
12657 2018-06-19 Richard Biener <rguenther@suse.de>
12659 PR tree-optimization/86179
12660 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
12661 after failed recognition.
12663 2018-06-18 Martin Sebor <msebor@redhat.com>
12665 PR middle-end/85602
12666 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
12667 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
12668 Handle integer subtraction.
12669 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
12670 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
12672 2018-06-18 David Malcolm <dmalcolm@redhat.com>
12674 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
12675 param from rtx to rtx_insn *.
12676 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
12678 (frv_ifcvt_modify_insn): Likwise.
12679 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
12680 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
12681 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
12682 as_a <rtx_insn *> cast to local "unprotected_region" once
12683 it's been established that it's not NULL or pc_rtx.
12684 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
12685 param "sethi" from rtx to rtx_insn *.
12686 (nds32_group_float_insns): Likewise for param "insn".
12687 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
12689 (vax_output_int_subtract): Likewise.
12690 * config/vax/vax.c (vax_output_int_add): Likewise for param
12692 (vax_output_int_subtract): Likewise.
12693 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
12694 (emit_pattern_after): Likewise for param "after".
12695 (emit_insn_after): Likewise.
12696 (emit_jump_insn_after): Likewise.
12697 (emit_call_insn_after): Likewise.
12698 (emit_debug_insn_after): Likewise.
12699 (emit_pattern_before): Likewise for param "before".
12700 (emit_insn_before): Likewise.
12701 (emit_jump_insn_before): Likewise.
12702 * final.c (get_insn_template): Likewise for param "insn", removing
12704 * output.h (get_insn_template): Likewise for 2nd param.
12705 * rtl.h (emit_insn_before): Likewise.
12706 (emit_jump_insn_before): Likewise.
12707 (emit_debug_insn_before_noloc): Likewise.
12708 (emit_insn_after): Likewise.
12709 (emit_jump_insn_after): Likewise.
12710 (emit_call_insn_after): Likewise.
12711 (emit_debug_insn_after): Likewise.
12712 (set_insn_deleted): Likewise for param.
12714 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
12717 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
12718 floating point modes, so that IFmode is numerically greater than
12719 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
12720 to declare the ordering. This prevents IFmode from being
12721 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
12722 machine. Include rs6000-modes.h to share the fractional values
12723 between genmodes* and the rest of the compiler.
12724 (IFmode): Likewise.
12725 (KFmode): Likewise.
12726 (TFmode): Likewise.
12727 * config/rs6000/rs6000-modes.h: New file.
12728 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
12729 meaning of rs6000_long_double_size so that 126..128 selects an
12730 appropriate 128-bit floating point type.
12731 (rs6000_option_override_internal): Likewise.
12732 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
12733 (TARGET_LONG_DOUBLE_128): Change the meaning of
12734 rs6000_long_double_size so that 126..128 selects an appropriate
12735 128-bit floating point type.
12736 (LONG_DOUBLE_TYPE_SIZE): Update comment.
12737 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
12738 source and destination to match the standard usage.
12739 (truncifkf2): Likewise.
12740 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
12741 ISA 2.07 to use an explicit clobber, instead of passing in a
12743 (copysign<mode>3_soft): Likewise.
12745 2018-06-18 David Malcolm <dmalcolm@redhat.com>
12747 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
12748 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
12749 (vect_slp_analyze_instance_dependence): Likewise.
12750 (vect_enhance_data_refs_alignment): Likewise.
12751 (vect_analyze_data_refs_alignment): Likewise.
12752 (vect_slp_analyze_and_verify_instance_alignment
12753 (vect_analyze_data_ref_accesses): Likewise.
12754 (vect_prune_runtime_alias_test_list): Likewise.
12755 (vect_analyze_data_refs): Likewise.
12756 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
12757 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
12758 (vect_analyze_scalar_cycles_1): Likewise.
12759 (vect_get_loop_niters): Likewise.
12760 (vect_analyze_loop_form_1): Likewise.
12761 (vect_update_vf_for_slp): Likewise.
12762 (vect_analyze_loop_operations): Likewise.
12763 (vect_analyze_loop): Likewise.
12764 (vectorizable_induction): Likewise.
12765 (vect_transform_loop): Likewise.
12766 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
12767 * tree-vect-slp.c (vect_analyze_slp): Likewise.
12768 (vect_make_slp_decision): Likewise.
12769 (vect_detect_hybrid_slp): Likewise.
12770 (vect_slp_analyze_operations): Likewise.
12771 (vect_slp_bb): Likewise.
12772 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
12773 (vectorizable_bswap): Likewise.
12774 (vectorizable_call): Likewise.
12775 (vectorizable_simd_clone_call): Likewise.
12776 (vectorizable_conversion): Likewise.
12777 (vectorizable_assignment): Likewise.
12778 (vectorizable_shift): Likewise.
12779 (vectorizable_operation): Likewise.
12780 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
12782 2018-06-18 Martin Sebor <msebor@redhat.com>
12784 PR tree-optimization/81384
12785 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
12786 * builtins.c (expand_builtin_strnlen): New function.
12787 (expand_builtin): Call it.
12788 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
12789 * builtins.def (BUILT_IN_STRNLEN): New.
12790 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
12791 Warn for bounds in excess of maximum object size.
12792 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
12793 single-value ranges. Handle strnlen.
12794 (handle_builtin_strlen): Handle strnlen.
12795 (strlen_check_and_optimize_stmt): Same.
12796 * doc/extend.texi (Other Builtins): Document strnlen.
12798 2018-06-18 Maya Rashish <coypu@sdf.org>
12800 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
12801 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
12802 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
12804 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
12806 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
12808 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12810 * tree.c (escaped_string::escape): Replace cast to char * by
12811 const_cast<char *> (unescaped).
12813 2018-06-18 Nick Clifton <nickc@redhat.com>
12816 * tree.c (escaped_string): New class. Converts an unescaped
12817 string into its escaped equivalent.
12818 (warn_deprecated_use): Use the new class to convert the
12819 deprecation message, if present.
12820 (test_escaped_strings): New self test.
12821 (test_c_tests): Add test_escaped_strings.
12822 * doc/extend.texi (deprecated): Add a note that the
12823 deprecation message is affected by the -fmessage-length
12824 option, and that control characters will be escaped.
12825 (#pragma GCC error): Document this pragma.
12826 (#pragma GCC warning): Likewise.
12827 * doc/invoke.texi (-fmessage-length): Document this option's
12828 effect on the #warning and #error preprocessor directives and
12829 the deprecated attribute.
12831 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
12833 * tree.c (decl_value_expr_lookup): Revert latest change.
12834 (decl_value_expr_insert): Likewise.
12836 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
12838 * gimplify.c (nonlocal_vlas): Delete.
12839 (nonlocal_vla_vars): Likewise.
12840 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
12842 (gimplify_body): Do not create and destroy nonlocal_vlas.
12843 * tree-nested.c: Include diagnostic.h.
12844 (use_pointer_in_frame): Tweak.
12845 (lookup_field_for_decl): Add assertion and declare the transformation.
12846 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
12847 internal error when the reference is in a wrong context. Do not
12848 create a debug decl by default.
12849 (note_nonlocal_block_vlas): Delete.
12850 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
12851 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
12852 create a debug decl by default.
12853 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
12854 call is in a wrong context.
12855 (fixup_vla_decls): New function.
12856 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
12857 debug variables were created.
12858 * tree.c (decl_value_expr_lookup): Add checking assertion.
12859 (decl_value_expr_insert): Likewise.
12861 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
12863 PR middle-end/82479
12864 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
12865 * tree-scalar-evolution.c (interpret_expr): Likewise.
12866 (expression_expensive_p): Likewise.
12867 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
12868 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
12869 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
12870 (ssa_defined_by_minus_one_stmt_p): New.
12872 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
12874 PR middle-end/64946
12875 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
12876 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
12877 * dojump.c (do_jump): Likewise.
12878 * expr.c (expand_expr_real_2): Check operand type's sign.
12879 * fold-const.c (const_unop): Handle ABSU_EXPR.
12880 (fold_abs_const): Likewise.
12881 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
12882 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
12883 (strip_sign_op_1): Likesise.
12884 * match.pd: Add new pattern to generate ABSU_EXPR.
12885 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
12886 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
12887 * tree-eh.c (operation_could_trap_helper_p): Likewise.
12888 * tree-inline.c (estimate_operator_cost): Likewise.
12889 * tree-pretty-print.c (dump_generic_node): Likewise.
12890 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
12891 * tree.def (ABSU_EXPR): New.
12893 2018-06-16 Jakub Jelinek <jakub@redhat.com>
12895 PR middle-end/86095
12896 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
12897 documented as preserved for backward compatibility only.
12898 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
12900 PR rtl-optimization/86108
12901 * bb-reorder.c (create_forwarder_block): Renamed to ...
12902 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
12903 jump from new landing pad to the second part.
12904 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
12907 2018-06-15 Jakub Jelinek <jakub@redhat.com>
12909 PR middle-end/85878
12910 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
12911 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
12912 Only call store_expr for halves if the mode is the same.
12914 PR middle-end/86123
12915 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
12916 Fix up comment formatting.
12918 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
12920 * typed-splay-tree.h (typed_splay_tree::remove): New function.
12921 (typed_splay_tree::closure,
12922 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
12923 (typed_splay_tree::typed_splay_tree,
12924 typed_splay_tree::operator =): Declared private.
12925 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
12926 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
12927 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
12928 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
12929 typed_splay_tree::splay_tree_splay,
12930 typed_splay_tree::splay_tree_foreach_helper,
12931 typed_splay_tree::splay_tree_insert,
12932 typed_splay_tree::splay_tree_remove,
12933 typed_splay_tree::splay_tree_lookup,
12934 typed_splay_tree::splay_tree_predecessor,
12935 typed_splay_tree::splay_tree_successor,
12936 typed_splay_tree::splay_tree_min,
12937 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
12938 (typed_splay_tree::root, typed_splay_tree::comp,
12939 typed_splay_tree::delete_key,
12940 typed_splay_tree::delete_value): New data members.
12941 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
12942 typed_splay_tree::remove.
12944 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
12946 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
12947 -mginv and -mno-ginv to the assembler.
12948 * config/mips/mips.opt (-mcrc): New option.
12949 (-mginv): Likewise.
12950 * doc/invoke.text (-mcrc): Document.
12951 (-mginv): Likewise.
12953 2018-06-15 Nick Clifton <nickc@redhat.com>
12956 * tree.c (escaped_string): New class. Converts an unescaped
12957 string into its escaped equivalent.
12958 (warn_deprecated_use): Use the new class to convert the
12959 deprecation message, if present.
12960 (test_escaped_strings): New self test.
12961 (test_c_tests): Add test_escaped_strings.
12962 * doc/extend.texi (deprecated): Add a note that the
12963 deprecation message is affected by the -fmessage-length
12964 option, and that control characters will be escaped.
12965 (#pragma GCC error): Document this pragma.
12966 (#pragma GCC warning): Likewise.
12967 * doc/invoke.texi (-fmessage-length): Document this option's
12968 effect on the #warning and #error preprocessor directives and
12969 the deprecated attribute.
12971 2018-06-15 Richard Biener <rguenther@suse.de>
12973 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
12974 here, also noting vector size used.
12975 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
12976 size used in MSG_OPTIMIZED_LOCATIONS dump.
12977 (pass_slp_vectorize::execute): Adjust.
12979 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
12982 * config/arc/arc.c (arc_return_address_register): Fix
12985 2018-06-15 Richard Biener <rguenther@suse.de>
12987 PR middle-end/86159
12988 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
12989 leave useless conversion stripping to force_gimple_operand_gsi.
12990 (gimplify_build2): Likewise.
12991 (gimplify_build1): Likewise.
12993 2018-06-15 Richard Biener <rguenther@suse.de>
12995 PR middle-end/86076
12996 * tree-cfg.c (move_stmt_op): unshare invariant addresses
12997 before adjusting their block.
12999 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
13001 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
13002 multilibs for *-*-rtems*.
13003 * config/riscv/t-rtems: New file.
13005 2018-06-14 Jakub Jelinek <jakub@redhat.com>
13007 PR middle-end/86122
13008 * match.pd ((A +- CST1) +- CST2): Punt if last resort
13009 unsigned_type_for returns NULL.
13012 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
13013 subregs of multi-word pseudos unless the float mode has word size.
13015 2018-06-14 Richard Biener <rguenther@suse.de>
13017 PR middle-end/86139
13018 * tree-vect-generic.c (build_word_mode_vector_type): Remove
13019 duplicate and harmful type_hash_canon.
13020 * tree.c (type_hash_canon): Assert we didn't find ourselves.
13022 2018-06-14 Richard Biener <rguenther@suse.de>
13025 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
13028 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
13030 * config/rtems.h (STDINT_LONG32): Define.
13032 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
13033 Prachi Godbole <prachi.godbole@imgtec.com>
13035 * config/mips/mips-cpus.def: Define P6600.
13036 * config/mips/mips-tables.opt: Regenerate.
13037 * config/mips/mips.c (mips_ucbranch_type): New enum.
13038 (mips_rtx_cost_data): Add support for P6600.
13039 (mips_issue_rate): Likewise.
13040 (mips_multipass_dfa_lookahead): Likewise.
13041 (mips_avoid_hazard): Likewise.
13042 (mips_reorg_process_insns): Likewise.
13043 (mips_classify_branch_p6600): New function.
13044 * config/mips/mips.h (TUNE_P6600): New define.
13045 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
13046 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
13047 * config/mips/mips.md: Include p6600.md.
13048 (processor): Add p6600.
13049 * config/mips/p6600.md: New file.
13050 * doc/invoke.texi: Add p6600 to supported architectures.
13052 2018-06-13 Martin Sebor <msebor@redhat.com>
13054 PR tree-optimization/86114
13055 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
13057 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
13059 2018-06-13 Richard Biener <rguenther@suse.de>
13061 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
13062 Properly set vector type of the intermediate stmt.
13063 * tree-vect-stmts.c (vectorizable_operation): The destination
13064 var always has vectype_out type.
13066 2018-06-13 Jeff Law <law@redhat.com>
13068 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
13069 integer 0 for argument to print_rtl_with_bb.
13070 (rl78_reorg): Likewise.
13072 2018-06-13 David Malcolm <dmalcolm@redhat.com>
13074 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
13075 from rtx to rtx_insn *.
13076 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
13078 (add_sched_insns_for_speculation): Likewise for local "target",
13079 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
13080 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
13081 from rtx_insn ** to rtx_code_label **.
13082 (reorg_emit_nops): Likewise.
13083 (c6x_reorg): Likewise for local "call_labels".
13084 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
13086 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
13087 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
13088 the loops over LABEL_REFS.
13089 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
13091 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
13092 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
13093 (split_branches): Strengthen local "olabel" from rtx to
13094 rtx_insn *, adding a safe_as_a cast.
13095 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
13097 (add_insn_after): Likewise for first two params.
13098 (add_insn_before): Likewise.
13099 (remove_insn): Likewise for param.
13100 (emit_pattern_before_noloc): Likewise for second and third params.
13101 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
13102 (emit_call_insn_before_noloc): Likewise.
13103 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
13105 (emit_barrier_before): Likewise.
13106 (emit_label_before): Strengthen "label" param from "rtx" to
13107 "rtx_code_label *". Strengthen "before" param from "rtx" to
13109 (emit_insn_after_1): Strengthen "after" param from "rtx" to
13111 (emit_pattern_after_noloc): Likewise.
13112 (emit_insn_after_noloc): Likewise.
13113 (emit_jump_insn_after_noloc): Likewise.
13114 (emit_call_insn_after_noloc): Likewise.
13115 (emit_debug_insn_after_noloc): Likewise.
13116 (emit_barrier_after): Likewise.
13117 (emit_label_after): Likewise for both params.
13118 (emit_pattern_after_setloc): Likewise for "after" param. Convert
13119 "loc" param from "int" to "location_t".
13120 (emit_insn_after_setloc): Likewise.
13121 (emit_jump_insn_after_setloc): Likewise.
13122 (emit_call_insn_after_setloc): Likewise.
13123 (emit_debug_insn_after_setloc): Likewise.
13124 (emit_pattern_before_setloc): Likewise for "before" param. Convert
13125 "loc" param from "int" to "location_t".
13126 (emit_pattern_before): Convert NULL_RTX to NULL.
13127 (emit_insn_before_setloc): Convert "loc" param from "int" to
13129 (emit_jump_insn_before_setloc): Likewise.
13130 (emit_call_insn_before_setloc): Likewise.
13131 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
13132 rtx_insn *. Convert "loc" param from "int" to "location_t".
13133 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
13134 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
13135 Convert 3rd param from "int" to "location_t".
13136 (emit_barrier_before, emit_barrier_after, next_real_insn):
13137 Strengthen param from rtx to rtx_insn *.
13138 (emit_label_before): Strengthen 1st param from "rtx" to
13139 "rtx_code_label *". Strengthen 2nd param from "rtx" to
13141 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
13142 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
13143 Strengthen 2nd param from "rtx" to "rtx_insn *".
13144 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
13145 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
13146 Likewise. Convert 3rd param from "int" to "location_t".
13147 (emit_label_after): Strengthen 1st param from "rtx" to
13148 "rtx_code_label *".
13149 (next_real_insn, remove_insn): Strengthen param from "rtx" to
13151 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
13152 from "rtx" to "rtx_insn *".
13154 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
13156 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
13157 bodies streamed in with -Q.
13158 * dumpfile.c (dump_files): Add lto-stream-out dump file.
13159 * dumpfile.h (tree_dump_index): Add lto_stream_out.
13160 * gimple-streamer-out.c: Include gimple-pretty-print.h
13161 (output_bb): Dump stmts streamed.
13162 * lto-section-out.c: Include print-tree.h
13163 (lto_begin_section): Dump sections created.
13164 (lto_output_decl_index): Dump decl encoded.
13165 * lto-streamer-out.c: Include print-tree.h
13166 (create_output_block): Dump output block created.
13167 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
13168 (output_function): Dump function output.
13169 (output_constructor): Dump constructor streamed.
13170 (write_global_stream): Output indexes encoded.
13171 (produce_asm_for_decls): Dump streams encoded.
13172 * lto-streamer.c (streamer_dump_file): New global var.
13173 * lto-streamer.h (streamer_dump_file): Declare.
13174 * passes.c (ipa_write_summaries): Initialize streamer dump.
13175 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
13178 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
13181 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
13182 offsets for register save directives. Emit a second batch of save
13183 directives, if need be, when the function accesses prior frames.
13185 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13187 * config/arc/fpu.md (fmasf4): Force operand to register.
13188 (fnmasf4): Likewise.
13190 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13192 * config/arc/arc-protos.h (arc_pad_return): Remove.
13193 * config/arc/arc.c (machine_function): Remove force_short_suffix
13195 (arc_print_operand): Adjust printing of '&'.
13196 (arc_verify_short): Remove conditional printing of short suffix.
13197 (arc_final_prescan_insn): Remove reference to size_reason.
13198 (pad_return): New function.
13199 (arc_reorg): Call pad_return.
13200 (arc_pad_return): Remove.
13201 (arc_init_machine_status): Remove reference to force_short_suffix.
13202 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
13203 (attr length): When attribute iscompact is true force to 2
13204 regardless; in the case of maybe check if we want to force the
13205 instruction to have 4 bytes length.
13206 (nopv): Change it to generate 4 byte long nop as well.
13207 (blockage): New pattern.
13208 (simple_return): Remove call to arc_pad_return.
13209 (p_return_i): Likewise.
13211 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13213 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
13215 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13217 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
13220 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13222 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
13223 for ARC700 and ARCv2.
13225 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
13228 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
13229 operands[2] instead of operands[1].
13232 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13234 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
13235 case, check whether the outer register overlaps an unallocatable
13236 register, not just whether it fits the required class.
13238 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13240 * poly-int.h (can_div_trunc_p): Add new overload in which all values
13242 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
13243 (memrefs_conflict_p): Likewise.
13244 (init_alias_analysis): Likewise.
13245 * cfgexpand.c (expand_debug_expr): Likewise.
13246 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
13247 * cse.c (fold_rtx): Likewise.
13248 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
13249 * expr.c (emit_block_move_hints): Likewise.
13250 (clear_storage_hints, push_block, emit_push_insn): Likewise.
13251 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
13252 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
13253 (emit_group_store): Likewise.
13254 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
13255 to read the PRE/POST_MODIFY increment.
13256 * calls.c (store_one_arg): Use strip_offset.
13257 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
13259 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
13261 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
13262 (simplify_binary_operation_1): Extend CONST_INT handling to
13264 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
13265 than a HOST_WIDE_INT.
13266 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
13268 (adjust_mems, add_stores): Update accodingly.
13269 (vt_canonicalize_addr): Track polynomial offsets.
13270 (emit_note_insn_var_location): Likewise.
13271 (vt_add_function_parameter): Likewise.
13272 (vt_initialize): Likewise.
13274 2018-06-12 Jeff Law <law@redhat.com>
13276 * config.gcc (alpha*-*-freebsd*): Remove.
13277 * config/alpha/freebsd.h: Remove.
13279 2018-06-12 David Malcolm <dmalcolm@redhat.com>
13282 * spellcheck-tree.c (levenshtein_distance): Rename to...
13283 (get_edit_distance): ...this, and update for underlying renaming.
13284 * spellcheck-tree.h (levenshtein_distance): Rename to...
13285 (get_edit_distance): ...this.
13286 * spellcheck.c (levenshtein_distance): Rename to...
13287 (get_edit_distance): ...this. Convert from Levenshtein distance
13288 to Damerau-Levenshtein distance by supporting transpositions of
13289 adjacent characters. Rename "v1" to "v_next" and "v0" to
13291 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
13292 (selftest::test_edit_distance_unit_test_oneway): ...this, and
13293 update for underlying renaming.
13294 (selftest::levenshtein_distance_unit_test): Rename to...
13295 (selftest::test_get_edit_distance_unit): ...this, and update for
13296 underlying renaming.
13297 (selftest::test_find_closest_string): Add example from PR 69968
13298 where transposition helps
13299 (selftest::test_metric_conditions): Update for renaming.
13300 (selftest::test_metric_conditions): Likewise.
13301 (selftest::spellcheck_c_tests): Likewise.
13302 * spellcheck.h (levenshtein_distance): Rename both overloads to...
13303 (get_edit_distance): ...this.
13304 (best_match::consider): Update for renaming.
13306 2018-06-12 Martin Sebor <msebor@redhat.com>
13308 PR tree-optimization/85259
13309 * builtins.c (compute_objsize): Handle constant offsets.
13310 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
13311 true iff a warning has been issued.
13312 * gimple.h (gimple_nonartificial_location): New function.
13313 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
13314 gimple_nonartificial_location and handle -Wno-system-headers.
13315 (handle_builtin_stxncpy): Same.
13317 2018-06-12 Martin Sebor <msebor@redhat.com>
13320 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
13322 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13324 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13325 BUILTIN_VEC_XST entries for pointer to double and long long.
13327 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
13330 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
13332 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
13335 2018-06-12 Martin Liska <mliska@suse.cz>
13337 * doc/options.texi: Document IntegerRange.
13339 2018-06-12 Martin Liska <mliska@suse.cz>
13341 * config/i386/i386.opt: Make MPX-related options as Deprecated.
13342 * opt-functions.awk: Handle Deprecated flag.
13343 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
13345 (read_cmdline_option): Report warning for a deprecated option.
13346 * opts.h (struct cl_option): Add new field cl_deprecated.
13347 (CL_ERR_DEPRECATED): New.
13349 2018-06-12 Martin Liska <mliska@suse.cz>
13351 * doc/options.texi: Document Deprecated option flag.
13353 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13355 * config/arc/arc-arch.h (arc_extras): New enum.
13356 (arc_cpu_t):Add field extra.
13357 (arc_cpu_types): Consider the extras.
13358 * config/arc/arc-cpus.def: Add extras info.
13359 * config/arc/arc-opts.h (processor_type): Consider extra field.
13360 * config/arc/arc.c (arc_override_options): Handle extra field.
13362 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13364 * config/arc/arc-arch.h: Update ARC_OPTX macro.
13365 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
13367 * config/arc/arc.c (arc_init): Update pic warning.
13368 (irq_range): Update irq range parsing warnings.
13369 (arc_override_options): Update various warning messages.
13370 (arc_handle_aux_attribute): Likewise.
13372 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
13374 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
13376 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
13378 * doc/sourcebuild.texi: Document usage of line number 0 in verify
13379 compiler messages directives.
13381 2018-06-12 Matthew Fortune <mfortune@gmail.com>
13383 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
13384 * config/mips/mips-tables.opt: Regenerate.
13385 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
13387 * doc/invoke.texi: Document -march=i6500.
13389 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
13391 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
13392 (i6400_gpmul): Add cpu_unit.
13393 (i6400_gpdiv): Likewise.
13394 (i6400_msa_add_d): Update reservations.
13395 (i6400_msa_int_add) Likewise.
13396 (i6400_msa_short_logic3) Likewise.
13397 (i6400_msa_short_logic2) Likewise.
13398 (i6400_msa_short_logic) Likewise.
13399 (i6400_msa_move) Likewise.
13400 (i6400_msa_cmp) Likewise.
13401 (i6400_msa_short_float2) Likewise.
13402 (i6400_msa_div_d) Likewise.
13403 (i6400_msa_long_logic1) Likewise.
13404 (i6400_msa_long_logic2) Likewise.
13405 (i6400_msa_mult) Likewise.
13406 (i6400_msa_long_float2) Likewise.
13407 (i6400_msa_long_float4) Likewise.
13408 (i6400_msa_long_float5) Likewise.
13409 (i6400_msa_long_float8) Likewise.
13410 (i6400_fpu_fadd): Include frint type.
13411 (i6400_fpu_store): New define_insn_reservation.
13412 (i6400_fpu_load): Likewise.
13413 (i6400_fpu_move): Likewise.
13414 (i6400_fpu_fcmp): Likewise.
13415 (i6400_fpu_fmadd): Likewise.
13416 (i6400_int_mult): Include imul3nc type and update reservation.
13417 (i6400_int_div): Include idiv3 type and update reservation.
13418 (i6400_int_load): Update to check type not move_type.
13419 (i6400_int_store): Likewise.
13420 (i6400_int_prefetch): Set zero latency.
13422 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
13424 * gcc.c: Document new %@{...} sequence.
13425 (LINK_COMMAND_SPEC): Use it for the -L switches.
13426 (cpp_unique_options): Use it for the -I switches.
13427 (at_file_argbuf): New global variable.
13428 (in_at_file): Likewise.
13429 (alloc_args): Create at_file_argbuf.
13430 (clear_args): Truncate at_file_argbuf.
13431 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
13432 (open_at_file): New function.
13433 (close_at_file): Likewise.
13434 (create_at_file): Delete.
13435 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
13438 (validate_switches_from_spec): Deal with %@{...} sequence.
13439 (validate_switches): Likewise.
13440 (driver::finalize): Call clear_args.
13442 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
13444 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
13446 2018-06-11 Martin Sebor <msebor@redhat.com>
13448 * doc/invoke.texi (-Wall): List -Wc++17-compat.
13449 (Wno-class-memaccess): Add @opindex.
13450 (Wno-templates, Wno-multiple-inheritance): Same.
13451 (Wno-virtual-inheritance, Wno-namespaces): Same.
13452 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
13453 (Wno-format-overflow, Wno-format-truncation): Same.
13454 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
13455 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
13456 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
13457 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
13458 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
13459 (Wno-misspelled-isr): Same.
13461 2018-06-11 Martin Sebor <msebor@redhat.com>
13463 * PR tree-optimization/86083
13464 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
13466 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
13468 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
13470 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
13473 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
13474 on the correct operand.
13475 (*movdi_internal64): Ditto.
13477 2018-06-11 Martin Liska <mliska@suse.cz>
13479 PR tree-optimization/86089
13480 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
13482 2018-06-11 Julia Koval <julia.koval@intel.com>
13484 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
13485 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
13486 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
13488 2018-06-11 Olivier Hainque <hainque@adacore.com>
13490 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
13491 for Ada with strict dwarf2.
13493 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
13496 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
13499 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
13501 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
13503 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
13505 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
13508 2018-06-08 Martin Liska <mliska@suse.cz>
13510 * tree-cfg.h (debug_function): Fix argument type to match
13513 2018-06-08 Martin Liska <mliska@suse.cz>
13515 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
13516 Remove usage of MPX-related (and removed) fields.
13517 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
13519 2018-06-08 David Malcolm <dmalcolm@redhat.com>
13521 * cfg.c (debug): Use TDF_NONE rather than 0.
13522 * cfghooks.c (debug): Likewise.
13523 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
13524 (struct dump_option_value_info): Convert to...
13525 (struct kv_pair): ...this template type.
13526 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
13528 (optinfo_verbosity_options): Likewise.
13529 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
13531 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
13532 than int for "optgroup_flags" param.
13533 (dump_generic_expr_loc): Use dump_flags_t rather than int for
13535 (dump_dec): Likewise.
13536 (dump_finish): Use TDF_NONE rather than 0.
13537 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
13538 rather than int for "optgroup_flags" param. Use TDF_NONE rather
13539 than 0. Update for change to option_ptr.
13540 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
13541 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
13542 0. Update for changes to optinfo_verbosity_options and
13544 (opt_info_switch_p): Convert optgroup_flags from int to
13546 (dump_basic_block): Use dump_flags_t rather than int
13547 for "dump_kind" param.
13548 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
13549 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
13550 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
13551 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
13552 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
13553 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
13554 TDF_NONE): Convert from macros to...
13555 (enum dump_flag): ...this new enum.
13556 (dump_flags_t): Update to use enum.
13557 (operator|, operator&, operator~, operator|=, operator&=):
13558 Implement for dump_flags_t.
13559 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
13560 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
13561 Convert from macros to...
13562 (enum optgroup_flag): ...this new enum.
13563 (optgroup_flags_t): New typedef.
13564 (operator|, operator|=): Implement for optgroup_flags_t.
13565 (struct dump_file_info): Convert field "alt_flags" to
13566 dump_flags_t. Convert field "optgroup_flags" to
13568 (dump_basic_block): Use dump_flags_t rather than int for param.
13569 (dump_generic_expr_loc): Likewise.
13570 (dump_dec): Likewise.
13571 (dump_register): Convert param "optgroup_flags" to
13573 (opt_info_enable_passes): Likewise.
13574 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
13576 * gimple-pretty-print.c (debug): Likewise.
13577 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
13578 (merged_store_group::apply_stores): Likewise.
13579 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
13580 * gimple.c (verify_gimple_pp): Likewise.
13581 * graphite-poly.c (print_pbb_body): Likewise.
13582 * passes.c (pass_manager::register_one_dump_file): Convert
13583 local "optgroup_flags" to optgroup_flags_t.
13584 * print-tree.c (print_node): Use TDF_NONE rather than 0.
13586 (debug_body): Likewise.
13587 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
13588 to optgroup_flags_t.
13589 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
13591 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
13592 (convert_mult_to_fma): Likewise.
13593 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
13594 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
13595 * tree-vect-data-refs.c (dump_lower_bound): Convert param
13596 "dump_kind" to dump_flags_t.
13598 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
13600 * config/rs6000/rs6000.c (min, max): Delete.
13602 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
13604 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
13605 -mabi=spe and -mabi=no-spe.
13607 2018-06-08 Martin Liska <mliska@suse.cz>
13609 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
13610 where we expect an existing summary.
13612 2018-06-08 Martin Liska <mliska@suse.cz>
13614 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
13615 * ipa-inline.h (estimate_edge_growth): Likewise.
13617 2018-06-08 Martin Liska <mliska@suse.cz>
13619 * cgraph.c (function_version_hasher::hash): Use
13620 cgraph_node::get_uid ().
13621 (function_version_hasher::equal):
13622 * cgraph.h (cgraph_node::get_uid): New method.
13623 * ipa-inline.c (update_caller_keys): Use
13624 cgraph_node::get_uid ().
13625 (update_callee_keys): Likewise.
13626 * ipa-utils.c (searchc): Likewise.
13627 (ipa_reduced_postorder): Likewise.
13628 * lto-cgraph.c (input_node): Likewise.
13629 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
13630 * symbol-summary.h (symtab_insertion): Likewise.
13631 (symtab_removal): Likewise.
13632 (symtab_duplication): Likewise.
13633 * tree-pretty-print.c (dump_function_header): Likewise.
13634 * tree-sra.c (convert_callers_for_node): Likewise.
13636 2018-06-08 Martin Liska <mliska@suse.cz>
13638 * cgraph.c (symbol_table::create_edge): Always assign a new
13640 (symbol_table::free_edge): Do not recycle numbers.
13641 * cgraph.h (cgraph_edge::get): New method.
13642 * symbol-summary.h (symtab_removal): Use it.
13643 (symtab_duplication): Likewise.
13644 (call_summary::hashable_uid): Remove.
13646 2018-06-08 Martin Liska <mliska@suse.cz>
13648 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
13649 (initialize_growth_caches): Remove.
13650 (free_growth_caches): Likewise.
13651 (do_estimate_edge_time): Use edge_growth_cache.
13652 (do_estimate_edge_size): Likewise.
13653 (do_estimate_edge_hints): Likewise.
13654 * ipa-inline.c (reset_edge_caches): Likewise.
13655 (recursive_inlining): Likewise.
13656 (inline_small_functions): Likewise.
13657 * ipa-inline.h (initialize_growth_caches): Remove.
13658 (estimate_edge_size): Likewise.
13659 (estimate_edge_time): Likewise.
13660 (estimate_edge_hints): Likewise.
13661 (reset_edge_growth_cache): Likewise.
13662 * symbol-summary.h (call_summary::remove): New method.
13664 2018-06-08 Martin Liska <mliska@suse.cz>
13666 * ipa-cp.c (class edge_clone_summary): New summary.
13667 (grow_edge_clone_vectors): Remove.
13668 (ipcp_edge_duplication_hook): Remove.
13669 (class edge_clone_summary_t): New call_summary class.
13670 (ipcp_edge_removal_hook): Remove.
13671 (edge_clone_summary_t::duplicate): New function.
13672 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
13673 (create_specialized_node): Likewise.
13674 (ipcp_driver): Initialize edge_clone_summaries and do not
13677 2018-06-08 Martin Liska <mliska@suse.cz>
13679 * symbol-summary.h (get): New function.
13680 (call_summary::m_initialize_when_cloning): New class member.
13682 2018-06-08 Martin Liska <mliska@suse.cz>
13684 * cgraph.c (cgraph_node::remove): Do not recycle uid.
13685 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
13686 (symbol_table::allocate_cgraph_symbol): Do not set uid.
13687 * passes.c (uid_hash_t): Record removed_nodes by their uids.
13688 (remove_cgraph_node_from_order): Use the removed_nodes set.
13689 (do_per_function_toporder): Likwise.
13690 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
13691 instead of summary_uid.
13692 (symtab_removal): Likewise.
13693 (symtab_duplication): Likewise.
13695 2018-06-08 Martin Liska <mliska@suse.cz>
13697 * ipa-cp.c (ipcp_store_bits_results): Use
13698 ipcp_transformation_sum.
13699 (ipcp_store_vr_results): Likewise.
13700 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
13702 (ipcp_transformation_initialize): ... this.
13703 (ipa_set_node_agg_value_chain):
13704 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
13705 (write_ipcp_transformation_info): Likewise.
13706 (read_ipcp_transformation_info): Likewise.
13707 (ipcp_update_bits): Likewise.
13708 (ipcp_update_vr): Likewise.
13709 (ipcp_transform_function): Likewise.
13710 * ipa-prop.h: Rename ipcp_transformation_summary to
13711 ipcp_transformation.
13712 (class ipcp_transformation_t): New function summary.
13713 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
13714 (ipa_get_agg_replacements_for_node): Likewise.
13716 2018-06-08 Martin Liska <mliska@suse.cz>
13718 * ipa-pure-const.c (struct funct_state_d): Do it class instead
13720 (class funct_state_summary_t): New function_summary class.
13721 (has_function_state): Remove.
13722 (get_function_state): Likewise.
13723 (set_function_state): Likewise.
13724 (add_new_function): Likewise.
13725 (funct_state_summary_t::insert): New function.
13726 (duplicate_node_data): Remove.
13727 (remove_node_data): Remove.
13728 (funct_state_summary_t::duplicate): New function.
13729 (register_hooks): Create new funct_state_summaries.
13730 (pure_const_generate_summary): Use it.
13731 (pure_const_write_summary): Likewise.
13732 (pure_const_read_summary): Likewise.
13733 (propagate_pure_const): Likewise.
13734 (propagate_nothrow): Likewise.
13735 (dump_malloc_lattice): Likewise.
13736 (propagate_malloc): Likewise.
13737 (execute): Do not register hooks, just remove summary
13739 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
13742 2018-06-08 Martin Liska <mliska@suse.cz>
13744 * ipa-reference.c (remove_node_data): Remove.
13745 (duplicate_node_data): Likewise.
13746 (class ipa_ref_var_info_summary_t): New class.
13747 (class ipa_ref_opt_summary_t): Likewise.
13748 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
13749 (get_reference_optimization_summary): Use
13750 ipa_ref_opt_sum_summaries.
13751 (set_reference_vars_info): Remove.
13752 (set_reference_optimization_summary): Likewise.
13753 (ipa_init): Create summaries.
13754 (init_function_info): Use function summary.
13755 (ipa_ref_opt_summary_t::duplicate): New function.
13756 (ipa_ref_opt_summary_t::remove): New function.
13757 (get_read_write_all_from_node): Fix GNU coding style.
13758 (propagate): Use function summary.
13759 (write_node_summary_p): Fix GNU coding style.
13760 (stream_out_bitmap): Likewise.
13761 (ipa_reference_read_optimization_summary): Use function summary.
13762 (ipa_reference_c_finalize): Do not release hooks.
13764 2018-06-08 Martin Liska <mliska@suse.cz>
13766 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
13767 (analyze_function_body): Extract multiple calls of get_create.
13768 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
13769 * ipa-inline.c (recursive_inlining): Use ::get method.
13770 * ipa-inline.h (estimate_edge_growth): Likewise.
13772 2018-06-08 Martin Liska <mliska@suse.cz>
13774 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
13776 (hsa_function_summary::hsa_function_summary): Use the new enum
13778 (hsa_gpu_implementation_p): Use hsa_summaries::get.
13779 * hsa-gen.c (hsa_get_host_function): Likewise.
13780 (get_brig_function_name): Likewise.
13781 * ipa-hsa.c (process_hsa_functions): Likewise.
13782 (ipa_hsa_write_summary): Likewise.
13783 * symbol-summary.h (symtab_duplication): Use ::get function/
13784 (get): New function.
13786 2018-06-08 Martin Liska <mliska@suse.cz>
13788 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
13790 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
13791 (hsa_register_kernel): Likewise.
13792 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
13793 * hsa-gen.c (hsa_get_host_function): Likewise.
13794 (get_brig_function_name): Likewise.
13795 (generate_hsa): Likewise.
13796 (pass_gen_hsail::execute): Likewise.
13797 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
13798 (devirtualization_time_bonus): Likewise.
13799 (ipcp_propagate_stage): Likewise.
13800 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
13801 (edge_set_predicate): Likewise.
13802 (evaluate_conditions_for_known_args): Likewise.
13803 (evaluate_properties_for_edge): Likewise.
13804 (ipa_fn_summary::reset): Likewise.
13805 (ipa_fn_summary_t::duplicate): Likewise.
13806 (dump_ipa_call_summary): Likewise.
13807 (ipa_dump_fn_summary): Likewise.
13808 (analyze_function_body): Likewise.
13809 (compute_fn_summary): Likewise.
13810 (estimate_edge_devirt_benefit): Likewise.
13811 (estimate_edge_size_and_time): Likewise.
13812 (estimate_calls_size_and_time): Likewise.
13813 (estimate_node_size_and_time): Likewise.
13814 (inline_update_callee_summaries): Likewise.
13815 (remap_edge_change_prob): Likewise.
13816 (remap_edge_summaries): Likewise.
13817 (ipa_merge_fn_summary_after_inlining): Likewise.
13818 (ipa_update_overall_fn_summary): Likewise.
13819 (read_ipa_call_summary): Likewise.
13820 (inline_read_section): Likewise.
13821 (write_ipa_call_summary): Likewise.
13822 (ipa_fn_summary_write): Likewise.
13823 (ipa_free_fn_summary): Likewise.
13824 * ipa-hsa.c (process_hsa_functions): Likewise.
13825 (ipa_hsa_write_summary): Likewise.
13826 (ipa_hsa_read_section): Likewise.
13827 * ipa-icf.c (sem_function::merge): Likewise.
13828 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
13829 (do_estimate_edge_time): Likewise.
13830 (estimate_size_after_inlining): Likewise.
13831 (estimate_growth): Likewise.
13832 (growth_likely_positive): Likewise.
13833 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
13834 (inline_call): Likewise.
13835 * ipa-inline.c (caller_growth_limits): Likewise.
13836 (can_inline_edge_p): Likewise.
13837 (can_inline_edge_by_limits_p): Likewise.
13838 (compute_uninlined_call_time): Likewise.
13839 (compute_inlined_call_time): Likewise.
13840 (want_inline_small_function_p): Likewise.
13841 (edge_badness): Likewise.
13842 (update_caller_keys): Likewise.
13843 (update_callee_keys): Likewise.
13844 (recursive_inlining): Likewise.
13845 (inline_small_functions): Likewise.
13846 (inline_to_all_callers_1): Likewise.
13847 (dump_overall_stats): Likewise.
13848 (early_inline_small_functions): Likewise.
13849 (early_inliner): Likewise.
13850 * ipa-inline.h (estimate_edge_growth): Likewise.
13851 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13852 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
13853 * ipa-prop.h (IPA_NODE_REF): Likewise.
13854 (IPA_EDGE_REF): Likewise.
13855 * ipa-pure-const.c (malloc_candidate_p): Likewise.
13856 (propagate_malloc): Likewise.
13857 * ipa-split.c (execute_split_functions): Likewise.
13858 * symbol-summary.h: Rename get to get_create.
13860 (get_create): Likewise.
13861 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
13863 2018-06-08 Martin Liska <mliska@suse.cz>
13865 * symbol-summary.h (release): Move definition out of class
13867 (symtab_removal): Likewise.
13868 (symtab_duplication): Likewise.
13870 2018-06-08 Martin Liska <mliska@suse.cz>
13872 * symbol-summary.h (function_summary): Move constructor
13873 implementation out of class declaration.
13874 (release): Likewise.
13875 (symtab_insertion): Likewise.
13876 (symtab_removal): Likewise.
13877 (symtab_duplication): Likewise.
13880 2018-06-08 Martin Liska <mliska@suse.cz>
13882 * Makefile.in: Remove support for MPX (macros, related functions,
13883 fields in cgraph_node, ...).
13884 * builtin-types.def (BT_BND): Likewise.
13885 (BT_FN_BND_CONST_PTR): Likewise.
13886 (BT_FN_CONST_PTR_BND): Likewise.
13887 (BT_FN_VOID_PTR_BND): Likewise.
13888 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
13889 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
13890 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
13891 (expand_builtin_mempcpy_with_bounds): Likewise.
13892 (expand_builtin_memset_with_bounds): Likewise.
13893 (expand_builtin_memset_args): Likewise.
13894 (std_expand_builtin_va_start): Likewise.
13895 (expand_builtin): Likewise.
13896 (expand_builtin_with_bounds): Likewise.
13897 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
13898 (DEF_LIB_BUILTIN_CHKP): Likewise.
13899 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
13900 (DEF_CHKP_BUILTIN): Likewise.
13901 (BUILT_IN_MEMCPY): Likewise.
13902 (BUILT_IN_MEMMOVE): Likewise.
13903 (BUILT_IN_MEMPCPY): Likewise.
13904 (BUILT_IN_MEMSET): Likewise.
13905 (BUILT_IN_STPCPY): Likewise.
13906 (BUILT_IN_STRCAT): Likewise.
13907 (BUILT_IN_STRCHR): Likewise.
13908 (BUILT_IN_STRCPY): Likewise.
13909 (BUILT_IN_STRLEN): Likewise.
13910 (BUILT_IN_MEMCPY_CHK): Likewise.
13911 (BUILT_IN_MEMMOVE_CHK): Likewise.
13912 (BUILT_IN_MEMPCPY_CHK): Likewise.
13913 (BUILT_IN_MEMSET_CHK): Likewise.
13914 (BUILT_IN_STPCPY_CHK): Likewise.
13915 (BUILT_IN_STRCAT_CHK): Likewise.
13916 (BUILT_IN_STRCPY_CHK): Likewise.
13917 * calls.c (store_bounds): Likewise.
13918 (emit_call_1): Likewise.
13919 (special_function_p): Likewise.
13920 (maybe_warn_nonstring_arg): Likewise.
13921 (initialize_argument_information): Likewise.
13922 (finalize_must_preallocate): Likewise.
13923 (compute_argument_addresses): Likewise.
13924 (expand_call): Likewise.
13925 * cfgexpand.c (expand_call_stmt): Likewise.
13926 (expand_return): Likewise.
13927 (expand_gimple_stmt_1): Likewise.
13928 (pass_expand::execute): Likewise.
13929 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
13930 (cgraph_node::remove): Likewise.
13931 (cgraph_node::dump): Likewise.
13932 (cgraph_node::verify_node): Likewise.
13933 * cgraph.h (chkp_function_instrumented_p): Likewise.
13934 (symtab_node::get_alias_target): Likewise.
13935 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
13936 (cgraph_local_p): Likewise.
13937 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
13938 (cgraph_edge::rebuild_references): Likewise.
13939 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
13940 (walk_polymorphic_call_targets): Likewise.
13941 (cgraph_node::expand_thunk): Likewise.
13942 (symbol_table::output_weakrefs): Likewise.
13943 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
13944 (ix86_handle_option): Likewise.
13945 * config/i386/constraints.md: Likewise.
13946 * config/i386/i386-builtin-types.def (BND): Likewise.
13950 * config/i386/i386-builtin.def (BDESC_END): Likewise.
13951 (BDESC_FIRST): Likewise.
13953 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
13954 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
13955 * config/i386/i386.c (enum reg_class): Likewise.
13956 (ix86_target_string): Likewise.
13957 (ix86_option_override_internal): Likewise.
13958 (ix86_conditional_register_usage): Likewise.
13959 (ix86_valid_target_attribute_inner_p): Likewise.
13960 (ix86_set_indirect_branch_type): Likewise.
13961 (ix86_set_current_function): Likewise.
13962 (ix86_function_arg_regno_p): Likewise.
13963 (init_cumulative_args): Likewise.
13964 (ix86_function_arg_advance): Likewise.
13965 (ix86_function_arg): Likewise.
13966 (ix86_pass_by_reference): Likewise.
13967 (ix86_function_value_regno_p): Likewise.
13968 (ix86_function_value_1): Likewise.
13969 (ix86_function_value_bounds): Likewise.
13970 (ix86_return_in_memory): Likewise.
13971 (ix86_setup_incoming_vararg_bounds): Likewise.
13972 (ix86_va_start): Likewise.
13973 (indirect_thunk_need_prefix): Likewise.
13974 (print_reg): Likewise.
13975 (ix86_print_operand): Likewise.
13976 (ix86_expand_call): Likewise.
13977 (ix86_output_function_return): Likewise.
13978 (reg_encoded_number): Likewise.
13979 (BDESC_VERIFYS): Likewise.
13980 (ix86_init_mpx_builtins): Likewise.
13981 (ix86_init_builtins): Likewise.
13982 (ix86_emit_cmove): Likewise.
13983 (ix86_emit_move_max): Likewise.
13984 (ix86_expand_builtin): Likewise.
13985 (ix86_builtin_mpx_function): Likewise.
13986 (ix86_get_arg_address_for_bt): Likewise.
13987 (ix86_load_bounds): Likewise.
13988 (ix86_store_bounds): Likewise.
13989 (ix86_load_returned_bounds): Likewise.
13990 (ix86_store_returned_bounds): Likewise.
13991 (ix86_class_likely_spilled_p): Likewise.
13992 (ix86_hard_regno_mode_ok): Likewise.
13993 (x86_order_regs_for_local_alloc): Likewise.
13994 (ix86_mitigate_rop): Likewise.
13995 (ix86_bnd_prefixed_insn_p): Likewise.
13996 (ix86_mpx_bound_mode): Likewise.
13997 (ix86_make_bounds_constant): Likewise.
13998 (ix86_initialize_bounds): Likewise.
13999 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
14000 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
14001 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
14002 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
14003 (TARGET_CHKP_BOUND_MODE): Likewise.
14004 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
14005 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
14006 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
14007 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
14008 * config/i386/i386.h (TARGET_MPX): Likewise.
14009 (TARGET_MPX_P): Likewise.
14010 (VALID_BND_REG_MODE): Likewise.
14011 (FIRST_BND_REG): Likewise.
14012 (LAST_BND_REG): Likewise.
14013 (enum reg_class): Likewise.
14014 (BND_REG_P): Likewise.
14015 (BND_REGNO_P): Likewise.
14016 (BNDmode): Likewise.
14017 (ADJUST_INSN_LENGTH): Likewise.
14018 * config/i386/i386.md: Likewise.
14019 * config/i386/i386.opt: Likewise.
14020 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
14021 (defined): Likewise.
14022 (LINK_MPX): Likewise.
14023 (MPX_SPEC): Likewise.
14024 (LIBMPX_SPEC): Likewise.
14025 (LIBMPXWRAPPERS_SPEC): Likewise.
14026 (CHKP_SPEC): Likewise.
14027 * config/i386/predicates.md: Likewise.
14028 * dbxout.c (dbxout_type): Likewise.
14029 * doc/extend.texi: Likewise.
14030 * doc/invoke.texi: Likewise.
14031 * doc/md.texi: Likewise.
14032 * doc/tm.texi: Likewise.
14033 * doc/tm.texi.in: Likewise.
14034 * dwarf2out.c (is_base_type): Likewise.
14035 (gen_formal_types_die): Likewise.
14036 (gen_subprogram_die): Likewise.
14037 (gen_type_die_with_usage): Likewise.
14038 (gen_decl_die): Likewise.
14039 (dwarf2out_late_global_decl): Likewise.
14040 * expr.c (expand_assignment): Likewise.
14041 (emit_storent_insn): Likewise.
14042 (store_expr_with_bounds): Likewise.
14043 (store_expr): Likewise.
14044 (expand_expr_real_1): Likewise.
14045 * expr.h (store_expr_with_bounds): Likewise.
14046 * function.c (use_register_for_decl): Likewise.
14047 (struct bounds_parm_data): Likewise.
14048 (assign_parms_augmented_arg_list): Likewise.
14049 (assign_parm_find_entry_rtl): Likewise.
14050 (assign_parm_is_stack_parm): Likewise.
14051 (assign_parm_load_bounds): Likewise.
14052 (assign_bounds): Likewise.
14053 (assign_parms): Likewise.
14054 (expand_function_start): Likewise.
14055 * gcc.c (CHKP_SPEC): Likewise.
14056 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
14057 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
14058 (wrestrict_dom_walker::check_call): Likewise.
14059 * gimple.c (gimple_build_call_from_tree): Likewise.
14060 * gimple.h (enum gf_mask): Likewise.
14061 (gimple_call_with_bounds_p): Likewise.
14062 (gimple_call_set_with_bounds): Likewise.
14063 * gimplify.c (gimplify_init_constructor): Likewise.
14064 * ipa-cp.c (initialize_node_lattices): Likewise.
14065 (propagate_constants_across_call): Likewise.
14066 (find_more_scalar_values_for_callers_subset): Likewise.
14067 * ipa-hsa.c (process_hsa_functions): Likewise.
14068 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
14069 * ipa-icf.c (sem_function::merge): Likewise.
14070 * ipa-inline.c (early_inliner): Likewise.
14071 * ipa-pure-const.c (warn_function_noreturn): Likewise.
14072 (warn_function_cold): Likewise.
14073 (propagate_pure_const): Likewise.
14074 * ipa-ref.h (enum GTY): Likewise.
14075 * ipa-split.c (find_retbnd): Likewise.
14076 (consider_split): Likewise.
14077 (split_function): Likewise.
14078 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
14079 * ipa.c (walk_polymorphic_call_targets): Likewise.
14080 (symbol_table::remove_unreachable_nodes): Likewise.
14081 (process_references): Likewise.
14082 (cgraph_build_static_cdtor_1): Likewise.
14083 * lto-cgraph.c (lto_output_node): Likewise.
14084 (output_refs): Likewise.
14085 (compute_ltrans_boundary): Likewise.
14086 (input_overwrite_node): Likewise.
14087 (input_node): Likewise.
14088 (input_cgraph_1): Likewise.
14089 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
14090 * passes.c (pass_manager::execute_early_local_passes): Likewise.
14091 (class pass_chkp_instrumentation_passes): Likewise.
14092 (make_pass_chkp_instrumentation_passes): Likewise.
14093 * passes.def: Likewise.
14094 * rtl.h (struct GTY): Likewise.
14095 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
14096 * stor-layout.c (layout_type): Likewise.
14097 * symtab.c: Likewise.
14098 * target.def: Likewise.
14099 * targhooks.c (default_chkp_bound_type): Likewise.
14100 (default_chkp_bound_mode): Likewise.
14101 (default_builtin_chkp_function): Likewise.
14102 (default_chkp_function_value_bounds): Likewise.
14103 (default_chkp_make_bounds_constant): Likewise.
14104 (default_chkp_initialize_bounds): Likewise.
14105 * targhooks.h (default_chkp_bound_type): Likewise.
14106 (default_chkp_bound_mode): Likewise.
14107 (default_builtin_chkp_function): Likewise.
14108 (default_chkp_function_value_bounds): Likewise.
14109 (default_chkp_make_bounds_constant): Likewise.
14110 (default_chkp_initialize_bounds): Likewise.
14111 * toplev.c (compile_file): Likewise.
14112 (process_options): Likewise.
14113 * tree-core.h (DEF_BUILTIN): Likewise.
14114 (DEF_BUILTIN_CHKP): Likewise.
14115 * tree-inline.c (declare_return_variable): Likewise.
14116 (remap_gimple_stmt): Likewise.
14117 (copy_bb): Likewise.
14118 (initialize_inlined_parameters): Likewise.
14119 (expand_call_inline): Likewise.
14120 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
14121 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
14122 (make_pass_ipa_chkp_produce_thunks): Likewise.
14123 (make_pass_chkp): Likewise.
14124 (make_pass_chkp_opt): Likewise.
14125 (make_pass_chkp_instrumentation_passes): Likewise.
14126 * tree-pretty-print.c (dump_generic_node): Likewise.
14127 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
14128 * tree-ssa-dce.c (propagate_necessity): Likewise.
14129 (eliminate_unnecessary_stmts): Likewise.
14130 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
14131 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
14132 * tree-ssa-sccvn.h: Likewise.
14133 * tree-ssa-strlen.c (get_string_length): Likewise.
14134 (valid_builtin_call): Likewise.
14135 (adjust_last_stmt): Likewise.
14136 (handle_builtin_strchr): Likewise.
14137 (handle_builtin_strcpy): Likewise.
14138 (handle_builtin_stxncpy): Likewise.
14139 (handle_builtin_memcpy): Likewise.
14140 (handle_builtin_strcat): Likewise.
14141 (strlen_check_and_optimize_stmt): Likewise.
14142 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
14143 * tree-streamer-in.c: Likewise.
14144 * tree-streamer.c (record_common_node): Likewise.
14145 * tree.c (tree_code_size): Likewise.
14146 (wide_int_to_tree_1): Likewise.
14147 (type_contains_placeholder_1): Likewise.
14148 (build_common_tree_nodes): Likewise.
14149 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
14150 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
14151 (POINTER_BOUNDS_P): Likewise.
14152 (BOUNDED_TYPE_P): Likewise.
14153 (BOUNDED_P): Likewise.
14154 (CALL_WITH_BOUNDS_P): Likewise.
14155 (pointer_bounds_type_node): Likewise.
14156 * value-prof.c (gimple_ic): Likewise.
14157 * var-tracking.c (vt_add_function_parameters): Likewise.
14158 * varasm.c (make_decl_rtl): Likewise.
14159 (assemble_start_function): Likewise.
14160 (output_constant): Likewise.
14161 (maybe_assemble_visibility): Likewise.
14162 * varpool.c (ctor_for_folding): Likewise.
14163 * chkp-builtins.def: Remove.
14164 * ipa-chkp.c: Remove.
14165 * ipa-chkp.h: Remove.
14166 * rtl-chkp.c: Remove.
14167 * rtl-chkp.h: Remove.
14168 * tree-chkp-opt.c: Remove.
14169 * tree-chkp.c: Remove.
14170 * tree-chkp.h: Remove.
14172 2018-06-07 Carl Love <cel@us.ibm.com>
14174 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14175 vextract_fp_from_shortl): Add BE support.
14177 2018-06-07 Paul Koning <ni1d@arrl.net>
14179 * compare-elim.c (try_merge_compare): Don't merge compare if
14180 address contains a side effect.
14181 (try_eliminate_compare): Likewise.
14183 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
14185 * config.gcc: Support "tremont".
14186 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
14187 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14189 * config/i386/i386.c (m_TREMONT): Define.
14190 (processor_target_table): Add "tremont".
14191 (PTA_TREMONT): Define.
14192 (ix86_lea_outperforms): Add TARGET_TREMONT.
14193 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
14194 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
14195 and M_INTEL_GOLDMONT_PLUS.
14196 (fold_builtin_cpu): Add "tremont".
14197 (ix86_add_stmt_cost): Add TARGET_TREMONT.
14198 (ix86_option_override_internal): Add "tremont".
14199 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
14200 (processor_type): Add PROCESSOR_TREMONT.
14201 * config/i386/x86-tune.def: Add m_TREMONT.
14202 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
14204 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14206 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
14207 symbol defined for msp430i* devices to be lower case.
14209 2018-06-07 Richard Biener <rguenther@suse.de>
14211 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
14212 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
14213 Properly wrap signed arithmetic if overflow wraps.
14215 2018-06-07 Jakub Jelinek <jakub@redhat.com>
14217 PR tree-optimization/69615
14218 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
14219 of a cast from a same precision integral SSA_NAME in a bb dominated
14220 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
14221 cast to utype if rhs2 has already a compatible type.
14223 2018-06-07 Richard Biener <rguenther@suse.de>
14225 PR tree-optimization/85935
14226 * graphite-scop-detection.c (find_params_in_bb): Analyze
14227 condition operands with respect to the correct loop. Assert
14228 the analysis doesn't fail.
14230 2018-06-04 Carl Love <cel@us.ibm.com>
14232 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
14233 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
14234 as it is slightly cheaper.
14235 (first_match_or_eos_index_<mode>):
14236 Calculate index using natural element order.
14237 (first_match_index_<mode>):
14238 Calculate index using natural element order.
14239 (first_match_or_eos_index_<mode>):
14240 Calculate index using natural order.
14241 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
14242 for BE and LE modes.
14243 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
14244 P9V_BUILTIN_VCLZLSBB_V16QI.
14245 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
14248 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
14250 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
14251 indentation and line wrap for many prototypes. Add missing
14252 @smallexample directives around block of prototypes for vec_xl and
14255 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
14257 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
14258 track if we pass or return IEEE 128-bit floating point.
14259 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
14260 C++ mangling that is compatible with GCC 8.1.
14261 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
14262 (init_cumulative_args): Note if we pass or return IEEE 128-bit
14263 floating point types.
14264 (rs6000_function_arg_advance_1): Likewise.
14265 (rs6000_mangle_type): Optionally generate mangled names that match
14266 what GCC 8.1 generated for IEEE 128-bit floating point types.
14267 (rs6000_globalize_decl_name): If we have an external function that
14268 passes or returns IEEE 128-bit types, generate a weak reference
14269 from the mangled name used in GCC 8.1 to the current mangled
14271 (rs6000_init_builtins): Make __ibm128 use the long double type if
14272 long double is IBM extended double. Make __float128 use the long
14273 double type if long double is IEEE 128-bit.
14276 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
14277 macro for __ibm128 built-in functions.
14278 (PACK_IF): Add __ibm128 pack/unpack functions.
14279 (UNPACK_IF): Likewise.
14280 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
14281 enable long double built-in functions if long double is IEEE
14282 128-bit floating point.
14283 (rs6000_invalid_builtin): Update long double built-in function
14285 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
14286 functions, adjust the built-in function to use the long double
14287 built-in function if __ibm128 and long double are the same type.
14288 * doc/extend.texi (PowerPC builtins): Update documention for
14289 __builtin_{,un}pack_longdouble. Add documentation for
14290 __builtin_{,un}pack_ibm128.
14292 2018-06-06 Jim Wilson <jimw@sifive.com>
14294 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
14295 (struct machine_function): New field interrupt_mode.
14296 (riscv_handle_type_attribute): New function. Add forward declaration.
14297 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
14298 (riscv_expand_epilogue): Check interrupt_mode field.
14299 (riscv_set_current_function): Check interrupt attribute args and
14300 set interrupt_mode field.
14301 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
14302 (riscv_sret, riscv_uret): New.
14303 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
14304 new arguments to interrupt attribute.
14306 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
14309 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
14310 Don't handle -mcpu=power8 if -mpower9-vector is also used.
14312 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
14314 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
14315 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
14316 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
14317 several redundant entries.
14319 2018-06-06 David Malcolm <dmalcolm@redhat.com>
14321 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
14322 type from "rtx" to "rtx_insn *".
14323 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
14324 for local "call_insn", removing cast.
14325 (ix86_expand_call): Likewise, introducing a "call_insn" local.
14327 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
14329 PR tree-optimization/86066
14330 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
14331 for BIT_INSERT_EXPR stores.
14333 2018-06-06 Richard Biener <rguenther@suse.de>
14335 PR tree-optimization/86062
14336 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
14337 component refs ontop
14338 of to be offsetted base.
14340 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14342 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
14343 to be static and remove check on interrupt attribute name.
14345 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
14347 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
14348 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
14350 2018-06-05 Steve Ellcey <sellcey@cavium.com>
14353 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
14355 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
14356 Remove second argument, change how error is called.
14357 (aarch64_layout_arg): Remove second argument from
14358 aarch64_err_no_fpadvsimd call.
14359 (aarch64_init_cumulative_args): Ditto.
14360 (aarch64_gimplify_va_arg_expr): Ditto.
14361 * config/aarch64/aarch64.md (mov<mode>): Ditto.
14363 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
14365 * config/i386/i386.md (simple_return_indirect_internal): New expander.
14366 (*simple_return_indirect_internal<mode>): Rename from
14367 simple_return_indirect_internal. Use W mode iterator.
14368 (rstorssp): New expander.
14369 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
14370 (clrssbsy): New expander.
14371 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
14373 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
14375 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
14377 (cmse_check_pointed_object): Likewise.
14379 2018-06-05 Martin Liska <mliska@suse.cz>
14381 PR gcov-profile/47618
14382 * doc/invoke.texi: Document how -fprofile-dir format
14385 2018-06-05 Richard Biener <rguenther@suse.de>
14387 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
14388 removal pretend DOM info isn't available so we do not update
14389 it and only remove edges, not dominated blocks. Actually free
14390 DOM info in case we removed something. Remove unreachable blocks.
14391 (mfb_keep_latches): Work with either DOM info or marked backedges.
14392 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
14393 first. Mark backedges if DOM info isn't available.
14394 (Re-)compute DOM info after cleanup_control_flow_pre.
14396 2018-06-05 Richard Biener <rguenther@suse.de>
14398 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
14399 (locus_discrim_hasher::hash): Adjust.
14400 (locus_discrim_hasher::equal): Likewise.
14401 (next_discriminator_for_locus): Work on line directly.
14402 (same_line_p): Pass in expanded locus1 as well.
14403 (assign_discriminators): Avoid redundant location expansions.
14405 2018-06-05 Richard Biener <rguenther@suse.de>
14407 PR tree-optimization/86046
14408 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
14409 if required after clearing TREE_ADDRESSABLE.
14411 2018-06-05 Richard Biener <rguenther@suse.de>
14413 PR tree-optimization/86047
14414 * tree-ssa-loop.c (for_each_index): Glob handling of all
14415 decls and constants and really handle all of them.
14417 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14420 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
14421 qualifier_void_pointer and qualifier_const_void_pointer.
14422 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
14423 (arm_init_builtins): Handle the above.
14424 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
14425 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
14426 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
14429 2018-06-05 Martin Liska <mliska@suse.cz>
14431 * auto-profile.c (read_autofdo_file): Do not use
14432 gcov_ctr_summary struct.
14433 (afdo_callsite_hot_enough_for_early_inline): Likewise.
14434 * coverage.c (struct counts_entry): Likewise.
14435 (read_counts_file): Read just single summary entry.
14436 (get_coverage_counts): Use gcov_summary struct.
14437 * coverage.h (get_coverage_counts): Likewise.
14438 * gcov-dump.c (dump_working_sets): Likewise.
14439 (tag_summary): Dump just single summary.
14440 * gcov-io.c (gcov_write_summary): Write just histogram
14442 (gcov_read_summary): Read just single summary.
14443 (compute_working_sets): Use gcov_summary struct.
14444 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
14445 of GCOV_COUNTERS_SUMMABLE.
14446 (GCOV_COUNTERS_SUMMABLE): Remove.
14447 (GCOV_FIRST_VALUE_COUNTER): Replace with
14448 GCOV_COUNTER_V_INTERVAL.
14449 (struct gcov_ctr_summary): Remove.
14450 (struct gcov_summary): Directly use fields of former
14452 (compute_working_sets): Use gcov_summary struct.
14453 * gcov.c (read_count_file): Do not use ctrs fields.
14454 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
14456 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
14458 * profile.c: Likewise.
14459 * profile.h: Likewise.
14461 2018-06-05 Martin Liska <mliska@suse.cz>
14463 PR gcov-profile/84846
14464 * gcov.c (output_lines): Print working directory only
14465 in intermediate format.
14467 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
14469 * config/s390/s390-builtin-types.def: Add void function type.
14470 * config/s390/s390-builtins.def: Use the function type for the
14473 2018-06-04 Jim Wilson <jimw@sifive.com>
14475 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
14477 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
14478 and maybe_eh_return. Change regno to unsigned int. Use new args to
14479 handle EH_RETURN_DATA_REGNO registers properly.
14480 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
14481 (riscv_expand_epilogue): Update comment. Change argument name and
14482 type. Update code to use new name and type. Pass new args to
14483 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
14485 * config/riscv/riscv.md (NORMAL_RETURN): New.
14486 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
14487 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
14488 (eh_return): Call gen_eh_return_internal and emit barrier.
14489 (eh_return_internal): Call riscv_expand_epilogue.
14491 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
14493 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
14494 bit_insertion field and declare can_be_merged_into method.
14495 (merged_store_group::can_be_merged_into): New method.
14496 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
14497 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
14498 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
14500 2018-06-04 Richard Biener <rguenther@suse.de>
14502 PR tree-optimization/85955
14503 * builtins.c (fold_builtin_sincos): Convert pointers to
14504 destination to appropriate type before dereferencing.
14506 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
14508 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
14510 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
14512 * expr.c (expand_expr_real_1): Force the operand into memory if
14513 its TYPE_MODE is BLKmode and if there is no integer mode for
14514 the number of bits being extracted.
14516 2018-06-04 Jakub Jelinek <jakub@redhat.com>
14520 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
14521 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
14523 2018-06-04 Richard Biener <rguenther@suse.de>
14525 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
14526 (cleanup_tree_cfg_noloop): ... single caller. Do
14527 start_recording_case_labels later.
14529 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
14531 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
14532 to _IMMINTRIN_H_INCLUDED.
14533 * config/i386/pconfigintrin.h: Ditto.
14534 * config/i386/waitpkgintrin.h: Ditto.
14535 * config/i386/immintrin.h: Add includes for sgxintrin.h,
14536 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
14537 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
14538 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
14539 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
14540 waitpkgintrin.h and cldemoteintrin.h.
14542 2018-06-04 Richard Biener <rguenther@suse.de>
14544 PR tree-optimization/86038
14545 * tracer.c (find_best_successor): Check probability for
14546 being initialized, bail out if not.
14548 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
14551 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
14552 of bits to ignore when comparing architectures.
14554 2018-06-04 Jakub Jelinek <jakub@redhat.com>
14556 PR tree-optimization/69615
14557 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
14558 maximum or minimum of the type, try to merge it also as if
14559 range1 is + [-, x - 1] or + [x + 1, -].
14562 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
14564 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
14566 PR tree-optimization/86034
14567 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
14568 the unsigned bitfield type in a bit insertion sequence if it does not
14569 have a larger precision than the bitfield size.
14570 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
14572 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
14574 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
14576 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
14578 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
14579 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
14580 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
14581 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
14583 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
14585 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
14586 Disable -fdelete-null-pointer-checks for ELF toolchain.
14588 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
14589 Kito Cheng <kito.cheng@gmail.com>
14591 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
14592 (nds32le-*-*, nds32be-*-*): Integrate checking process.
14593 (nds32*-*-*): Add glibc and uclibc conditions.
14594 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
14595 (TARGET_EXCEPT_UNWIND_INFO): Define.
14596 * config/nds32/elf.h: New file.
14597 * config/nds32/linux.h: New file.
14598 * config/nds32/nds32-elf.opt: New file.
14599 * config/nds32/nds32-linux.opt: New file.
14600 * config/nds32/nds32-fp-as-gp.c
14601 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
14602 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
14604 (nds32_asm_file_end): Ditto.
14605 (nds32_print_operand): Ditto.
14606 (nds32_insert_attributes): Ditto.
14607 (nds32_init_libfuncs): New function.
14608 (TARGET_HAVE_TLS): Define.
14609 (TARGET_INIT_LIBFUNCS): Define.
14610 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
14612 (TARGET_ELF): Apply different mcmodel setting.
14613 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
14614 been migrated into elf.h and linux.h files.
14615 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
14616 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
14617 (mcmodel): The content has been migrated into nds32-elf.opt and
14618 nds32-linux.opt files.
14619 * config/nds32/t-elf: New file.
14620 * config/nds32/t-linux: New file.
14622 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
14623 Shiva Chen <shiva0217@gmail.com>
14625 * config/nds32/constants.md (unspec_volatile_element): Add
14626 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
14627 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
14629 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
14630 (make_pass_nds32_fp_as_gp): Declare.
14631 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
14633 (nds32_asm_function_end_prologue): Remove unused asm output.
14634 (nds32_asm_function_begin_epilogue): Remove unused asm output.
14635 (nds32_asm_file_start): Output necessary fp_as_gp information.
14636 (nds32_option_override): Adjust register usage.
14637 (nds32_expand_prologue): Consider fp_as_gp situation.
14638 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
14639 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
14642 (simple_return): Ditto.
14643 (omit_fp_begin): Output special directive for fp_as_gp.
14644 (omit_fp_end): Output special directive for fp_as_gp.
14645 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
14646 mforbid-fp-as-gp): New options.
14648 2018-06-01 Mark Wielaard <mark@klomp.org>
14650 * dwarf2out.c (dwarf2out_finish): Remove generation of
14651 DW_AT_loclists_base.
14653 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
14655 * gimple-ssa-store-merging.c: Include gimple-fold.h.
14656 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
14657 (struct merged_store_group): Add bit_insertion field.
14658 (dump_char_array): Use standard hexadecimal format.
14659 (merged_store_group::merged_store_group): Set bit_insertion to false.
14660 (merged_store_group::apply_stores): Use optimal buffer size. Deal
14661 with BIT_INSERT_EXPR stores. Move up code updating the mask and
14662 also print the mask in the dump file.
14663 (pass_store_merging::gate): Minor tweak.
14664 (imm_store_chain_info::coalesce_immediate): Fix wrong association
14665 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
14666 stores with INTEGER_CST stores.
14667 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
14668 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
14669 and use it throughout. Generate bit insertion sequences if need be.
14670 (pass_store_merging::process_store): Remove redundant condition.
14671 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
14673 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
14675 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
14676 the 128-bit floating point types. Fix function comment.
14678 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14680 * config/aarch64/aarch64-simd.md
14681 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
14683 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
14686 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
14688 PR tree-optimization/85989
14689 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
14691 (backprop::intersect_uses): Check it when deciding whether this
14692 is a backedge reference.
14693 (backprop::process_block): Add each phi to m_visited_phis
14694 after visiting it, then clear it at the end.
14696 2018-06-01 Richard Biener <rguenther@suse.de>
14698 * tree-vectorizer.h (vect_dr_stmt): New function.
14699 (vect_get_load_cost): Adjust.
14700 (vect_get_store_cost): Likewise.
14701 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
14702 Use vect_dr_stmt instead of DR_SMTT.
14703 (vect_record_base_alignments): Likewise.
14704 (vect_calculate_target_alignment): Likewise.
14705 (vect_compute_data_ref_alignment): Likewise and make static.
14706 (vect_update_misalignment_for_peel): Likewise.
14707 (vect_verify_datarefs_alignment): Likewise.
14708 (vector_alignment_reachable_p): Likewise.
14709 (vect_get_data_access_cost): Likewise. Pass down
14710 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
14711 (vect_get_peeling_costs_all_drs): Likewise.
14712 (vect_peeling_hash_get_lowest_cost): Likewise.
14713 (vect_enhance_data_refs_alignment): Likewise.
14714 (vect_find_same_alignment_drs): Likewise.
14715 (vect_analyze_data_refs_alignment): Likewise.
14716 (vect_analyze_group_access_1): Likewise.
14717 (vect_analyze_group_access): Likewise.
14718 (vect_analyze_data_ref_access): Likewise.
14719 (vect_analyze_data_ref_accesses): Likewise.
14720 (vect_vfa_segment_size): Likewise.
14721 (vect_small_gap_p): Likewise.
14722 (vectorizable_with_step_bound_p): Likewise.
14723 (vect_prune_runtime_alias_test_list): Likewise.
14724 (vect_analyze_data_refs): Likewise.
14725 (vect_supportable_dr_alignment): Likewise.
14726 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
14727 (vect_gen_prolog_loop_niters): Likewise.
14728 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
14729 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
14731 (vect_recog_mask_conversion_pattern): Likewise.
14732 (vect_try_gather_scatter_pattern): Likewise.
14733 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
14734 to vect_get_store_cost.
14735 (vect_get_store_cost): Get stmt_info instead of DR.
14736 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
14737 (vect_get_load_cost): Get stmt_info instead of DR.
14739 2018-06-01 Richard Biener <rguenther@suse.de>
14741 PR middle-end/86017
14742 * gimple-fold.c (var_decl_component_p): Also allow offsetted
14743 vars wrapped in MEM_REFs.
14745 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
14747 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
14748 Fix subreg tests so that we only return a choice between
14749 GENERAL_REGS and FP_REGS if the original classes included both.
14751 2018-06-01 Richard Biener <rguenther@suse.de>
14754 * tree-ssa-structalias.c (get_function_part_constraint):
14755 Handle NULL fi->decl.
14756 (find_func_aliases_for_call): Properly handle indirect
14757 fi from direct call.
14758 (find_func_clobbers): Likewise.
14759 (ipa_pta_execute): Likewise.
14760 (create_variable_info_for): For functions that are ifunc_resolver
14761 resolve to a varinfo that contains the result of the resolver call.
14762 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
14765 2018-05-31 Michael Collison <michael.collison@arm.com>
14767 * config/aarch64/aarch64.md:
14768 (*fix_to_zero_extenddfdi2): New pattern.
14769 * gcc.target/aarch64/fix_extend1.c: New testcase.
14771 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
14773 PR middle-end/78809
14774 PR middle-end/83026
14775 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
14776 and BUILT_IN_STRNCMP_EQ.
14777 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
14778 BUILT_IN_STRNCMP_EQ.
14779 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
14780 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
14781 (gimple_fold_builtin): Likewise.
14782 * tree-ssa-strlen.c (compute_string_length): New function.
14783 (determine_min_obsize): New function.
14784 (handle_builtin_string_cmp): New function to handle calls to
14785 string compare functions.
14786 (strlen_optimize_stmt): Add handling to builtin string compare
14788 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
14789 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
14790 * tree.c (build_common_builtin_nodes): Add new defines of
14791 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
14793 2018-05-31 Jakub Jelinek <jakub@redhat.com>
14796 * bb-reorder.c (pass_partition_blocks::gate): Return false for
14797 functions with naked attribute.
14799 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
14801 * config/i386/sse.md (avx_vec_concat<mode>):
14802 Substitute concat_tg_mode mode attribute with xtg_mode.
14803 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
14804 (concat_tg_mode): Remove mode attribute.
14806 2018-05-31 Martin Sebor <msebor@redhat.com>
14809 * calls.c (alloc_max_size): Correct a logic error/typo.
14810 Treat excessive arguments as infinite. Warn for invalid arguments.
14811 * doc/invoke.texi (-Walloc-size-larger-than): Update.
14813 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
14816 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
14817 and movx for Haswell.
14819 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
14820 Cesar Philippidis <cesar@codesourcery.com>
14822 PR middle-end/85879
14823 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
14824 when emitting error on private/firstprivate reductions.
14825 * omp-low.c (lower_omp_target): Avoid reference-type processing
14826 on pointers for firstprivate clause.
14828 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
14830 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
14833 * config/aarch64/aarch64-simd.md
14834 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
14835 (aarch64_ld1_x3_<mode>): Likewise
14836 (aarch64_st1x2<VALLDIF:mode>): Likewise
14837 (aarch64_st1_x2_<mode>): Likewise
14838 (aarch64_st1x3<VALLDIF:mode>): Likewise
14839 (aarch64_st1_x3_<mode>): Likewise
14840 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
14841 (vld1_s8_x3): Likewise.
14842 (vld1_u16_x3): Likewise.
14843 (vld1_s16_x3): Likewise.
14844 (vld1_u32_x3): Likewise.
14845 (vld1_s32_x3): Likewise.
14846 (vld1_u64_x3): Likewise.
14847 (vld1_s64_x3): Likewise.
14848 (vld1_f16_x3): Likewise.
14849 (vld1_f32_x3): Likewise.
14850 (vld1_f64_x3): Likewise.
14851 (vld1_p8_x3): Likewise.
14852 (vld1_p16_x3): Likewise.
14853 (vld1_p64_x3): Likewise.
14854 (vld1q_u8_x3): Likewise.
14855 (vld1q_s8_x3): Likewise.
14856 (vld1q_u16_x3): Likewise.
14857 (vld1q_s16_x3): Likewise.
14858 (vld1q_u32_x3): Likewise.
14859 (vld1q_s32_x3): Likewise.
14860 (vld1q_u64_x3): Likewise.
14861 (vld1q_s64_x3): Likewise.
14862 (vld1q_f16_x3): Likewise.
14863 (vld1q_f32_x3): Likewise.
14864 (vld1q_f64_x3): Likewise.
14865 (vld1q_p8_x3): Likewise.
14866 (vld1q_p16_x3): Likewise.
14867 (vld1q_p64_x3): Likewise.
14868 (vst1_s64_x2): Likewise.
14869 (vst1_u64_x2): Likewise.
14870 (vst1_f64_x2): Likewise.
14871 (vst1_s8_x2): Likewise.
14872 (vst1_p8_x2): Likewise.
14873 (vst1_s16_x2): Likewise.
14874 (vst1_p16_x2): Likewise.
14875 (vst1_s32_x2): Likewise.
14876 (vst1_u8_x2): Likewise.
14877 (vst1_u16_x2): Likewise.
14878 (vst1_u32_x2): Likewise.
14879 (vst1_f16_x2): Likewise.
14880 (vst1_f32_x2): Likewise.
14881 (vst1_p64_x2): Likewise.
14882 (vst1q_s8_x2): Likewise.
14883 (vst1q_p8_x2): Likewise.
14884 (vst1q_s16_x2): Likewise.
14885 (vst1q_p16_x2): Likewise.
14886 (vst1q_s32_x2): Likewise.
14887 (vst1q_s64_x2): Likewise.
14888 (vst1q_u8_x2): Likewise.
14889 (vst1q_u16_x2): Likewise.
14890 (vst1q_u32_x2): Likewise.
14891 (vst1q_u64_x2): Likewise.
14892 (vst1q_f16_x2): Likewise.
14893 (vst1q_f32_x2): Likewise.
14894 (vst1q_f64_x2): Likewise.
14895 (vst1q_p64_x2): Likewise.
14896 (vst1_s64_x3): Likewise.
14897 (vst1_u64_x3): Likewise.
14898 (vst1_f64_x3): Likewise.
14899 (vst1_s8_x3): Likewise.
14900 (vst1_p8_x3): Likewise.
14901 (vst1_s16_x3): Likewise.
14902 (vst1_p16_x3): Likewise.
14903 (vst1_s32_x3): Likewise.
14904 (vst1_u8_x3): Likewise.
14905 (vst1_u16_x3): Likewise.
14906 (vst1_u32_x3): Likewise.
14907 (vst1_f16_x3): Likewise.
14908 (vst1_f32_x3): Likewise.
14909 (vst1_p64_x3): Likewise.
14910 (vst1q_s8_x3): Likewise.
14911 (vst1q_p8_x3): Likewise.
14912 (vst1q_s16_x3): Likewise.
14913 (vst1q_p16_x3): Likewise.
14914 (vst1q_s32_x3): Likewise.
14915 (vst1q_s64_x3): Likewise.
14916 (vst1q_u8_x3): Likewise.
14917 (vst1q_u16_x3): Likewise.
14918 (vst1q_u32_x3): Likewise.
14919 (vst1q_u64_x3): Likewise.
14920 (vst1q_f16_x3): Likewise.
14921 (vst1q_f32_x3): Likewise.
14922 (vst1q_f64_x3): Likewise.
14923 (vst1q_p64_x3): Likewise.
14925 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14927 * config/msp430/msp430.c (msp430_output_labelref): Prepend
14928 user_label_prefix to name.
14930 * tree-core.h: Update comment about the format of NAME string
14931 passed to handler in attribute_spec.
14933 * config/msp430/msp430.md: Remove erroneous subreg expression from
14934 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
14935 zero_extend{q,h}isi2.
14937 2018-05-30 Borislav Petkov <bp@suse.de>
14939 * doc/extend.texi: Document some architecture specific
14940 constraints and sort entries.
14942 2018-05-30 Martin Sebor <msebor@redhat.com>
14944 PR middle-end/85369
14945 * builtins.c (expand_builtin_stpcpy_1): New function.
14946 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
14947 only if the former succeeds.
14949 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
14951 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
14954 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
14956 * doc/invoke.texi (-flinker-output): Document
14958 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
14960 * passes.c (ipa_write_summaries): Only modify statements if body
14962 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
14963 incrementally linking.
14964 (ipa_passes): Likewise.
14965 * lto-cgraph.c (lto_output_node): When incrementally linking do not
14966 pass down resolution info.
14967 * common.opt (flag_incremental_link): Update info.
14968 * gcc.c (plugin specs): Turn flinker-output=* to
14969 -plugin-opt=-linker-output-known
14970 * toplev.c (compile_file): Also cut compilation when doing incremental
14972 * flag-types. (enum lto_partition_model): Add
14973 LTO_LINKER_OUTPUT_NOLTOREL.
14974 (invoke.texi): Add -flinker-output docs.
14975 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
14976 link same way as WPA; do not stream in dead initializers.
14978 * dwarf2out.c (dwarf2out_die_ref_for_decl,
14979 darf2out_register_external_decl): Support incremental link.
14981 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
14983 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
14985 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
14987 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
14988 it down to simple_object_copy_lto_debug_sections.
14989 (run_gcc): Determine incremental LTO link time and configure
14990 lto1 into non-wpa mode, disable renaming of debug sections.
14992 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
14994 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
14995 descriptions of various incorrectly documented functions.
14997 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
15000 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
15001 address check not strict.
15003 2018-05-30 Richard Biener <rguenther@suse.de>
15005 PR tree-optimization/85964
15006 * tracer.c (better_p): Drop initialized count check, we only
15007 call the function with initialized counts now.
15008 (find_best_successor): Do find a best edge if one
15009 has uninitialized count.
15010 (find_best_predecessor): Likewise. Do BB frequency check only
15011 if count is initialized.
15013 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
15015 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
15016 (aarch64_ldrstr_offset_compare): New.
15017 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
15018 load/store orderings.
15019 (aarch64_gen_adjusted_ldpstp): Likewise.
15021 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
15023 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
15024 Check for subset of GENERAL_REGS and FP_REGS.
15025 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
15028 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
15030 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
15031 and wi::to_poly_offset. Add the current offset and then check
15032 whether the sum fits, rather than using an unchecked addition of
15033 a checked term. Check for a shwi rather than a uhwi.
15034 * expr.c (get_bit_range): Use tree_to_poly_uint64.
15035 (store_constructor): Use poly_int_tree_p.
15036 (expand_expr_real_1): Likewise.
15037 * function.c (assign_temp): Likewise.
15038 * fold-const.c (const_binop): Use poly_int_tree_p and
15039 wi::to_poly_offset.
15040 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
15042 * ipa-icf-gimple.c (func_checker::compare_operand): Use
15043 to_poly_offset for MEM offsets.
15044 * ipa-icf.c (sem_variable::equals): Likewise.
15045 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
15046 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
15047 wi::to_poly_offset for BIT_FIELD_REF offsets.
15048 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
15049 wi::to_poly_offset.
15050 * var-tracking.c (emit_note_insn_var_location): Use
15051 tree_to_poly_uint64.
15053 2018-05-29 Jim Wilson <jimw@sifive.com>
15055 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
15057 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
15060 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
15061 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
15063 (sse4_1_round<mode>2): Use nonimmediate_operand
15064 for operand 1 predicate.
15066 2018-05-29 Martin Sebor <msebor@redhat.com>
15067 Richard Biener <rguenther@suse.de>
15070 * calls.c (get_size_range): Call determine_value_range instead
15071 of get_value_range..
15072 * tree-vrp.h (determine_value_range): Declared new function.
15073 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
15075 2018-05-29 Richard Biener <rguenther@suse.de>
15077 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
15078 sure to use non-pattern stmts for get_earlier_stmt arguments.
15079 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
15080 called on pattern stmts.
15081 (get_later_stmt): Likewise.
15083 2018-05-29 Martin Liska <mliska@suse.cz>
15085 PR gcov-profile/85759
15086 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
15089 2018-05-29 Jakub Jelinek <jakub@redhat.com>
15091 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
15093 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
15097 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
15098 VEC_PACK_FLOAT_EXPR): New tree codes.
15099 * tree-pretty-print.c (op_code_prio): Handle
15100 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
15101 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
15102 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
15103 * tree-inline.c (estimate_operator_cost): Likewise.
15104 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
15105 * fold-const.c (const_binop): Likewise.
15106 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
15107 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
15108 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
15109 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
15110 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
15111 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
15112 * expr.c (expand_expr_real_2): Likewise.
15113 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
15114 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
15115 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
15117 * optabs.c (expand_widen_pattern_expr): For
15118 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
15119 sign from result type rather than operand's type.
15120 (expand_binop_directly): For vec_packu_float_optab and
15121 vec_packs_float_optab allow result type to be different from operand's
15123 * optabs-tree.c (optab_for_tree_code): Handle
15124 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
15125 VEC_PACK_FLOAT_EXPR. Formatting fixes.
15126 * tree-vect-generic.c (expand_vector_operations_1): Handle
15127 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
15128 VEC_PACK_FLOAT_EXPR.
15129 * tree-vect-stmts.c (supportable_widening_operation): Handle
15131 (supportable_narrowing_operation): Handle FLOAT_EXPR.
15132 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
15133 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
15134 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
15135 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
15137 (vec_pack<floatprefix>_float_<mode>): New expander.
15138 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
15140 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
15141 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
15142 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
15143 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
15144 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
15146 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
15147 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
15148 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
15149 VEC_PACK_FLOAT_EXPR): Document.
15151 2018-05-29 Richard Biener <rguenther@suse.de>
15153 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
15155 (stmt_vec_info_vec): Make pointer.
15156 (init_stmt_vec_info_vec): Remove.
15157 (free_stmt_vec_info_vec): Likewise.
15158 (set_stmt_vec_info_vec): New function.
15159 (free_stmt_vec_infos): Likewise.
15160 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
15161 (set_vinfo_for_stmt): Likewise.
15162 (get_earlier_stmt): Likewise.
15163 (get_later_stmt): Likewise.
15164 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
15165 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
15166 (vec_info::~vec_info): Free stmt_vec_infos.
15167 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
15168 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
15169 (pass_slp_vectorize::execute): Likewise.
15170 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
15171 (free_stmt_vec_info_vec): Likewise.
15172 (set_stmt_vec_info_vec): New function.
15173 (free_stmt_vec_infos): Likewise.
15174 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
15175 the global stmt_vec_info_vec.
15176 * tree-parloops.c (gather_scalar_reductions): Use
15177 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
15180 2018-05-29 Richard Biener <rguenther@suse.de>
15182 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
15184 2018-05-29 Martin Liska <mliska@suse.cz>
15185 David Malcolm <dmalcolm@redhat.com>
15187 * vec.c (test_reverse): New.
15188 (vec_c_tests): Add new test.
15189 * vec.h (vl_ptr>::reverse): New function.
15191 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
15193 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
15195 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
15198 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
15200 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
15202 2018-05-28 Richard Biener <rguenther@suse.de>
15204 PR tree-optimization/85933
15205 * tree-vect-data-refs.c (vect_record_base_alignments): Only
15206 look at stmts marked as vectorizable.
15208 2018-05-28 Richard Biener <rguenther@suse.de>
15210 PR tree-optimization/85934
15211 * tree-vect-generic.c (expand_vector_operations_1): Hoist
15212 vector boolean check before scalar optimization.
15214 2018-05-28 Jakub Jelinek <jakub@redhat.com>
15216 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
15219 2018-05-28 Mark Wielaard <mark@klomp.org>
15221 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
15222 if it is an expression containing a minus sign.
15224 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
15226 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
15228 2018-05-27 Paul Koning <ni1d@arrl.net>
15230 * config/pdp11/pdp11.md (truncsihi2): Remove.
15232 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
15233 Chung-Ju Wu <jasonwucj@gmail.com>
15235 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
15237 (unaligned_store_dw): Ditto.
15238 * config/nds32/nds32-memory-manipulation.c
15239 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
15240 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
15241 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
15242 (emit_setmem_word_loop): Rename to ...
15243 (emit_setmem_doubleword_loop): ... this.
15244 (nds32_gen_dup_4_byte_to_word_value): New function.
15245 (nds32_gen_dup_8_byte_to_double_word_value): New function.
15246 (nds32_expand_setmem_loop): Refine implementation.
15247 (nds32_expand_setmem_loop_v3m): Ditto.
15248 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
15251 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
15253 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
15255 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
15257 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
15258 (nds32_init_machine_status): Initialize machine->attr_naked_p and
15259 machine->attr_no_prologue_p.
15260 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
15261 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
15262 (nds32_expand_epilogue): Consider attr_naked_p.
15263 (nds32_expand_epilogue_v3pop): Likewise.
15264 (nds32_can_use_return_insn): Likewise.
15265 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
15266 attr_no_prologue_p fields.
15267 * config/nds32/nds32.opt (mret-in-naked-func): New option.
15269 2018-05-27 Jakub Jelinek <jakub@redhat.com>
15272 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
15274 * config/i386/sse.md
15275 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
15277 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
15279 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
15281 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
15283 (*<floatsuffix>floatv2div2sf2): Rename to ...
15284 (*float<floatunssuffix>v2div2sf2): ... this.
15285 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
15286 (float<floatunssuffix>v2div2sf2_mask): ... this.
15287 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
15288 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
15289 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
15291 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
15293 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
15295 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
15297 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
15299 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
15301 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
15302 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
15303 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
15304 gen_ufix_truncv8dfv8si2.
15305 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
15306 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
15307 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
15308 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
15309 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
15310 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
15311 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
15312 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
15314 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
15318 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
15320 2018-05-25 Jim Wilson <jimw@sifive.com>
15322 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
15323 * config/riscv/riscv.c (struct machine_function): Add
15324 interrupt_handler_p and attribute_checked_p fields.
15325 (riscv_attribute_table): Add interrupt.
15326 (riscv_interrupt_type_p): New.
15327 (riscv_save_reg_p): Save extra regs for interrupt handler.
15328 (riscv_use_save_libcall): Return false for interrupt handler.
15329 (riscv_first_stack_step): Add forward declaration.
15330 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
15331 for interrupt handler with large frame. Use it for saved reg list.
15332 (riscv_expand_prologue): Move flag_stack_usage_info support to
15333 eliminate duplication.
15334 (riscv_expand_epilogue): Generate mret for interrupt handler.
15335 (riscv_epilogue_uses): New.
15336 (riscv_can_use_return_insn): Return false for interrupt handler.
15337 (riscv_function_ok_for_sibcall): Likewise.
15338 (riscv_set_current_function): Add interrupt handler support.
15339 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
15340 * config/riscv/riscv.md (UNSPECV_MRET): New.
15342 (riscv_frflags, riscv_fsflags): Use tab after opcode.
15344 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
15346 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
15348 PR tree-optimization/85712
15349 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
15350 this candidate has already been replaced in-situ by a copy.
15352 2018-05-25 Jason Merrill <jason@redhat.com>
15354 PR c++/80485 - inline function non-zero address.
15355 * symtab.c (nonzero_address): Check DECL_COMDAT.
15357 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
15360 * config/alpha/alpha.md (ashlsi3): New insn pattern.
15361 (*ashlsi_se): Rename from *ashldi_se. Define as sign
15362 extension of SImode operation. Use const123_operand predicate.
15363 (*saddsi_1): Remove.
15364 (*saddl_se_1): Ditto.
15365 (*ssubsi_1): Ditto.
15366 (*ssubl_se_1): Ditto.
15367 * config/alpha/predicates.md (const123_operand): New predicate.
15368 * config/alpha/constraints.md (P): Use IN_RANGE.
15370 2018-05-25 Richard Biener <rguenther@suse.de>
15372 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
15374 (ref_maybe_used_by_stmt_p): Likewise.
15375 (stmt_may_clobber_ref_p): Likewise.
15376 (stmt_may_clobber_ref_p_1): Likewise.
15377 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
15379 (ref_maybe_used_by_stmt_p): Likewise.
15380 (stmt_may_clobber_ref_p): Likewise.
15381 (stmt_may_clobber_ref_p_1): Likewise.
15382 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
15383 the alias oracle to disambiguate DRs with stmts DR analysis
15385 (vect_analyze_data_refs): Do not give up on not analyzable
15386 DRs for BB vectorization. Remove code truncating the dataref
15389 2018-05-25 Jakub Jelinek <jakub@redhat.com>
15392 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
15393 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
15394 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
15396 2018-05-25 Richard Biener <rguenther@suse.de>
15398 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
15399 function, combining stmt data ref gathering and fatal analysis
15401 (vect_analyze_data_refs): Remove now redudnant code and simplify.
15402 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
15403 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
15404 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
15405 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
15407 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15409 PR tree-optimization/85720
15410 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
15411 SCC if all partitions are builtins.
15412 (version_loop_by_alias_check): New parameter. Generate cancelable
15413 runtime alias check if all partitions are builtins.
15414 (distribute_loop): Update call to above function.
15416 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15418 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
15419 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
15420 (parm_default_def_partition_arg): Ditto.
15421 (set_parm_default_def_partition): Ditto.
15422 (get_parm_default_def_partitions): Ditto and make it static.
15423 (get_undefined_value_partitions): Ditto and make it static.
15424 (remove_ssa_form): Refactor call to init_var_map here.
15425 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
15426 computation for loop region.
15427 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
15428 (register_default_def): Delete.
15429 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
15430 (parm_default_def_partition_arg): Ditto.
15431 (set_parm_default_def_partition): Ditto.
15432 (get_parm_default_def_partitions): Ditto and make it static.
15433 (get_undefined_value_partitions): Ditto and make it static.
15434 (coalesce_with_default, coalesce_with_default): Update comment.
15435 (create_coalesce_list_for_region): New func factored out from
15436 create_outofssa_var_map.
15437 (populate_coalesce_list_for_outofssa): New func factored out from
15438 create_outofssa_var_map and coalesce_ssa_name.
15439 (create_outofssa_var_map): Delete.
15440 (coalesce_ssa_name): Refactor to support live range computation.
15441 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
15442 (get_parm_default_def_partitions): Delete.
15443 (get_undefined_value_partitions): Ditto.
15444 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
15445 computation for loop region.
15446 (new_tree_live_info, loe_visit_block): Ditto.
15447 (live_worklist, set_var_live_on_entry): Ditto.
15448 (calculate_live_on_exit, verify_live_on_entry): Ditto.
15449 * tree-ssa-live.h (struct _var_map): New fields.
15450 (init_var_map): Change decl.
15451 (region_contains_p): New.
15453 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15455 * tree-ssa-live.h (live_merge_and_clear): Delete.
15457 2018-05-25 Richard Biener <rguenther@suse.de>
15460 * tree-dump.c (dequeue_and_dump): Remove access to removed
15461 operand 2 of a SWITCH_EXPR.
15463 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15465 * doc/sourcebuild.texi (vect_double_cond_arith): Include
15466 multiplication and division.
15467 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
15468 (cond_udiv@var{m}, cond_umod@var{m}): Document.
15469 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
15470 (cond_udiv_optab, cond_umod_optab): New optabs.
15471 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
15472 (IFN_COND_RDIV): New internal functions.
15473 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
15474 TRUNC_MOD_EXPR and RDIV_EXPR.
15475 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
15476 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
15478 (SVE_INT_BINARY): Include mult.
15479 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
15480 (optab, sve_int_op): Handle mult.
15481 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
15483 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
15484 for SVE_INT_BINARY_SD.
15486 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15488 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
15489 (optab, sve_int_op): Handle div and udiv.
15490 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
15491 for SVE_INT_BINARY_SD.
15492 (*<optab><mode>3): New insn for the same.
15494 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15496 * tree-vect-patterns.c: Include predict.h.
15497 (vect_recog_divmod_pattern): Restrict check for division support
15498 to when optimizing for size.
15500 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15502 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
15503 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
15504 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
15505 (gimple_match_op::set_op): Likewise.
15506 (gimple_resimplify4): Declare.
15507 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
15508 (expr::gen_transform): Likewise.
15509 (decision_tree::gen): Generate a simplification routine for 4 operands.
15510 * gimple-match-head.c (gimple_simplify): Add an overload for
15511 4 operands. In the top-level function, handle up to 4 call
15512 arguments and call gimple_resimplify4.
15513 (gimple_resimplify4): New function.
15514 (build_call_internal): Pass a fourth operand.
15515 (maybe_push_to_seq): Likewise.
15516 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
15517 Fold VEC_COND_EXPRs of an operation and a default value into
15518 an IFN_COND_* function if possible.
15519 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
15521 (SVE_COND_FP_BINARY): Include them.
15522 (optab, sve_fp_op): Handle them.
15523 (SVE_INT_BINARY_REV): New code iterator.
15524 (SVE_COND_FP_BINARY_REV): New int iterator.
15525 (commutative): New int attribute.
15526 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
15528 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
15530 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
15531 (*cond_<optab><mode>): New patterns for reversed operands.
15533 2018-05-25 Richard Biener <rguenther@suse.de>
15535 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
15536 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
15537 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
15538 (STMT_VINFO_GROUPED_ACCESS): Adjust.
15539 * tree-vect-data-refs.c (everywhere): Adjust users.
15540 * tree-vect-loop.c (everywhere): Likewise.
15541 * tree-vect-slp.c (everywhere): Likewise.
15542 * tree-vect-stmts.c (everywhere): Likewise.
15543 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
15545 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15547 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
15549 (gcc_cv_as_section_exclude): ... this.
15550 Try Solaris as #exclude syntax.
15551 * configure: Regenerate.
15552 * config.in: Regenerate.
15553 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
15555 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
15556 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
15558 * varasm.c (default_elf_asm_named_section): Don't check if
15559 HAVE_GAS_SECTION_EXCLUDE is defined.
15561 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15563 * doc/md.texi: Update the documentation of the cond_* optabs
15564 to mention the new final operand. Fix GET_MODE_NUNITS call.
15565 Describe the scalar case too.
15566 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
15567 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
15569 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
15570 (get_conditional_internal_fn): Update comment.
15571 * tree-vect-loop.c (vectorizable_reduction): Pass the original
15572 accumulator value as a final argument to conditional functions.
15573 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
15574 a define_expand and add an "else" operand. Assert for now that
15575 the else operand is equal to operand 2. Use SVE_INT_BINARY and
15576 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
15577 (*cond_<optab><mode>): New patterns.
15578 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
15579 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
15580 (UNSPEC_COND_EOR): Delete.
15581 (optab): Remove associated mappings.
15582 (SVE_INT_BINARY): New code iterator.
15583 (sve_int_op): Remove int attribute and add "minus" to the code
15585 (SVE_COND_INT_OP): Delete.
15586 (SVE_COND_FP_OP): Rename to...
15587 (SVE_COND_FP_BINARY): ...this.
15589 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15591 * optabs.c (can_reuse_operands_p): New function.
15592 (maybe_legitimize_operands): Try to reuse the results for
15595 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
15597 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
15598 Add {q} suffix to insn mnemonic.
15600 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
15602 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
15603 (msp430_warn_func_return): New.
15605 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
15607 * fold-const.c (tree_nonzero_bits): New function.
15608 * fold-const.h (tree_nonzero_bits): Likewise.
15609 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
15610 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
15612 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
15616 * varasm.c (assemble_alias): Check ifunc_resolver only on
15619 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
15622 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
15623 when memory input operand is handled.
15625 2018-05-24 Luis Machado <luis.machado@linaro.org>
15627 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
15629 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
15631 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
15633 * match.pd: Delay FMA folds until after vectorization.
15635 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
15638 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
15639 address check not strict.
15641 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
15643 * gimple-match.h (gimple_match_op): New class.
15644 (mprts_hook): Replace parameters with a gimple_match_op *.
15645 (maybe_build_generic_op): Likewise.
15646 (gimple_simplified_result_is_gimple_val): Replace parameters with
15647 a const gimple_match_op *.
15648 (gimple_simplify): Replace code_helper * and tree * parameters with
15649 a gimple_match_op * parameter.
15650 (gimple_resimplify1): Replace code_helper *, tree and tree *
15651 parameters with a gimple_match_op * parameter.
15652 (gimple_resimplify2): Likewise.
15653 (gimple_resimplify3): Likewise.
15654 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
15655 parameters with a gimple_match_op * parameter.
15656 * gimple-match-head.c (gimple_simplify): Change prototypes of
15657 auto-generated functions to take a gimple_match_op * instead of
15658 separate code_helper * and tree * parameters. Make the same
15659 change in the top-level overload and update calls to the
15660 gimple_resimplify routines. Update calls to the auto-generated
15661 functions and to maybe_push_res_to_seq in the publicly-facing
15662 operation-specific gimple_simplify overloads.
15663 (gimple_match_op::MAX_NUM_OPS): Define.
15664 (gimple_resimplify1): Replace rcode and ops with a single res_op
15665 parameter. Update call to gimple_simplify.
15666 (gimple_resimplify2): Likewise.
15667 (gimple_resimplify3): Likewise.
15668 (mprts_hook): Replace parameters with a gimple_match_op *.
15669 (maybe_build_generic_op): Likewise.
15670 (build_call_internal): Replace type, nargs and ops with
15671 a gimple_match_op *.
15672 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
15673 with a single gimple_match_op *. Update calls to mprts_hook,
15674 build_call_internal and gimple_simplified_result_is_gimple_val.
15675 Factor out code that is common to the tree_code and combined_fn cases.
15676 * genmatch.c (expr::gen_transform): Replace tem_code and
15677 tem_ops with a gimple_match_op called tem_op. Update calls
15678 to the gimple_resimplify functions and maybe_push_res_to_seq.
15679 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
15680 res_ops. Update call to the gimple_resimplify functions.
15681 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
15682 (decision_tree::gen): Make the functions take a gimple_match_op *
15683 called res_op instead of separate res_code and res_ops parameters.
15684 Update call accordingly.
15685 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
15686 and ops with a single res_op parameter. Update calls to
15687 maybe_build_generic_op and maybe_push_res_to_seq.
15688 (fold_stmt_1): Update calls to gimple_simplify and
15689 replace_stmt_with_simplification.
15690 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
15691 and gimple_simplified_result_is_gimple_val.
15692 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
15694 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
15695 with a gimple_match_op *.
15696 (vn_nary_build_or_lookup): Likewise. Update call to
15697 vn_nary_build_or_lookup_1.
15698 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
15699 gimple_match_op *. Update calls to the gimple_resimplify routines
15700 and to gimple_simplified_result_is_gimple_val.
15701 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
15702 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
15703 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
15704 (visit_nary_op): Likewise.
15705 (visit_reference_op_load): Likewise.
15707 2018-05-23 Luis Machado <luis.machado@linaro.org>
15709 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
15710 modifier for printing the step amount.
15712 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15715 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
15718 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
15720 * doc/sourcebuild.texi (Endianness): New subsubsection.
15722 2018-05-23 Luis Machado <luis.machado@linaro.org>
15724 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
15725 <prefetch_dynamic_strides>: New const bool field.
15726 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
15727 prefetch_dynamic_strides.
15728 (exynosm1_prefetch_tune): Likewise.
15729 (thunderxt88_prefetch_tune): Likewise.
15730 (thunderx_prefetch_tune): Likewise.
15731 (thunderx2t99_prefetch_tune): Likewise.
15732 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
15734 (aarch64_override_options_internal): Update to set
15735 PARAM_PREFETCH_DYNAMIC_STRIDES.
15736 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
15737 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
15738 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
15739 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
15740 prefetch-dynamic-strides setting.
15742 2018-05-23 Luis Machado <luis.machado@linaro.org>
15744 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
15745 <minimum_stride>: New const int field.
15746 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
15747 minimum_stride field defaulting to -1.
15748 (exynosm1_prefetch_tune): Likewise.
15749 (thunderxt88_prefetch_tune): Likewise.
15750 (thunderx_prefetch_tune): Likewise.
15751 (thunderx2t99_prefetch_tune): Likewise.
15752 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
15753 <default_opt_level>: Set to 3.
15754 (aarch64_override_options_internal): Update to set
15755 PARAM_PREFETCH_MINIMUM_STRIDE.
15756 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
15757 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
15758 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
15759 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
15760 stride is constant and is below the minimum stride threshold.
15762 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15764 * config/arm/arm-cpus.in (mode26): Delete.
15765 (armv4): Delete mode26 reference.
15766 * config/arm/arm.c (arm_configure_build_target): Delete use of
15769 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
15771 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
15773 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
15774 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
15776 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
15777 (floatunsdidf2): Ditto.
15779 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
15781 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
15782 (fixuns_trunc<mode>si2_avx512f): Ditto.
15783 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
15784 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
15785 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
15787 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
15789 PR rtl-optimization/79985
15790 * df-scan.c (df_insn_refs_collect): Remove special case for
15791 global registers and asm statements.
15793 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
15795 * extend.texi (Global Register Variables): Rewrite the bullet list.
15796 Note that the register is available for allocation. Note that access
15797 via inline asm must use constraints. Add note about async-signal
15798 handlers. Remove paragraph about automagic register selection.
15800 2018-05-23 Richard Biener <rguenther@suse.de>
15802 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
15803 of fixed offset from memset VN.
15805 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
15807 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
15808 first_interp field.
15809 (alloc_cand_and_find_basis): Initialize first_interp field.
15810 (slsr_process_mul): Modify first_interp field.
15811 (slsr_process_add): Likewise.
15812 (slsr_process_cast): Modify first_interp field for each new
15814 (slsr_process_copy): Likewise.
15815 (dump_candidate): Dump first_interp field.
15816 (replace_mult_candidate): Process all interpretations, not just
15818 (replace_rhs_if_not_dup): Likewise.
15819 (replace_one_candidate): Likewise.
15821 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
15823 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
15825 (aarch64_needs_frame_chain): New function.
15826 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
15828 2018-05-23 Sudakshina Das <sudi.das@arm.com>
15831 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
15832 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
15833 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
15834 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
15835 as true for strict-align.
15836 (aarch64_can_inline_p): Perform checks even when callee has no
15837 attributes to check for strict alignment.
15838 * doc/extend.texi (AArch64 Function Attributes): Document
15840 * doc/invoke.texi: (AArch64 Options): Likewise.
15842 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
15844 PR tree-optimization/85853
15845 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
15846 the handling of the root of the node to...
15847 (vect_slp_analyze_node_operations_1): ...this new function,
15848 and run the whole thing with the child nodes' def types
15849 set according to their SLP node's def type.
15851 2018-05-23 Richard Biener <rguenther@suse.de>
15853 PR middle-end/85874
15854 * tree-data-ref.c (create_runtime_alias_checks): Defer
15855 and ignore overflow warnings.
15857 2018-05-23 Yury Gribov <tetra2005@gmail.com>
15859 PR tree-optimization/85822
15860 * tree-vrp.c (is_masked_range_test): Fix handling of negative
15863 2018-05-23 Richard Biener <rguenther@suse.de>
15865 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
15866 memset constants via native_interpret_expr.
15868 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
15871 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
15873 (cgraph_node::create_alias): Likewise.
15874 (cgraph_node::get_availability): Check ifunc_resolver instead
15875 of looking up ifunc attribute.
15876 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
15877 * varasm.c (do_assemble_alias): Likewise.
15878 (assemble_alias): Likewise.
15879 (default_binds_local_p_3): Likewise.
15880 * cgraph.h (cgraph_node): Add ifunc_resolver.
15881 (cgraph_node::only_called_directly_or_aliased_p): Return false
15882 for IFUNC resolver.
15883 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
15885 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
15886 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
15887 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
15888 instead of looking up ifunc attribute.
15890 2018-05-22 Luis Machado <luis.machado@linaro.org>
15892 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
15894 2018-05-22 Martin Sebor <msebor@redhat.com>
15896 PR middle-end/85359
15897 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
15898 only when expasion succeeds.
15899 (expand_builtin_strcmp): Same.
15900 (expand_builtin_strncmp): Same.
15902 2018-05-22 Martin Sebor <msebor@redhat.com>
15904 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
15906 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
15907 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15909 * config/aarch64/aarch64-ldpstp.md: Replace uses of
15910 aarch64_mem_pair_operand with memory_operand and delete operand swapping
15912 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
15913 Add check for legitimate_address.
15914 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
15915 (aarch64_swap_ldrstr_operands): New.
15916 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
15919 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
15920 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15922 * config/aarch64/aarch64.md: New patterns to generate stp
15924 (store_pair_sw, store_pair_dw): New patterns to generate stp for
15925 single words and double words.
15926 (load_pair_sw, load_pair_dw): Likewise.
15927 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
15929 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
15931 * config/aarch64/aarch64-ldpstp.md: Modify peephole
15932 for different mode ldpstp and add peephole for merged zero stores.
15933 Likewise for loads.
15934 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
15936 (aarch64_gen_store_pair): Rename calls to match new patterns.
15937 (aarch64_gen_load_pair): Rename calls to match new patterns.
15938 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
15939 (load_pair<DREG:mode><DREG2:mode>): ... This.
15940 (store_pair<mode>): Rename to...
15941 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
15942 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
15943 New mode iterators.
15944 (V_INT_EQUIV): Handle SImode.
15945 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
15948 2018-05-22 Martin Sebor <msebor@redhat.com>
15951 * calls.c (maybe_warn_nonstring_arg): Use string length to set
15952 or ajust the presumed bound on an operation to avoid unnecessary
15955 2018-05-22 Martin Sebor <msebor@redhat.com>
15957 PR tree-optimization/85826
15958 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
15959 assuming that a DECL necesarily has a constant size.
15961 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
15963 PR middle-end/85862
15964 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
15966 2018-05-22 Richard Biener <rguenther@suse.de>
15968 PR tree-optimization/85834
15969 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
15970 non-constant and non-zero memset arguments.
15972 2018-05-22 Martin Liska <mliska@suse.cz>
15975 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
15977 2018-05-22 Richard Biener <rguenther@suse.de>
15979 PR tree-optimization/85863
15980 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
15981 comparisons when vectype is specified.
15982 (vectorizable_condition): Do not specify vectype for
15983 vect_is_simple_cond when SLP vectorizing.
15985 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
15988 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
15989 define __ibm128 as long double.
15990 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
15991 as a distinct type when IEEE 128-bit support is enabled.
15992 (init_float128_ieee): Fix up conversions between IFmode and IEEE
15993 128-bit types to use the correct functions.
15994 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
15995 convert between 128-bit floating point types that have different
15996 modes but the same representation, instead of using gen_lowpart to
15998 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
16000 (IFKF_reg): New attributes to give the register constraints for
16002 (extend<mode>tf2_internal): New insns to mark an explicit
16003 conversion between 128-bit floating point types that have a
16004 different mode but share the same representation.
16006 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
16008 PR tree-optimization/85814
16009 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
16010 a null return from get_strinfo when unsharing the next
16011 strinfo in the chain.
16013 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
16016 * varasm.c (weak_finish): Clean up weak_decls.
16018 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16020 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
16021 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
16022 UNSPEC_UADALP values.
16023 * config/aarch64/iterators.md (ABAL): New int iterator.
16026 (sur): Add mappings for the above.
16027 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
16029 (aarch64_<sur>abal<mode>_4): Likewise.
16030 (aarch64_<sur>adalp<mode>_3): Likewise.
16031 (<sur>sadv16qi): New define_expand.
16033 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
16035 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
16036 (*movdf_internal): Ditto.
16037 (*rcpsf2_sse): Ditto.
16038 (*rsqrtsf2_sse): Ditto.
16039 (*sqrt<mode>2_sse): Ditto.
16041 2018-05-21 Tamar Christina <tamar.christina@arm.com>
16043 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
16045 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
16046 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
16047 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
16048 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
16050 * config/aarch64/arm_neon.h: Likewise.
16051 * config/aarch64/iterators.md (VQ_I): New.
16053 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
16055 * config.gcc: Add arc/t-multilib-linux to tmake_file for
16057 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
16060 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
16062 * config/nds32/constraints.md (S): New constraint.
16063 * config/nds32/nds32.md (call_internal): Use constraint S.
16064 (call_value_internal): Likewise.
16065 (sibcall_internal): Likewise.
16066 (sibcall_value_internal): Likewise.
16068 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
16069 Chung-Ju Wu <jasonwucj@gmail.com>
16071 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
16072 into consideration.
16074 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
16075 Chung-Ju Wu <jasonwucj@gmail.com>
16077 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
16078 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
16079 (nds32_rtx_costs_impl): Simplify.
16080 (nds32_address_cost_impl): Simplify.
16081 (nds32_init_rtx_costs): New function.
16082 (nds32_rtx_costs_speed_prefer): Likewise.
16083 (nds32_rtx_costs_size_prefer): Likewise.
16084 (nds32_address_cost_speed_prefer): Likewise.
16085 (nds32_address_cost_speed_fwprop): Likewise.
16086 (nds32_address_cost_size_prefer): Likewise.
16087 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
16088 * config/nds32/nds32.c (nds32_option_override): Use
16089 nds32_init_rtx_costs function.
16091 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
16093 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
16094 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
16095 (TARGET_PIPELINE_N8): Likewise.
16096 (TARGET_PIPELINE_N10): Likewise.
16097 (TARGET_PIPELINE_N13): Likewise.
16098 (TARGET_PIPELINE_GRAYWOLF): Likewise.
16100 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
16102 * config/nds32/nds32-fpu.md: Update copyright year.
16104 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16106 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
16108 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16110 * config/nds32/nds32.c
16111 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
16112 * config/nds32/nds32.opt (minline-asm-r15): New option.
16114 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16116 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
16118 * config/nds32/nds32.md (abssi2): New pattern.
16120 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
16122 * config/i386/i386.md (rex64namesuffix): New mode attribute.
16123 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
16124 Merge insn pattern from sse_cvtsi2ss<round_name> and
16125 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
16126 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
16127 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
16128 using SWI48 mode iterator.
16129 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
16130 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
16131 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
16132 pattern from sse_cvttss2si<round_saeonly_name>
16133 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
16134 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
16135 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
16136 using SWI48 mode iterator.
16137 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
16138 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
16139 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
16140 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
16141 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
16142 using SWI48 mode iterator.
16143 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
16144 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
16145 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
16146 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
16147 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
16148 SWI48 mode iterator.
16149 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
16150 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
16151 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
16152 pattern from sse_cvttsd2si<round_saeonly_name>
16153 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
16155 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16157 * config/nds32/nds32-md-auxiliary.c
16158 (nds32_valid_smw_lwm_base_p): Refine.
16159 (nds32_output_smw_single_word): Refine.
16160 (nds32_output_smw_double_word): New.
16161 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
16163 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16165 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
16166 (nds32_output_stack_pop): Refine.
16167 (nds32_expand_unaligned_load): Refine.
16168 (nds32_expand_unaligned_store): Refine.
16170 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16171 Chung-Ju Wu <jasonwucj@gmail.com>
16173 * config/nds32/constants.md: Add TP_REGNUM constant.
16174 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
16175 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
16177 * config/nds32/nds32-doubleword.md: Consider flag_pic.
16178 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
16179 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
16180 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
16181 and PIC code generation.
16182 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
16184 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
16186 * config/nds32/nds32.md: Support TLS and PIC.
16187 * config/nds32/nds32.c: Support TLS and PIC.
16188 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
16189 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
16192 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16194 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
16195 mode with E_ prefix.
16197 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16198 Chung-Ju Wu <jasonwucj@gmail.com>
16200 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
16201 * config/nds32/nds32-md-auxiliary.c
16202 (symbolic_reference_mentioned_p): New.
16203 (nds32_legitimize_ict_address): New.
16204 (nds32_expand_ict_move): New.
16205 (nds32_indirect_call_referenced_p): New.
16206 (nds32_symbol_binds_local_p): Delete.
16207 (nds32_long_call_p): Modify.
16208 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
16209 * config/nds32/nds32-protos.h
16210 (symbolic_reference_mentioned_p): Declare.
16211 (nds32_legitimize_ict_address): Declare.
16212 (nds32_expand_ict_move): Declare.
16213 (nds32_indirect_call_referenced_p): Declare.
16214 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
16215 (nds32_relax_group): Use nds32_ict_const_p as condition.
16216 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
16217 (nds32_asm_file_start): Output ict_model directive in asm code.
16218 (nds32_legitimate_address_p): Consider indirect call.
16219 (nds32_print_operand): Consider indirect call.
16220 (nds32_print_operand_address): Consider indirect call.
16221 (nds32_insert_attributes): Handle "indirect_call" attribute.
16222 (TARGET_LEGITIMATE_ADDRESS_P): Define.
16223 (TARGET_LEGITIMATE_CONSTANT_P): Define.
16224 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
16225 (TARGET_DELEGITIMIZE_ADDRESS): Define.
16226 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
16227 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
16228 (TARGET_ICT_MODEL_SMALL): Define.
16229 (TARGET_ICT_MODEL_LARGE): Define.
16230 * config/nds32/nds32.md (movsi): Consider ict model.
16231 (call, call_value): Consider ict model.
16232 (sibcall, sibcall_value): Consider ict model.
16233 * config/nds32/nds32.opt (mict-model): New option.
16234 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
16237 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
16238 Monk Chiang <sh.chiang04@gmail.com>
16239 Jim Wilson <jimw@sifive.com>
16241 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
16242 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
16243 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
16244 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
16245 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
16246 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
16247 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
16248 compute save_libcall_adjustment properly.
16249 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
16250 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
16251 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
16252 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
16253 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
16254 (ABI_SPEC): Handle mabi=ilp32e.
16255 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
16256 (RVE): Add RVE mask.
16257 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
16258 <-march>: Add rv32e as an example.
16260 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
16263 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
16264 (intra_create_variable_infos): Handle C++ constructors.
16266 2018-05-18 Martin Liska <mliska@suse.cz>
16268 * passes.def: Remove a redundant pass.
16270 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
16273 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
16275 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16277 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
16279 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
16281 (armv2, armv2a, armv3, armv3m): Delete architectures.
16282 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
16283 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
16284 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
16286 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
16287 (*mulsidi3adddi): Likewise.
16288 (mulsidi3): Likewise.
16289 (*mulsidi3_nov6): Likewise.
16290 (umulsidi3): Likewise.
16291 (umulsidi3_nov6): Likewise.
16292 (umaddsidi4): Likewise.
16293 (*umulsidi3adddi): Likewise.
16294 (smulsi3_highpart): Likewise.
16295 (*smulsi3_highpart_nov6): Likewise.
16296 (umulsi3_highpart): Likewise.
16297 (*umulsi3_highpart_nov6): Likewise.
16298 * config/arm/arm.h (arm_arch3m): Delete.
16299 * config/arm/arm.c (arm_arch3m): Delete.
16300 (arm_option_override_internal): Update armv3-related comment.
16301 (arm_configure_build_target): Delete use of isa_bit_mode32.
16302 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
16303 (arm_rtx_costs_internal): Delete check of arm_arch3m.
16304 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
16305 (mulsa3): Likewise.
16306 (mulusa3): Likewise.
16307 * config/arm/arm-protos.h (arm_arch3m): Delete.
16308 * config/arm/arm-tables.opt: Regenerate.
16309 * config/arm/arm-tune.md: Likewise.
16310 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
16311 deleted architectures.
16313 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16315 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
16316 (armv5t, armv5te): New features.
16317 (ARMv5, ARMv5e): Delete fgroups.
16318 (ARMv5t, ARMv5te): Adjust for above changes.
16319 (ARMv6m): Likewise.
16320 (armv5, armv5e): Delete arches.
16321 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
16323 (*call_reg_arm): Likewise.
16324 (*call_value_reg_armv5): Likewise.
16325 (*call_value_reg_arm): Likewise.
16326 (*call_symbol): Likewise.
16327 (*call_value_symbol): Likewise.
16328 (*sibcall_insn): Likewise.
16329 (*sibcall_value_insn): Likewise.
16330 (clzsi2): Likewise.
16331 (prefetch): Likewise.
16332 (define_split and define_peephole2 dependent on arm_arch5):
16334 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
16336 (TARGET_ARM_QBIT): Likewise.
16337 (TARGET_DSP_MULTIPLY): Likewise.
16338 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
16339 (arm_arch5, arm_arch5e): Delete.
16340 (arm_arch5t, arm_arch5te): Declare.
16341 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
16342 (arm_arch5t): Declare.
16343 (arm_option_reconfigure_globals): Update for the above.
16344 (arm_options_perform_arch_sanity_checks): Update comment, replace
16345 use of arm_arch5 with arm_arch5t.
16346 (use_return_insn): Likewise.
16347 (arm_emit_call_insn): Likewise.
16348 (output_return_instruction): Likewise.
16349 (arm_final_prescan_insn): Likewise.
16350 (arm_coproc_builtin_available): Likewise.
16351 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
16352 arm_arch5e with arm_arch5t and arm_arch5te.
16353 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
16354 (arm_arch5t, arm_arch5te): Declare.
16355 * config/arm/arm-tables.opt: Regenerate.
16356 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
16357 * config/arm/t-multilib: Likewise.
16358 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
16359 instead of arm_arch5.
16360 (*call_reg_thumb1): Likewise.
16361 (*call_value_reg_thumb1_v5): Likewise.
16362 (*call_value_reg_thumb1): Likewise.
16363 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
16365 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
16367 2018-05-18 Martin Liska <mliska@suse.cz>
16369 PR gcov-profile/84846
16370 * doc/gcov.texi: Document -t option of gcov tool.
16372 2018-05-18 Martin Liska <mliska@suse.cz>
16374 PR gcov-profile/84846
16375 * gcov.c (print_usage): Add new -t option.
16376 (process_args): Handle the option.
16377 (generate_results): Use stdout as output when requested by
16380 2018-05-18 Martin Liska <mliska@suse.cz>
16382 PR gcov-profile/84846
16383 * coverage.c (coverage_init): Write PWD to .gcno file.
16384 * doc/gcov.texi: Document how working directory is printed.
16385 * gcov-dump.c (dump_gcov_file): Print PWD.
16386 * gcov.c (output_intermediate_file): Likewise.
16387 (read_graph_file): Read PWD string.
16388 (output_lines): Print PWD.
16390 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16392 PR middle-end/85817
16393 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
16394 for retval and return false if all args to phi are zero.
16396 2018-05-18 Richard Biener <rguenther@suse.de>
16398 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
16400 (evrp_dom_walker::before_dom_children): Call it.
16402 2018-05-18 Richard Biener <rguenther@suse.de>
16404 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
16405 results when processing array refs with variable index.
16407 2018-05-18 Toon Moene <toon@moene.org>
16409 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
16410 directly after that of -floop-interchange. Indicate that both
16411 options are enabled by default when specifying -O3.
16413 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16415 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
16416 iterator. Delete separate integer-mode vec_set<mode> expander.
16417 (aarch64_simd_vec_setv2di): Delete.
16418 (vec_setv2di): Delete.
16419 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
16420 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
16421 the "w, r" alternative.
16423 2018-05-18 Martin Liska <mliska@suse.cz>
16425 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
16426 * tree-pass.h (make_pass_lower_switch_O0): New function.
16427 * tree-switch-conversion.c (node_has_low_bound): Remove.
16428 (node_has_high_bound): Likewise.
16429 (node_is_bounded): Likewise.
16430 (class pass_lower_switch): Make it a template type and create
16432 (pass_lower_switch::execute): Add template argument.
16433 (make_pass_lower_switch): New function.
16434 (make_pass_lower_switch_O0): New function.
16435 (do_jump_if_equal): Remove.
16436 (emit_case_nodes): Simplify to just handle all 3 cases and leave
16437 all the hard work to tree optimization passes.
16439 2018-05-18 Martin Liska <mliska@suse.cz>
16441 * dbgcnt.c (limit_low): Renamed from limit.
16442 (limit_high): New variable.
16443 (dbg_cnt_is_enabled): Check for upper limit.
16444 (dbg_cnt): Adjust dumping.
16445 (dbg_cnt_set_limit_by_index): Add new argument for high
16447 (dbg_cnt_set_limit_by_name): Likewise.
16448 (dbg_cnt_process_single_pair): Parse new format.
16449 (dbg_cnt_process_opt): Use strtok.
16450 (dbg_cnt_list_all_counters): Remove 'value' and add
16452 * doc/invoke.texi: Document changes.
16454 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
16456 * doc/sourcebuild.texi (scalar_all_fma): Document.
16457 * tree.def (FMA_EXPR): Delete.
16458 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
16459 * internal-fn.c (ternary_direct): New macro.
16460 (expand_ternary_optab_fn): Likewise.
16461 (direct_ternary_optab_supported_p): Likewise.
16462 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
16463 * builtins.c (fold_builtin_fma): Delete.
16464 (fold_builtin_3): Don't call it.
16465 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
16466 * expr.c (expand_expr_real_2): Likewise.
16467 * fold-const.c (operand_equal_p): Likewise.
16468 (fold_ternary_loc): Likewise.
16469 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
16470 * gimple.c (DEFTREECODE): Likewise.
16471 * gimplify.c (gimplify_expr): Likewise.
16472 * optabs-tree.c (optab_for_tree_code): Likewise.
16473 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16474 * tree-eh.c (operation_could_trap_p): Likewise.
16475 (stmt_could_throw_1_p): Likewise.
16476 * tree-inline.c (estimate_operator_cost): Likewise.
16477 * tree-pretty-print.c (dump_generic_node): Likewise.
16478 (op_code_prio): Likewise.
16479 * tree-ssa-loop-im.c (stmt_cost): Likewise.
16480 * tree-ssa-operands.c (get_expr_operands): Likewise.
16481 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
16482 * fold-const-call.h (fold_fma): Delete.
16483 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
16484 CFN_FNMA and CFN_FNMS.
16485 (fold_fma): Delete.
16486 * genmatch.c (combined_fn): New enum.
16487 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
16488 (commutative_op): New function.
16489 (commutate): Use it. Handle more than 2 operands.
16490 (dt_operand::gen_gimple_expr): Use commutative_op.
16491 (parser::parse_expr): Allow :c to be used with non-binary
16492 operators if the commutative operand is known.
16493 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
16494 CFN_FMS, CFN_FNMA and CFN_FNMS.
16495 (backprop::process_assign_use): Remove FMA_EXPR handling.
16496 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
16497 (gen_hsa_fma): New function.
16498 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
16499 IFN_FNMA and IFN_FNMS.
16500 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
16501 * gimple-fold.h (follow_all_ssa_edges): Declare.
16502 * gimple-fold.c (follow_all_ssa_edges): New function.
16503 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
16504 gimple_build interface and use follow_all_ssa_edges to fold the result.
16505 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
16506 instead of checking for optabs directly.
16507 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
16508 rather than FMA_EXPRs.
16509 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
16510 call to IFN_FMA instead of an FMA_EXPR.
16512 2018-05-17 Jim Wilson <jimw@sifive.com>
16514 * expr.c (do_tablejump): When converting index to Pmode, if we have a
16515 sign extended promoted subreg, and the range does not have the sign bit
16516 set, then do a sign extend.
16518 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
16519 test, check for sign extended subreg and/or constant operands, and
16520 do a sign extend in that case.
16522 2018-05-17 Steve Ellcey <sellcey@cavium.com>
16524 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
16525 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
16527 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
16528 Change logics_shift_reg to logics_shift_imm.
16529 (thunderx2t99_fp_loadpair_basic): Delete.
16530 (thunderx2t99_fp_storepair_basic): Delete.
16531 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
16532 (thunderx2t99_asimd_polynomial): Delete.
16533 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
16534 and neon_fp_mul_d_scalar_q.
16535 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
16536 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
16537 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
16538 (thunderx2t99_asimd_lut): Add missing tbl types.
16539 (thunderx2t99_asimd_ext): Delete.
16540 (thunderx2t99_asimd_load1_1_mult): Delete.
16541 (thunderx2t99_asimd_load1_2_mult): Delete.
16542 (thunderx2t99_asimd_load1_ldp): New.
16543 (thunderx2t99_asimd_load1): New.
16544 (thunderx2t99_asimd_load2): Add missing *load2* types.
16545 (thunderx2t99_asimd_load3): New.
16546 (thunderx2t99_asimd_load4): New.
16547 (thunderx2t99_asimd_store1_1_mult): Delete.
16548 (thunderx2t99_asimd_store1_2_mult): Delete.
16549 (thunderx2t99_asimd_store2_mult): Delete.
16550 (thunderx2t99_asimd_store2_onelane): Delete.
16551 (thunderx2t99_asimd_store_stp): New.
16552 (thunderx2t99_asimd_store1): New.
16553 (thunderx2t99_asimd_store2): New.
16554 (thunderx2t99_asimd_store3): New.
16555 (thunderx2t99_asimd_store4): New.
16557 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
16559 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
16560 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
16562 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
16563 Segher Boessenkool <segher@kernel.crashing.org>
16566 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
16569 2018-05-17 Richard Biener <rguenther@suse.de>
16571 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
16572 for pruning loop and prune defs feeding only already visited PHIs.
16574 2018-05-17 Richard Biener <rguenther@suse.de>
16576 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
16578 2018-05-17 Bin Cheng <bin.cheng@arm.com>
16579 Richard Biener <rguenther@suse.de>
16581 PR tree-optimization/85793
16582 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
16583 for VMAT_ELEMENTWISE.
16585 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
16587 * internal-fn.h (lookup_internal_fn): Declare
16588 * internal-fn.c (lookup_internal_fn): New function.
16589 * gimple.c (gimple_build_call_from_tree): Handle calls to
16590 internal functions.
16591 * gimple-pretty-print.c (dump_gimple_call): Print "." before
16592 internal function names.
16593 * tree-pretty-print.c (dump_generic_node): Likewise.
16594 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
16596 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
16598 * gimple-fold.h (gimple_build): Make the function forms take
16599 combined_fn rather than built_in_function.
16600 (gimple_simplify): Likewise.
16601 * gimple-match-head.c (gimple_simplify): Likewise.
16602 * gimple-fold.c (gimple_build): Likewise.
16603 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
16604 rather than gimple_build_call_internal.
16605 (get_initial_defs_for_reduction): Likewise.
16606 (vect_create_epilog_for_reduction): Likewise.
16607 (vectorizable_live_operation): Likewise.
16609 2018-05-17 Martin Liska <mliska@suse.cz>
16611 * gimple-ssa-sprintf.c (format_directive): Do not use
16612 space in between 'G_' and '('.
16614 2018-05-17 Jakub Jelinek <jakub@redhat.com>
16617 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
16618 even if the mask is not all ones.
16621 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
16623 (ix86_gimple_fold_builtin): Likewise.
16626 * config/i386/i386.c: Include tree-vector-builder.h.
16627 (ix86_vector_shift_count): New function.
16628 (ix86_fold_builtin): Fold shift builtins by scalar count.
16629 (ix86_gimple_fold_builtin): Likewise.
16631 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
16632 _mm512_setzero): New intrinsics.
16634 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
16635 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16637 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
16638 code generation for cases where splatting a value is not useful.
16639 * simplify-rtx.c (simplify_ternary_operation): Simplify
16640 vec_merge across a vec_duplicate and a paradoxical subreg forming
16641 a vector mode to a vec_concat.
16643 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
16645 * config.gcc: Support "goldmont-plus".
16646 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
16648 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
16649 PROCESSOR_GOLDMONT_PLUS.
16650 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
16651 (processor_target_table): Add "goldmont-plus".
16652 (PTA_GOLDMONT_PLUS): Define.
16653 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
16654 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
16655 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
16656 (fold_builtin_cpu): Add "goldmont-plus".
16657 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
16658 (ix86_option_override_internal): Add "goldmont-plus".
16659 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
16660 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
16661 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
16662 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
16664 2018-05-17 Richard Biener <rguenther@suse.de>
16666 PR tree-optimization/85757
16667 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
16668 remove defs that only feed that PHI from further processing.
16670 2018-05-16 Jim Wilson <jimw@sifive.com>
16672 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
16674 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
16676 2018-05-16 Mark Wielaard <mark@klomp.org>
16678 * dwarf2out.c (count_index_strings): New function.
16679 (output_indirect_strings): Call count_index_strings and generate
16680 header for dwarf_version >= 5.
16682 2018-05-16 Mark Wielaard <mark@klomp.org>
16684 * dwarf2out.c (dwarf_FORM): New function.
16685 (set_indirect_string): Use dwarf_FORM.
16686 (reset_indirect_string): Likewise.
16687 (size_of_die): Likewise.
16688 (value_format): Likewise.
16689 (output_die): Likewise.
16690 (add_skeleton_AT_string): Likewise.
16691 (output_macinfo_op): Likewise.
16692 (index_string): Likewise.
16693 (output_index_string_offset): Likewise.
16694 (output_index_string): Likewise.
16695 (count_index_strings): Likewise.
16697 2018-05-16 Carl Love <cel@us.ibm.com>
16699 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
16700 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
16702 2018-05-16 Martin Jambor <mjambor@suse.cz>
16704 * ipa-prop.c (ipa_free_all_edge_args): Remove.
16705 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
16707 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
16709 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
16710 (fnma<mode>4): Likewise.
16711 (fms<mode>4): Likewise.
16712 (fnms<mode>4): Likewise.
16713 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
16714 (aarch64_fnma<mode>4): Likewise.
16715 (aarch64_fms<mode>4): Likewise.
16716 (aarch64_fnms<mode>4): Likewise.
16717 (aarch64_fnmadd<mode>4): Likewise.
16719 2018-05-16 Jason Merrill <jason@redhat.com>
16721 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
16723 2018-05-16 Richard Biener <rguenther@suse.de>
16725 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
16726 (dump_stmt_cost): Declare.
16727 (add_stmt_cost): Dump cost we add.
16728 (add_stmt_costs): New function.
16729 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
16730 No longer exported.
16731 (vect_analyze_stmt): Adjust prototype.
16732 (vectorizable_condition): Likewise.
16733 (vectorizable_live_operation): Likewise.
16734 (vectorizable_reduction): Likewise.
16735 (vectorizable_induction): Likewise.
16736 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
16737 cost vector to pass to vectorizable_ and record afterwards.
16738 (vect_model_reduction_cost): Take cost vector argument and adjust.
16739 (vect_model_induction_cost): Likewise.
16740 (vectorizable_reduction): Likewise.
16741 (vectorizable_induction): Likewise.
16742 (vectorizable_live_operation): Likewise.
16743 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
16744 SLP_TREE_NUMBER_OF_VEC_STMTS.
16745 (vect_analyze_slp_cost_1): Remove.
16746 (vect_analyze_slp_cost): Likewise.
16747 (vect_slp_analyze_node_operations): Take visited args and
16748 a target cost vector. Avoid processing already visited stmt sets.
16749 (vect_slp_analyze_operations): Use a local cost vector to gather
16750 costs and register those of non-discarded instances.
16751 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
16752 (vect_schedule_slp_instance): Remove copying of
16753 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
16755 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
16756 adding cost. Record cost entry location.
16757 (vect_prologue_cost_for_slp_op): Function to compute cost of
16758 a constant or invariant generated for SLP vect in the prologue,
16759 split out from vect_analyze_slp_cost_1.
16760 (vect_model_simple_cost): Make static. Adjust for SLP costing.
16761 (vect_model_promotion_demotion_cost): Likewise.
16762 (vect_model_store_cost): Likewise, make static.
16763 (vect_model_load_cost): Likewise.
16764 (vectorizable_bswap): Add cost vector arg and adjust.
16765 (vectorizable_call): Likewise.
16766 (vectorizable_simd_clone_call): Likewise.
16767 (vectorizable_conversion): Likewise.
16768 (vectorizable_assignment): Likewise.
16769 (vectorizable_shift): Likewise.
16770 (vectorizable_operation): Likewise.
16771 (vectorizable_store): Likewise.
16772 (vectorizable_load): Likewise.
16773 (vectorizable_condition): Likewise.
16774 (vectorizable_comparison): Likewise.
16775 (can_vectorize_live_stmts): Likewise.
16776 (vect_analyze_stmt): Likewise.
16777 (vect_transform_stmt): Adjust calls to vectorizable_*.
16778 * tree-vectorizer.c: Include gimple-pretty-print.h.
16779 (dump_stmt_cost): New function.
16781 2018-05-16 Richard Biener <rguenther@suse.de>
16783 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
16784 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
16785 * tree-ssa-dse.c: Include tree-ssa-loop.h.
16786 (check_name): New callback.
16787 (dse_classify_store): Track cycles via a visited bitmap of PHI
16788 defs and simplify handling of in-loop and across loop dead stores
16789 and properly fail for loop-variant refs. Handle byte-tracking with
16790 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
16793 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
16795 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
16796 (vect_get_mask_type_for_stmt): Likewise.
16797 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
16799 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
16800 to determine the statement's vector type and the vector type that
16801 should be used for calculating nunits. Deal with cases in which
16802 the type has to be deferred.
16803 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
16804 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
16805 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
16806 (vect_determine_vf_for_stmt): New functions, split out from...
16807 (vect_determine_vectorization_factor): ...here.
16808 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
16809 (vect_get_mask_type_for_stmt): New functions, split out from
16810 vect_determine_vectorization_factor.
16812 2018-05-16 Richard Biener <rguenther@suse.de>
16814 * tree-cfg.c (verify_gimple_assign_ternary): Properly
16815 verify the [VEC_]COND_EXPR embedded comparison.
16817 2018-05-15 Martin Sebor <msebor@redhat.com>
16819 PR tree-optimization/85753
16820 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
16821 RECORD_TYPE in addition to ARRAY_TYPE.
16823 2018-05-15 Martin Sebor <msebor@redhat.com>
16825 PR middle-end/85643
16826 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
16828 2018-05-15 Richard Biener <rguenther@suse.de>
16830 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
16831 add by_clobber_p one. Change algorithm to collect all defs
16832 representing uses we need to walk and try reducing them to
16833 a single one before failing.
16834 (dse_dom_walker::dse_optimize_stmt): Adjust.
16836 2018-05-13 Mark Wielaard <mark@klomp.org>
16838 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
16839 (size_of_loc_descr): Likewise.
16840 (output_loc_operands): Likewise.
16841 (output_loc_operands_raw): Likewise.
16842 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
16843 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
16844 (hash_loc_operands): Likewise.
16845 (compare_loc_operands): Likewise.
16847 2018-05-14 Mark Wielaard <mark@klomp.org>
16849 * dwarf2out.c (count_index_addrs): New function.
16850 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
16852 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16854 PR tree-optimization/83648
16855 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
16856 return value as malloc candidate.
16858 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16861 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
16862 param as true in call to suggest_attribute.
16864 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
16866 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
16867 -mreadonly-in-sdata.
16869 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16871 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
16873 (aarch64_crypto_aesd_fused): Likewise.
16875 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
16877 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
16878 (movsi_aarch64): Likewise.
16879 (load_pairsi): Likewise.
16880 (load_pairdi): Likewise.
16881 (store_pairsi): Likewise.
16882 (store_pairdi): Likewise.
16883 (load_pairsf): Likewise.
16884 (load_pairdf): Likewise.
16885 (store_pairsf): Likewise.
16886 (store_pairdf): Likewise.
16887 (zero_extend): Likewise.
16888 (trunc): Swap alternatives.
16889 (fcvt_target): Add '?' to prefer w over r.
16891 2018-05-14 Jakub Jelinek <jakub@redhat.com>
16894 * config/i386/i386.md: Disallow non-commutative arithmetics in
16895 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
16896 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
16897 in the peephole2 before it.
16899 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
16901 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
16902 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
16903 (ix86_handle_option): Handle -mcldemote.
16904 * config.gcc: New header.
16905 * config/i386/cldemoteintrin.h: New file.
16906 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
16907 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
16909 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
16910 OPTION_MASK_ISA_CLDEMOTE.
16911 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
16912 (ix86_valid_target_attribute_inner_p): Ditto.
16913 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
16914 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
16915 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
16916 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
16917 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
16919 * config/i386/i386.opt: Add -mcldemote.
16920 * config/i386/x86intrin.h: New header.
16921 * doc/invoke.texi: Add -mcldemote.
16923 2018-05-14 Richard Biener <rguenther@suse.de>
16925 * doc/match-and-simplify.texi: Adjust :s documentation.
16927 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
16929 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
16930 intended memcpy size.
16931 (REORDER_45): Likewise.
16933 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
16935 * sort.cc: New file.
16936 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
16937 * vec.c (qsort_chk): Use gcc_qsort.
16938 * Makefile.in (OBJS-libcommon): Add sort.o.
16939 (build/sort.o): New target. Use it...
16940 (BUILD_RTL): ... here, and...
16941 (build/gencfn-macros): ... here, and...
16942 (build/genmatch): ... here.
16944 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
16945 Chung-Ju Wu <jasonwucj@gmail.com>
16947 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
16948 * config/nds32/nds32-graywolf.md: New file.
16949 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
16950 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
16952 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
16953 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
16954 * config/nds32/nds32.md (pipeline_model): Add graywolf.
16955 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
16956 * config/nds32/pipelines.md: Include n15 settings.
16958 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
16959 Chung-Ju Wu <jasonwucj@gmail.com>
16961 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
16962 * config/nds32/nds32-n13.md: New file.
16963 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
16964 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
16966 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
16967 * config/nds32/nds32.md (pipeline_model): Add n13.
16968 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
16969 * config/nds32/pipelines.md: Include n13 settings.
16971 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
16972 Chung-Ju Wu <jasonwucj@gmail.com>
16974 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
16975 * config/nds32/nds32-n10.md: New file.
16976 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
16977 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
16979 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
16980 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
16981 * config/nds32/nds32.md (pipeline_model): Add n10.
16982 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
16983 * config/nds32/pipelines.md: Include n10 settings.
16985 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
16986 Kito Cheng <kito.cheng@gmail.com>
16987 Chung-Ju Wu <jasonwucj@gmail.com>
16989 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
16990 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
16991 Add enum values for DSP extension instructions.
16992 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
16994 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
16995 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
16996 New code iterators.
16997 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
16998 * config/nds32/nds32-dspext.md: New file for DSP implementation.
16999 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
17000 * config/nds32/nds32-intrinsic.md: Likewise.
17001 * config/nds32/nds32_intrinsic.h: Likewise.
17002 * config/nds32/nds32-md-auxiliary.c: Likewise.
17003 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
17004 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
17005 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
17006 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
17007 * config/nds32/nds32-protos.h: New declarations for DSP extension.
17008 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
17009 TYPE_DMAC in switch statement.
17010 * config/nds32/nds32.c: New checking and implementation for DSP
17011 extension instructions.
17012 * config/nds32/nds32.h: Likewise.
17013 * config/nds32/nds32.md: Likewise.
17014 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
17015 * config/nds32/predicates.md: Implement new predicates for DSP
17018 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17020 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
17021 Reformat alternatives and attributes so it is easier to identify
17022 which constraints/attributes go with which instruction.
17023 (mov<mode>_hardfloat32, FMOVE64): Likewise.
17024 (mov<mode>_softfloat32, FMOVE64): Likewise.
17025 (mov<mode>_hardfloat64, FMOVE64): Likewise.
17026 (mov<mode>_softfloat64, FMOVE64): Likewise.
17028 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
17030 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17032 (Basic PowerPC Built-in Functions): The new name of the
17033 subsection previously known as "PowerPC Built-in Functions".
17034 (Basic PowerPC Built-in Functions Available on all Configurations):
17036 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
17037 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
17038 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
17039 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
17041 2018-05-11 Martin Jambor <mjambor@suse.cz>
17044 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
17047 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
17050 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
17052 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
17054 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
17055 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
17056 (ix86_handle_option): Handle -mwaitpkg.
17057 * config.gcc: New header.
17058 * config/i386/cpuid.h (bit_WAITPKG): New bit.
17059 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
17060 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
17062 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17063 OPTION_MASK_ISA_WAITPKG.
17064 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
17065 (ix86_option_override_internal): Add PTA_WAITPKG.
17066 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
17067 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
17068 IX86_BUILTIN_TPAUSE.
17069 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
17070 __builtin_ia32_umwait and __builtin_ia32_tpause.
17071 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
17072 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
17073 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
17074 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
17075 UNSPECV_TPAUSE): New.
17076 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
17077 * config/i386/i386.opt: Add -mwaitpkg.
17078 * config/i386/waitpkgintrin.h: New file.
17079 * config/i386/x86intrin.h: New header.
17080 * doc/invoke.texi: Add -mwaitpkg.
17082 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
17085 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
17087 (cortex-m0): Use armv6s-m isa.
17088 (cortex-m0plus): Likewise.
17089 (cortex-m1): Likewise.
17090 (cortex-m0.small-multiply): Likewise.
17091 (cortex-m0plus.small-multiply): Likewise.
17092 (cortex-m1.small-multiply): Likewise.
17094 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
17095 Jakub Jelinek <jakub@redhat.com>
17097 PR tree-optimization/85692
17098 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
17099 source permute as well.
17101 2018-05-11 Martin Liska <mliska@suse.cz>
17104 * doc/extend.texi: Document LLVM style format for no_sanitize
17107 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
17109 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
17110 mode_supports_vsx_dform_quad to mode_supports_dq_form.
17111 (mode_supports_vsx_dform_quad): Likewise.
17112 (mode_supports_vmx_dform): Move these functions to be next to the
17113 other mode_supports functions.
17114 (mode_supports_dq_form): Likewise.
17115 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
17116 mode_supports_dq_form.
17117 (reg_offset_addressing_ok_p): Likewise.
17118 (offsettable_ok_by_alignment): Likewise.
17119 (rs6000_legitimate_offset_address_p): Likewise.
17120 (legitimate_lo_sum_address_p): Likewise.
17121 (rs6000_legitimize_address): Likewise.
17122 (rs6000_legitimize_reload_address): Likewise.
17123 (rs6000_secondary_reload_inner): Likewise.
17124 (rs6000_preferred_reload_class): Likewise.
17125 (rs6000_output_move_128bit): Likewise.
17127 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
17129 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
17130 Generate SImode target register for null target.
17131 <case IX86_BUILTIN_XGETBV>: Ditto.
17132 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
17133 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
17135 2018-05-10 Carl Love <cel@us.ibm.com>
17137 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
17138 dcbtt and dcbtstt if operands[2] is 0.
17140 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
17143 * config/i386/sse.md (usadv64qi): New expander.
17145 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
17147 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
17148 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
17149 -maltivec=be support.
17150 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
17151 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
17152 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17153 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
17154 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
17155 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
17156 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
17157 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
17158 altivec_vsumsws): Adjust.
17159 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
17160 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
17161 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
17163 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
17164 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
17165 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
17166 (altivec_lve<VI_char>x): Delete expand.
17167 (*altivec_lve<VI_char>x_internal): Rename to...
17168 (altivec_lve<VI_char>x): ... this.
17169 (altivec_lvxl_<mode>): Delete expand.
17170 (*altivec_lvxl_<mode>_internal): Rename to ...
17171 (altivec_lvxl_<mode>): ... this.
17172 (altivec_stvxl_<mode>): Delete expand.
17173 (*altivec_stvxl_<mode>_internal): Rename to ...
17174 (altivec_stvxl_<mode>): ... this.
17175 (altivec_stve<VI_char>x): Delete expand.
17176 (*altivec_stve<VI_char>x_internal): Rename to ...
17177 (altivec_stve<VI_char>x): ... this.
17178 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
17179 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
17180 reduc_plus_scal_<mode>): Adjust.
17181 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
17183 (rs6000_cpu_cpp_builtins): Adjust.
17184 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
17185 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
17186 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
17187 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
17188 -maltivec=be support.
17189 (rs6000_split_vec_extract_var): Adjust.
17190 (rs6000_split_v4si_init): Adjust.
17191 (swap_selector_for_mode): Delete.
17192 (altivec_expand_lvx_be, altivec_expand_stvx_be,
17193 altivec_expand_stvex_be): Delete.
17194 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
17195 -maltivec=be support.
17196 (rs6000_gimple_fold_builtin): Ditto.
17197 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
17199 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
17200 (TARGET_DIRECT_MOVE_64BIT): Adjust.
17201 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
17202 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
17203 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
17204 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
17205 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
17206 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
17207 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
17208 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
17209 anonymous split): Adjust.
17210 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
17211 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
17213 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
17215 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
17216 when --with-gxx-include-dir is also specified.
17217 * configure: Regenerate.
17219 2018-05-09 Jim Wilson <jimw@sifive.com>
17222 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
17223 * config/riscv/t-withmultilib: New.
17224 * config/riscv/withmultilib.h: New.
17225 * doc/install.texi: Document RISC-V --with-multilib-list support.
17227 2018-05-09 Richard Biener <rguenther@suse.de>
17229 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
17231 (vect_bb_vectorization_profitable_p): Adjust. Compute
17232 actual scalar cost using the cost vector and the add_stmt_cost
17235 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17237 PR rtl-optimization/85645
17238 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
17239 in the REG_CFA_REGISTER note for LR, don't leave it empty.
17241 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17243 PR rtl-optimization/85645
17244 * shrink-wrap.c (spread_components): Return a boolean saying if
17245 anything was changed.
17246 (try_shrink_wrapping_separate): Iterate spread_components until
17247 nothing changes anymore.
17249 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17251 PR rtl-optimization/85645
17252 * regrename.c (build_def_use): Also kill the chains that include the
17253 destination of a REG_CFA_REGISTER note.
17255 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17257 PR rtl-optimization/85645
17258 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
17259 insn that has a REG_CFA_REGISTER note.
17261 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
17263 * cfgexpand.c (expand_clobber): New function.
17264 (expand_gimple_stmt_1): Use it.
17265 * tree-vect-stmts.c (vect_clobber_variable): New function,
17267 (vectorizable_simd_clone_call): ...here.
17268 (vectorizable_store): Emit a clobber either side of an
17269 IFN_STORE_LANES sequence.
17270 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
17272 2018-05-09 Tom de Vries <tom@codesourcery.com>
17275 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
17276 (define_insn "trap_if_false"): Add exit after trap.
17278 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
17280 PR rtl-optimization/85638
17281 * bb-reorder.c: Include common/common-target.h.
17282 (create_forwarder_block): New function extracted from...
17283 (fix_up_crossing_landing_pad): ...here. Rename into...
17284 (dw2_fix_up_crossing_landing_pad): ...this.
17285 (sjlj_fix_up_crossing_landing_pad): New function.
17286 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
17287 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
17288 from both partitions and exit the loop after one iteration.
17290 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17293 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17295 (Basic PowerPC Built-in Functions): The new name of the
17296 subsection previously known as "PowerPC Built-in Functions".
17297 (Basic PowerPC Built-in Functions Available on all Configurations):
17299 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
17301 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
17303 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
17305 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
17308 2018-05-08 Jim Wilson <jimw@sifive.com>
17310 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
17311 (LD_EMUL_SUFFIX): New.
17312 (LINK_SPEC): Use it.
17314 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17316 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17318 (Basic PowerPC Built-in Functions): The new name of the
17319 subsection previously known as "PowerPC Built-in Functions".
17320 (Basic PowerPC Built-in Functions Available on all Configurations):
17322 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
17324 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
17326 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
17328 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
17331 2018-05-08 Jakub Jelinek <jakub@redhat.com>
17334 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
17335 after cmpelim optimization.
17337 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
17339 * config.gcc: Support "goldmont".
17340 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
17341 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17342 PROCESSOR_GOLDMONT.
17343 * config/i386/i386.c (m_GOLDMONT): Define.
17344 (processor_target_table): Add "goldmont".
17345 (PTA_GOLDMONT): Define.
17346 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
17347 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
17348 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
17349 (fold_builtin_cpu): Add "goldmont".
17350 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
17351 (ix86_option_override_internal): Add "goldmont".
17352 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
17353 (processor_type): Add PROCESSOR_GOLDMONT.
17354 * config/i386/i386.md: Add CPU "glm".
17355 * config/i386/glm.md: New file.
17356 * config/i386/x86-tune.def: Add m_GOLDMONT.
17357 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
17359 2018-05-08 Jakub Jelinek <jakub@redhat.com>
17362 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
17364 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
17365 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
17366 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
17369 * config/i386/i386.c (ix86_fold_builtin): Handle
17370 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
17373 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
17374 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
17376 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
17379 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
17380 (check_arch): Likewise.
17381 (check_fpu): Return the result rather than printing it.
17382 (end arch): Fix operator precedence.
17383 (end cpu): Likewise.
17384 (END): Print the result from check_fpu.
17386 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
17387 Alan Hayward <alan.hayward@arm.com>
17388 David Sherwood <david.sherwood@arm.com>
17390 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
17391 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
17392 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
17393 (*fcmuo<mode>_and): New patterns.
17395 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
17397 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
17398 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
17399 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
17400 (cmp_op, sve_imm_con): New code attributes.
17401 (SVE_COND_INT_CMP, imm_con): Delete.
17402 (cmp_op): Remove above unspecs from int attribute.
17403 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
17405 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
17406 comparison-specific unspecs.
17407 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
17408 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
17409 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
17410 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
17411 (*vec_fcm<cmp_op><mode>): Rename to...
17412 (*fcm<cmp_op><mode>): ...this and adjust likewise.
17413 (*vec_fcmuo<mode>): Rename to...
17414 (*fcmuo<mode>): ...this and adjust likewise.
17415 (*pred_fcm<cmp_op><mode>): New pattern.
17416 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
17417 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
17419 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
17421 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
17422 (aarch64_emit_sve_predicated_cond): New function.
17423 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
17424 (aarch64_emit_unspec_cond_or): Replace with...
17425 (aarch64_emit_sve_or_conds): ...this new function. Use
17426 aarch64_emit_sve_ptrue_op for the individual comparisons and
17427 aarch64_emit_binop to OR them together.
17428 (aarch64_emit_inverted_unspec_cond): Replace with...
17429 (aarch64_emit_sve_inverted_cond): ...this new function. Use
17430 aarch64_emit_sve_ptrue_op for the comparison and
17431 aarch64_emit_unop to invert the result.
17432 (aarch64_expand_sve_vec_cmp_float): Update after the above
17433 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
17435 2018-05-07 Nathan Sidwell <nathan@acm.org>
17437 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
17438 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
17439 (Backwards Compatibility): Likewise.
17441 2018-05-07 Luis Machado <luis.machado@linaro.org>
17445 2018-05-07 Luis Machado <luis.machado@linaro.org>
17447 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17448 <prefetch_dynamic_strides>: New const bool field.
17449 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17450 prefetch_dynamic_strides.
17451 (exynosm1_prefetch_tune): Likewise.
17452 (thunderxt88_prefetch_tune): Likewise.
17453 (thunderx_prefetch_tune): Likewise.
17454 (thunderx2t99_prefetch_tune): Likewise.
17455 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
17457 (aarch64_override_options_internal): Update to set
17458 PARAM_PREFETCH_DYNAMIC_STRIDES.
17459 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
17460 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
17461 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
17462 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
17463 prefetch-dynamic-strides setting.
17465 2018-05-07 Luis Machado <luis.machado@linaro.org>
17467 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17468 <minimum_stride>: New const int field.
17469 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17470 minimum_stride field.
17471 (exynosm1_prefetch_tune): Likewise.
17472 (thunderxt88_prefetch_tune): Likewise.
17473 (thunderx_prefetch_tune): Likewise.
17474 (thunderx2t99_prefetch_tune): Likewise.
17475 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
17476 (aarch64_override_options_internal): Update to set
17477 PARAM_PREFETCH_MINIMUM_STRIDE.
17478 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
17479 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
17480 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
17481 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
17482 stride is constant and is below the minimum stride threshold.
17484 2018-05-07 Luis Machado <luis.machado@linaro.org>
17486 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
17489 2018-05-07 Luis Machado <luis.machado@linaro.org>
17491 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17492 <prefetch_dynamic_strides>: New const bool field.
17493 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17494 prefetch_dynamic_strides.
17495 (exynosm1_prefetch_tune): Likewise.
17496 (thunderxt88_prefetch_tune): Likewise.
17497 (thunderx_prefetch_tune): Likewise.
17498 (thunderx2t99_prefetch_tune): Likewise.
17499 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
17501 (aarch64_override_options_internal): Update to set
17502 PARAM_PREFETCH_DYNAMIC_STRIDES.
17503 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
17504 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
17505 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
17506 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
17507 prefetch-dynamic-strides setting.
17509 2018-05-07 Luis Machado <luis.machado@linaro.org>
17511 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17512 <minimum_stride>: New const int field.
17513 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17514 minimum_stride field.
17515 (exynosm1_prefetch_tune): Likewise.
17516 (thunderxt88_prefetch_tune): Likewise.
17517 (thunderx_prefetch_tune): Likewise.
17518 (thunderx2t99_prefetch_tune): Likewise.
17519 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
17520 (aarch64_override_options_internal): Update to set
17521 PARAM_PREFETCH_MINIMUM_STRIDE.
17522 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
17523 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
17524 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
17525 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
17526 stride is constant and is below the minimum stride threshold.
17528 2018-05-06 Jakub Jelinek <jakub@redhat.com>
17531 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
17532 the type is addressable. Don't force op into register if it has
17535 2018-05-05 Roland McGrath <mcgrathr@google.com>
17538 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
17539 any section for which we don't know a specific type it should have,
17540 regardless of name. Previously this was done only for the exact
17541 names ".init_array", ".fini_array", and ".preinit_array".
17542 (default_elf_asm_named_section): Add comment about
17543 relationship with default_section_type_flags and SECTION_NOTYPE.
17544 (get_section): Don't consider it a type conflict if one side has
17545 SECTION_NOTYPE and the other doesn't, as long as neither has the
17546 SECTION_BSS et al used in the default_section_type_flags logic.
17548 2018-05-05 Tom de Vries <tom@codesourcery.com>
17551 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
17552 (workaround_barsyncs): New function.
17553 (nvptx_reorg): Use workaround_barsyncs.
17554 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
17555 (define_expand "nvptx_membar_cta"): New define_expand.
17556 (define_insn "*nvptx_membar_cta"): New insn.
17558 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17560 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
17561 To improve optimization opportunities.
17562 * builtin-types.def: The new needed builtin types for the above.
17564 2018-05-04 Richard Biener <rguenther@suse.de>
17566 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
17567 * gimple-ssa-store-merging.c
17568 (imm_store_chain_info::output_merged_store): Remove redundant create,
17569 release split_store vector contents on failure.
17570 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
17571 scalar stmt vector on cache hit.
17573 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
17575 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
17577 * config.gcc (powerpc-xilinx-eabi*): Remove.
17578 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
17580 (fusion_addis_mem_combo_load): Ditto.
17581 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
17583 (rs6000_cpu_cpp_builtins): Ditto.
17584 * config/rs6000/rs6000-linux.c
17585 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
17586 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
17587 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
17589 (rs6000_setup_reg_addr_masks): Ditto.
17590 (rs6000_init_hard_regno_mode_ok): Ditto.
17591 (rs6000_option_override_internal): Ditto.
17592 (legitimate_lo_sum_address_p): Ditto.
17593 (rs6000_legitimize_address): Ditto.
17594 (rs6000_legitimize_reload_address): Ditto.
17595 (rs6000_legitimate_address_p): Ditto.
17596 (abi_v4_pass_in_fpr): Ditto.
17597 (setup_incoming_varargs): Ditto.
17598 (rs6000_gimplify_va_arg): Ditto.
17599 (rs6000_split_multireg_move): Ditto.
17600 (rs6000_savres_strategy): Ditto.
17601 (rs6000_emit_prologue_components): Ditto.
17602 (rs6000_emit_epilogue_components): Ditto.
17603 (rs6000_emit_prologue): Ditto.
17604 (rs6000_emit_epilogue): Ditto.
17605 (rs6000_elf_file_end): Ditto.
17606 (rs6000_function_value): Ditto.
17607 (rs6000_libcall_value): Ditto.
17608 * config/rs6000/rs6000.h: Ditto.
17609 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
17610 (TARGET_MINMAX): ... this. New.
17611 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
17612 * config/rs6000/rs6000.md: Remove Xilinx FP support.
17613 (*movsi_internal1_single): Delete.
17614 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
17615 mfpu=, mxilinx-fpu): Delete.
17616 * config/rs6000/singlefp.h: Delete.
17617 * config/rs6000/sysv4.h: Remove Xilinx FP support.
17618 * config/rs6000/t-rs6000: Ditto.
17619 * config/rs6000/t-xilinx: Delete.
17620 * config/rs6000/titan.md: Adjust for fp_type removal.
17621 * config/rs6000/vsx.md: Remove Xilinx FP support.
17622 (VStype_simple): Delete.
17623 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
17624 * config/rs6000/xfpu.h: Delete.
17625 * config/rs6000/xfpu.md: Delete.
17626 * config/rs6000/xilinx.h: Delete.
17627 * config/rs6000/xilinx.opt: Delete.
17628 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
17629 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
17631 2018-05-04 Tom de Vries <tom@codesourcery.com>
17634 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
17637 2018-05-04 Richard Biener <rguenther@suse.de>
17639 PR middle-end/85627
17640 * tree-complex.c (update_complex_assignment): We are always in SSA form.
17641 (expand_complex_div_wide): Likewise.
17642 (expand_complex_operations_1): Likewise.
17643 (expand_complex_libcall): Preserve EH info of the original stmt.
17644 (tree_lower_complex): Handle removed blocks.
17645 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
17646 on complex multiplication and division libcall builtins.
17648 2018-05-04 Richard Biener <rguenther@suse.de>
17650 PR middle-end/85574
17651 * fold-const.c (negate_expr_p): Restrict negation of operand
17652 zero of a division to when we know that can happen without
17654 (fold_negate_expr_1): Likewise.
17656 2018-05-04 Jakub Jelinek <jakub@redhat.com>
17659 * real.h (real_nextafter): Declare.
17660 * real.c (real_nextafter): New function.
17661 * fold-const-call.c (fold_const_nextafter): New function.
17662 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
17663 CASE_CFN_NEXTTOWARD.
17664 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
17665 even when arg1_mode is different from arg0_mode.
17667 2018-05-03 Nathan Sidwell <nathan@acm.org>
17669 * doc/extend.texi (Deprecated Features): Remove
17670 -ffriend-injection.
17671 (Backwards Compatibility): Likewise.
17672 * doc/invoke.texi (C++ Language Options): Likewise.
17673 (C++ Dialect Options): Likewise.
17675 2018-05-03 Jakub Jelinek <jakub@redhat.com>
17678 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
17679 _mm512_mask_mullox_epi64): New intrinsics.
17681 2018-05-03 Tom de Vries <tom@codesourcery.com>
17684 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
17685 dump files): Add offload-tree.
17687 2018-05-03 Richard Biener <rguenther@suse.de>
17689 PR tree-optimization/85615
17690 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
17691 to loops not nested in BBs loop father to avoid creating multi-entry
17694 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17696 PR tree-optimization/70291
17697 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
17698 arguments. Change return type to tree. Emit libcall as a new
17699 statement rather than replacing existing one when inplace_p is true.
17700 (expand_complex_multiplication_components): New function.
17701 (expand_complex_multiplication): Expand floating-point complex
17702 multiplication using the above.
17703 (expand_complex_division): Rename inner_type parameter to type.
17704 Update expand_complex_libcall call-site.
17705 (expand_complex_operations_1): Update expand_complex_multiplication
17706 and expand_complex_division call-sites.
17708 2018-05-02 Jakub Jelinek <jakub@redhat.com>
17711 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
17712 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
17713 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
17714 the highest significant bit of the shift count mask is clear. In
17715 check whether and[sq]i3 is needed verify that all significant bits
17716 of the shift count other than the highest are set.
17718 2018-05-02 Tom de Vries <tom@codesourcery.com>
17721 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
17722 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
17723 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
17724 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
17725 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
17726 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
17727 __builtin_goacc_parlevel_size.
17729 2018-05-02 Richard Biener <rguenther@suse.de>
17731 PR tree-optimization/85597
17732 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
17733 do not use split vect_get_vec_defs call but call vect_get_slp_defs
17736 2018-05-02 Tom de Vries <tom@codesourcery.com>
17739 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
17740 dump files): Add ltrans-tree.
17742 2018-05-02 Tom de Vries <tom@codesourcery.com>
17745 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
17746 dump files): Add wpa-ipa.
17748 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
17750 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
17751 powerpc*-*-linux*paired* target.
17752 * config/rs6000/750cl.h: Delete.
17753 * config/rs6000/paired.h: Delete.
17754 * config/rs6000/paired.md: Delete.
17755 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
17757 * config/rs6000/rs6000-builtin.def: Remove paired float support.
17758 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
17759 comment. Remove paired float support.
17760 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
17761 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
17763 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
17764 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
17766 * config/rs6000/rs6000.c: Remove paired float support.
17767 (paired_expand_vector_init, paired_expand_vector_move,
17768 paired_emit_vector_compare, paired_emit_vector_cond_expr,
17769 (paired_expand_lv_builtin, paired_expand_stv_builtin,
17770 paired_expand_builtin, paired_expand_predicate_builtin,
17771 paired_init_builtins): Delete.
17772 * config/rs6000/rs6000.h: Remove paired float support.
17773 * config/rs6000/rs6000.md: Remove paired float support.
17774 (move_from_CR_ov_bit): Delete.
17775 * config/rs6000/rs6000.opt (mpaired): Delete.
17776 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
17777 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
17779 2018-05-02 Richard Biener <rguenther@suse.de>
17781 PR middle-end/85567
17782 * gimplify.c (gimplify_save_expr): When in SSA form allow
17783 SAVE_EXPRs to compute to SSA vars.
17785 2018-05-02 Jakub Jelinek <jakub@redhat.com>
17788 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
17789 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
17790 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
17791 clobber operands[2], instead use a new pseudo. Formatting fixes.
17793 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
17795 PR tree-optimization/85586
17796 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
17797 exit early for statements in the same group if the accesses are
17800 2018-05-02 Tom de Vries <tom@codesourcery.com>
17803 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
17806 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
17808 PR tree-optimization/85143
17809 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
17811 2018-05-01 Tom de Vries <tom@codesourcery.com>
17814 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
17815 not found" error message.
17817 2018-05-01 Tom de Vries <tom@codesourcery.com>
17820 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
17821 * vec.c (test_ordered_remove_if): New function.
17822 (vec_c_tests): Call test_ordered_remove_if.
17823 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
17824 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
17825 * tree-vect-patterns.c (vect_pattern_recog_1): Use
17826 VEC_ORDERED_REMOVE_IF.
17828 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17830 PR tree-optimization/82665
17831 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
17832 pointer subtraction where arguments come from a memchr call.
17834 2018-05-01 Jakub Jelinek <jakub@redhat.com>
17836 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
17837 --push-state --as-needed and --pop-state instead of --as-needed and
17838 --no-as-needed if ld supports it.
17839 * configure: Regenerated.
17842 * doc/install.texi2html: Replace _002d with - and _002a with * in
17843 generated html files using sed.
17845 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17848 * gcc-rich-location.c (blank_line_before_p): New function.
17849 (use_new_line): New function.
17850 (gcc_rich_location::add_fixit_insert_formatted): New function.
17851 * gcc-rich-location.h
17852 (gcc_rich_location::add_fixit_insert_formatted): New function.
17854 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17856 * selftest.c (assert_streq): Rename "expected" and "actual" to
17857 "val1" and "val2". Extend NULL-handling to cover both inputs
17858 symmetrically, while still requiring both to be non-NULL for a pass.
17859 * selftest.h (assert_streq): Rename "expected" and "actual" to
17861 (ASSERT_EQ): Likewise.
17862 (ASSERT_EQ_AT): Likewise.
17863 (ASSERT_KNOWN_EQ): Likewise.
17864 (ASSERT_KNOWN_EQ_AT): Likewise.
17865 (ASSERT_NE): Likewise.
17866 (ASSERT_MAYBE_NE): Likewise.
17867 (ASSERT_MAYBE_NE_AT): Likewise.
17868 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
17869 the assertion to pass.
17870 (ASSERT_STREQ_AT): Likewise.
17872 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
17874 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
17875 interaction with -pie.
17877 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17879 * selftest.h: Fix alphabetization of per-source-file selftest
17882 2018-04-30 Jason Merrill <jason@redhat.com>
17884 PR c++/61982 - dead stores to destroyed objects.
17885 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
17888 2018-04-30 Jason Merrill <jason@redhat.com>
17890 * tree.c (build_clobber): New.
17891 * tree.h: Declare it.
17892 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
17894 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17896 * diagnostic-show-locus.c (layout::layout): Update for
17897 location_get_source_line returning a char_span.
17898 (struct char_span): Move to input.h.
17899 (struct correction): Update for fields in char_span becoming
17901 (struct source_line): Update for location_get_source_line
17902 returning a char_span.
17903 (layout::print_line): Likewise.
17904 * edit-context.c (edited_file::print_content): Likewise.
17905 (edited_file::print_diff_hunk): Likewise.
17906 (edited_file::print_run_of_changed_lines): Likewise.
17907 (edited_file::get_num_lines): Likewise.
17908 (edited_line::edited_line): Likewise.
17909 * final.c (asm_show_source): Likewise.
17910 * input.c (location_get_source_line): Convert return type
17911 from const char * to char_span, losing the final "line_len"
17913 (dump_location_info): Update for the above.
17914 (get_substring_ranges_for_loc): Likewise. Use a char_span
17915 when handling the literal within the line.
17916 (test_reading_source_line): Update for location_get_source_line
17917 returning a char_span.
17918 * input.h (class char_span): Move here from
17919 diagnostic-show-locus.c, converting from a struct to a class.
17920 Make data members private.
17921 (char_span::operator bool): New.
17922 (char_span::length): New.
17923 (char_span::get_buffer): New.
17924 (char_span::operator[]): New.
17925 (char_span::subspan): Make const.
17926 (char_span::xstrdup): New.
17927 (location_get_source_line): Convert return type from const char *
17928 to char_span, losing the final "line_size" param.
17930 2018-04-30 Jan Hubicka <jh@suse.cz>
17932 * lto-wrapper.c (ltrans_priorities): New static var.
17933 (cmp_priority): New.
17934 (run_gcc): Read priorities and if doing parallel build order
17935 the Makefile by them.
17937 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17939 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
17941 2018-04-30 Richard Biener <rguenther@suse.de>
17943 * tree-cfg.c (verify_address): Remove base argument, add
17944 flag whether to check TREE_ADDRESSABLE and do that.
17945 (verify_expr): Remove.
17946 (verify_types_in_gimple_reference): Add pieces from verify_expr.
17947 (verify_gimple_assign_single): Likewise.
17948 (verify_gimple_switch): Likewise.
17949 (verify_expr_location_1): Dereference tp once. Add (disabled)
17950 piece from verify_expr.
17951 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
17953 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
17955 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
17957 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
17959 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
17960 (small_data_pattern): Likewise.
17961 (arc_rewrite_small_data): Likewise.
17962 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
17963 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
17964 (get_symbol_alignment): New function.
17965 (legitimate_small_data_address_p): Likewise.
17966 (legitimate_scaled_address): Update, call
17967 legitimate_small_data_address_p.
17968 (output_sdata): New static variable.
17969 (arc_print_operand): Update how we handle small data operands.
17970 (arc_print_operand_address): Likewise.
17971 (arc_legitimate_address_p): Update, use
17972 legitimate_small_data_address_p.
17973 (arc_rewrite_small_data_p): Remove.
17974 (arc_rewrite_small_data_1): Likewise.
17975 (arc_rewrite_small_data): Likewise.
17976 (small_data_pattern): Likewise.
17977 (compact_sda_memory_operand): Update to use
17978 legitimate_small_data_address_p and get_symbol_alignment.
17979 (prepare_move_operands): Don't rewite sdata pattern.
17980 (prepare_extend_operands): Remove.
17981 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
17983 (zero_extendqisi2): Likewise.
17984 (zero_extendhisi2): Likewise.
17985 (extendqihi2): Likewise.
17986 (extendqisi2): Likewise.
17987 (extendhisi2): Likewise.
17988 (addsi3): Likewise.
17989 (subsi3): Likewise.
17990 (andsi3): Likewise.
17991 * config/arc/constraints.md (Usd): Change it to memory constraint.
17993 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
17995 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
17996 as source of std instructions.
17997 * config/arc/arc.md (movsi_insn): Update pattern predicate to
17998 allow 6-bit constants as source for store instructions.
17999 (movdi_insn): Update instruction pattern to allow 6-bit constants
18000 as source for store instructions.
18002 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
18004 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
18006 2018-04-30 Nathan Sidwell <nathan@acm.org>
18007 Sandra Loosemore <sandra@codesourcery.com>
18009 * dumpfile.c (dump_open): Allow '-' for stdout.
18010 * doc/invoke.texi (Developer Options): Document dump filename
18011 determination early. Document stdin/stdout selection.
18013 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
18015 Microblaze Target: PIC data text relative
18017 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
18018 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
18020 * config/microblaze/microblaze.h (microblaze_constant_address_p):
18021 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
18022 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
18023 New addressing mode for data-text relative position indepenedent code.
18024 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
18025 'ADDRESS_SYMBOLIC_TXT_REL'.
18026 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
18027 (microblaze_legitimate_pic_operand): Exclude function calls from
18028 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
18029 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
18031 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
18032 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
18033 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
18034 for 'address + offset'.
18035 (microblaze_expand_prologue): Add new function prologue call for
18037 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
18038 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
18039 table in case of TARGET_PIC_DATA_TEXT_REL.
18040 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
18041 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
18042 Add new macros 'UNSPEC_TEXT',
18043 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
18044 + exclude function calls from 'UNSPEC_PLT' in case of data text
18046 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
18047 new target hook for generating address diff vector tables in case of
18049 * doc/tm.texi : Regenerate.
18050 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
18051 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
18052 of addr diff vector generation.
18053 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
18054 target hook definition.
18055 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
18056 Add default function for generate_pic_addr_diff_vec -> flag_pic.
18057 * doc/invoke.texi (Add new pic option): Add new microblaze pic
18058 option for data text relative.
18060 2018-04-30 Richard Biener <rguenther@suse.de>
18062 * tree-chrec.h (evolution_function_is_constant_p): Remove
18064 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
18066 2018-04-30 Richard Biener <rguenther@suse.de>
18069 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
18071 2018-04-30 Richard Biener <rguenther@suse.de>
18073 PR tree-optimization/28364
18074 PR tree-optimization/85275
18075 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
18076 copying first exit test.
18078 2018-04-28 Mark Wielaard <mark@klomp.org>
18080 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
18081 dwarf_version >= 5.
18082 (dwarf_AT): Handle DW_AT_addr_base.
18083 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
18085 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
18088 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
18089 (*ashl<dwi>3_doubleword_mask_1): Ditto.
18090 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
18091 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
18093 2018-04-28 Richard Biener <rguenther@suse.de>
18095 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
18096 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
18097 to reflect use. Only add interesting stmts.
18099 2018-04-27 Martin Jambor <mjambor@suse.cz>
18102 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
18103 the jump function allows for passing through aggregate values.
18105 2018-04-27 David Malcolm <dmalcolm@redhat.com>
18107 * input.h (in_system_header_at): Convert from macro to inline
18109 (from_macro_expansion_at): Likewise.
18110 (from_macro_definition_at): Likewise.
18112 2018-04-27 Jeff Law <law@redhat.com>
18114 * config.gcc: Mark tile* targets as deprecated/obsolete.
18116 2018-04-27 Richard Biener <rguenther@suse.de>
18118 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
18121 2018-04-27 Richard Biener <rguenther@suse.de>
18123 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
18125 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
18127 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
18128 with Yd constraint. Set "preferred_for_speed" attribute from
18129 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
18130 with Yd constraint.
18131 (*movdi_internal): Ditto.
18132 (movti_interunit splitters): Remove
18133 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
18134 (movdi_interunit splitters): Ditto.
18135 * config/i386/constraints.md (Ye): Remove.
18136 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
18138 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18141 * config/aarch64/constraints.md (Usg): Limit to 31.
18142 (Usj): Limit to 63.
18144 2018-04-27 Jakub Jelinek <jakub@redhat.com>
18146 PR tree-optimization/85529
18147 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
18148 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
18149 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
18150 zero extension or masking of the MSB bit.
18151 (optimize_range_tests): Add FIRST_BB argument, pass it through
18152 to optimize_range_tests_var_bound.
18153 (maybe_optimize_range_tests, reassociate_bb): Adjust
18154 optimize_range_tests callers.
18156 2018-04-26 Richard Biener <rguenther@suse.de>
18157 Jakub Jelinek <jakub@redhat.com>
18159 * cgraph.h (symbol_table): Just declare debug method here.
18160 * symtab.c (symbol_table::debug): Define.
18162 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
18164 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
18166 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
18168 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
18169 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
18170 (*movdi_internal): Substitute Yi and Yj constraint with x
18171 and Ym and Yn constraint with y constraint. Update "isa"
18172 attribute and set "preferred_for_speed" attribute from
18173 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
18174 (*movsi_internal): Ditto.
18175 (*movdf_internal): Ditto.
18176 (*movsf_internal): Ditto.
18177 (*zero_extendsidi2): Ditto.
18178 * config/i386/sse.md (vec_set<mode>_0): Ditto.
18179 (sse2_loadld): Ditto.
18180 (*vec_extract<ssevecmodelower>_0): Ditto.
18181 (*vec_extractv4si_0_zext_sse4): Ditto.
18182 (vec_concatv2di): Ditto.
18183 (*vec_dup<mode>): Ditto.
18184 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
18185 * config/i386/constraints.md (Yi): Remove.
18190 2018-04-26 Nathan Sidwell <nathan@acm.org>
18192 * dumpfile.c (dump_open): New.
18193 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
18194 (dump_finish): Detect stdio/stderr by value not name.
18196 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
18198 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
18200 2018-04-26 Tom de Vries <tom@codesourcery.com>
18203 * config/nvptx/nvptx.c (verify_neutering_jumps)
18204 (verify_neutering_labels): New function
18205 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
18207 2018-04-26 Tom de Vries <tom@codesourcery.com>
18210 * config/nvptx/nvptx.c (needs_neutering_p): New function.
18211 (nvptx_single): Use needs_neutering_p to skip over insns that do not
18214 2018-04-26 Richard Biener <rguenther@suse.de>
18215 Tom de Vries <tom@codesourcery.com>
18218 * lto-streamer-out.c (output_function): Fixup loops if required to match
18219 discovery done in the reader.
18221 2018-04-26 Richard Biener <rguenther@suse.de>
18223 PR tree-optimization/85116
18224 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
18225 have a loop exit from the single latch predecessor. Remove
18226 case of header with just condition.
18227 (ch_base::copy_headers): Exclude infinite loops from any
18229 (pass_ch::execute): Record exits.
18231 2018-04-26 Richard Biener <rguenther@suse.de>
18233 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
18234 prologue cost vector and pass it to vect_get_load_cost.
18235 (vect_get_peeling_costs_all_drs): Likewise.
18236 (vect_peeling_hash_get_lowest_cost): Likewise.
18237 (vect_enhance_data_refs_alignment): Likewise.
18239 2018-04-26 Richard Biener <rguenther@suse.de>
18241 PR middle-end/85450
18242 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
18243 checking of integer<->pointer conversions.
18244 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
18245 sign-/zero-extending pointer types.
18246 (expand_omp_for_static_chunk): Likewise.
18248 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
18249 Jean Lee <xiaoyur347@gmail.com>
18251 * config/mips/mips.c (mips_asan_shadow_offset): New function.
18252 (TARGET_ASAN_SHADOW_OFFSET): Define.
18253 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
18254 true for -fsanitize=address.
18256 2018-04-25 Mark Wielaard <mark@klomp.org>
18258 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
18261 2018-04-25 Jakub Jelinek <jakub@redhat.com>
18263 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
18264 than "alu", remove explicit "memory" and "imm_disp" attributes.
18265 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
18267 PR middle-end/85414
18268 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
18269 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
18270 gen_lowpart_no_emit.
18272 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
18275 * config/i386/i386.c (ix86_expand_builtin): Change memory
18276 operand to XI, extend p0 to Pmode.
18277 * config/i386/i386.md: Change unspec volatile and operand
18278 1 mode to XI, change operand 0 mode to P.
18280 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18282 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
18283 GET_MODE_MASK before any checking.
18284 (nds32_can_use_bset_p): Likewise.
18285 (nds32_can_use_btgl_p): Likewise.
18287 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18289 * config/nds32/nds32-doubleword.md: New define_split pattern for
18290 illegal register number.
18292 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18294 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
18296 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18298 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
18300 2018-04-25 Richard Biener <rguenther@suse.de>
18302 * lto-streamer.h (LTO_major_version): Bump to 8.
18304 2018-04-25 Jakub Jelinek <jakub@redhat.com>
18306 * BASE-VER: Set to 9.0.0.
18308 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
18310 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
18311 in __abskf2 and __powikf2.
18313 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18316 * config/aarch64/constraints.md (Usg, Usj): New constraints.
18317 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
18318 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
18319 Use the above on operand 2. Reindent.
18320 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
18322 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
18325 * common/config/i386/i386-common.c (ix86_handle_option): Don't
18327 * config/i386/i386.opt (mcet): Removed.
18328 * doc/install.texi: Remove -mcet documentation.
18329 * doc/invoke.texi: Likewise.
18331 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
18334 * doc/install.texi: Remove -mcet from bootstrap-cet.
18336 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18339 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
18340 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
18344 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
18345 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
18346 containing a CONST_VECTOR.
18348 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
18350 * doc/install.texi: Update newlib dependency for nvptx.
18352 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18355 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
18356 instead of INTVAL when shifting x left.
18358 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
18360 PR tree-optimization/85478
18361 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
18362 vect_grouped_store_supported for single element vectors.
18364 2018-04-24 Richard Biener <rguenther@suse.de>
18367 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
18368 load cost increase to the case of non-constant step.
18370 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18373 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
18374 destination if any_malformed_asm.
18376 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
18378 PR middle-end/85496
18379 * expr.c (store_field): In the bitfield case, if the value comes from
18380 a function call and is returned in registers by means of a PARALLEL,
18381 do not change the mode of the temporary unless BLKmode and VOIDmode.
18383 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
18385 PR rtl-optimization/85423
18386 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
18387 dependencies to debug insns when the previous insn is non-debug.
18389 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
18391 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
18392 enums into a single definition.
18393 (fls): Fix predicates and printing.
18396 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
18398 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
18399 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
18400 and short u6 immediate.
18401 (check_if_valid_sleep_operand): Remove.
18402 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
18404 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18406 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
18407 flag_always_save_lp condition.
18408 * config/nds32/nds32.opt (malways-save-lp): New option.
18410 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18412 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
18413 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
18414 * config/nds32/nds32.h
18415 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
18416 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
18418 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18420 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
18421 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
18423 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18424 Chung-Ju Wu <jasonwucj@gmail.com>
18426 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
18428 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
18429 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
18431 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18433 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
18435 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18437 * config/nds32/nds32-protos.h (nds32_data_alignment,
18438 nds32_local_alignment): Declare.
18439 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
18440 nds32_local_alignment): New functions.
18441 (TARGET_CONSTANT_ALIGNMENT): Define.
18442 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
18444 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18446 * config/nds32/nds32.c
18447 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
18448 (TARGET_MODES_TIEABLE_P): Likewise.
18450 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18452 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
18453 level Ofast and Og.
18455 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
18456 Chung-Ju Wu <jasonwucj@gmail.com>
18458 * config/nds32/constants.md (unspec_volatile_element): Add enum values
18459 for unaligned access.
18460 * config/nds32/nds32-intrinsic.c: Implementation of expanding
18462 * config/nds32/nds32-intrinsic.md: Likewise.
18463 * config/nds32/nds32_intrinsic.h: Likewise.
18464 * config/nds32/nds32.h (nds32_builtins): Likewise.
18465 * config/nds32/nds32.opt (munaligned-access): New option.
18466 * config/nds32/nds32.c (nds32_asm_file_start): Display
18467 flag_unaligned_access status.
18469 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
18471 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
18472 -mno-relax is present.
18473 * config/riscv/linux.h (LINK_SPEC): Ditto.
18475 2018-04-20 Martin Sebor <msebor@redhat.com>
18478 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
18480 (gimple_fold_builtin_stxcpy_chk): Same.
18481 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
18483 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
18486 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
18487 __powikf2 when long double is IEEE 128-bit.
18489 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
18491 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
18492 step to make sure stack always aligned.
18494 2018-04-20 Carl Love <cel@us.ibm.com>
18497 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
18498 size check for arg0.
18500 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
18501 Tom de Vries <tom@codesourcery.com>
18504 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
18505 Emit insns for calls too.
18506 (nvptx_find_par): Always look for worker-level predecessor insn.
18507 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
18509 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
18510 (nvptx_process_pars): Propagate frames for calls.
18512 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
18515 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
18517 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
18518 (ix86_handle_option): Don't handle OPT_mibt.
18519 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
18521 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
18523 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
18524 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
18525 (ix86_target_macros): Define __CET__ with flag_cf_protection
18526 for -fcf-protection.
18527 * config/i386/i386.c (isa2_opts): Remove -mibt.
18528 * config/i386/i386.h (TARGET_IBT): Removed.
18529 (TARGET_IBT_P): Likewise.
18530 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
18531 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
18532 * config/i386/i386.opt (mcet): Update help message.
18533 (mshstk): Likewise.
18535 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
18536 -mcet as an alias for -mshstk.
18538 2018-04-20 Richard Biener <rguenther@suse.de>
18540 PR middle-end/85475
18541 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
18542 complexity by forcing a single use of the multiply operand.
18544 2018-04-20 Martin Jambor <mjambor@suse.cz>
18547 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
18548 recursion dependency to only apply to non-clones.
18550 2018-04-20 Martin Jambor <mjambor@suse.cz>
18553 * ipa-cp.c (create_specialized_node): Check that clones of
18554 self-recursive edges exist during IPA-CP.
18556 2018-04-19 Toon Moene <toon@moene.org>
18558 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
18561 2018-04-19 Jakub Jelinek <jakub@redhat.com>
18563 PR tree-optimization/85467
18564 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
18565 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
18566 VECTOR_CST element to type.
18568 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
18571 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
18572 * config/i386/i386.md (builtin_setjmp_setup): Removed.
18573 (builtin_longjmp): Likewise.
18574 (save_stack_nonlocal): New pattern.
18575 (restore_stack_nonlocal): Likewise.
18577 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
18580 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
18581 Replace ASM_OUTPUT_LABEL with fprintf.
18583 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
18586 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
18587 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
18588 * config/i386/i386-c.c (ix86_target_macros_internal): Also
18589 define __IBT__ and __SHSTK__ for -fcf-protection.
18590 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
18592 (ix86_trampoline_init): Likewise.
18593 (x86_output_mi_thunk): Likewise.
18594 (ix86_notrack_prefixed_insn_p): Likewise.
18595 (ix86_option_override_internal): Don't disallow -fcf-protection.
18596 * config/i386/i386.md (rdssp<mode>): Also enable for
18598 (incssp<mode>): Likewise.
18599 (nop_endbr): Likewise.
18600 * config/i386/i386.opt (mcet): Change help message to built-in
18603 (mshstk): Likewise.
18604 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
18605 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
18606 enable CET built-in functions.
18608 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
18610 * common/config/i386/i386-common.c
18611 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
18612 OPTION_MASK_ISA_MOVDIRI_UNSET,
18613 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
18614 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
18615 * config.gcc (movdirintrin.h): New header.
18616 * config/i386/cpuid.h (bit_MOVDIRI,
18617 bit_MOVDIR64B): New bits.
18618 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
18620 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
18621 (VOID, PVOID, PCVOID)): New function types.
18622 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
18623 __builtin_ia32_directstoreu_u64,
18624 __builtin_ia32_movdir64b): New builtins.
18625 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
18626 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
18628 (ix86_valid_target_attribute_inner_p): Ditto.
18629 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
18630 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
18631 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
18632 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
18633 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
18634 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
18635 (movdiri<mode>, movdir64b_<mode>): New.
18636 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
18637 * config/i386/immintrin.h: Include movdirintrin.h.
18638 * config/i386/movdirintrin.h: New file.
18639 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
18641 2018-04-19 Richard Biener <rguenther@suse.de>
18643 PR middle-end/85455
18644 * cfg.c (clear_bb_flags): When loop state says we have
18645 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
18647 2018-04-19 Richard Biener <rguenther@suse.de>
18649 PR tree-optimization/84737
18650 * tree-vect-data-refs.c (vect_copy_ref_info): New function
18651 copying restrict info.
18652 (vect_setup_realignment): Use it.
18653 * tree-vectorizer.h (vect_copy_ref_info): Declare.
18654 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
18655 the first DR to all generated stores.
18656 (vectorizable_load): Likewise for loads.
18658 2018-04-19 Jakub Jelinek <jakub@redhat.com>
18660 PR tree-optimization/85446
18661 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
18662 the integral and pointer types to have the same precision.
18664 * doc/install.texi: Document --disable-cet being the default and
18667 2018-04-18 Martin Liska <mliska@suse.cz>
18669 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
18672 2018-04-18 Martin Liska <mliska@suse.cz>
18675 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
18678 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
18679 arguments if they are comparable.
18681 2018-04-18 Martin Liska <mliska@suse.cz>
18684 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
18687 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
18690 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
18693 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
18694 ENDBR after calling __morestack.
18696 2018-04-18 David Malcolm <dmalcolm@redhat.com>
18699 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
18700 by using gcc_base_ver to generate a gcc_driver_version, and use
18701 it when generating GCC_DRIVER_NAME.
18702 * configure: Regenerate.
18704 2018-04-18 Jakub Jelinek <jakub@redhat.com>
18707 * config.gcc: Obsolete powerpc*-*-*spe*.
18709 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18712 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
18713 (stabstr_D): Change type of unum from unsigned int to
18714 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
18717 2018-04-17 Jim Wilson <jimw@sifive.com>
18720 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
18721 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
18722 Set arg_pointer_offset after using pretend_args_size.
18724 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18726 PR rtl-optimization/85431
18727 * dse.c (record_store): Ignore zero width stores.
18730 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
18731 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
18732 __builtin_stack_restore rather than after it.
18733 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
18734 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
18735 argument instead of virtual_dynamic_stack_rtx.
18737 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
18739 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
18741 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18742 Add note to error message to explain internal mapping of overloaded
18743 built-in function name to non-overloaded built-in function name.
18744 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
18747 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
18750 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
18751 where the inputs overlap with the output.
18753 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18755 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
18756 (=v, v) alternative and explicit "memory" attribute.
18757 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
18758 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
18760 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
18761 "sselog1" type instead of "sselog".
18762 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
18763 "sselog". Remove explicit "memory" attribute.
18764 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
18765 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
18767 (vec_extract_hi_v32hi): Merge all alternatives into one, use
18768 "sselog1" type instead of "sselog". Remove explicit "memory"
18770 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
18771 use "sselog1" type instead of "sselog". Remove explicit "memory"
18773 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
18774 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
18776 (vec_extract_hi_v64qi): Merge all alternatives into one, use
18777 "sselog1" type instead of "sselog". Remove explicit "memory"
18779 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
18780 use "sselog1" type instead of "sselog". Remove explicit "memory"
18784 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
18786 PR middle-end/85414
18787 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
18790 2018-04-17 Martin Jambor <mjambor@suse.cz>
18793 * ipa-cp.c (create_specialized_node): Call
18794 expand_all_artificial_thunks if necessary.
18796 2018-04-17 Martin Liska <mliska@suse.cz>
18799 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
18800 in message, remote space in between '_G' and '('.
18802 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18805 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
18806 avx512f_vmcmp<mode>3<round_saeonly_name>,
18807 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
18808 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
18809 avx512f_rndscale<mode><round_saeonly_name>,
18810 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
18811 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
18812 Use %<iptr>2 instead of %2 for -masm=intel.
18813 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
18814 avx512f_vcvttss2usi<round_saeonly_name>,
18815 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
18817 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
18818 avx512f_vcvttsd2usi<round_saeonly_name>,
18819 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
18820 Use %q1 instead of %1 for -masm=intel.
18821 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
18822 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
18823 of %3 for -masm=intel.
18824 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
18826 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
18828 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
18830 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
18831 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
18833 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
18835 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
18836 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
18837 %g1 and one with %0 and %1.
18838 (avx512er_vmrcp28<mode><round_saeonly_name>,
18839 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
18840 %1 for -masm=intel.
18841 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
18842 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
18843 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
18844 of %0 and %{%4%} for -masm=intel.
18845 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
18846 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
18847 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
18848 order of %0 and %{%5%}%{z%} for -masm=intel.
18850 2018-04-17 Jan Hubicka <jh@suse.cz>
18853 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
18855 2018-04-17 Martin Liska <mliska@suse.cz>
18858 * multiple_target.c (create_dispatcher_calls): Set apostrophes
18859 for target_clone error message. Make default implementation
18860 clone to be a local declaration.
18861 (separate_attrs): Add new argument and check for an empty
18863 (expand_target_clones): Handle it.
18864 (ipa_target_clone): Make redirection just for target_clones
18867 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
18868 Tom de Vries <tom@codesourcery.com>
18870 PR middle-end/84955
18871 * omp-expand.c (expand_oacc_for): Add dummy false branch for
18872 tiled basic blocks without omp continue statements.
18874 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
18877 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
18878 vec_extract expression as having side effects to make sure it gets
18881 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
18884 * config/i386/i386.c (get_builtin_code_for_version): Check
18887 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
18890 * config.gcc: Support "skylake".
18891 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
18893 * config/i386/i386.c (m_SKYLAKE): Define.
18894 (processor_target_table): Add "skylake".
18895 (ix86_option_override_internal): Add "skylake".
18896 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
18897 PROCESSOR_CANNONLAKE.
18898 (get_builtin_code_for_version): Fix priority for
18899 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
18900 PROCESSOR_SKYLAKE-AVX512.
18901 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
18902 (processor_type): Add PROCESSOR_SKYLAKE.
18904 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
18905 Jason Merrill <jason@redhat.com>
18908 * convert.c (convert_to_integer_1): Use direct recursion for
18909 enumeral types and types with a precision less than the number
18910 of bits in their mode.
18912 2018-04-16 Julia Koval <julia.koval@intel.com>
18915 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
18916 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
18918 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
18921 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
18922 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
18923 and -mno-direct-move.
18925 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
18928 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
18929 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
18930 Ensure negative shifts result in {0}.
18932 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
18934 PR rtl-optimization/79916
18935 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
18936 regs (if any) to define how to gnerate SD moves when LRA is in
18939 2018-04-13 Jakub Jelinek <jakub@redhat.com>
18941 PR rtl-optimization/85393
18942 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
18943 * except.c (expand_dw2_landing_pad_for_region): Make static.
18944 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
18945 a label and unconditional jump to old_bb, rather than
18946 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
18949 PR rtl-optimization/85376
18950 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
18951 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
18952 instead of a specific value.
18954 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
18955 Bin Cheng <bin.cheng@arm.com>
18957 PR tree-optimization/82965
18958 PR tree-optimization/83991
18959 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
18960 by_profile_only parameter.
18961 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
18962 information if the loop was predicted to iterate too many times.
18963 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
18965 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
18968 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
18971 2018-04-13 Martin Liska <mliska@suse.cz>
18972 Jakub Jelinek <jakub@redhat.com>
18974 PR middle-end/81657
18975 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
18976 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
18977 * builtins.c (expand_builtin_memory_copy_args): Use
18978 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
18979 handle dest_addr == pc_rtx.
18981 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
18984 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
18985 asked to not generate direct moves.
18986 (fix_trunc<mode>si2_stfiwx): Similar.
18987 (fix_trunc<mode>si2_internal): Similar.
18989 2018-04-12 Jakub Jelinek <jakub@redhat.com>
18992 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
18993 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
18994 lookup if dest in some wider mode is known to be const0_rtx and
18995 if so, record permanent equivalence for it to be ZERO_EXTEND of
18996 the narrower mode destination.
18998 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
19000 * lto-streamer-out.c (output_function): Revert 259346.
19001 * omp-expand.c (expand_oacc_for): Likewise.
19003 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
19005 PR rtl-optimization/85354
19006 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
19007 * sel-sched.c (sel_global_init): ... here.
19009 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
19012 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
19013 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
19014 mode for PE-COFF targets.
19015 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
19016 (i386_pe_asm_lto_end): Likewise.
19017 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
19018 (TARGET_ASM_LTO_END): Likewise.
19019 * config/i386/winnt.c (saved_debug_info_level): New static variable.
19020 (i386_pe_asm_lto_start): New function.
19021 (i386_pe_asm_lto_end): Likewise.
19023 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
19024 Richard Biener <rguenther@suse.de>
19026 PR middle-end/84955
19027 * lto-streamer-out.c (output_function): Fix CFG loop state before
19029 * omp-expand.c (expand_oacc_for): Handle calls to internal
19030 functions like regular functions.
19032 2018-04-12 Richard Biener <rguenther@suse.de>
19035 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
19036 for the early LTO debug to properly generate references to it
19037 during DIE emission. Do not re-use that for the skeleton for
19039 (dwarf2out_early_finish): Likewise.
19041 2018-04-12 Jakub Jelinek <jakub@redhat.com>
19044 * config/i386/sse.md
19045 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
19046 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
19047 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
19048 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
19049 and output is a reg, avoid creating invalid lowpart subreg, but
19050 instead split into a 512-bit move. Don't split if not AVX512VL,
19051 input is xmm16+ reg and output is a mem.
19052 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
19053 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
19054 xmm16+ reg and output is a mem.
19056 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19058 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
19059 also for flag_dwarf2_cfi_asm.
19061 2018-04-12 Jakub Jelinek <jakub@redhat.com>
19063 PR rtl-optimization/85342
19064 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
19065 a bool scalar var inside of the loop instead. Don't try to update
19066 recog_data.operand after failed apply_change_group.
19068 2018-04-12 Tom de Vries <tom@codesourcery.com>
19071 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
19072 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
19073 array with flexible array member as array without given dimension.
19074 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
19075 argument for undefined param to true.
19077 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
19080 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
19081 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
19082 from PowerPC section.
19083 * config/rs6000/sysv4.opt (mcall-): Improve help text.
19084 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
19085 help text that is too long.
19086 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
19087 help text that is too long.
19088 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
19089 help text that is too long.
19091 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
19093 * config/alpha/alpha.md (stack_probe_internal): Rename
19094 from "probe_stack". Update all callers.
19096 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19098 PR rtl-optimization/84566
19099 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
19100 sched_macro_fuse_insns.
19102 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19105 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
19106 (compute_block_dependences): ... from here.
19108 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19110 PR tree-optimization/85331
19111 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
19112 from int to HOST_WIDE_INT.
19114 2018-04-11 Martin Jambor <mjambor@suse.cz>
19117 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
19118 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
19119 not the same as the source val.
19120 (cgraph_edge_brings_value_p): New parameter.
19121 (gather_edges_for_value): Pass destination value to
19122 cgraph_edge_brings_value_p.
19123 (perhaps_add_new_callers): Likewise.
19124 (get_info_about_necessary_edges): Likewise and exclude values brought
19125 only by self-recursive edges.
19126 (create_specialized_node): Redirect only clones of self-calling edges.
19127 (+self_recursive_pass_through_p): New function.
19128 (find_more_scalar_values_for_callers_subset): Use it.
19129 (find_aggregate_values_for_callers_subset): Likewise.
19130 (known_aggs_to_agg_replacement_list): Removed.
19131 (decide_whether_version_node): Re-calculate known constants for all
19132 remaining context clones.
19134 2018-04-11 Richard Biener <rguenther@suse.de>
19137 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
19138 from early DWARF output.
19139 (dwarf2out_early_finish): Output line info unconditionally into
19140 early DWARF and add reference to it.
19142 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19145 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
19146 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
19147 other than V2DFmode using iptr mode attribute.
19148 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
19150 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19152 PR rtl-optimization/84659
19153 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
19155 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19158 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
19160 (output_loc_list): Pass address of a dummy size variable even in the
19161 locview handling loop.
19162 (index_location_lists): Add comment on why skip_loc_list_entry can't
19165 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
19168 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
19171 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
19174 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
19175 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
19176 and -mstring-compare-inline-limit.
19178 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19181 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
19182 for stack clash protection in a register whenever we need it to be in
19185 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19187 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
19188 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
19190 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19193 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
19195 (mlong-double-): Ditto.
19196 * config/rs6000/sysv4.opt (msdata=): Ditto.
19197 (mtls-size=): Ditto.
19199 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19201 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
19202 erroneous entries for
19203 "vector int vec_ldl (int, long int *)", and
19204 "vector unsigned int vec_ldl (int, unsigned long int *)".
19205 Add comments and entries for
19206 "vector bool char vec_ldl (int, bool char *)",
19207 "vector bool short vec_ldl (int, bool short *)",
19208 "vector bool int vec_ldl (int, bool int *)",
19209 "vector bool long long vec_ldl (int, bool long long *)",
19210 "vector pixel vec_ldl (int, pixel *)",
19211 "vector long long vec_ldl (int, long long *)",
19212 "vector unsigned long long vec_ldl (int, unsigned long long *)".
19213 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
19214 type tree bool_long_long_type_node and correct definition of
19215 bool_V2DI_type_node to make reference to this new type tree.
19216 (rs6000_mangle_type): Replace erroneous reference to
19217 bool_long_type_node with bool_long_long_type_node.
19218 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
19219 comments to emphasize sign distinctions for char and int types and
19220 replace RS6000_BTI_bool_long constant with
19221 RS6000_BTI_bool_long_long constant. Also add comment to restrict
19222 use of RS6000_BTI_pixel.
19223 (bool_long_type_node): Remove this macro definition.
19224 (bool_long_long_type_node): New macro definition
19226 2018-04-10 Jakub Jelinek <jakub@redhat.com>
19228 PR rtl-optimization/85300
19229 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
19230 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
19231 simplify_unary_operation fails.
19233 2018-04-10 Martin Liska <mliska@suse.cz>
19235 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
19236 cgraph_edge and ipa_ref.
19238 2018-04-10 Jakub Jelinek <jakub@redhat.com>
19242 * config/i386/sse.md
19243 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
19244 computation of the VEC_MERGE selector from mask.
19245 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
19246 Fix decoding of the VEC_MERGE selector into mask.
19248 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
19250 PR tree-optimization/85286
19251 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
19253 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
19255 * final.c (final_1): Set insn_last_address as well as
19256 insn_current_address.
19258 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19261 * explow.c (emit_stack_probe): Call validize_mem on memory location
19262 before passing it to gen_probe_stack. Create address operand and
19263 legitimize it for the probe_stack_address case.
19265 2018-04-09 Jan Hubicka <jh@suse.cz>
19268 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
19269 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
19270 * tree.c (free_lang_data_in_type): Fix handling of binfos;
19272 (free_lang_data): Rebuild type inheritance graph.
19274 2018-04-09 Martin Sebor <msebor@redhat.com>
19276 * invoke.texi (-finline-small-functions): Mention other optimization
19278 (-findirect-inlining, -fpartial-inlining): Same.
19279 (-finline-functions-called-once): Same.
19280 (-freorder-blocks-and-partition): Same.
19282 2018-04-09 Jan Hubicka <jh@suse.cz>
19285 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
19286 jumps; choose last target that matches the criteria (i.e.
19287 no partition changes for non-crossing jumps).
19288 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
19289 support for redirecting crossing jumps to non-crossing.
19291 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
19293 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
19294 also for naked functions.
19296 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
19298 * config/arc/arc.md (add_shift): New pattern.
19299 (add_shift2): Likewise.
19300 (sub_shift): Likewise.
19301 (sub_shift_cmp0_noout): Likewise.
19302 (compare_si_ashiftsi): Likewise.
19303 (xbfu_cmp0_noout): New combine pattern.
19304 (xbfu_cmp0"): Likewise.
19305 (movsi_set_cc_insn): Place the predicable variant first.
19306 (commutative_binary_cmp0_noout): Remove clobber.
19307 (commutative_binary_cmp0): New pattern.
19308 (noncommutative_binary_cmp0): Likewise.
19309 (noncommutative_binary_cmp0_noout): Likewise.
19310 (noncommutative_binary_comparison_result_used): Removed.
19311 (rsub_cmp0): New pattern.
19312 (rsub_cmp0_noout): Likewise.
19313 (extzvsi): Changed, keep only meaningful variants.
19314 (SQH, SEZ): New iterators.
19315 (SQH_postfix): New mode attribute.
19316 (SEZ_prefix): New code attribute.
19317 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
19318 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
19319 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
19320 of numerical value.
19321 (noncommutative_operator): Check the availability of barrel
19324 2018-04-09 Richard Biener <rguenther@suse.de>
19326 PR tree-optimization/85284
19327 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
19328 Only use the niter constraining form of simple_iv when the exit
19329 is always executed.
19331 2018-04-09 Tom de Vries <tom@codesourcery.com>
19334 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
19335 (define_expand "*memory_barrier"): New define_expand.
19336 (define_insn "memory_barrier"): New insn.
19338 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19340 PR rtl-optimization/80463
19341 PR rtl-optimization/83972
19342 PR rtl-optimization/83480
19344 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
19345 correct producer for the insn.
19346 (tidy_control_flow): Fixup seqnos in case of debug insns.
19348 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19350 PR rtl-optimization/83913
19352 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
19353 different sched-times when merging exprs.
19355 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19357 PR rtl-optimization/83962
19359 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
19360 tidy_fallthru_edge and tidy_control_flow.
19362 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19364 PR rtl-optimization/83530
19366 * sel-sched.c (force_next_insn): New global variable.
19367 (remove_insn_for_debug): When force_next_insn is true, also leave only
19368 next insn in the ready list.
19369 (sel_sched_region): When the region wasn't scheduled, make another pass
19370 over it with force_next_insn set to 1.
19372 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
19374 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
19376 * config/nds32/constants.md (unspec_volatile_element): Add enum values
19377 for interrupt control.
19378 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
19379 functions for interrupt control.
19380 * config/nds32/nds32-intrinsic.md: Likewise.
19381 * config/nds32/nds32_intrinsic.h: Likewise.
19382 * config/nds32/nds32.h (nds32_builtins): Likewise.
19384 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
19386 * config/nds32/nds32.c (nds32_init_machine_status,
19387 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
19388 strict_aligned_p field.
19389 (nds32_expand_to_rtl_hook): New function.
19390 (TARGET_EXPAND_TO_RTL_HOOK): Define.
19391 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
19393 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19394 Chung-Ju Wu <jasonwucj@gmail.com>
19396 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
19397 * config/nds32/nds32-n7.md: New file.
19398 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
19399 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
19401 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
19402 * config/nds32/nds32.md (pipeline_model): Add n7.
19403 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
19404 * config/nds32/pipelines.md: Include n7 settings.
19406 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19407 Chung-Ju Wu <jasonwucj@gmail.com>
19409 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
19410 * config/nds32/nds32-e8.md: New file.
19411 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
19412 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
19414 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
19415 * config/nds32/nds32.md (pipeline_model): Add e8.
19416 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
19417 * config/nds32/pipelines.md: Include e8 settings.
19419 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19420 Chung-Ju Wu <jasonwucj@gmail.com>
19422 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
19423 * config/nds32/nds32-n8.md: New file.
19424 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
19425 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
19427 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
19428 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
19429 * config/nds32/nds32.md (pipeline_model): Add n8.
19430 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
19431 * config/nds32/pipelines.md: Include n8 settings.
19433 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19434 Chung-Ju Wu <jasonwucj@gmail.com>
19436 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
19437 * config/nds32/nds32-n9-2r1w.md: New file.
19438 * config/nds32/nds32-n9-3r2w.md: New file.
19439 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
19440 nds32_register_ports): New or modify for cpu n9.
19441 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
19443 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
19444 * config/nds32/nds32-utils.c: New file.
19445 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
19446 TARGET_MUL_SLOW): Define.
19447 * config/nds32/nds32.md (pipeline_model): New attribute.
19448 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
19449 New options that support cpu n9.
19450 * config/nds32/pipelines.md: Include n9 settings.
19451 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
19453 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
19455 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
19456 information if necessary.
19457 (output_cond_branch_compare_zero): Likewise.
19458 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
19459 (nds32_target_alignment): Refine for alignment.
19460 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
19461 (FUNCTION_BOUNDARY): Modify.
19462 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
19464 * config/nds32/nds32.opt (malways-align, malign-functions): New.
19466 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
19468 * config/nds32/constants.md (unspec_volatile_element): Add values for
19469 TLB operation and data prefetch.
19470 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
19471 functions for TLB operation and data prefetch.
19472 * config/nds32/nds32-intrinsic.md: Likewise.
19473 * config/nds32/nds32_intrinsic.h: Likewise.
19474 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
19475 (nds32_print_operand): Likewise.
19476 * config/nds32/nds32.h (nds32_builtins): Likewise.
19478 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
19479 Andrew Pinski <pinsika@gcc.gnu.org>
19481 PR middle-end/82976
19482 * match.pd: Use constant_boolean_node of correct type instead of
19483 boolean_true_node or boolean_false_node for simplifying
19484 pointer comparisons to zero.
19486 2018-04-07 Jakub Jelinek <jakub@redhat.com>
19488 PR tree-optimization/80021
19489 * tree.c (verify_type_variant): Make error call in verify_variant_match
19490 translatable and remove final full stop.
19492 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19494 * config/nds32/constants.md (unspec_volatile_element): Add
19495 UNSPEC_VOLATILE_EH_RETURN.
19496 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
19497 nds32_output_stack_pop): Support dwarf exception handling process.
19498 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
19499 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
19500 exception handling process.
19501 (nds32_compute_stack_frame): Likewise.
19502 (nds32_return_addr_rtx): Likewise.
19503 (nds32_initial_elimination_offset): Likewise.
19504 (nds32_expand_prologue): Likewise.
19505 (nds32_expand_epilogue): Likewise.
19506 (nds32_dynamic_chain_address): New function.
19507 * config/nds32/nds32.h (machine_function): Add fields for dwarf
19508 exception handling.
19509 (DYNAMIC_CHAIN_ADDRESS): Define.
19510 (EH_RETURN_DATA_REGNO): Define.
19511 (EH_RETURN_STACKADJ_RTX): Define.
19512 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
19513 patterns for dwarf exception handling.
19515 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19517 * config/nds32/nds32.h: Clean up obsolete macros.
19519 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19521 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
19522 Add enum values for particular instructions.
19523 * config/nds32/nds32-intrinsic.c: Implementation of expanding
19524 particular intrinsic functions.
19525 * config/nds32/nds32-intrinsic.md: Likewise.
19526 * config/nds32/nds32_intrinsic.h: Likewise.
19527 * config/nds32/nds32.h (nds32_builtins): Likewise.
19528 * config/nds32/nds32.md (type): Add pbsad and pbsada.
19529 (btst, ave): New patterns for particular instructions.
19531 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19533 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
19534 Add enum values for atomic load/store and memory sync.
19535 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
19537 * config/nds32/nds32-intrinsic.md: Likewise.
19538 * config/nds32/nds32_intrinsic.h: Likewise.
19539 * config/nds32/nds32.h (nds32_builtins): Likewise.
19541 2018-04-07 Jakub Jelinek <jakub@redhat.com>
19543 PR tree-optimization/85257
19544 * fold-const.c (native_encode_vector): If not all elts could fit
19545 and off is -1, return 0 rather than offset.
19546 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
19547 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
19548 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
19549 adjust buffer in native_interpret_expr call.
19551 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19553 * config/nds32/constants.md (unspec_volatile_element): Add cache
19554 control enum values.
19555 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
19556 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
19557 * config/nds32/nds32.c (nds32_cctl_names): New.
19558 (nds32_print_operand): Handle cache control register names.
19559 * config/nds32/nds32.h (nds32_builtins): New enum values.
19560 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
19562 * config/nds32/nds32.md (type): Add mmu.
19563 * config/nds32/pipelines.md (simple_insn): Add mmu.
19565 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19567 * config/nds32/nds32.md (type): Remove call.
19568 * config/nds32/pipelines.md (simple_insn): Likewise.
19570 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19572 * config/nds32/constants.md (unspec_volatile_element): Add
19573 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
19574 UNSPEC_VOLATILE_FMFCFG.
19575 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
19576 description for fmfcfg and fmfcsr.
19577 (bdesc_1arg): Add fmtcsr.
19578 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
19579 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
19580 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
19581 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
19582 unspec_fmfcfg): New patterns.
19583 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
19584 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
19585 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
19586 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
19587 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
19588 __nds32__fmfcfg): Define.
19590 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19592 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
19593 intrinsic register names.
19594 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
19595 intrinsic register enum values and macros.
19597 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19599 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
19600 for load/store addressing form.
19601 (nds32_print_operand_address): Likewise.
19603 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
19606 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
19607 based on LABEL_REF. Remove useless assertion.
19608 (pic_address_needs_scratch): Fix formatting.
19609 (sparc_legitimize_pic_address): Minor tweaks.
19610 (sparc_delegitimize_address): Adjust assertion accordingly.
19611 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
19612 into symbolic_operand.
19613 (movsi_high_pic_label_ref): Likewise.
19614 (movsi_lo_sum_pic_label_ref): Likewise.
19615 (movdi_pic_label_ref): Likewise.
19616 (movdi_high_pic_label_ref): Likewise.
19617 (movdi_lo_sum_pic_label_ref): Likewise.
19619 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
19621 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
19622 custom LIB_SPEC setup.
19624 2018-04-06 Ruslan Bukin <br@bsdpad.com>
19625 Kito Cheng <kito.cheng@gmail.com>
19627 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
19628 * config/riscv/freebsd.h: New.
19630 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
19632 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
19633 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
19636 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
19637 Kito Cheng <kito.cheng@gmail.com>
19639 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
19640 nds32_output_call, nds32_symbol_binds_local_p): New functions.
19641 * config/nds32/nds32-protos.h (nds32_output_call,
19642 nds32_output_return): Declare.
19643 * config/nds32/nds32.md: Refine all the call and return patterns.
19645 2018-04-06 Jakub Jelinek <jakub@redhat.com>
19648 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
19649 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
19651 PR rtl-optimization/84872
19652 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
19653 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
19654 EDGE_CROSSING edge.
19656 2018-04-06 Tamar Christina <tamar.christina@arm.com>
19658 * expr.c (copy_blkmode_to_reg): Revert 254862.
19659 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
19661 2018-04-06 Richard Biener <rguenther@suse.de>
19663 PR middle-end/85244
19664 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
19665 after seeing a component reference with an adjacent field. Treat
19666 refs to arrays at struct end of external decls similar to
19667 refs to unconstrained commons.
19669 2018-04-06 Jakub Jelinek <jakub@redhat.com>
19672 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
19673 look through SAVE_EXPRs with non-side-effects argument. Adjust
19675 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
19678 2018-04-06 Richard Biener <rguenther@suse.de>
19680 PR middle-end/85180
19681 * alias.c (find_base_term): New wrapper around find_base_term
19682 unwinding CSELIB_VAL_PTR changes.
19683 (find_base_term): Do not restore CSELIB_VAL_PTR during the
19686 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19688 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
19690 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
19691 constant definitions.
19692 ("nop"): lr 0,0 -> nopr r0
19693 ("nop_lr0", "nop_lr1"): New insn definitions.
19695 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
19697 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
19698 NDS32_V3PUSH_AVAILABLE_P macro.
19700 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
19701 Chung-Ju Wu <jasonwucj@gmail.com>
19703 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
19704 (nds32*-*-*): Add float and fpu_config into supported_defaults.
19705 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
19706 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
19707 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
19708 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
19709 * config/nds32/constraints.md: New constraints and checking for hard
19710 float configuration.
19711 * config/nds32/iterators.md: New mode iterator and attribute for hard
19712 float configuration.
19713 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
19715 * config/nds32/nds32-fpu.md: New file.
19716 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
19717 deal with hard float code generation.
19718 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
19720 (abi_type, float_reg_number): New enum type.
19721 * config/nds32/nds32-predicates.c: New predicates for hard float.
19722 * config/nds32/nds32-protos.h: Declare functions for hard float.
19723 * config/nds32/nds32.c: Implementation for hard float configuration.
19724 * config/nds32/nds32.h: Definitions for hard float configuration.
19725 * config/nds32/nds32.md: Include hard float machine description and
19726 modify patterns for hard float configuration.
19727 * config/nds32/nds32.opt: New options for hard float configuration.
19728 * config/nds32/predicates.md: New predicates for hard float
19731 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
19733 * common/config/nds32/nds32-common.c
19734 (nds32_option_optimization_table): Enable -mreleax-hint by default.
19736 2018-04-05 Jakub Jelinek <jakub@redhat.com>
19738 PR middle-end/85195
19739 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
19740 CONSTRUCTOR_ELT (ctor, ...)->value.
19742 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
19745 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
19747 2018-04-05 Tom de Vries <tom@codesourcery.com>
19750 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
19753 2018-04-05 Shiva Chen <shiva0217@gmail.com>
19754 Kito Cheng <kito.cheng@gmail.com>
19756 * config/nds32/constraints.md (U33): Fine-tune checking condition.
19757 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
19758 * config/nds32/nds32.h (nds32_16bit_address_type): Add
19759 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
19761 2018-04-05 Shiva Chen <shiva0217@gmail.com>
19762 Kito Cheng <kito.cheng@gmail.com>
19764 * config/nds32/constraints.md (Ufe): New memory constraint.
19765 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
19766 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
19767 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
19769 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
19770 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
19772 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19774 * config/nds32/nds32.md: Use optimize_size in the condition for
19775 alu-shift instructions.
19777 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19779 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
19781 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19783 * config/nds32/nds32.md (negsi2): Refine pattern.
19785 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
19786 Chung-Ju Wu <jasonwucj@gmail.com>
19788 * config/nds32/iterators.md (shift_rotate): New code iterator.
19789 (shift): New code attribute.
19790 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
19791 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
19792 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
19793 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
19794 bit-wise operations.
19795 (andsi3, *andsi3): Ditto.
19796 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
19797 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
19798 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
19799 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
19800 nds32_ior_operand, nds32_xor_operand): New predicates.
19802 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19804 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
19805 (addsi3, subsi3): ... this.
19807 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19809 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
19811 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19813 * config/nds32/nds32.md: Adjust indention.
19815 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
19817 * config/nds32/nds32.md (feature): New attribute.
19819 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19821 * config/nds32/nds32.md (subtype): New attribute.
19823 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19826 * config/arm/arm-builtins.c (arm_expand_builtin): Change
19827 expansion to perform a bitwise AND of the argument followed by a
19828 boolean negation of the result.
19830 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
19832 PR rtl-optimization/84878
19833 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
19834 the basic block. Assert the use reference is not artificial and that
19835 it has an associated insn.
19837 2018-04-04 Michael Matz <matz@suse.de>
19839 * builtins.c (compute_objsize): Pass correct operand
19840 to array_at_struct_end_p.
19842 2018-04-04 Richard Biener <rguenther@suse.de>
19845 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
19846 from contexts for DINFO_LEVEL_TERSE and below.
19848 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
19850 * config/nds32/nds32-doubleword.md (move_<mode>): Require
19851 resiter_operand condition.
19852 * config/nds32/nds32.md (*move<mode>): Ditto.
19854 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
19855 Monk Chiang <sh.chiang04@gmail.com>
19857 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
19859 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
19861 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
19863 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
19864 Kito Cheng <kito.cheng@gmail.com>
19866 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
19867 nds32_cond_code_str, output_cond_branch,
19868 output_cond_branch_compare_zero, nds32_expand_cbranch,
19869 nds32_expand_cstore, nds32_expand_movcc,
19870 nds32_output_cbranchsi4_equality_zero,
19871 nds32_output_cbranchsi4_equality_reg,
19872 nds32_output_cbranchsi4_equality_reg_or_const_int,
19873 nds32_output_cbranchsi4_greater_less_zero: New functions.
19874 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
19875 nds32_expand_cstore, nds32_expand_movcc,
19876 nds32_output_cbranchsi4_equality_zero,
19877 nds32_output_cbranchsi4_equality_reg,
19878 nds32_output_cbranchsi4_equality_reg_or_const_int,
19879 nds32_output_cbranchsi4_greater_less_zero): Declare.
19880 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
19881 nds32_rimm11s_operand): New predicates.
19882 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
19883 * config/nds32/nds32.md: Rewrite all the branch and conditional move
19886 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
19888 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
19889 * config/nds32/nds32.md: Ditto.
19890 * config/nds32/pipelines.md: Ditto.
19892 2018-04-04 Richard Biener <rguenther@suse.de>
19894 PR tree-optimization/85168
19895 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
19896 propagating abnormals.
19898 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
19900 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
19902 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
19903 Kito Cheng <kito.cheng@gmail.com>
19905 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
19906 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
19907 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
19908 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
19909 * config/nds32/nds32.md (sibcall_internal): New.
19910 (sibcall_register): Remove.
19911 (sibcall_immediate): Remove.
19912 (sibcall_value_internal): New.
19913 (sibcall_value_register): Remove.
19914 (sibcall_value_immediate): Remove.
19915 * config/nds32/predicates.md (nds32_general_register_operand): New.
19916 (nds32_call_address_operand): New.
19918 2018-04-03 Jakub Jelinek <jakub@redhat.com>
19920 PR rtl-optimization/85167
19921 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
19922 bb_defs if *split_p, instead preinitialize it to NULL.
19924 PR tree-optimization/85156
19925 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
19926 evaluating the argument multiple times.
19928 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
19930 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
19932 (_mm_cvtpd_ps): Likewise.
19933 (_mm_cvttpd_epi32): Likewise.
19934 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
19935 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
19936 vector, pixel, and bool following altivec.h include.
19938 2018-04-03 Martin Sebor <msebor@redhat.com>
19940 * doc/extend.texi (Common Function Attributes): Clarify.
19941 (const attribute): Likewise.
19942 (pure attribute): Likewise.
19944 2018-04-03 Jakub Jelinek <jakub@redhat.com>
19947 * config/i386/i386.c (ix86_expand_vector_set): Use
19948 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
19950 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
19952 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
19953 instructions when changing rounding bits to preserve precision bits
19954 in the x87 control word.
19956 2018-04-03 Martin Liska <mliska@suse.cz>
19958 PR tree-optimization/82491
19959 * rtl.h (strip_offset_and_add): Replace += suboffset with
19960 poly_uint64 () + suboffset.
19962 2018-03-29 Martin Liska <mliska@suse.cz>
19963 Martin Jambor <mjambor@suse.cz>
19966 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
19967 param_type is not an integral or pointer type.
19969 2018-04-03 Richard Biener <rguenther@suse.de>
19971 * sese.h (recompute_all_dominators): Remove.
19973 2018-04-02 Martin Sebor <msebor@redhat.com>
19975 * doc/invoke.texi (-Wrestrict): Fix typos.
19977 2018-04-02 Jim Wilson <jimw@sifive.com>
19979 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
19980 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
19981 (<optab>di3, <optab>si3_extend): Likewise.
19982 (<optab>si3_mask, <optab>si3_mask_1): New.
19983 (<optab>di3_mask, <optab>di3_mask_1): New.
19984 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
19985 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
19986 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
19988 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
19990 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
19993 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
19995 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
19996 (nds32_canonicalize_comparison): New function.
19998 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
19999 Kito Cheng <kito.cheng@gmail.com>
20000 Kuan-Lin Chen <kuanlinchentw@gmail.com>
20002 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
20003 * config/nds32/constants.md (unspec_volatile_element): Add
20004 UNSPEC_VOLATILE_RELAX_GROUP.
20005 * config/nds32/nds32-relax-opt.c: New file.
20006 * config/nds32/nds32-predicates.c
20007 (nds32_symbol_load_store_p): New function.
20008 * config/nds32/nds32-protos.h
20009 (nds32_symbol_load_store_p): Declare function.
20010 (make_pass_nds32_relax_opt): Declare new rtl pass function.
20011 * config/nds32/nds32.c
20012 (nds32_register_pass): New function to register pass.
20013 (nds32_register_passes): New function to register passes.
20014 * config/nds32/nds32.md (relax_group): New pattern.
20015 * config/nds32/nds32.opt (mrelax-hint): New option.
20016 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
20018 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
20020 * config/nds32/t-nds32: Modify files dependency.
20022 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
20024 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
20025 (PROFILE_HOOK): Define its implementation.
20027 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
20029 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
20030 type and 32-bit size.
20032 2018-04-01 Jakub Jelinek <jakub@redhat.com>
20034 PR middle-end/85090
20035 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
20036 (V_128_256): New mode iterator.
20037 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
20038 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
20039 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
20041 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
20042 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
20044 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
20047 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
20048 NaN inputs correctly.
20050 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
20053 * config/rs6000/vsx.md (??r): New mode attribute.
20054 (*vsx_mov<mode>_64bit): Use it.
20055 (*vsx_mov<mode>_32bit): Likewise.
20057 2018-03-30 Martin Sebor <msebor@redhat.com>
20059 PR tree-optimization/84818
20060 * builtins.c (check_access): Use warning_n.
20062 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20065 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
20067 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
20070 2018-03-30 Julia Koval <julia.koval@intel.com>
20073 * x86-tune.def (movx, partial_reg_dependency): Enable for
20076 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
20078 PR inline-asm/84985
20079 * lra-constraints.c (process_alt_operands): Move setting
20080 this_alternative_matches below.
20082 2018-03-29 Martin Liska <mliska@suse.cz>
20085 * doc/invoke.texi: Document how LTO works with debug info.
20086 Describe auto-load support of binutils. Mention 'x86-64'
20087 as valid option value of -march option.
20089 2018-03-29 Jakub Jelinek <jakub@redhat.com>
20091 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
20094 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
20095 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
20096 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
20099 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
20102 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
20103 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
20104 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
20105 for RS6000_BTM_POWERPC64.
20106 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
20107 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
20108 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
20111 (DIVDEU): Likewise.
20113 2018-03-28 Carl Love <cel@us.ibm.com>
20116 2017-09-27 Carl Love <cel@us.ibm.com>
20118 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
20119 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
20120 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
20123 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20125 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
20126 instead of __vector bool.
20127 (_mm_max_pu8): Likewise.
20128 (_mm_min_pi16): Likewise.
20130 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
20133 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
20134 (DIVWEUO): Likewise.
20135 (DIVDEO): Likewise.
20136 (DIVDEUO): Likewise.
20137 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
20138 DIVWEUO and DIVDEUO.
20139 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
20140 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
20141 (div_extend): Likewise.
20142 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
20144 (__builtin_divweuo): Likewise.
20145 (__builtin_divdeo): Likewise.
20146 (__builtin_divdeuo): Likewise.
20148 2018-03-28 Jakub Jelinek <jakub@redhat.com>
20151 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
20152 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
20154 PR tree-optimization/82004
20155 * gimple-match-head.c (optimize_pow_to_exp): New function.
20156 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
20157 Don't fold to exp if optimize_pow_to_exp is false.
20159 2018-03-28 Martin Liska <mliska@suse.cz>
20162 * calls.c (initialize_argument_information): Fix trailing space.
20163 * common.opt: Fix typo and provide better explanation for
20164 -fsanitize-coverage option.
20165 * config/i386/i386.opt: Fix typo.
20167 2018-03-28 Jakub Jelinek <jakub@redhat.com>
20168 Martin Liska <mliska@suse.cz>
20171 * gimplify.c (asan_poison_variable): Don't do the check for
20172 gimplify_omp_ctxp here.
20173 (gimplify_decl_expr): Do it here.
20174 (gimplify_target_expr): Likewise.
20176 2018-03-28 Martin Liska <mliska@suse.cz>
20179 * config/i386/i386.c (ix86_function_arg_advance): Do not call
20180 chkp_type_bounds_count if MPX is not enabled.
20182 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
20184 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
20186 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20189 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
20190 function to create the function decl for complex long double
20191 multiply and divide for -mabi=ieeelongdouble.
20192 (init_float128_ieee): Call it.
20194 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
20197 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
20198 -fcf-protection=branch -mibt.
20199 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
20201 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20204 * config/arm/arm.c (arm_valid_symbolic_address): Handle
20205 arm_word_relocations.
20207 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
20210 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
20211 extern array declarations.
20213 2018-03-27 Richard Biener <rguenther@suse.de>
20215 PR middle-end/84067
20216 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
20217 explicit single_use checks.
20219 2018-03-27 Richard Biener <rguenther@suse.de>
20221 PR tree-optimization/85082
20222 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
20225 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20227 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
20228 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
20229 Turn on fasynchronous-unwind-tables and funwind-tables.
20231 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
20234 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
20235 (*bmi_blsr_<mode>_ccz): Ditto.
20237 2018-03-26 Tom de Vries <tom@codesourcery.com>
20239 PR tree-optimization/85063
20240 * omp-general.c (offloading_function_p): New function. Factor out
20242 * omp-offload.c (pass_omp_target_link::gate): ... here.
20243 * omp-general.h (offloading_function_p): Declare.
20244 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
20245 with attribute omp declare target for offloading functions.
20247 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
20249 PR tree-optimization/84005
20250 * tree-data-ref.h (get_base_for_alignment): Declare.
20251 * tree-data-ref.c (get_base_for_alignment_1): New function.
20252 (get_base_for_alignment): Likewise.
20253 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
20254 get_base_for_alignment to find a suitable base object, instead
20255 of always using drb->base_address.
20257 2018-03-23 Jakub Jelinek <jakub@redhat.com>
20259 PR inline-asm/85022
20260 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
20261 known size by default.
20263 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
20265 PR inline-asm/85030
20266 * lra-constraints.c (process_alt_operands): Don't match BLKmode
20267 and non BLKmode operands.
20269 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20272 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
20273 Clean up attributes.
20275 2018-03-23 Richard Biener <rguenther@suse.de>
20278 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
20279 we are going to emit early debug for LTO.
20281 2018-03-23 Jakub Jelinek <jakub@redhat.com>
20283 PR inline-asm/85034
20284 * function.c (match_asm_constraints_1): Don't optimize if input
20285 doesn't satisfy general_operand predicate for output's mode.
20287 PR inline-asm/85022
20288 * alias.c (write_dependence_p): Don't require for x_canonicalized
20289 non-VOIDmode if x has VOIDmode.
20292 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
20293 just don't try to optimize it rather than assert it never happens.
20295 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
20297 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
20298 macro expansions for definition of ST_INTERNAL_<mode> and
20299 LD_INTERNAL_<mode> builtins.
20300 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
20302 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
20304 (altivec_expand_st_builtin): Likewise.
20305 (altivec_expand_builtin): Remove calls to deleted functions.
20306 (rs6000_address_for_altivec): Delete this function.
20307 * config/rs6000/vector.md: Remove expands for
20308 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
20310 2018-03-22 Sudakshina Das <sudi.das@arm.com>
20313 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
20314 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
20315 re-computing once computed.
20316 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
20317 (arm_init_machine_status): Initialize
20318 machine->static_chain_stack_bytes.
20320 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
20323 * doc/extend.texi: Add four new prototypes for vec_ld.
20324 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
20325 definitions for more logical presentation.
20326 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
20327 entries for V1TI variants of __builtin_altivec_ld builtin.
20328 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
20329 handling of V1TI variant of LVX icode pattern.
20330 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
20331 (rs6000_gimple_fold_builtin): Likewise.
20332 (altivec_init_builtins): Add code to define
20333 __builtin_altivec_lvx_v1ti function.
20335 2018-03-22 Jakub Jelinek <jakub@redhat.com>
20337 PR inline-asm/84941
20338 * function.c (match_asm_constraints_1): Don't do the optimization
20339 if input isn't a REG, SUBREG, MEM or constant.
20341 2018-03-22 Tom de Vries <tom@codesourcery.com>
20343 PR tree-optimization/84956
20344 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
20345 bb_has_abnormal_pred.
20347 2018-03-22 Jakub Jelinek <jakub@redhat.com>
20350 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
20351 DECL_INITIAL (decl) to decl at the end.
20352 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
20353 adjust the comment.
20355 2018-03-21 Joseph Myers <joseph@codesourcery.com>
20357 * doc/extend.texi (__builtin_tgmath): Document when complex
20358 integer types are treated as _Complex _Float64.
20360 2018-03-21 Tom de Vries <tom@codesourcery.com>
20362 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
20364 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20366 PR tree-optimization/84960
20367 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
20368 if it is ENTRY block, move them into single succ of ENTRY in that case.
20370 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
20372 PR tree-optimization/84811
20373 * poly-int.h (poly_span_traits): Remove the T3 parameter and
20374 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
20375 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
20376 (known_subrange_p): Update accordingly. Cast each value involved
20377 in the size comparison, rather than casting the result of the
20380 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20382 PR tree-optimization/84982
20383 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
20384 by flipping the least significant bit rather than all bits from
20385 bitpos to bitpos + bitsize - 1.
20387 2018-03-21 Nathan Sidwell <nathan@acm.org>
20389 * doc/extend.texi (Deprecated Features): Remove mention of
20390 long-deleted deprecations.
20392 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20395 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
20396 * configure: Regenerate.
20398 2018-03-21 Tom de Vries <tom@codesourcery.com>
20400 PR tree-optimization/83126
20401 * tree-parloops.c (num_phis): New function.
20402 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
20404 2018-03-21 Nathan Sidwell <nathan@acm.org>
20406 * doc/extend.texi (Deprecated Features): Update deprecated flags,
20407 mention anon-struct/union members and trailing attributes.
20409 2018-03-21 Bin Cheng <bin.cheng@arm.com>
20411 PR tree-optimization/84969
20412 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
20413 builtin memset partitions if they set different rhs values.
20415 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20417 PR rtl-optimization/84989
20418 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
20419 VEC_DUPLICATE with scalar result mode.
20421 2018-03-21 Martin Liska <mliska@suse.cz>
20424 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
20425 not intended return statement.
20427 2018-03-21 Martin Liska <mliska@suse.cz>
20430 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
20431 (chkp_find_bound_slots_1): Limit number of iterations.
20433 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
20436 * Minor grammar fixes for x86 options.
20438 2018-03-20 Jakub Jelinek <jakub@redhat.com>
20441 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
20442 holding REG_CFA_RESTORE notes, instead turn them into a USE.
20444 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
20447 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
20448 (altivec_lvx_<mode>_1op): Likewise.
20449 (altivec_stvx_<mode>_2op): Likewise.
20450 (altivec_stvx_<mode>_1op): Likewise.
20451 (altivec_lvx_<VM2:mode>): New define_expand.
20452 (altivec_stvx_<VM2:mode>): Likewise.
20453 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
20454 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
20455 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
20456 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
20457 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
20458 (rs6000_gen_lvx): Likewise.
20459 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
20460 (altivec_expand_stv_builtin): Likewise.
20461 (altivec_expand_builtin): Likewise.
20462 * config/rs6000/vector.md: Likewise.
20464 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20467 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
20470 2018-03-20 Richard Biener <rguenther@suse.de>
20473 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
20474 sign-conversions as zero, fall back to standard scalar_stmt
20477 2018-03-20 Martin Liska <mliska@suse.cz>
20480 * predict.c (rebuild_frequencies): Handle case when we have
20481 PROFILE_ABSENT, but flag_guess_branch_prob is false.
20483 2018-03-20 Jakub Jelinek <jakub@redhat.com>
20486 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
20487 flag_section_anchors.
20488 * varasm.c (use_blocks_for_decl_p): Remove hack for
20489 dw2_force_const_mem.
20492 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
20494 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
20495 be created, use lowpart_subreg of operands[0] rather than operands[0]
20497 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
20498 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
20499 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
20500 and n constraint instead of aarch64_shift_imm_di and Usd.
20501 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
20502 (*aarch64_<optab>_reg_minus<mode>3): ... this.
20504 2018-03-20 Sudakshina Das <sudi.das@arm.com>
20507 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
20508 to favor GPR over NEON registers.
20509 (<shift>di3_neon): Likewise.
20511 2018-03-20 Tom de Vries <tom@codesourcery.com>
20514 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
20515 (nvptx_process_pars): Emit bar.sync asap and alap.
20517 2018-03-20 Tom de Vries <tom@codesourcery.com>
20520 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
20521 seen_label if seen_label is already set.
20523 2018-03-20 Jakub Jelinek <jakub@redhat.com>
20526 * config/i386/i386.c (fold_builtin_cpu): For features above 31
20527 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
20528 Use 1U instead of 1. Formatting fixes.
20531 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
20532 instead of TREE_TYPE (s1) for the return value.
20534 2018-03-19 Jakub Jelinek <jakub@redhat.com>
20536 PR tree-optimization/84946
20537 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
20538 bitsize + bitsize in poly_uint64 rather than poly_int64.
20541 * dwarf2asm.c: Include fold-const.c.
20542 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
20543 of decl rather than decl itself.
20545 PR rtl-optimization/84643
20546 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
20548 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
20551 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
20552 calling assemble_variable.
20554 2018-03-19 Sudakshina Das <sudi.das@arm.com>
20557 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
20558 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
20560 2018-03-19 Jim Wilson <jimw@sifive.com>
20563 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
20564 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
20565 (riscv_first_stack_step): Likewise.
20566 (riscv_option_override): Use STACK_BOUNDARY instead of
20567 MIN_STACK_BOUNDARY.
20568 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
20569 MIN_STACK_BOUNDARY.
20570 (BIGGEST_ALIGNMENT): Set to 128.
20571 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
20572 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
20575 2018-03-19 Richard Biener <rguenther@suse.de>
20577 PR tree-optimization/84933
20578 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
20579 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
20581 2018-03-19 Richard Biener <rguenther@suse.de>
20583 PR tree-optimization/84859
20584 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
20585 (cond_if_else_store_replacement): Perform sinking operation on
20586 single-store BBs regardless of MAX_STORES_TO_SINK setting.
20587 Generalize what a BB with a single eligible store is.
20589 2018-03-19 Richard Biener <rguenther@suse.de>
20591 PR tree-optimization/84929
20592 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
20593 chrec_is_positive against non-chrec arg.
20595 2018-03-19 Tamar Christina <tamar.christina@arm.com>
20598 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
20600 2018-03-18 Martin Liska <mliska@suse.cz>
20602 PR rtl-optimization/84635
20603 * regrename.c (build_def_use): Use matches_mode only when
20606 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
20608 PR tree-optimization/84913
20609 * tree-vect-loop.c (vectorizable_reduction): Don't try to
20610 vectorize chains of COND_EXPRs.
20612 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
20614 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
20616 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
20618 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
20620 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
20622 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
20624 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
20625 Kito Cheng <kito.cheng@gmail.com>
20627 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
20628 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
20629 (nds32_adjust_reg_alloc_order): New function.
20630 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
20632 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
20634 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
20635 nds32_print_operand, nds32_print_operand_address): Use
20636 HOST_WIDE_INT_PRINT_DEC instead.
20638 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
20640 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
20642 2018-03-17 Jakub Jelinek <jakub@redhat.com>
20645 * config/i386/i386.c (initial_ix86_tune_features,
20646 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
20647 unsigned long long.
20648 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
20649 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
20650 rather than 1u << ix86_tune. Formatting fix.
20651 (ix86_option_override_internal): Change ix86_arch_mask from
20652 unsigned int to unsigned HOST_WIDE_INT, initialize to
20653 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
20654 (ix86_function_specific_restore): Likewise.
20656 2018-03-16 Jakub Jelinek <jakub@redhat.com>
20659 * postreload.c (reload_combine_recognize_pattern): Perform
20660 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
20661 truncate_int_for_mode the result for the destination's mode.
20664 * hsa-gen.c (mem_type_for_type): Fix comment typo.
20665 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
20667 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
20670 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
20673 * lra-assigns.c (lra_split_hard_reg_for): Don't use
20674 regno_allocno_class_array and sorted_pseudos.
20675 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
20676 insns where regno is used.
20678 2018-03-16 Martin Liska <mliska@suse.cz>
20681 * multiple_target.c (create_dispatcher_calls): Redirect
20682 reference in the symbol table.
20684 2018-03-16 Martin Liska <mliska@suse.cz>
20687 * multiple_target.c (create_dispatcher_calls): Redirect also
20690 2018-03-16 Jakub Jelinek <jakub@redhat.com>
20694 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
20695 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
20696 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
20698 2018-03-16 Julia Koval <julia.koval@intel.com>
20700 * doc/invoke.texi (Skylake Server): Add CLWB.
20701 Cannonlake): Remove CLWB.
20703 2018-03-16 Jakub Jelinek <jakub@redhat.com>
20705 PR tree-optimization/84841
20706 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
20708 (FLOAT_ONE_CONST_TYPE): Define.
20709 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
20710 (sort_by_operand_rank): Put entries with higher constant_type last
20711 rather than first to match comments.
20713 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
20715 * config/nios2/nios2.md (movsi_internal): Fix thinko in
20718 2018-03-15 Jakub Jelinek <jakub@redhat.com>
20721 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
20722 check and use address of target always.
20724 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
20727 * config/i386/i386.c (indirect_thunk_needed): Update comments.
20728 (indirect_thunk_bnd_needed): Likewise.
20729 (indirect_thunks_used): Likewise.
20730 (indirect_thunks_bnd_used): Likewise.
20731 (indirect_return_needed): New.
20732 (indirect_return_bnd_needed): Likewise.
20733 (output_indirect_thunk_function): Add a bool argument for
20735 (output_indirect_thunk_function): Don't generate alias for
20736 function return thunk.
20737 (ix86_code_end): Call output_indirect_thunk_function to generate
20738 function return thunks.
20739 (ix86_output_function_return): Set indirect_return_bnd_needed
20740 and indirect_return_needed instead of indirect_thunk_bnd_needed
20741 and indirect_thunk_needed.
20743 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
20745 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
20746 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
20747 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
20749 2018-03-15 David Malcolm <dmalcolm@redhat.com>
20750 Paul Hua <paul.hua.gm@gmail.com>
20753 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
20755 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
20757 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
20758 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
20759 resp. SFmode cases.
20761 2018-03-15 Tamar Christina <tamar.christina@arm.com>
20764 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
20765 instead of GET_MODE_SIZE when comparing Units.
20767 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
20770 * varasm.c (hash_section): Return an unchangeble hash value
20771 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
20772 Return !aarch64_can_use_per_function_literal_pools_p ().
20774 2018-03-15 Jakub Jelinek <jakub@redhat.com>
20777 * optabs.c (emit_conditional_move): Pass address of cmode's copy
20778 rather than address of cmode as last argument to prepare_cmp_insn.
20780 2018-03-15 Julia Koval <julia.koval@intel.com>
20782 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
20783 F_AVX512VNNI, F_AVX512BITALG): New.
20785 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
20788 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
20789 insn for floating-point loads and stores.
20791 2018-03-14 Carl Love <cel@us.ibm.com>
20793 * config/rs6000/rs6000-c.c: Add macro definitions for
20794 ALTIVEC_BUILTIN_VEC_PERMXOR.
20795 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
20796 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
20797 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
20798 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
20800 * config/doc/extend.texi: Add prototypes for vec_permxor.
20802 2018-03-14 David Malcolm <dmalcolm@redhat.com>
20805 * diagnostic-show-locus.c (class layout_point): Convert m_line
20806 from int to linenum_type.
20807 (line_span::comparator): Use linenum "compare" function when
20808 comparing line numbers.
20809 (test_line_span): New function.
20810 (layout_range::contains_point): Convert param "row" from int to
20812 (layout_range::intersects_line_p): Likewise.
20813 (layout::will_show_line_p): Likewise.
20814 (layout::print_source_line): Likewise.
20815 (layout::should_print_annotation_line_p): Likewise.
20816 (layout::print_annotation_line): Likewise.
20817 (layout::print_leading_fixits): Likewise.
20818 (layout::annotation_line_showed_range_p): Likewise.
20819 (struct line_corrections): Likewise for field m_row.
20820 (line_corrections::line_corrections): Likewise for param "row".
20821 (layout::print_trailing_fixits): Likewise.
20822 (layout::get_state_at_point): Likewise.
20823 (layout::get_x_bound_for_row): Likewise.
20824 (layout::print_line): Likewise.
20825 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
20826 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
20827 * input.c (selftest::test_linenum_comparisons): New function.
20828 (selftest::input_c_tests): Call it.
20829 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
20830 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
20831 * selftest.h (ASSERT_GT): New macro.
20832 (ASSERT_GT_AT): New macro.
20833 (ASSERT_LT): New macro.
20834 (ASSERT_LT_AT): New macro.
20836 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
20838 PR rtl-optimization/84780
20839 * combine.c (distribute_links): Don't make a link based on pc_rtx.
20841 2018-03-14 Martin Liska <mliska@suse.cz>
20843 * tree.c (record_node_allocation_statistics): Use
20844 get_stats_node_kind.
20845 (get_stats_node_kind): New function extracted from
20846 record_node_allocation_statistics.
20847 (free_node): Use get_stats_node_kind.
20849 2018-03-14 Richard Biener <rguenther@suse.de>
20851 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
20852 that the value-set of ANTIC_IN doesn't grow.
20855 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
20857 (BB_VISITED_WITH_VISITED_SUCCS): New define.
20858 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
20860 2018-03-14 Julia Koval <julia.koval@intel.com>
20862 * config.gcc (icelake-client, icelake-server): New.
20864 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
20865 (initial_ix86_arch_features): Ditto.
20866 (PTA_SKYLAKE): Add SGX.
20867 (PTA_ICELAKE): Remove.
20868 (PTA_ICELAKE_CLIENT): New.
20869 (PTA_ICELAKE_SERVER): New.
20870 (ix86_option_override_internal): Split up icelake on icelake client and
20872 (get_builtin_code_for_version): Ditto.
20873 (fold_builtin_cpu): Ditto.
20874 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
20875 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
20876 * config/i386/i386.h (processor_type): Ditto.
20877 * doc/invoke.texi: Ditto.
20879 2018-03-14 Jakub Jelinek <jakub@redhat.com>
20882 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
20883 INTEGER_CST offset, add it together with bitpos / 8 and
20884 sign extend based on POINTER_SIZE.
20888 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20891 * config/i386/constraints.md (Yc): New register constraint.
20892 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
20893 Use Yc constraint for alternative 2 of operand 0. Remove
20894 preferred_for_speed attribute.
20896 2018-03-14 Richard Biener <rguenther@suse.de>
20898 PR tree-optimization/84830
20899 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
20900 with the old one to avoid oscillations.
20902 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
20905 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
20907 (assign_by_spills): Return a flag of reload assignment failure.
20908 Do not process the reload assignment failures. Do not spill other
20909 reload pseudos if they has the same reg class. Update n if
20911 (lra_assign): Add a return arg. Set up from the result of
20912 assign_by_spills call.
20913 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
20914 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
20915 usage_insns if it is not NULL.
20916 (spill_hard_reg_in_range): New function.
20917 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
20918 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
20919 function prototypes.
20920 (lra_assign): Change prototype.
20921 * lra.c (lra): Add code to deal with fails by splitting hard reg
20924 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
20926 * config/riscv/riscv.opt (mrelax): New option.
20927 * config/riscv/riscv.c (riscv_file_start): Emit ".option
20928 "norelax" when riscv_mrelax is disabled.
20929 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
20931 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20934 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
20935 reassociation for int modes.
20937 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
20939 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
20940 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
20942 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
20943 * config/aarch64/aarch64-sve.md
20944 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
20945 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
20946 (*extend<mode><Vwide>2): Rename to...
20947 (aarch64_sve_extend<mode><Vwide>2): ...this.
20948 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
20949 renaming the old pattern to...
20950 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
20952 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
20953 define_expand, renaming the old pattern to...
20954 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
20955 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
20956 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
20957 account when deciding which SVE instruction the optab should use.
20958 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
20960 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
20962 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
20963 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
20964 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
20965 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
20966 (tlsdesc_small_<mode>): Turn a define_expand and use
20967 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
20968 (tlsdesc_small_advsimd_<mode>): ...this.
20969 (tlsdesc_small_sve_<mode>): New pattern.
20971 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
20973 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
20974 (UNSPEC_UMUL_HIGHPART): New constants.
20975 (MUL_HIGHPART): New int iteraor.
20976 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
20977 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
20979 (*<su>mul<mode>3_highpart): New define_insn.
20981 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
20984 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
20987 2018-03-13 Martin Liska <mliska@suse.cz>
20990 * (sem_item_optimizer::sem_item_optimizer): Initialize new
20992 (sem_item_optimizer::~sem_item_optimizer): Release it.
20993 (sem_item_optimizer::merge_classes): Register variable aliases.
20994 (sem_item_optimizer::fixup_pt_set): New function.
20995 (sem_item_optimizer::fixup_points_to_sets): Likewise.
20996 * ipa-icf.h: Declare new variables and functions.
20998 2018-03-13 Jakub Jelinek <jakub@redhat.com>
21000 PR middle-end/84834
21001 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
21002 integer_pow2p@2 and test integer_pow2p in condition.
21003 (A < 0 ? C : 0): Similarly for @1.
21005 PR middle-end/84831
21006 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
21007 characters starting at p contain '\0' character, don't look beyond
21011 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
21012 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
21015 * reg-stack.c (change_stack): Change update_end var from int to
21016 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
21017 also call set_block_for_insn on the newly added insns and rescan.
21020 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
21021 on the last operand.
21024 * tree.c (stabilize_reference_1): Return save_expr (e) for
21025 STATEMENT_LIST even if it doesn't have side-effects.
21027 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
21029 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
21031 2018-03-12 Renlin Li <renlin.li@arm.com>
21033 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
21034 aarch64_output_scalar_simd_mov_immediate.
21036 2018-03-12 Martin Sebor <msebor@redhat.com>
21038 PR tree-optimization/83456
21039 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
21040 for perfectly overlapping calls to memcpy.
21041 (gimple_fold_builtin_memory_chk): Same.
21042 (gimple_fold_builtin_strcpy): Handle no-warning.
21043 (gimple_fold_builtin_stxcpy_chk): Same.
21044 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
21046 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
21048 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
21049 parameter. Use it for SFmode.
21050 (rs6000_function_arg_advance_1): Adjust.
21051 (rs6000_function_arg): Adjust.
21052 (rs6000_gimplify_va_arg): Pass false for that new parameter.
21054 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
21056 PR rtl-optimization/84169
21057 PR rtl-optimization/84780
21058 * combine.c (can_combine_p): Check for a 2-insn combination whether
21059 the destination register is used between the two insns, too.
21061 2018-03-12 Richard Biener <rguenther@suse.de>
21063 PR tree-optimization/84803
21064 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
21065 for refs DR analysis didn't process.
21067 2018-03-12 Richard Biener <rguenther@suse.de>
21069 PR tree-optimization/84777
21070 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
21071 force-vectorize loops ignore whether we are optimizing for size.
21073 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
21075 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
21076 (TARGET_MD_ASM_ADJUST): Define.
21078 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
21079 Kito Cheng <kito.cheng@gmail.com>
21080 Chung-Ju Wu <jasonwucj@gmail.com>
21082 * config/nds32/nds32.c (nds32_compute_stack_frame,
21083 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
21084 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
21085 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
21086 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
21087 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
21088 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
21089 * config/nds32/nds32.md (prologue, epilogue): Use macro
21090 NDS32_V3PUSH_AVAILABLE_P to do checking.
21092 2018-03-11 Jakub Jelinek <jakub@redhat.com>
21095 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
21096 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
21097 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
21098 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
21099 addition of most attributes on !orig_type_die or the attribute not
21100 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
21102 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21103 Chung-Ju Wu <jasonwucj@gmail.com>
21105 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
21106 __NDS32_VH__ macro.
21107 * config/nds32/nds32.opt (mvh): New option.
21109 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21110 Chung-Ju Wu <jasonwucj@gmail.com>
21112 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
21114 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
21115 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
21118 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21119 Chung-Ju Wu <jasonwucj@gmail.com>
21121 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
21123 * config/nds32/nds32-multiple.md (strlensi): New pattern.
21124 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
21126 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
21127 Kito Cheng <kito.cheng@gmail.com>
21128 Chung-Ju Wu <jasonwucj@gmail.com>
21130 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
21131 UNSPEC_FFMISM and UNSPEC_FLMISM.
21132 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
21133 for ffb, ffmism and flmism.
21134 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
21135 (unspec_ffmism): Ditto.
21136 (unspec_flmism): Ditto.
21137 (nds32_expand_builtin_impl): Check if string extension is available.
21138 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
21139 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
21141 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
21144 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
21147 * lra-assigns.c (assign_by_spills): Return a flag of reload
21148 assignment failure. Do not process the reload assignment
21149 failures. Do not spill other reload pseudos if they has the same
21151 (lra_assign): Add a return arg. Set up from the result of
21152 assign_by_spills call.
21153 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
21154 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
21155 usage_insns if it is not NULL.
21156 (spill_hard_reg_in_range): New function.
21157 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
21158 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
21159 function prototypes.
21160 (lra_assign): Change prototype.
21161 * lra.c (lra): Add code to deal with fails by splitting hard reg
21164 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
21167 * config/i386/i386.opt: Replace Enforcment with Enforcement.
21169 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
21172 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
21173 (dw_val_node): Add val_symbolic_view.
21174 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
21175 (symview_upper_bound): New.
21176 (new_line_info_table): Initialize symviews_since_reset.
21177 (dwarf2out_source_line): Count symviews_since_reset and set
21178 symview_upper_bound.
21179 (dw_val_equal_p): Handle symview.
21180 (add_AT_symview): New.
21181 (print_dw_val): Handle symview.
21182 (attr_checksum, attr_checksum_ordered): Likewise.
21183 (same_dw_val_p, size_of_die): Likewise.
21184 (value_format, output_die): Likewise.
21185 (add_high_low_attributes): Use add_AT_symview for entry_view.
21186 (dwarf2out_finish): Reset symview_upper_bound, clear
21189 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
21192 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
21193 Add strict argument and use it.
21194 (rs6000_split_multireg_move): Update for new strict argument.
21195 (mem_operand_gpr): Disallow all non-offsettable addresses.
21196 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
21198 2018-03-09 Jakub Jelinek <jakub@redhat.com>
21201 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
21202 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
21203 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
21206 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
21207 decl, use remap_type if we want to use the type.
21209 2018-03-09 Martin Sebor <msebor@redhat.com>
21211 PR tree-optimization/84526
21212 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
21214 (builtin_access::generic_overlap): Be prepared to handle non-array
21217 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
21219 PR rtl-optimization/84682
21220 * lra-constraints.c (process_address_1): Check is_address flag
21221 for address constraints.
21222 (process_alt_operands): Likewise.
21223 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
21224 preprocess_constraints.
21225 * recog.h (preprocess_constraints): Add oploc parameter.
21227 * recog.c (preprocess_constraints): Test address_operand for
21228 CT_ADDRESS constraints.
21230 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
21233 * lra-assigns.c (assign_by_spills): Return a flag of reload
21234 assignment failure. Do not process the reload assignment
21235 failures. Do not spill other reload pseudos if they has the same
21237 (lra_assign): Add a return arg. Set up from the result of
21238 assign_by_spills call.
21239 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
21240 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
21241 usage_insns if it is not NULL.
21242 (spill_hard_reg_in_range): New function.
21243 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
21244 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
21245 function prototypes.
21246 (lra_assign): Change prototype.
21247 * lra.c (lra): Add code to deal with fails by splitting hard reg
21250 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21253 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
21254 Accept complain bool parameter. Only emit errors if it is true.
21255 (arm_parse_cpu_option_name): Likewise.
21256 (arm_target_thumb_only): Adjust callers of the above.
21257 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
21258 prototype to take a default true bool parameter.
21259 (arm_parse_arch_option_name): Likewise.
21261 2018-03-09 David Malcolm <dmalcolm@redhat.com>
21262 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
21266 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
21267 * configure: Regenerate.
21268 * configure.ac ("linker --version-script option"): New.
21269 ("linker soname option"): New.
21271 2018-03-09 Richard Biener <rguenther@suse.de>
21273 PR tree-optimization/84775
21274 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
21275 immediate uses of predicate stmts and mark them modified.
21278 PR tree-optimization/84178
21279 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
21281 (version_loop_for_if_conversion): Delay update_ssa call.
21282 (tree_if_conversion): Delay update_ssa until after predicate
21285 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
21288 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
21289 when the function accesses prior frames.
21291 2018-03-08 Jakub Jelinek <jakub@redhat.com>
21294 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
21295 gen_llsym, otherwise call maybe_gen_llsym.
21297 PR inline-asm/84742
21298 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
21299 has ',' character inside of it.
21301 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21304 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
21305 as clobbering CC_REGNUM.
21307 2018-03-08 Richard Biener <rguenther@suse.de>
21309 PR middle-end/84552
21310 * tree-scalar-evolution.c: Include tree-into-ssa.h.
21311 (follow_copies_to_constant): Do not follow SSA names registered
21314 2018-03-08 Richard Biener <rguenther@suse.de>
21316 PR tree-optimization/84178
21317 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
21319 (version_loop_for_if_conversion): Delay update_ssa call.
21320 (tree_if_conversion): Delay update_ssa until after predicate
21323 2018-03-08 David Malcolm <dmalcolm@redhat.com>
21325 PR tree-optimization/84178
21326 * tree-if-conv.c (release_bb_predicate): Remove the
21327 the assertion that the stmts have NULL use_ops.
21328 Discard the statements, asserting that they haven't
21329 yet been added to a BB.
21331 2018-03-08 Richard Biener <rguenther@suse.de>
21333 PR tree-optimization/84746
21334 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
21335 (phi_translate): Pass in destination ANTIC_OUT set.
21336 (phi_translate_1): Likewise. For a simplified result lookup
21337 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
21338 (phi_translate_set): Adjust.
21339 (do_pre_regular_insertion): Likewise.
21340 (do_pre_partial_partial_insertion): Likewise.
21342 2018-03-08 Martin Liska <mliska@suse.cz>
21344 PR gcov-profile/84735
21345 * doc/gcov.texi: Document usage of profile files.
21346 * gcov-io.h: Document changes in the format.
21348 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
21352 * dwarf2out.c (struct dw_line_info_table): Update comments for
21354 (FORCE_RESET_NEXT_VIEW): New.
21355 (FORCE_RESETTING_VIEW_P): New.
21356 (RESETTING_VIEW_P): Check for -1 too.
21357 (ZERO_VIEW_P): Likewise.
21358 (new_line_info_table): Force-reset next view.
21359 (dwarf2out_begin_function): Likewise.
21360 (dwarf2out_source_line): Simplify zero_view_p initialization.
21361 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
21362 view directly. Omit view when omitting .loc at line 0.
21364 2018-03-08 Jakub Jelinek <jakub@redhat.com>
21366 PR tree-optimization/84740
21367 * tree-switch-conversion.c (process_switch): Call build_constructors
21368 only if info.phi_count is non-zero.
21370 PR tree-optimization/84739
21371 * tree-tailcall.c (find_tail_calls): Check call arguments against
21372 DECL_ARGUMENTS (current_function_decl) rather than
21373 DECL_ARGUMENTS (func) when checking for tail recursion.
21375 2018-03-07 Jakub Jelinek <jakub@redhat.com>
21377 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
21378 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
21379 Volker Reichelt's entry and add entries for people that perform
21380 GCC fuzzy testing and report numerous bugs.
21382 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21385 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
21386 readonly data in sdata, if that is disabled.
21387 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
21388 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
21389 -mreadonly-in-sdata option.
21391 2018-03-07 Martin Sebor <msebor@redhat.com>
21393 PR tree-optimization/84468
21394 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
21395 basic block when looking for nul assignment.
21397 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
21400 * except.h (output_function_exception_table): Adjust prototype.
21401 * except.c (output_function_exception_table): Remove FNNAME parameter
21402 and add SECTION parameter. Ouput one part of the table at a time.
21403 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
21404 the first part of the exception table and emit unwind directives.
21405 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
21406 (i386_pe_seh_cold_init): Likewise.
21407 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
21408 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
21409 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
21410 (ix86_output_call_insn): Emit a nop in one more case for SEH.
21411 * config/i386/winnt.c: Include except.h.
21412 (struct seh_frame_state): Add reg_offset, after_prologue and
21413 in_cold_section fields.
21414 (i386_pe_seh_end_prologue): Set seh->after_prologue.
21415 (i386_pe_seh_cold_init): New function.
21416 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
21417 to seh->in_cold_section.
21418 (seh_emit_push): Record the offset of the push.
21419 (seh_emit_save): Record the offet of the save.
21420 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
21421 Test seh->after_prologue to disregard the epilogue.
21422 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
21423 (i386_pe_end_cold_function): New function.
21425 2018-03-07 Jakub Jelinek <jakub@redhat.com>
21428 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
21429 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
21432 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
21434 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
21435 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
21437 PR middle-end/84723
21438 * multiple_target.c: Include tree-inline.h and intl.h.
21439 (expand_target_clones): Diagnose and fail if node->definition and
21440 !tree_versionable_function_p (node->decl).
21442 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
21444 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
21446 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
21447 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
21448 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
21450 2018-03-06 Jakub Jelinek <jakub@redhat.com>
21453 * combine.c (try_combine): Use reg_or_subregno instead of handling
21454 just paradoxical SUBREGs and REGs.
21456 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
21458 * config/arc/arc.c (arc_finalize_pic): Remove function.
21459 (arc_must_save_register): We use single base PIC register, remove
21460 checks to save/restore the PIC register.
21461 (arc_expand_prologue): Likewise.
21462 * config/arc/arc-protos.h (arc_set_default_type_attributes):
21464 (arc_verify_short): Likewise.
21465 (arc_attr_type): Likewise.
21466 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
21467 (walk_stores): Likewise.
21468 (arc_address_cost): Make it static.
21469 (arc_verify_short): Likewise.
21470 (branch_dest): Likewise.
21471 (arc_attr_type): Likewise.
21472 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
21473 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
21474 (arc_final_prescan_insn): Remove inserting the nops due to
21475 hardware hazards. It is done in reorg step.
21476 (insn_length_variant_t): Remove.
21477 (insn_length_parameters_t): Likewise.
21478 (arc_insn_length_parameters): Likewise.
21479 (arc_get_insn_variants): Likewise.
21480 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
21482 2018-03-06 Jakub Jelinek <jakub@redhat.com>
21484 PR inline-asm/84683
21485 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
21488 PR tree-optimization/84687
21489 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
21491 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
21493 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21495 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
21496 Rename to ppc_speculation_barrier.
21497 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
21498 __builtin_ppc_speculation_barrier.
21500 2018-03-05 Jakub Jelinek <jakub@redhat.com>
21503 * combine.c (combine_simplify_rtx): Don't try to simplify if
21504 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
21507 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
21509 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
21510 to 32 bytes when compiling for POWER9.
21512 2018-03-05 Jakub Jelinek <jakub@redhat.com>
21515 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
21516 regparm >= 3 with no arg reg available also for calls with
21517 flag_force_indirect_call. Pass decl to ix86_function_regparm.
21520 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
21522 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
21524 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
21527 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
21529 2018-03-05 Richard Biener <rguenther@suse.de>
21531 PR tree-optimization/84486
21532 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
21533 When inserting a __builtin_assume_aligned call set the LHS
21534 SSA name alignment info accordingly.
21536 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
21538 PR tree-optimization/84114
21539 * config/aarch64/aarch64.c (aarch64_reassociation_width)
21540 Avoid reassociation of FLOAT_MODE addition.
21542 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
21544 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
21545 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
21546 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
21547 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
21548 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
21549 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
21550 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
21552 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
21553 __builtin_ia32_wbinvd): New builtins.
21554 (SPECIAL_ARGS2): New.
21555 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
21556 (SPECIAL_ARGS2): New.
21557 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
21558 (ix86_valid_target_attribute_inner_p): Ditto.
21559 (ix86_init_mmx_sse_builtins): Add special_args2.
21560 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
21561 TARGET_WBNOINVD_P): New.
21562 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
21563 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
21564 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
21565 * config/i386/immintrin.h (_wbinvd): New intrinsic.
21566 * config/i386/pconfigintrin.h: New file.
21567 * config/i386/wbnoinvdintrin.h: Ditto.
21568 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
21570 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
21572 2018-03-05 Richard Biener <rguenther@suse.de>
21574 PR tree-optimization/84670
21575 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
21577 (BB_VISITED_WITH_VISITED_SUCCS): New define.
21578 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
21579 (compute_antic_aux): Only assert the number of values in ANTIC_IN
21580 doesn't grow if all successors (recursively) were visited at least
21583 2018-03-05 Richard Biener <rguenther@suse.de>
21585 PR tree-optimization/84650
21586 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
21587 if executed in the loop pipeline.
21589 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
21591 * doc/configfiles.texi (Configuration Files): Move info about
21592 conditionalizing $target-protos.h to...
21593 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
21594 differs from $target-protos.h.
21596 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
21597 Chung-Ju Wu <jasonwucj@gmail.com>
21599 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
21600 * config/nds32/nds32-multiple.md (setmemsi): Define.
21601 * config/nds32/nds32-memory-manipulation.c
21602 (nds32_gen_dup_4_byte_to_word_value): New.
21603 (emit_setmem_word_loop): New.
21604 (emit_setmem_byte_loop): New.
21605 (nds32_expand_setmem_loop): New.
21606 (nds32_expand_setmem_loop_v3m): New.
21607 (nds32_expand_setmem_unroll): New.
21608 (nds32_expand_setmem): New.
21610 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
21611 Chung-Ju Wu <jasonwucj@gmail.com>
21613 * config/nds32/nds32-memory-manipulation.c
21614 (nds32_emit_load_store): New.
21615 (nds32_emit_post_inc_load_store): New.
21616 (nds32_emit_mem_move): New.
21617 (nds32_emit_mem_move_block): New.
21618 (nds32_expand_movmemsi_loop_unknown_size): New.
21619 (nds32_expand_movmemsi_loop_known_size): New.
21620 (nds32_expand_movmemsi_loop): New.
21621 (nds32_expand_movmemsi_unroll): New.
21622 (nds32_expand_movmemqi): Rename ...
21623 (nds32_expand_movmemsi): ... to this.
21624 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
21625 (movmemsi): ... to this.
21626 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
21627 (nds32_expand_movmemsi): ... to this.
21629 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
21630 Monk Chiang <sh.chiang04@gmail.com>
21631 Chung-Ju Wu <jasonwucj@gmail.com>
21633 * config/nds32/nds32-protos.h
21634 (nds32_expand_load_multiple): New arguments.
21635 (nds32_expand_store_multiple): Ditto.
21636 (nds32_valid_multiple_load_store): Rename ...
21637 (nds32_valid_multiple_load_store_p): ... to this.
21638 * config/nds32/nds32-memory-manipulation.c
21639 (nds32_expand_load_multiple): Refine implementation.
21640 (nds32_expand_store_multiple): Ditto.
21641 * config/nds32/nds32-multiple.md
21642 (load_multiple): Update nds32_expand_load_multiple interface.
21643 (store_multiple): Update nds32_expand_store_multiple interface.
21644 * config/nds32/nds32-predicates.c
21645 (nds32_valid_multiple_load_store): Rename ...
21646 (nds32_valid_multiple_load_store_p): ... to this and refine
21648 * config/nds32/predicates.md
21649 (nds32_load_multiple_and_update_address_operation): New predicate.
21650 (nds32_store_multiple_and_update_address_operation): New predicate.
21652 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
21653 Chung-Ju Wu <jasonwucj@gmail.com>
21655 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
21656 (combo): New attribute.
21657 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
21659 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
21661 * config/nds32/nds32.opt: Change -mcmodel= default value.
21663 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
21664 Monk Chiang <sh.chiang04@gmail.com>
21665 Chung-Ju Wu <jasonwucj@gmail.com>
21667 * config/nds32/constants.md (unspec_element): New enum.
21668 * config/nds32/constraints.md (Umw): New constraint.
21669 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
21670 * config/nds32/nds32-intrinsic.md: Likewise.
21671 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
21672 (nds32_valid_smw_lwm_base_p): New.
21673 (nds32_output_smw_single_word): New.
21674 (nds32_output_lmw_single_word): New.
21675 (nds32_expand_unaligned_load): New.
21676 (nds32_expand_unaligned_store): New.
21677 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
21678 (nds32_output_smw_single_word): Declare.
21679 (nds32_output_lmw_single_word): Declare.
21680 (nds32_expand_unaligned_load): Declare.
21681 (nds32_expand_unaligned_store): Declare.
21682 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
21683 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
21684 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
21685 NDS32_BUILTIN_UASTORE_DW.
21686 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
21689 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
21690 Kito Cheng <kito.cheng@gmail.com>
21691 Chung-Ju Wu <jasonwucj@gmail.com>
21693 * config/nds32/nds32-intrinsic.c
21694 (nds32_expand_builtin_null_ftype_reg): Delete.
21695 (nds32_expand_builtin_reg_ftype_imm): Ditto.
21696 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
21697 (nds32_read_argument): New.
21698 (nds32_legitimize_target): Ditto.
21699 (nds32_legitimize_argument): Ditto.
21700 (nds32_check_constant_argument): Ditto.
21701 (nds32_expand_unop_builtin): Ditto.
21702 (nds32_expand_unopimm_builtin): Ditto.
21703 (nds32_expand_binop_builtin): Ditto.
21704 (nds32_builtin_decl_impl): Ditto.
21705 (builtin_description): Ditto.
21706 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
21707 (nds32_init_builtins_impl): Ditto.
21708 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
21709 (nds32_builtin_decl): New.
21710 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
21711 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
21713 2018-03-02 Jeff Law <law@redhat.com>
21715 * reorg.c (stop_search_p): Handle DEBUG_INSN.
21716 (redundant_insn, fill_simple_delay_slots): Likewise.
21717 (fill_slots_from_thread): Likewise.
21718 * resource.c (mark_referenced_resources): Likewise.
21719 (mark_set_resources, find_dead_or_set_registers): Likewise.
21721 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21723 * substring-locations.h (format_warning_va): Formatting fix for
21724 ATTRIBUTE_GCC_DIAG.
21725 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
21727 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
21728 * substring-locations.c: Include intl.h.
21729 (format_warning_va): Turned into small wrapper around
21730 format_warning_n_va, renamed to ...
21731 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
21732 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
21734 (format_warning_at_substring_n): New function.
21735 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
21736 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
21737 format_warning_at_substring with just a shorter name instead of
21738 const function pointer.
21739 (fmtwarn_n): New function.
21740 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
21741 appropriate, get rid of all the fmtstr temporaries, move conditionals
21742 with G_() wrapped string literals directly into fmtwarn arguments,
21743 cast dir.len to (int), formatting fixes.
21745 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
21747 * doc/invoke.texi: Remove "Cilk Plus" references.
21749 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21750 Richard Biener <rguenther@suse.de>
21753 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
21754 for error or warning attributes if CALL_FROM_THUNK_P is set.
21757 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21760 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
21761 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
21764 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
21765 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
21767 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
21768 instead of -1U in last predictors element's probability member.
21770 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
21773 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
21774 arguments if they are comparable.
21776 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
21778 PR tree-optimization/84634
21779 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
21780 masks and masked_loop_p with a single loop_masks, making sure it's
21781 null for bb vectorization.
21783 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
21785 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
21786 (vect_analyze_data_ref_access): Use loop->safe_len rather than
21787 loop->force_vectorize to check whether there is no alias.
21789 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21792 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
21794 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
21796 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
21797 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
21798 instead of a loop around prev_real_insn.
21799 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
21802 PR inline-asm/84625
21803 * config/i386/i386.c (ix86_print_operand): Use conditional
21804 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
21807 2018-03-02 Richard Biener <rguenther@suse.de>
21809 PR tree-optimization/84427
21810 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
21811 (bitmap_set_subtract_values): Rewrite to handle multiple
21814 (prune_clobbered_mems): Likewise.
21815 (phi_translate): Take edge instead of pred/phiblock.
21816 (phi_translate_1): Likewise.
21817 (phi_translate_set): Likewise. Insert all translated
21818 exprs for a value into the set, keeping possibly multiple
21819 expressions per value.
21820 (compute_antic_aux): Adjust for phi_translate changes.
21821 When intersecting union the expressions and prune those
21822 not in the final value set, keeping possibly multiple
21823 expressions per value. Do not use value-insertion
21824 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
21825 all expressions. Add verification that the value-sets
21826 only shrink during iteration.
21827 (compute_partial_antic_aux): Adjust for the phi_translate changes.
21828 (do_pre_regular_insertion): Likewise.
21829 (do_pre_partial_partial_insertion): Likewise.
21831 2018-03-02 Richard Biener <rguenther@suse.de>
21834 * config/darwin.c (saved_debug_info_level): New static global.
21835 (darwin_asm_lto_start): Disable debug info generation for LTO out.
21836 (darwin_asm_lto_end): Restore debug info generation settings.
21838 2018-03-01 Martin Liska <mliska@suse.cz>
21841 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
21842 volatile arguments.
21844 2018-03-01 Richard Biener <rguenther@suse.de>
21847 * dwarf2out.c (gen_variable_die): Properly handle late VLA
21848 type annotation with LTO when debug was disabled at compile-time.
21850 2018-03-01 Matthew Fortune <mfortune@gmail.com>
21852 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
21854 (mips_final_postscan_insn): Likewise.
21856 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
21858 PR rtl-optimization/84528
21859 * alias.c (init_alias_target): Add commentary.
21860 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
21861 a unique base value if the frame pointer is not eliminated
21862 to the stack pointer.
21864 2018-03-01 Tom de Vries <tom@codesourcery.com>
21866 PR rtl-optimization/83327
21867 * lra-int.h (hard_regs_spilled_into): Declare.
21868 * lra.c (hard_regs_spilled_into): Define.
21869 (init_reg_info): Init hard_regs_spilled_into.
21870 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
21871 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
21872 (process_bb_lives): Handle hard_regs_spilled_into.
21873 (lra_create_live_ranges_1): Before doing liveness propagation, clear
21874 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
21876 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
21878 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
21879 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
21880 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
21881 * config/rs6000/aix72.h: New file.
21883 2018-02-28 Jakub Jelinek <jakub@redhat.com>
21885 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
21886 instead of warning_at with conditional singular and plural messages
21890 * stor-layout.c (update_alignment_for_field): For
21891 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
21892 && !DECL_PACKED (field), do the alignment update, just use
21893 only desired_align instead of MAX (type_align, desired_align)
21895 (place_field): Don't do known_align < desired_align handling
21896 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
21897 is non-NULL, instead do it after rli->prev_field handling and
21898 only if not within a bitfield word. For DECL_PACKED (field)
21899 use type_align of BITS_PER_UNIT.
21901 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
21903 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
21904 superfluous parentheses and trailing spaces.
21906 2018-02-28 Richard Biener <rguenther@suse.de>
21908 PR tree-optimization/84584
21909 * graphite-scop-detection.c (scop_detection::add_scop): Discard
21910 SCoPs with fake exit edge.
21912 2018-02-28 Martin Liska <mliska@suse.cz>
21915 * timevar.c (timer::print): Fix format to properly print 100%
21918 2018-02-28 Richard Biener <rguenther@suse.de>
21920 PR middle-end/84607
21921 * genmatch.c (capture_info::walk_match): Do not mark
21922 captured expressions without operands as expr_p given
21923 they act more like predicates and should be subject to
21924 "lost tail" side-effect preserving.
21926 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
21928 PR rtl-optimization/81611
21929 * auto-inc-dec.c (attempt_change): Move dead note from
21930 mem_insn if it's the next use of regno
21931 (find_address): Take address use of reg holding
21932 non-incremented value. Add parm to limit search to the named
21934 (merge_in_block): Attempt to use a mem insn that is the next
21935 use of the original regno.
21937 2018-02-27 Martin Sebor <msebor@redhat.com>
21940 * doc/invoke.texi (-Wmissing-attributes): New option.
21941 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
21943 2018-02-27 Martin Sebor <msebor@redhat.com>
21945 PR translation/84207
21946 * diagnostic-core.h (warning_n, error_n, inform_n): Change
21947 n argument to unsigned HOST_WIDE_INT.
21948 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
21949 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
21950 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
21951 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
21953 2018-02-27 Richard Biener <rguenther@suse.de>
21955 PR tree-optimization/84512
21956 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
21957 Do not use the estimate returned from record_stmt_cost for
21958 the scalar iteration cost but sum properly using add_stmt_cost.
21960 2018-02-27 Richard Biener <rguenther@suse.de>
21962 PR tree-optimization/84466
21963 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
21964 Adjust last change to less strictly validate use operands.
21966 2018-02-27 Martin Liska <mliska@suse.cz>
21968 PR gcov-profile/84548
21969 * gcov.c (process_file): Allow partial overlap and consider it
21970 also as group functions.
21971 (output_lines): Properly calculate range of lines for a group.
21973 2018-02-27 Martin Liska <mliska@suse.cz>
21975 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
21976 'ggc' suffixes. Change first column width.
21977 (timer::print): Fix formatting of the column.
21979 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
21981 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
21982 preserve inline entry blocks for the sake of debug inline
21983 entry point markers alone.
21984 (remove_unused_locals): Suggest in comments a better place to
21985 force the preservation of inline entry blocks that are
21986 otherwise unused, but do not preserve them.
21988 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
21990 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
21992 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
21995 * config/i386/constraints.md (Bs): Replace
21996 ix86_indirect_branch_register with
21997 TARGET_INDIRECT_BRANCH_REGISTER.
21999 * config/i386/i386.md (indirect_jump): Likewise.
22000 (tablejump): Likewise.
22001 (*sibcall_memory): Likewise.
22002 (*sibcall_value_memory): Likewise.
22003 Peepholes of indirect call and jump via memory: Likewise.
22004 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
22005 (*sibcall_value_GOT_32): Likewise.
22006 * config/i386/predicates.md (indirect_branch_operand): Likewise.
22007 (GOT_memory_operand): Likewise.
22008 (call_insn_operand): Likewise.
22009 (sibcall_insn_operand): Likewise.
22010 (GOT32_symbol_operand): Likewise.
22011 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
22013 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
22015 PR rtl-optimization/83496
22016 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
22017 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
22018 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
22019 redundant insn, if any.
22020 (relax_delay_slots): Likewise.
22021 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
22023 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
22025 PR tree-optimization/83965
22026 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
22027 that grouped statements are part of a reduction chain. Return
22028 true if the statement is not marked as a reduction itself but
22029 is part of a group.
22030 (vect_recog_dot_prod_pattern): Don't check whether the statement
22031 is part of a group here.
22032 (vect_recog_sad_pattern): Likewise.
22033 (vect_recog_widen_sum_pattern): Likewise.
22035 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
22038 * final.c (rest_of_clean_state): Also look for calls inside sequences.
22040 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
22043 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
22045 (ix86_output_indirect_function_return): New prototype.
22046 (ix86_split_simple_return_pop_internal): Likewise.
22047 * config/i386/i386.c (indirect_return_via_cx): New.
22048 (indirect_return_via_cx_bnd): Likewise.
22049 (indirect_thunk_name): Handle return va CX_REG.
22050 (output_indirect_thunk_function): Create alias for
22051 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
22052 (ix86_output_indirect_jmp): Remove the bool argument.
22053 (ix86_output_indirect_function_return): New function.
22054 (ix86_split_simple_return_pop_internal): Likewise.
22055 * config/i386/i386.md (*indirect_jump): Don't pass false
22056 to ix86_output_indirect_jmp.
22057 (*tablejump_1): Likewise.
22058 (simple_return_pop_internal): Change it to define_insn_and_split.
22059 Call ix86_split_simple_return_pop_internal to split it for
22060 -mfunction-return=.
22061 (simple_return_indirect_internal): Call
22062 ix86_output_indirect_function_return instead of
22063 ix86_output_indirect_jmp.
22065 2018-02-26 Jakub Jelinek <jakub@redhat.com>
22068 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
22069 memset and value initialization afterwards.
22071 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
22073 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
22075 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22078 * common/config/aarch64/aarch64-common.c
22079 (aarch_option_optimization_table[]): Switch
22080 off fomit-frame-pointer
22082 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
22083 Chung-Ju Wu <jasonwucj@gmail.com>
22085 * config/nds32/nds32-multiple.md (load_multiple): Disallow
22087 (store_multiple): Ditto.
22089 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
22091 * config.gcc: Add --with-cpu support for nds32 target.
22092 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
22093 * config/nds32/nds32.opt: Add -mcpu= option.
22095 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
22097 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
22098 isel=yes): Warn for these deprecated options.
22100 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
22102 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
22103 ISA_2_5_MASKS_EMBEDDED.
22105 2018-02-23 Jakub Jelinek <jakub@redhat.com>
22107 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
22108 p->max as pointers rather than using iterative_hash_expr.
22110 2018-02-23 Carl Love <cel@us.ibm.com>
22112 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
22113 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
22115 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
22116 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
22117 P8V_BUILTIN_VEC_VUNSIGNED2.
22119 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
22122 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
22123 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
22125 * lra-constraints.c (curr_insn_transform): Set up
22126 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
22128 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
22129 * lra-eliminations.c (spill_pseudos): Ditto.
22130 (process_insn_for_elimination): Ditto.
22131 * lra-lives.c (reg_early_clobber_p): Use the new macros.
22132 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
22133 LRA_NON_CLOBBERED_ALT.
22135 2018-02-22 Martin Sebor <msebor@redhat.com>
22137 PR tree-optimization/84480
22138 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
22139 to maybe_diag_stxncpy_trunc. Call it.
22140 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
22141 from gimple_fold_builtin_strcpy. Print inlining stack.
22142 (handle_builtin_stxncpy): Print inlining stack.
22143 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
22145 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
22148 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
22149 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
22150 and -fcheck-pointer-bounds are used together.
22151 (indirect_thunk_prefix): New enum.
22152 (indirect_thunk_need_prefix): New function.
22153 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
22154 "_nt" instead of "_bnd" for NOTRACK prefix.
22155 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
22156 (output_indirect_thunk_function): Likewise.
22158 (ix86_code_end): Update output_indirect_thunk_function calls.
22159 (ix86_output_indirect_branch_via_reg): Replace
22160 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
22161 (ix86_output_indirect_branch_via_push): Likewise.
22162 (ix86_output_function_return): Likewise.
22163 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
22164 incompatible with -fcf-protection=branch and
22165 -fcheck-pointer-bounds.
22167 2018-02-22 Steve Ellcey <sellcey@cavium.com>
22170 * config/aarch64/aarch64.c (aarch64_print_address_internal):
22171 Change gcc_assert call to output_operand_lossage.
22173 2018-02-22 Steve Ellcey <sellcey@cavium.com>
22175 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
22177 2018-02-22 DJ Delorie <dj@redhat.com>
22178 Sebastian Perta <sebastian.perta@renesas.com>
22179 Oleg Endo <olegendo@gcc.gnu.org>
22181 * config/rx/rx.c (rx_rtx_costs): New function.
22182 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
22184 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
22186 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
22188 2018-02-22 Martin Liska <mliska@suse.cz>
22191 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
22192 Add "native" as a possible value.
22194 2018-02-22 Martin Liska <mliska@suse.cz>
22197 * config/i386/i386.c (ix86_option_override_internal):
22198 Add "native" as a possible value for -march and -mtune.
22200 2018-02-22 Jakub Jelinek <jakub@redhat.com>
22203 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
22204 to all type variants.
22206 PR tree-optimization/84503
22207 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
22208 width as info->bitpos + info->bitsize - start.
22209 (merged_store_group::merge_overlapping): Simplify width computation.
22210 (check_no_overlap): New function.
22211 (imm_store_chain_info::try_coalesce_bswap): Compute expected
22212 start + width and last_order of the group, fail if check_no_overlap
22214 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
22215 to group if check_no_overlap fails.
22217 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
22219 * config/rs6000/altivec.md: Delete contraint arguments to
22220 define_expand, define_split, and define_peephole2, and in
22221 define_insn_and_split if always unused.
22222 * config/rs6000/darwin.md: Ditto.
22223 * config/rs6000/dfp.md: Ditto.
22224 * config/rs6000/rs6000.md: Ditto.
22225 * config/rs6000/sync.md: Ditto.
22226 * config/rs6000/vector.md: Ditto.
22227 * config/rs6000/vsx.md: Ditto.
22229 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
22231 * config/rs6000/altivec.md: Write output control strings as braced
22232 blocks instead of double-quoted strings.
22233 * config/rs6000/darwin.md: Ditto.
22234 * config/rs6000/rs6000.md: Ditto.
22235 * config/rs6000/vector.md: Ditto.
22236 * config/rs6000/vsx.md: Ditto.
22238 2018-02-21 Jason Merrill <jason@redhat.com>
22240 PR c++/84314 - ICE with templates and fastcall attribute.
22241 * attribs.c (build_type_attribute_qual_variant): Remove assert.
22243 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
22245 * ipa-cp.c (determine_versionability): Fix comment typos.
22247 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
22250 * ipa-cp.c (determine_versionability): Do not version functions caling
22253 2018-02-21 Martin Liska <mliska@suse.cz>
22256 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
22257 Add "native" as a possible value.
22258 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
22259 the macro when native cpu detection is available.
22261 2018-02-21 Martin Liska <mliska@suse.cz>
22264 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
22265 Add "native" as a possible value.
22266 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
22267 when native cpu detection is available.
22269 2018-02-21 Jakub Jelinek <jakub@redhat.com>
22270 Martin Sebor <msebor@redhat.com>
22272 PR tree-optimization/84478
22273 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
22275 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
22276 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
22277 support which is conservatively correct, for 2 only stay conservative
22278 for maxlen. Formatting and comment capitalization fixes. Add STRICT
22279 argument to the 2 argument get_range_strlen, adjust 6 arg
22280 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
22282 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
22283 (gimple_fold_builtin_strlen): Pass true as last argument to
22286 2018-02-20 Martin Sebor <msebor@redhat.com>
22288 PR middle-end/84095
22289 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
22290 (builtin_memref::set_base_and_offset): Same. Handle inner references.
22291 (builtin_memref::builtin_memref): Factor out parts into
22292 set_base_and_offset and call it.
22294 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
22296 PR middle-end/84406
22297 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
22298 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
22299 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
22300 search at the associated MODE_INT.
22302 2018-02-20 Jeff Law <law@redhat.com>
22304 PR middle-end/82123
22305 PR tree-optimization/81592
22306 PR middle-end/79257
22307 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
22308 for range data rather than using global data.
22309 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
22310 range data rather than using global data.
22311 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
22312 pass it to children as needed.
22313 (struct directive::fmtresult): Similarly.
22314 (struct directive::set_width): Similarly.
22315 (struct directive::set_precision): Similarly.
22316 (format_integer, format_directive, parse_directive): Similarly.
22317 (format_none): Accept unnamed vr_values parameter.
22318 (format_percent, format_floating, format_character): Similarly.
22319 (format_string, format_plain): Similarly.
22320 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
22321 the EVRP range analyzer for range data rather than using global data.
22322 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
22323 gimple-ssa-evrp-analyze.h
22324 (class sprintf_dom_walker): Add after_dom_children member function.
22325 Add evrp_range_analyzer member.
22326 (sprintf_dom_walker::before_dom_children): Call into the EVRP
22327 range analyzer as needed.
22328 (sprintf_dom_walker::after_dom_children): New member function.
22329 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
22331 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
22332 (evrp_range_analyzer::pop_to_marker): Likewise.
22334 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
22336 PR tree-optimization/84419
22337 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
22338 with the required type if its current type is compatible but
22341 2018-02-20 Jakub Jelinek <jakub@redhat.com>
22343 PR middle-end/82004
22344 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
22345 after vectorization.
22347 2018-02-20 Martin Liska <mliska@suse.cz>
22350 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
22351 possible values if we don't have a hint.
22353 2018-02-20 Martin Liska <mliska@suse.cz>
22357 * final.c (shorten_branches): Build align_tab array with one
22359 * opts.c (finish_options): Add alignment option limit check.
22360 (MAX_CODE_ALIGN): Likewise.
22361 (MAX_CODE_ALIGN_VALUE): Likewise.
22362 * doc/invoke.texi: Document maximum allowed option value for
22363 all -falign-* options.
22365 2018-02-19 Jakub Jelinek <jakub@redhat.com>
22368 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
22369 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
22370 * var-tracking.c (emit_note_insn_var_location): Remove all references
22371 to NOTE_INSN_CALL_ARG_LOCATION.
22372 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
22373 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
22374 Use copy_rtx_if_shared.
22375 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
22376 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
22377 (dwarf2out_var_location): Remove handling of
22378 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
22380 * final.c (final_scan_insn): Remove all references to
22381 NOTE_INSN_CALL_ARG_LOCATION.
22382 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
22383 before dumping final insns.
22384 * except.c (emit_note_eh_region_end): Remove all references to
22385 NOTE_INSN_CALL_ARG_LOCATION.
22386 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
22387 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
22388 * config/arc/arc.c (hwloop_optimize): Likewise.
22389 * config/arm/arm.c (create_fix_barrier): Likewise.
22390 * config/s390/s390.c (s390_chunkify_start): Likewise.
22391 * config/sh/sh.c (find_barrier): Likewise.
22392 * config/i386/i386.c (rest_of_insert_endbranch,
22393 ix86_seh_fixup_eh_fallthru): Likewise.
22394 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
22395 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
22396 * config/frv/frv.c (frv_function_prologue): Likewise.
22397 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
22399 (note_outside_basic_block_p): Remove all references to
22400 NOTE_INSN_CALL_ARG_LOCATION.
22401 * gengtype.c (adjust_field_rtx_def): Likewise.
22402 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
22404 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
22405 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
22408 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
22411 PR tree-optimization/84452
22412 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
22413 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
22416 2018-02-19 Martin Liska <mliska@suse.cz>
22419 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
22421 2018-02-19 Martin Liska <mliska@suse.cz>
22422 Richard Sandiford <richard.sandiford@linaro.org>
22424 PR tree-optimization/82491
22425 * gimple-fold.c (get_base_constructor): Make earlier bail out
22428 2018-02-19 Carl Love <cel@us.ibm.com>
22430 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
22431 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
22433 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
22434 P8V_BUILTIN_VEC_NEG.
22436 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
22438 * config/rl78/rl78.md (movdf): New define expand.
22440 2018-02-19 Martin Liska <mliska@suse.cz>
22443 * doc/invoke.texi: Fix typo.
22444 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
22446 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
22448 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
22449 handle rs6000_single_float and rs6000_double_float specially for
22452 2018-02-16 Jeff Law <law@redhat.com>
22454 * config/rx/rx.c (add_pop_cfi_notes): New function.;
22455 (pop_regs): Use it.
22457 2018-02-16 Jakub Jelinek <jakub@redhat.com>
22460 * ipa-inline.c (inline_small_functions): Fix a typo.
22462 2018-02-16 Nathan Sidwell <nathan@acm.org>
22464 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
22466 2018-02-16 Carl Love <cel@us.ibm.com>
22468 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
22469 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
22470 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
22471 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
22472 expansion to P8V_BUILTIN_VEC_FLOAT2.
22474 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
22476 PR rtl-optimization/70023
22477 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
22478 src_regno into account.
22480 2018-02-16 Carl Love <cel@us.ibm.com>
22482 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
22483 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
22484 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
22485 * config/rs6000/rs6000.c: Remove case statements for
22486 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
22487 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
22488 and P9V_BUILTIN_VEC_VINSERT4B.
22489 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
22490 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
22491 * config/rs6000/vsx.md:
22492 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
22495 2018-02-16 Carl Love <cel@us.ibm.com>
22497 * config/rs6000/altivec.h: Add builtin names vec_extract4b
22499 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
22501 * config/rs6000/rs6000-c.c: Add the definitions for
22502 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
22503 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
22504 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
22505 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
22506 definition for insert4b and define insn *insert3b_internal.
22507 * doc/extend.texi: Add documentation for vec_extract4b.
22509 2018-02-16 Nathan Sidwell <nathan@acm.org>
22511 * doc/extend.texi (Backwards Compatibility): Mention friend
22512 injection. Note for-scope is deprecated.
22513 * doc/invoke.texi (-ffriend-injection): Deprecate.
22515 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
22517 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
22518 that moved to I2, also allow destinations that are a paradoxical
22519 subreg (instead of a normal reg).
22521 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
22524 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
22527 2018-02-16 Richard Biener <rguenther@suse.de>
22529 PR tree-optimization/84037
22530 PR tree-optimization/84016
22532 * config/i386/i386.c (ix86_builtin_vectorization_cost):
22533 Adjust vec_construct for the fact we need additional higher latency
22534 128bit inserts for AVX256 and AVX512 vector builds.
22535 (ix86_add_stmt_cost): Scale vector construction cost for
22538 2018-02-16 Richard Biener <rguenther@suse.de>
22540 PR tree-optimization/84417
22541 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
22542 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
22543 (non_rewritable_lvalue_p): Likewise, use poly-ints.
22545 2018-02-16 Martin Liska <mliska@suse.cz>
22548 * internal-fn.def (ASAN_CHECK): Set proper flags.
22549 (ASAN_MARK): Likewise.
22551 2018-02-16 Julia Koval <julia.koval@intel.com>
22553 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
22554 from PTA_CANNONLAKE.
22556 2018-02-16 Jakub Jelinek <jakub@redhat.com>
22559 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
22560 Use ++iter rather than iter++ for std::list iterators.
22561 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
22562 defer deleting them until all nodes in the forest are processed. Do
22563 free even leaf nodes. Change to_process into auto_vec.
22566 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
22567 * vec.h (vec_default_construct): Use memset instead of placement new
22568 if BROKEN_VALUE_INITIALIZATION is defined.
22569 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
22570 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
22573 PR rtl-optimization/83723
22574 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
22575 * lra.c (lra_substitute_pseudo): Likewise. If true, use
22576 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
22578 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
22580 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
22582 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
22584 PR rtl-optimization/81443
22585 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
22586 from inner REGs to paradoxical SUBREGs.
22588 2018-02-16 Richard Biener <rguenther@suse.de>
22590 PR tree-optimization/84399
22591 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
22592 For operands we can analyze at their definition make sure we can
22593 analyze them at each use as well.
22595 2018-02-16 Richard Biener <rguenther@suse.de>
22597 PR tree-optimization/84190
22598 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
22599 volatile accesses if the decl isn't volatile.
22601 2018-02-15 Jason Merrill <jason@redhat.com>
22603 PR c++/84314 - ICE with templates and fastcall attribute.
22604 * attribs.c (build_type_attribute_qual_variant): Don't clobber
22605 TYPE_CANONICAL on an existing type.
22607 2018-02-15 Jakub Jelinek <jakub@redhat.com>
22609 PR tree-optimization/84383
22610 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
22611 dstoff nor call operand_equal_p if dstbase is NULL.
22613 PR tree-optimization/84334
22614 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
22615 also a CONSTANT_CLASS_P, punt.
22617 2018-02-14 Jim Wilson <jimw@sifive.com>
22619 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
22620 first SMALL_OPERAND check. New local min_second_step. Move assert
22621 to where locals are set. Add TARGET_RVC support.
22622 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
22624 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
22626 * doc/invoke.texi: Correct -Wformat-overflow code sample.
22628 2018-02-14 Martin Sebor <msebor@redhat.com>
22630 PR tree-optimization/83698
22631 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
22632 arrays constrain the offset range to their bounds.
22633 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
22634 (builtin_access::overlap): Avoid setting the size of overlap if it's
22636 (maybe_diag_overlap): Also consider arrays when deciding what values
22637 of offsets to include in diagnostics.
22639 2018-02-14 Martin Sebor <msebor@redhat.com>
22642 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
22643 that correspond to the kind of a declaration.
22645 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
22648 * config/pa/pa.md: Load address of PIC label using the linkage table
22649 if the label is nonlocal.
22651 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
22653 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
22654 warning message if user requests -maltivec=be.
22655 * doc/invoke.texi: Document deprecation of -maltivec=be.
22657 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
22660 * config/rs6000/rs6000-c.c: Update definitions for
22661 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
22662 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
22664 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
22667 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
22668 add _get_ssp intrinsics. Remove argument from
22669 __builtin_ia32_rdssp[d|q].
22670 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
22671 * config/i386/i386-builtin.def: Remove argument from
22672 __builtin_ia32_rdssp[d|q].
22673 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
22674 ix86_expand_special_args_builtin for _rdssp[d|q].
22675 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
22676 Clear register before usage.
22677 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
22678 Add documentation for new _get_ssp and _inc_ssp intrinsics.
22680 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
22682 PR tree-optimization/84357
22683 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
22684 operand 1 of an ARRAY_REF too.
22686 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
22689 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
22690 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
22692 (set_of_reg): New struct.
22693 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
22694 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
22695 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
22697 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
22698 Split into bitclr, bitset, bitinvert patterns if appropriate.
22699 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
22700 use rx_fuse_in_memory_bitop.
22701 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
22702 to named insn, correct maximum insn length.
22704 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
22707 * machmode.def: Define a complex mode for PARTIAL_INT.
22708 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
22710 * doc/rtl.texi: Document CSPImode.
22711 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
22713 (msp430_hard_regno_nregs_with_padding): Likewise.
22715 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
22718 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
22720 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
22722 PR rtl-optimization/84169
22723 * combine.c (try_combine): New variable split_i2i3. Set it to true if
22724 we generated a parallel as new i3 and we split that to new i2 and i3
22725 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
22726 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
22727 those to i2, not i1. Partially rewrite this scan code.
22729 2018-02-13 Jakub Jelinek <jakub@redhat.com>
22732 * stor-layout.c (place_field): For variable length fields, adjust
22733 offset_align afterwards not just based on the field's alignment,
22734 but also on the size.
22736 PR middle-end/84309
22737 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
22738 of exps and logs in the use_exp2 case.
22740 2018-02-13 Jeff Law <law@redhat.com>
22742 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
22743 entry for "vector".
22745 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
22748 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
22752 * common.opt (gas-loc-support, gas-locview-support): New.
22753 (ginline-points, ginternal-reset-location-views): New.
22754 * doc/invoke.texi: Document them. Use @itemx where intended.
22755 (gvariable-location-views): Adjust.
22756 * target.def (reset_location_view): New.
22757 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
22758 (TARGET_RESET_LOCATION_VIEW): New.
22759 * doc/tm.texi: Rebuilt.
22760 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
22761 (dwarf2out_default_as_locview_support): New.
22762 (output_asm_line_debug_info): Use option variables.
22763 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
22764 (output_loc_list): Likewise.
22765 (add_high_low_attributes): Check option variables.
22766 Don't output entry view attribute in strict mode.
22767 (gen_inlined_subroutine_die): Check option variables.
22768 (dwarf2out_inline_entry): Likewise.
22769 (init_sections_and_labels): Likewise.
22770 (dwarf2out_early_finish): Likewise.
22771 (maybe_reset_location_view): New, from...
22772 (dwarf2out_var_location): ... here. Call it.
22773 * debug.h (dwarf2out_default_as_loc_support): Declare.
22774 (dwarf2out_default_as_locview_support): Declare.
22775 * hooks.c (hook_int_rtx_insn_0): New.
22776 * hooks.h (hook_int_rtx_insn_0): Declare.
22777 * toplev.c (process_options): Take -gas-loc-support and
22778 -gas-locview-support from dwarf2out. Enable
22779 -gvariable-location-views by default only with locview
22780 assembler support. Enable -ginternal-reset-location-views by
22781 default only if the target defines the corresponding hook.
22782 Enable -ginline-points by default if location views are
22783 enabled; force it disabled if statement frontiers are
22785 * tree-inline.c (expand_call_inline): Check option variables.
22786 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
22788 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
22790 PR tree-optimization/84321
22791 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
22792 handling. Also check whether the anti-range contains any values
22793 that satisfy the mask; switch to a VR_RANGE if not.
22795 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
22798 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
22800 2018-02-13 Martin Jambor <mjambor@suse.cz>
22803 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
22804 of call statements, also set location of a load to a temporary.
22806 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
22808 * config/rl78/rl78.c (add_vector_labels): New function.
22809 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
22810 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
22811 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
22812 which checks that no arguments are passed.
22813 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
22814 * doc/extend.texi: Documentation for the new attribute.
22816 2018-02-13 Andreas Schwab <schwab@suse.de>
22818 * config/riscv/linux.h (CPP_SPEC): Define.
22820 2018-02-13 Jakub Jelinek <jakub@redhat.com>
22823 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
22824 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
22825 OPTION_MASK_ISA_AES as first argument to def_builtin_const
22826 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
22827 instead of OPTION_MASK_ISA_PCLMUL as first argument to
22828 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
22829 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
22830 temporarily for AES and PCLMUL builtins.
22832 PR tree-optimization/84339
22833 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
22834 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
22837 PR middle-end/84309
22838 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
22839 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
22840 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
22842 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
22844 * omp-simd-clone.h: New file.
22845 * omp-simd-clone.c: Include omp-simd-clone.h.
22846 (expand_simd_clones): No longer static.
22847 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
22848 cgraph.h and omp-simd-clone.h.
22849 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
22850 (vect_recog_widen_shift_pattern): Formatting fix.
22851 (vect_pattern_recog_1): Don't check optab for calls.
22854 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
22855 operands[2] into a REG before using gen_lowpart on it.
22857 2018-02-12 Jeff Law <law@redhat.com>
22860 * config/sh/sh.c (find_barrier): Consider a sibling call
22863 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
22864 successfully back substituting a reg.
22866 2018-02-12 Richard Biener <rguenther@suse.de>
22868 PR tree-optimization/84037
22869 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
22870 parameter, move visited init to caller.
22871 (vect_slp_analyze_operations): Separate cost from validity
22872 check, initialize visited once for all instances.
22873 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
22875 * tree-vect-stmts.c (vect_model_simple_cost): Make early
22877 (vect_model_promotion_demotion_cost): Likewise.
22878 (vectorizable_bswap): Guard cost modeling with !slp_node
22879 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
22881 (vectorizable_call): Likewise.
22882 (vectorizable_conversion): Likewise.
22883 (vectorizable_assignment): Likewise.
22884 (vectorizable_shift): Likewise.
22885 (vectorizable_operation): Likewise.
22886 (vectorizable_store): Likewise.
22887 (vectorizable_load): Likewise.
22888 (vectorizable_condition): Likewise.
22889 (vectorizable_comparison): Likewise.
22891 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
22894 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
22895 (ASAN_MARK): Fix fnspec to account for return value, change pointer
22896 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
22898 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
22900 PR middle-end/83665
22901 * params.def (inline-min-speedup): Increase from 8 to 15.
22902 (max-inline-insns-auto): Decrease from 40 to 30.
22903 * ipa-split.c (consider_split): Add some buffer for function to
22904 be considered inlining candidate.
22905 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
22908 2018-02-12 Richard Biener <rguenther@suse.de>
22910 PR tree-optimization/84037
22911 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
22912 matched stmts if we cannot swap the non-matched ones.
22914 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
22916 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
22917 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
22918 _mm_maskz_scalef_round_ss): New intrinsics.
22919 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
22920 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
22921 __builtin_ia32_scalefss_round): Remove.
22922 (__builtin_ia32_scalefsd_mask_round,
22923 __builtin_ia32_scalefss_mask_round): New intrinsics.
22924 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
22925 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
22926 ((match_operand:VF_128 2 "<round_nimm_predicate>"
22927 "<round_constraint>")): Changed to ...
22928 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
22929 "<round_scalar_constraint>")): ... this.
22930 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
22931 %0, %1, %2<round_op3>}"): Changed to ...
22932 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
22933 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
22934 %2<round_scalar_mask_op3>}"): ... this.
22935 * config/i386/subst.md (round_scalar_nimm_predicate): New.
22937 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
22939 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
22940 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
22941 (_mm_maskz_sqrt_round_ss): New intrinsics.
22942 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
22943 (__builtin_ia32_sqrtsd_mask_round)
22944 (__builtin_ia32_sqrtss_mask_round): New builtins.
22945 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
22946 (__builtin_ia32_sqrtss_round): Remove.
22947 (__builtin_ia32_sqrtsd_mask_round)
22948 (__builtin_ia32_sqrtss_mask_round): New builtins.
22949 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
22950 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
22951 ((match_operand:VF_128 1 "vector_operand"
22952 "xBm,<round_constraint>")): Changed to ...
22953 ((match_operand:VF_128 1 "vector_operand"
22954 "xBm,<round_scalar_constraint>")): ... this.
22955 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
22956 %0, %2, %<iptr>1<round_op3>}): Changed to ...
22957 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
22958 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
22959 %<iptr>1<round_scalar_mask_op3>}): ... this.
22960 ((set_attr "prefix" "<round_prefix>")): Changed to ...
22961 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
22963 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
22966 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
22967 Cast vec_cmpeq result to correct type.
22968 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
22969 Cast vec_cmpgt result to correct type.
22971 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
22973 * final.c (final_scan_insn_1): Renamed from...
22974 (final_scan_insn): ... this. New wrapper, to recover
22975 seen from the outermost call in recursive ones.
22976 * config/sparc/sparc.c (output_return): Drop seen from call.
22977 (output_sibcall): Likewise.
22978 * config/visium/visium.c (output_branch): Likewise.
22980 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
22982 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
22985 2018-02-10 Alan Modra <amodra@gmail.com>
22988 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
22989 Specify LR as an input.
22991 2018-02-10 Jakub Jelinek <jakub@redhat.com>
22994 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
22995 remove_member_access_dummy_vars): New functions.
22996 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
22997 lower_omp_1, execute_lower_omp): Use them.
22999 PR rtl-optimization/84308
23000 * shrink-wrap.c (spread_components): Release todo vector.
23002 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
23004 PR rtl-optimization/57193
23005 * ira-color.c (struct allocno_color_data): Add member
23006 conflict_allocno_hard_prefs.
23007 (update_conflict_allocno_hard_prefs): New.
23008 (bucket_allocno_compare_func): Add a preference based on
23009 conflict_allocno_hard_prefs.
23010 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
23011 (color_allocnos): Remove a dead code. Initiate
23012 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
23014 2018-02-09 Jakub Jelinek <jakub@redhat.com>
23017 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
23018 constraint from =wa to wa. Avoid a subreg on the output operand,
23019 instead use a pseudo and subreg it in a move.
23020 (p9_xxbrd_<mode>): Changed to ...
23021 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
23022 (p9_xxbrd_v2df): New expander.
23023 (p9_xxbrw_<mode>): Changed to ...
23024 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
23025 (p9_xxbrw_v4sf): New expander.
23027 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
23029 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
23031 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
23034 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
23035 multiply in 32-bit mode.
23036 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
23037 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
23040 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
23042 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
23043 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
23044 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
23045 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
23047 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
23050 * dwarf2out.c (is_trivial_indirect_ref): New function.
23051 (dwarf2out_late_global_decl): Do not generate a location
23052 attribute for variables that have a non-trivial DECL_VALUE_EXPR
23053 and that are not defined in the current unit.
23055 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
23057 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
23058 instead of a libcall for UNORDERED.
23060 2018-02-09 Tamar Christina <tamar.christina@arm.com>
23063 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
23064 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
23066 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23069 * config/s390/s390.c (s390_set_current_function): Invoke
23070 s390_indirect_branch_settings also if fndecl didn't change.
23072 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
23074 * config/rs6000/rs6000.md (blockage): Set length to zero.
23076 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
23078 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
23080 2018-02-09 Jakub Jelinek <jakub@redhat.com>
23083 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
23084 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
23088 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
23089 PARALLEL incoming that failed vt_get_decl_and_offset check.
23091 PR middle-end/84237
23092 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
23093 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
23095 (get_variable_section): For decls in named .bss* sections pass true as
23096 second argument to bss_initializer_p.
23098 2018-02-09 Marek Polacek <polacek@redhat.com>
23099 Jakub Jelinek <jakub@redhat.com>
23102 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
23103 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
23104 Sync some changes from cxx_fold_indirect_ref.
23106 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
23108 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
23110 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
23111 (BLOCK_INLINE_ENTRY_LABEL): New.
23112 (dwarf2out_var_location): Disregard inline entry markers.
23113 (inline_entry_data): New struct.
23114 (inline_entry_data_hasher): New hashtable type.
23115 (inline_entry_data_hasher::hash): New.
23116 (inline_entry_data_hasher::equal): New.
23117 (inline_entry_data_table): New variable.
23118 (add_high_low_attributes): Add DW_AT_entry_pc and
23119 DW_AT_GNU_entry_view attributes if a pending entry is found
23120 in inline_entry_data_table. Add old entry_pc attribute only
23121 if debug nonbinding markers are disabled.
23122 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
23123 markers are enabled.
23124 (block_within_block_p, dwarf2out_inline_entry): New.
23125 (dwarf2out_finish): Check that no entries remained in
23126 inline_entry_data_table.
23127 * final.c (reemit_insn_block_notes): Handle inline entry notes.
23128 (final_scan_insn, notice_source_line): Likewise.
23129 (rest_of_clean_state): Skip inline entry markers.
23130 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
23132 * gimple.c (gimple_build_debug_inline_entry): New.
23133 * gimple.h (enum gimple_debug_subcode): Add
23134 GIMPLE_DEBUG_INLINE_ENTRY.
23135 (gimple_build_debug_inline_entry): Declare.
23136 (gimple_debug_inline_entry_p): New.
23137 (gimple_debug_nonbind_marker_p): Adjust.
23138 * insn-notes.def (INLINE_ENTRY): New.
23139 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
23140 inline entry marker notes.
23141 (print_insn): Likewise.
23142 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
23143 (INSN_DEBUG_MARKER_KIND): Likewise.
23144 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
23145 * tree-inline.c (expand_call_inline): Build and insert
23146 debug_inline_entry stmt.
23147 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
23148 inline entry blocks early, if nonbind markers are enabled.
23149 (dump_scope_block): Dump fragment info.
23150 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
23151 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
23152 (gimple_build_debug_inline_entry): New.
23153 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
23154 Enable/disable inline entry points too.
23155 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
23156 (DEBUG_INSN): Describe inline entry markers.
23158 * common.opt (gvariable-location-views): New.
23159 (gvariable-location-views=incompat5): New.
23160 * config.in: Rebuilt.
23161 * configure: Rebuilt.
23162 * configure.ac: Test assembler for view support.
23163 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
23164 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
23165 * dwarf2out.c (var_loc_view): New typedef.
23166 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
23167 (dwarf2out_locviews_in_attribute): New.
23168 (dwarf2out_locviews_in_loclist): New.
23169 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
23170 (enum dw_line_info_opcode): Add LI_adv_address.
23171 (struct dw_line_info_table): Add view.
23172 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
23173 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
23174 (zero_view_p): New variable.
23175 (ZERO_VIEW_P): New macro.
23176 (output_asm_line_debug_info): New.
23177 (struct var_loc_node): Add view.
23178 (add_AT_view_list, AT_loc_list): New.
23179 (add_var_loc_to_decl): Add view param. Test it against last.
23180 (new_loc_list): Add view params. Record them.
23181 (AT_loc_list_ptr): Handle loc and view lists.
23182 (view_list_to_loc_list_val_node): New.
23183 (print_dw_val): Handle dw_val_class_view_list.
23184 (size_of_die): Likewise.
23185 (value_format): Likewise.
23186 (loc_list_has_views): New.
23187 (gen_llsym): Set vl_symbol too.
23188 (maybe_gen_llsym, skip_loc_list_entry): New.
23189 (dwarf2out_maybe_output_loclist_view_pair): New.
23190 (output_loc_list): Output view list or entries too.
23191 (output_view_list_offset): New.
23192 (output_die): Handle dw_val_class_view_list.
23193 (output_dwarf_version): New.
23194 (output_compilation_unit_header): Use it.
23195 (output_skeleton_debug_sections): Likewise.
23196 (output_rnglists, output_line_info): Likewise.
23197 (output_pubnames, output_aranges): Update version comments.
23198 (output_one_line_info_table): Output view numbers in asm comments.
23199 (dw_loc_list): Determine current endview, pass it to new_loc_list.
23200 Call maybe_gen_llsym.
23201 (loc_list_from_tree_1): Adjust.
23202 (add_AT_location_description): Create view list attribute if
23203 needed, check it's absent otherwise.
23204 (convert_cfa_to_fb_loc_list): Adjust.
23205 (maybe_emit_file): Call output_asm_line_debug_info for test.
23206 (dwarf2out_var_location): Reset views as needed. Precompute
23207 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
23208 attribute. Set view.
23209 (new_line_info_table): Reset next view.
23210 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
23211 (dwarf2out_source_line): Likewise. Output view resets and labels to
23212 the assembler, or select appropriate line info opcodes.
23213 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
23214 (optimize_string_length): Catch it. Adjust.
23215 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
23216 dw_val_class_view_list, and remove it if no longer needed.
23217 (hash_loc_list): Hash view numbers.
23218 (loc_list_hasher::equal): Compare them.
23219 (optimize_location_lists): Check whether a view list symbol is
23220 needed, and whether the locview attribute is present, and
23221 whether they match. Remove the locview attribute if no longer
23223 (index_location_lists): Call skip_loc_list_entry for test.
23224 (dwarf2out_finish): Call output_asm_line_debug_info for test.
23225 Use output_dwarf_version.
23226 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
23227 (struct dw_val_node): Add val_view_list.
23228 * final.c (SEEN_NEXT_VIEW): New.
23229 (set_next_view_needed): New.
23230 (clear_next_view_needed): New.
23231 (maybe_output_next_view): New.
23232 (final_start_function): Rename to...
23233 (final_start_function_1): ... this. Take pointer to FIRST,
23234 add SEEN parameter. Emit param bindings in the initial view.
23235 (final_start_function): Reintroduce SEEN-less interface.
23236 (final): Rename to...
23237 (final_1): ... this. Take SEEN parameter. Output final pending
23238 next view at the end.
23239 (final): Reintroduce seen-less interface.
23240 (final_scan_insn): Output pending next view before switching
23241 sections or ending a block. Mark the next view as needed when
23242 outputting variable locations. Notify debug backend of section
23243 changes, and of location view changes.
23244 (rest_of_handle_final): Adjust.
23245 * toplev.c (process_options): Autodetect value for debug variable
23246 location views option. Warn on incompat5 without -gdwarf-5.
23247 * doc/invoke.texi (gvariable-location-views): New.
23248 (gvariable-location-views=incompat5): New.
23249 (gno-variable-location-views): New.
23251 2018-02-08 David Malcolm <dmalcolm@redhat.com>
23253 PR tree-optimization/84136
23254 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
23255 that the result of find_edge is non-NULL.
23257 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
23260 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
23261 storing integer register in SImode. Fix cost of 256 and 512
23262 byte aligned SSE register store.
23264 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
23266 * config/i386/i386.c (ix86_multiplication_cost): Fix
23267 multiplication cost for TARGET_AVX512DQ.
23269 2018-02-08 Marek Polacek <polacek@redhat.com>
23271 PR tree-optimization/84238
23272 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
23275 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23277 PR tree-optimization/84265
23278 * tree-vect-stmts.c (vectorizable_store): Don't treat
23279 VMAT_CONTIGUOUS accesses as grouped.
23280 (vectorizable_load): Likewise.
23282 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23284 PR tree-optimization/81635
23285 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
23286 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
23287 (test_round_for_mask): New functions.
23288 (wide_int_cc_tests): Call test_round_for_mask.
23289 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
23290 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
23291 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
23292 range returned by get_range_info.
23294 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
23297 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
23298 * symtab.c: Include builtins.h
23299 (symtab_node::output_to_lto_symbol_table_p): Move here
23300 from lto-streamer-out.c:output_symbol_p.
23301 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
23302 (output_symbol_p): Move all logic to symtab.c
23303 (produce_symtab): Update.
23305 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23307 * config/s390/s390-opts.h (enum indirect_branch): Define.
23308 * config/s390/s390-protos.h (s390_return_addr_from_memory)
23309 (s390_indirect_branch_via_thunk)
23310 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
23311 (enum s390_indirect_branch_type): Define.
23312 * config/s390/s390.c (struct s390_frame_layout, struct
23313 machine_function): Remove.
23314 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
23315 (indirect_branch_table_label_no, indirect_branch_table_name):
23317 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
23318 (enum s390_indirect_branch_option): Define.
23319 (s390_return_addr_from_memory): New function.
23320 (s390_handle_string_attribute): New function.
23321 (s390_attribute_table): Add new attribute handler.
23322 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
23323 (s390_indirect_branch_via_thunk): New function.
23324 (s390_indirect_branch_via_inline_thunk): New function.
23325 (s390_function_ok_for_sibcall): When jumping via thunk disallow
23326 sibling call optimization for non z10 compiles.
23327 (s390_emit_call): Force indirect branch target to be a single
23328 register. Add r1 clobber for non-z10 compiles.
23329 (s390_emit_epilogue): Emit return jump via return_use expander.
23330 (s390_reorg): Handle JUMP_INSNs as execute targets.
23331 (s390_option_override_internal): Perform validity checks for the
23332 new command line options.
23333 (s390_indirect_branch_attrvalue): New function.
23334 (s390_indirect_branch_settings): New function.
23335 (s390_set_current_function): Invoke s390_indirect_branch_settings.
23336 (s390_output_indirect_thunk_function): New function.
23337 (s390_code_end): Implement target hook.
23338 (s390_case_values_threshold): Implement target hook.
23339 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
23341 * config/s390/s390.h (struct s390_frame_layout)
23342 (struct machine_function): Move here from s390.c.
23343 (TARGET_INDIRECT_BRANCH_NOBP_RET)
23344 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
23345 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
23346 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
23347 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
23348 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
23349 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
23350 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
23351 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
23352 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
23353 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
23354 (mnemonic attribute): Add values which aren't recognized
23356 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
23357 pattern for branch conversion. Fix mnemonic attribute.
23358 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
23359 indirect branch via thunk if requested.
23360 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
23361 ("*indirect_jump"): Disable for branch conversion using out of
23363 ("indirect_jump_via_thunk<mode>_z10")
23364 ("indirect_jump_via_thunk<mode>")
23365 ("indirect_jump_via_inlinethunk<mode>_z10")
23366 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
23367 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
23368 ("casesi_jump_via_inlinethunk<mode>_z10")
23369 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
23370 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
23371 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
23372 ("*indirect2_jump"): Disable for branch conversion.
23373 ("casesi_jump"): Turn into expander and expand patterns for branch
23375 ("return_use"): New expander.
23376 ("*return"): Emit return via thunk and rename it to ...
23377 ("*return<mode>"): ... this one.
23378 * config/s390/s390.opt: Add new options and and enum for the
23381 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23383 * lra-constraints.c (match_reload): Unconditionally use
23384 gen_lowpart_SUBREG, rather than selecting between that
23385 and equivalent gen_rtx_SUBREG code.
23387 2018-02-08 Richard Biener <rguenther@suse.de>
23389 PR tree-optimization/84233
23390 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
23391 changed flag instead of boguously re-using phi_inserted.
23393 2018-02-08 Martin Jambor <mjambor@suse.cz>
23395 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
23396 static local variables.
23398 2018-02-08 Richard Biener <rguenther@suse.de>
23400 PR tree-optimization/84278
23401 * tree-vect-stmts.c (vectorizable_store): When looking for
23402 smaller vector types to perform grouped strided loads/stores
23403 make sure the mode is supported by the target.
23404 (vectorizable_load): Likewise.
23406 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
23408 * config/aarch64/aarch64.c (aarch64_components_for_bb):
23409 Increase LDP/STP opportunities by adding adjacent callee-saves.
23411 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
23413 PR rtl-optimization/84068
23414 PR rtl-optimization/83459
23415 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
23417 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
23419 PR tree-optimization/84224
23420 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
23421 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
23422 non-zero arguments.
23424 2018-02-07 Iain Sandoe <iain@codesourcery.com>
23427 * config/rs6000/altivec.md (*restore_world): Remove LR use.
23428 * config/rs6000/predicates.md (restore_world_operation): Adjust op
23429 count, remove one USE.
23431 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23433 * doc/install.texi (Configuration): Document the
23434 --with-long-double-format={ibm,ieee} PowerPC configuration
23438 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
23439 Convert from define_expand to be define_insn_and_split. Rework
23440 float/double/_Float128 conversions to QI/HI/SImode to work with
23441 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
23442 conversions to QI/HImode types did a store and then a load to
23443 truncate the value. For conversions to VSX registers, don't split
23444 the insn, instead emit the code directly. Use the code iterator
23445 any_fix to combine signed and unsigned conversions.
23446 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
23447 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
23448 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
23449 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
23450 (fix_<mode>di2_hw): Likewise.
23451 (fixuns_<mode>di2_hw): Likewise.
23452 (fix_<mode>si2_hw): Likewise.
23453 (fixuns_<mode>si2_hw): Likewise.
23454 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
23455 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
23456 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
23457 fix<uns>_trunc<SFDF:mode>si2_p8.
23458 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
23459 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23460 (fix<uns>_<mode>_mem): Likewise.
23461 (fctiw<u>z_<mode>_mem): Likewise.
23462 (fix<uns>_<mode>_mem): Likewise.
23463 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
23464 the register allocator from doing a direct move to the GPRs to do
23465 a store, and instead use the ISA 3.0 store byte/half-word from
23466 vector register instruction. For IEEE 128-bit floating point,
23467 also optimize stores of 32-bit ints.
23468 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
23470 2018-02-07 Alan Hayward <alan.hayward@arm.com>
23472 * genextract.c (push_pathstr_operand): New function to support
23474 (walk_rtx): Call push_pathstr_operand.
23475 (print_path): Support [a-zA-Z].
23477 2018-02-07 Richard Biener <rguenther@suse.de>
23479 PR tree-optimization/84037
23480 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
23481 (cse_and_gimplify_to_preheader): Declare.
23482 (vect_get_place_in_interleaving_chain): Likewise.
23483 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
23485 (_loop_vec_info::~_loop_vec_info): Delete it.
23486 (cse_and_gimplify_to_preheader): New function.
23487 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
23488 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
23489 (vectorizable_load): Likewise. For grouped stores always base
23490 the IV on the first element.
23491 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
23492 condition before gimplifying.
23494 2018-02-07 Jakub Jelinek <jakub@redhat.com>
23496 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
23497 *DIV_EXPR and *MOD_EXPR.
23499 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
23502 * config/i386/i386.c (ix86_option_override_internal): Mask out
23503 the CF_SET bit when checking -fcf-protection.
23505 2018-02-07 Tom de Vries <tom@codesourcery.com>
23508 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
23511 2018-02-07 Richard Biener <rguenther@suse.de>
23513 PR tree-optimization/84204
23514 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
23517 PR tree-optimization/84205
23518 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
23519 special-case isl_ast_op_zdiv_r.
23521 PR tree-optimization/84223
23522 * graphite-scop-detection.c (gather_bbs::before_dom_children):
23523 Only add conditions from within the region.
23524 (gather_bbs::after_dom_children): Adjust.
23526 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
23529 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
23530 * config/avr/avr.md: Only post-reload split REG-REG moves if
23531 either register is GENERAL_REG_P.
23533 2018-02-07 Jakub Jelinek <jakub@redhat.com>
23535 PR tree-optimization/84235
23536 * tree-ssa-scopedtables.c
23537 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
23538 if the subtraction is performed in floating point type where NaNs are
23539 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
23540 build 1. Formatting fix.
23542 2018-02-06 Jakub Jelinek <jakub@redhat.com>
23545 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
23546 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
23547 and skip it regardless of bb boundaries. Use CALL_P macro,
23548 don't test INSN_P (insn) together with CALL_P or JUMP_P check
23549 unnecessarily, formatting fix.
23551 2018-02-06 Michael Collison <michael.collison@arm.com>
23553 * config/arm/thumb2.md:
23554 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
23555 (*thumb_mov_notscc): Ditto.
23557 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
23560 * config/rs6000/rs6000.md (su code attribute): Use "u" for
23561 unsigned_fix, not "s".
23563 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23565 * configure.ac (gcc_fn_eh_frame_ro): New function.
23566 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
23567 correct .eh_frame permissions.
23568 * configure: Regenerate.
23570 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
23572 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
23573 irrelevant options.
23575 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23577 * config/rs6000/rs6000.c (rs6000_option_override_internal):
23578 Display warning message for -mno-speculate-indirect-jumps.
23580 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
23582 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
23584 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
23586 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
23588 PR tree-optimization/84225
23589 * tree-eh.c (find_trapping_overflow): Only call
23590 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
23592 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
23595 * config/i386/i386.c: Reimplement the check of possible options
23596 -mibt/-mshstk conbination. Change error messages.
23597 * doc/invoke.texi: Fix a typo: remove extra '='.
23599 2018-02-06 Marek Polacek <polacek@redhat.com>
23601 PR tree-optimization/84228
23602 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
23604 2018-02-06 Tamar Christina <tamar.christina@arm.com>
23607 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
23608 emitted arch directives.
23609 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
23610 __ARM_FEATURE_COPROC before changing architectures.
23612 2018-02-06 Richard Biener <rguenther@suse.de>
23614 * config/i386/i386.c (print_reg): Fix typo.
23615 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
23617 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
23619 * configure: Regenerate.
23621 2018-02-05 Martin Sebor <msebor@redhat.com>
23623 PR tree-optimization/83369
23624 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
23627 2018-02-05 Martin Liska <mliska@suse.cz>
23629 * doc/invoke.texi: Cherry-pick upstream r323995.
23631 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
23633 * ira.c (ira_init_register_move_cost): Adjust comment.
23635 2018-02-05 Martin Liska <mliska@suse.cz>
23637 PR gcov-profile/84137
23638 * doc/gcov.texi: Fix typo in documentation.
23640 2018-02-05 Martin Liska <mliska@suse.cz>
23642 PR gcov-profile/83879
23643 * doc/gcov.texi: Document necessity of --dynamic-list-data when
23644 using dlopen functionality.
23646 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
23648 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
23649 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
23650 _mm_maskz_range_ss, _mm_mask_range_round_ss,
23651 _mm_maskz_range_round_ss): New intrinsics.
23652 (__builtin_ia32_rangesd128_round)
23653 (__builtin_ia32_rangess128_round): Remove.
23654 (__builtin_ia32_rangesd128_mask_round,
23655 __builtin_ia32_rangess128_mask_round): New builtins.
23656 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
23657 __builtin_ia32_rangess128_round): Remove.
23658 (__builtin_ia32_rangesd128_mask_round,
23659 __builtin_ia32_rangess128_mask_round): New builtins.
23660 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
23661 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
23662 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
23663 "<round_saeonly_constraint>")): Changed to ...
23664 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
23665 "<round_saeonly_scalar_constraint>")): ... this.
23666 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
23667 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
23668 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
23669 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
23670 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
23672 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
23674 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
23676 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
23677 Remove all values except native, 8540 and 8548.
23679 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
23681 * config/i386/i386.c (ix86_output_function_return): Pass
23682 INVALID_REGNUM, instead of -1, as invalid register number to
23683 indirect_thunk_name and output_indirect_thunk.
23685 2018-02-02 Julia Koval <julia.koval@intel.com>
23687 * config.gcc: Add -march=icelake.
23688 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
23689 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
23690 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
23691 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
23692 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
23693 (processor_target_table): Add icelake.
23694 (ix86_option_override_internal): Handle new PTAs.
23695 (get_builtin_code_for_version): Handle icelake.
23696 (M_INTEL_COREI7_ICELAKE): New.
23697 (fold_builtin_cpu): Handle icelake.
23698 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
23699 * doc/invoke.texi: Add -march=icelake.
23701 2018-02-02 Julia Koval <julia.koval@intel.com>
23703 * config/i386/i386.c (ix86_option_override_internal): Change flags type
23704 to wide_int_bitmask.
23705 * wide-int-bitmask.h: New.
23707 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
23710 * config/i386/i386.md: Replace Pmode with word_mode in
23711 builtin_setjmp_setup and builtin_longjmp to support x32.
23713 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
23717 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
23719 (rs6000_supported_cpu_names): New static variable.
23720 (linux_cpu_translation_table): Likewise.
23721 (elf_platform) <cpu>: Define new static variable and use it.
23722 Translate kernel AT_PLATFORM name to canonical name if needed.
23723 Error if platform name is unknown.
23725 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
23728 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
23730 2018-02-01 Jeff Law <law@redhat.com>
23733 * config/i386/i386.c (release_scratch_register_on_entry): Add new
23734 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
23735 the scratch if RELEASE_VIA_POP is false.
23736 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
23737 If we have to save a temporary register, decrement SIZE appropriately.
23738 Pass new arguments to release_scratch_register_on_entry.
23739 (ix86_adjust_stack_and_probe): Likewise.
23740 (ix86_emit_probe_stack_range): Pass new arguments to
23741 release_scratch_register_on_entry.
23743 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
23745 PR rtl-optimization/84157
23746 * combine.c (change_zero_ext): Use REG_P predicate in
23747 front of HARD_REGISTER_P predicate.
23749 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
23751 * config/avr/avr.c (avr_option_override): Move disabling of
23752 -fdelete-null-pointer-checks to...
23753 * common/config/avr/avr-common.c (avr_option_optimization_table):
23756 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23758 PR tree-optimization/81635
23759 * tree-data-ref.c (split_constant_offset_1): For types that
23760 wrap on overflow, try to use range info to prove that wrapping
23763 2018-02-01 Renlin Li <renlin.li@arm.com>
23766 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
23767 TAILCALL_ADDR_REGS.
23768 (aarch64_register_move_cost): Likewise.
23769 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
23770 TAILCALL_ADDR_REGS.
23771 (REG_CLASS_NAMES): Likewise.
23772 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
23773 TAILCALL_ADDR_REGS. Remove IP registers.
23774 * config/aarch64/aarch64.md (Ucs): Update register constraint.
23776 2018-02-01 Richard Biener <rguenther@suse.de>
23778 * domwalk.h (dom_walker::dom_walker): Add additional constructor
23779 for specifying RPO order and allow NULL for that.
23780 * domwalk.c (dom_walker::dom_walker): Likewise.
23781 (dom_walker::walk): Handle NULL RPO order.
23782 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
23784 (rewrite_update_dom_walker): Likewise.
23785 (mark_def_dom_walker): Likewise.
23787 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23789 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
23790 (aarch64_maybe_expand_sve_subreg_move): Declare.
23791 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
23792 * config/aarch64/predicates.md (aarch64_any_register_operand): New
23794 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
23795 that are semantically a reverse operation.
23796 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
23797 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
23798 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
23800 (aarch64_can_change_mode_class): For big-endian, forbid changes
23801 between two SVE modes if they have different element sizes.
23803 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23805 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
23806 the TImode handling for big-endian targets.
23808 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23810 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
23811 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
23813 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
23814 Remove BSWAP handing for big-endian targets and use the form of
23815 LD1RQ appropariate for the mode.
23817 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23819 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
23820 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
23821 duplicated element.
23823 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23826 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
23827 check for operands that need to go through aarch64_sve_reload_be.
23829 2018-02-01 Jakub Jelinek <jakub@redhat.com>
23831 PR tree-optimization/81661
23832 PR tree-optimization/84117
23833 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
23834 * tree-eh.c: Include gimplify.h.
23835 (find_trapping_overflow, replace_trapping_overflow,
23836 rewrite_to_non_trapping_overflow): New functions.
23837 * tree-vect-loop.c: Include tree-eh.h.
23838 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
23839 * tree-data-ref.c: Include tree-eh.h.
23840 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
23842 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
23844 PR rtl-optimization/84123
23845 * combine.c (change_zero_ext): Check if hard register satisfies
23846 can_change_dest_mode before calling gen_lowpart_SUBREG.
23848 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
23851 * ira.c (ira_init_register_move_cost): Remove assert.
23853 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
23855 PR rtl-optimization/84071
23856 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
23857 * doc/tm.texi: Regenerate.
23859 2018-01-31 Richard Biener <rguenther@suse.de>
23861 PR tree-optimization/84132
23862 * tree-data-ref.c (analyze_miv_subscript): Properly
23863 check whether evolution_function_is_affine_multivariate_p
23864 before calling gcd_of_steps_may_divide_p.
23866 2018-01-31 Julia Koval <julia.koval@intel.com>
23869 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
23870 * config/i386/i386.md (rdpid_rex64) New.
23871 (rdpid): Make 32bit only.
23873 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
23876 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
23877 an IDENTIFIER_NODE for FUNCTION_TYPE's.
23879 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
23882 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
23884 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
23886 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
23888 PR rtl-optimization/84071
23889 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
23890 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
23892 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
23894 * config/arc/arc.c (arc_handle_aux_attribute): New function.
23895 (arc_attribute_table): Add 'aux' attribute.
23896 (arc_in_small_data_p): Consider aux like variables.
23897 (arc_is_aux_reg_p): New function.
23898 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
23899 (arc_get_aux_arg): New function.
23900 (prepare_move_operands): Handle aux-register access.
23901 (arc_handle_aux_attribute): New function.
23902 * doc/extend.texi (ARC Variable attributes): Add subsection.
23904 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
23906 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
23907 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
23908 (arc_attribute_table): Add 'uncached' attribute.
23909 (arc_print_operand): Print '.di' flag for uncached memory
23911 (arc_in_small_data_p): Do not consider for small data the uncached
23913 (arc_is_uncached_mem_p): New function.
23914 * config/arc/predicates.md (compact_store_memory_operand): Check
23915 for uncached memory accesses.
23916 (nonvol_nonimm_operand): Likewise.
23917 * doc/extend.texi (ARC Type Attribute): New subsection.
23919 2018-01-31 Jakub Jelinek <jakub@redhat.com>
23922 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
23923 falign-loops=): Add Optimization flag.
23925 2018-01-30 Jeff Law <law@redhat.com>
23928 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
23929 INT_REGISTERS_SAVED. Check it prior to calling
23930 get_scratch_register_on_entry.
23931 (ix86_adjust_stack_and_probe): Similarly.
23932 (ix86_emit_probe_stack_range): Similarly.
23933 (ix86_expand_prologue): Corresponding changes.
23935 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23938 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
23939 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
23941 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
23944 * lra-constraints.c (curr_insn_transform): Process AND in the
23947 2018-01-30 Jakub Jelinek <jakub@redhat.com>
23949 PR rtl-optimization/83986
23950 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
23951 dependence against last_pending_memory_flush in addition to
23952 pending_jump_insns.
23954 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
23956 PR tree-optimization/81611
23957 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
23960 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23963 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
23966 2018-01-30 Richard Biener <rguenther@suse.de>
23967 Jakub Jelinek <jakub@redhat.com>
23969 PR tree-optimization/84111
23970 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
23971 inner loops added during recursion, as they don't have up-to-date
23974 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
23977 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
23978 (can_inline_edge_by_limits_p): ... here.
23979 (can_early_inline_edge_p, check_callers,
23980 update_caller_keys, update_callee_keys, recursive_inlining,
23981 add_new_edges_to_heap, speculation_useful_p,
23982 inline_small_functions,
23983 inline_small_functions, flatten_function,
23984 inline_to_all_callers_1): Update.
23986 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
23988 * profile-count.c (profile_count::combine_with_ipa_count): Handle
23991 2018-01-30 Richard Biener <rguenther@suse.de>
23993 PR tree-optimization/83008
23994 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
23995 invariant and constant vector uses in stmts when they need
23996 more than one stmt.
23998 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24001 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
24002 * configure: Regenerate.
24004 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24006 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
24008 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
24009 Use gen_rtx_REG rather than gen_lowpart.
24011 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24013 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
24014 rather than 0 when creating partial subregs.
24016 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24018 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
24021 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
24024 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
24025 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
24026 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
24027 flags. This restores the settings used before the 2017-07-24.
24028 Turning off pre increment/decrement/modify allows IVOPTS to
24029 optimize DF/SF loops where the index is an int.
24031 2018-01-29 Richard Biener <rguenther@suse.de>
24032 Kelvin Nilsen <kelvin@gcc.gnu.org>
24035 * tree-vect-stmts.c (vectorizable_call): Don't call
24036 targetm.vectorize_builtin_md_vectorized_function if callee is
24039 2018-01-22 Carl Love <cel@us.ibm.com>
24041 * doc/extend.tex: Fix typo in second arg in
24042 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
24044 2018-01-29 Richard Biener <rguenther@suse.de>
24046 PR tree-optimization/84086
24047 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
24048 (flush_ssaname_freelist): When SSA names were released reset
24049 the SCEV hash table.
24051 2018-01-29 Richard Biener <rguenther@suse.de>
24053 PR tree-optimization/84057
24054 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
24055 removed paths when removing edges.
24057 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
24059 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
24060 -mfunction-return=@var{choice}.
24062 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
24064 PR diagnostic/84034
24065 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
24066 Handle CR like TAB.
24067 (layout::print_source_line): Likewise.
24068 (test_get_line_width_without_trailing_whitespace): Add test cases.
24070 2018-01-27 Jakub Jelinek <jakub@redhat.com>
24072 PR middle-end/84040
24073 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
24076 2018-01-26 Jim Wilson <jimw@sifive.com>
24078 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
24080 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
24083 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24085 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
24086 and CMP + SUB-immediate -> SUBS.
24088 2018-01-26 Martin Sebor <msebor@redhat.com>
24090 PR tree-optimization/83896
24091 * tree-ssa-strlen.c (get_string_len): Rename...
24092 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
24093 Avoid assuming length is constant.
24094 (handle_char_store): Use HOST_WIDE_INT for string length.
24096 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
24099 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
24100 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
24102 2018-01-26 Richard Biener <rguenther@suse.de>
24104 PR rtl-optimization/84003
24105 * dse.c (record_store): Only record redundant stores when
24106 the earlier store aliases at least all accesses the later one does.
24108 2018-01-26 Jakub Jelinek <jakub@redhat.com>
24110 PR rtl-optimization/83985
24111 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
24112 REG_CFA_RESTORE insns.
24113 (delete_unmarked_insns): Don't ignore separate shrink wrapping
24114 REG_CFA_RESTORE insns here.
24117 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
24118 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
24120 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24122 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
24123 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
24124 (arc_init): Likewise.
24125 (arc_override_options): Likewise.
24126 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
24128 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
24130 * config/arc/arc.h (TARGET_DBNZ): Define.
24131 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
24132 properly set the tune attribute.
24133 (dbnz): Use TARGET_DBNZ guard.
24134 * config/arc/arc.opt (mtune): Add core3 option.
24136 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24138 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
24139 recognize new pic like addresses.
24140 (arc_delegitimize_address): Clean up.
24142 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24144 * config/arc/arc-arches.def: Option mrf16 valid for all
24146 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
24147 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
24148 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
24149 * config/arc/arc-tables.opt: Regenerate.
24150 * config/arc/arc.c (arc_conditional_register_usage): Handle
24151 reduced register file case.
24152 (arc_file_start): Set must have build attributes.
24153 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
24154 mrf16 option value.
24155 * config/arc/arc.opt (mrf16): Add new option.
24156 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
24157 * config/arc/genmultilib.awk: Handle new mrf16 option.
24158 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
24159 * config/arc/t-multilib: Regenerate.
24160 * doc/invoke.texi (ARC Options): Document mrf16 option.
24162 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24164 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
24165 * config/arc/arc.c (arc_handle_secure_attribute): New function.
24166 (arc_attribute_table): Add 'secure_call' attribute.
24167 (arc_print_operand): Print secure call operand.
24168 (arc_function_ok_for_sibcall): Don't optimize tail calls when
24170 (arc_is_secure_call_p): New function. * config/arc/arc.md
24171 (call_i): Add support for sjli instruction.
24172 (call_value_i): Likewise.
24173 * config/arc/constraints.md (Csc): New constraint.
24175 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24176 John Eric Martin <John.Martin@emmicro-us.com>
24178 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
24179 * config/arc/arc.c (_arc_jli_section): New struct.
24180 (arc_jli_section): New type.
24181 (rc_jli_sections): New static variable.
24182 (arc_handle_jli_attribute): New function.
24183 (arc_attribute_table): Add jli_always and jli_fixed attribute.
24184 (arc_file_end): New function.
24185 (TARGET_ASM_FILE_END): Define.
24186 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
24187 (arc_add_jli_section): New function.
24188 (jli_call_scan): Likewise.
24189 (arc_reorg): Call jli_call_scan.
24190 (arc_output_addsi): Remove 'S' from printing asm operand.
24191 (arc_is_jli_call_p): New function.
24192 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
24194 (movhi_insn): Likewise.
24195 (movsi_insn): Likewise.
24196 (movsi_set_cc_insn): Likewise.
24197 (loadqi_update): Likewise.
24198 (load_zeroextendqisi_update): Likewise.
24199 (load_signextendqisi_update): Likewise.
24200 (loadhi_update): Likewise.
24201 (load_zeroextendhisi_update): Likewise.
24202 (load_signextendhisi_update): Likewise.
24203 (loadsi_update): Likewise.
24204 (loadsf_update): Likewise.
24205 (movsicc_insn): Likewise.
24206 (bset_insn): Likewise.
24207 (bxor_insn): Likewise.
24208 (bclr_insn): Likewise.
24209 (bmsk_insn): Likewise.
24210 (bicsi3_insn): Likewise.
24211 (cmpsi_cc_c_insn): Likewise.
24212 (movsi_ne): Likewise.
24213 (movsi_cond_exec): Likewise.
24214 (clrsbsi2): Likewise.
24215 (norm_f): Likewise.
24225 (call_i): Remove 'S' asm letter, add jli instruction.
24226 (call_value_i): Likewise.
24227 * config/arc/arc.op (mjli-always): New option.
24228 * config/arc/constraints.md (Cji): New constraint.
24229 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
24231 (subsf3_fpx): Likewise.
24232 (mulsf3_fpx): Likewise.
24233 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
24235 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
24236 function attrbutes.
24237 * doc/invoke.texi (ARC): Document mjli-always option.
24239 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
24241 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
24242 avoid addition with 0 and use incw and decw where possible.
24244 2018-01-26 Richard Biener <rguenther@suse.de>
24246 PR tree-optimization/81082
24247 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
24248 association if it requires casting to unsigned.
24249 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
24250 from fold_plusminus_mult_expr to catch important cases late when
24251 range info is available.
24253 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24255 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
24256 * configure.ac (hidden_linkonce): New test.
24257 * configure: Regenerate.
24258 * config.in: Regenerate.
24260 2018-01-26 Julia Koval <julia.koval@intel.com>
24262 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
24263 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
24264 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
24265 _mm_mask_bitshuffle_epi64_mask): Fix type.
24266 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
24267 USI_FTYPE_V4DI_V4DI_USI): Remove.
24268 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
24269 __builtin_ia32_vpshufbitqmb256_mask,
24270 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
24271 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
24272 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
24274 2018-01-26 Alan Modra <amodra@gmail.com>
24277 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
24278 UNSPEC_VBPERMQ. Sort other unspecs.
24280 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
24282 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
24284 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
24286 PR middle-end/83055
24287 * predict.c (drop_profile): Do not push/pop cfun; update also
24289 (handle_missing_profiles): Fix logic looking for zero profiles.
24291 2018-01-25 Jakub Jelinek <jakub@redhat.com>
24293 PR middle-end/83977
24294 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
24295 on functions with #pragma omp declare simd or functions with simd
24297 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
24298 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
24299 Remove trailing \n from warning_at calls.
24301 2018-01-25 Tom de Vries <tom@codesourcery.com>
24304 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
24305 for neutered workers.
24307 2018-01-24 Joseph Myers <joseph@codesourcery.com>
24310 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
24311 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
24313 2018-01-24 Jeff Law <law@redhat.com>
24316 * i386.c (get_probe_interval): Move to earlier point.
24317 (ix86_compute_frame_layout): If -fstack-clash-protection and
24318 the frame is larger than the probe interval, then use pushes
24319 to save registers rather than reg->mem moves.
24320 (ix86_expand_prologue): Remove conditional for int_registers_saved
24323 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
24326 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
24327 min/max for never referenced object.
24329 2018-01-24 Jakub Jelinek <jakub@redhat.com>
24331 PR middle-end/83977
24332 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
24334 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
24335 attributes from DECL_ATTRIBUTES (decl) without affecting
24336 DECL_ATTRIBUTES (current_function_decl).
24337 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
24338 functions with non-NULL DECL_ABSTRACT_ORIGIN.
24340 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
24342 PR tree-optimization/83979
24343 * fold-const.c (fold_comparison): Use constant_boolean_node
24344 instead of boolean_{true,false}_node.
24346 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
24348 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
24351 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24353 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
24354 Simplify the clause that sets the length attribute.
24355 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
24356 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
24357 clause that sets the length attribute.
24358 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
24360 2018-01-24 Tom de Vries <tom@codesourcery.com>
24363 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
24364 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
24365 Add strict parameter.
24366 (prevent_branch_around_nothing): Insert dummy insn between branch to
24367 label and label with no ptx insn inbetween.
24368 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
24370 2018-01-24 Tom de Vries <tom@codesourcery.com>
24373 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
24374 for neutered threads in warp.
24375 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
24377 2018-01-24 Richard Biener <rguenther@suse.de>
24379 PR tree-optimization/83176
24380 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
24383 2018-01-24 Richard Biener <rguenther@suse.de>
24385 PR tree-optimization/82819
24386 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
24387 code generating pluses that are no-ops in the target precision.
24389 2018-01-24 Richard Biener <rguenther@suse.de>
24391 PR middle-end/84000
24392 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
24394 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
24396 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
24397 to merge probabilities.
24398 * predict.c (probably_never_executed): Also mark as cold functions
24399 with global 0 profile and guessed local profile.
24400 * profile-count.c (profile_probability::combine_with_count): New
24402 * profile-count.h (profile_probability::operator*,
24403 profile_probability::operator*=, profile_probability::operator/,
24404 profile_probability::operator/=): Reduce precision to adjusted
24405 and set value to guessed on contradictory divisions.
24406 (profile_probability::combine_with_freq): Remove.
24407 (profile_probability::combine_wiht_count): Declare.
24408 (profile_count::force_nonzero):: Set to adjusted.
24409 (profile_count::probability_in):: Set quality to adjusted.
24410 * tree-ssa-tail-merge.c (replace_block_by): Use
24411 combine_with_count.
24413 2018-01-23 Andrew Waterman <andrew@sifive.com>
24414 Jim Wilson <jimw@sifive.com>
24416 * config/riscv/riscv.c (riscv_stack_boundary): New.
24417 (riscv_option_override): Set riscv_stack_boundary. Handle
24418 riscv_preferred_stack_boundary_arg.
24419 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
24420 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
24421 (STACK_BOUNDARY): Set to riscv_stack_boundary.
24422 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
24423 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
24424 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
24426 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
24429 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
24430 of struct ix86_frame.
24431 (ix86_expand_epilogue): Likewise. Add a local variable for
24432 the reg_save_offset field in struct ix86_frame.
24434 2018-01-23 Bin Cheng <bin.cheng@arm.com>
24436 PR tree-optimization/82604
24437 * tree-loop-distribution.c (enum partition_kind): New enum item
24438 PKIND_PARTIAL_MEMSET.
24439 (partition_builtin_p): Support above new enum item.
24440 (generate_code_for_partition): Ditto.
24441 (compute_access_range): Differentiate cases that equality can be
24442 proven at all loops, the innermost loops or no loops.
24443 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
24444 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
24445 (finalize_partitions, distribute_loop): Don't fuse partition of
24446 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
24447 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
24448 parloop is enabled.
24450 2018-01-23 Martin Liska <mliska@suse.cz>
24452 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
24453 order to ignore the predictor.
24454 (PRED_POLYMORPHIC_CALL): Likewise.
24455 (PRED_RECURSIVE_CALL): Likewise.
24457 2018-01-23 Martin Liska <mliska@suse.cz>
24459 * tree-profile.c (tree_profiling): Print function header to
24460 aware reader which function we are working on.
24461 * value-prof.c (gimple_find_values_to_profile): Do not print
24462 not interesting value histograms.
24464 2018-01-23 Martin Liska <mliska@suse.cz>
24466 * profile-count.h (enum profile_quality): Add
24467 profile_uninitialized as the first value. Do not number values
24468 as they are zero based.
24469 (profile_count::verify): Update sanity check.
24470 (profile_probability::verify): Likewise.
24472 2018-01-23 Nathan Sidwell <nathan@acm.org>
24474 * doc/invoke.texi (ffor-scope): Deprecate.
24476 2018-01-23 David Malcolm <dmalcolm@redhat.com>
24478 PR tree-optimization/83510
24479 * domwalk.c (set_all_edges_as_executable): New function.
24480 (dom_walker::dom_walker): Convert bool param
24481 "skip_unreachable_blocks" to enum reachability. Move setup of
24482 edge flags to set_all_edges_as_executable and only do it when
24483 reachability is REACHABLE_BLOCKS.
24484 * domwalk.h (enum dom_walker::reachability): New enum.
24485 (dom_walker::dom_walker): Convert bool param
24486 "skip_unreachable_blocks" to enum reachability.
24487 (set_all_edges_as_executable): New decl.
24488 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
24489 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
24491 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
24492 but converting true to REACHABLE_BLOCKS.
24493 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
24495 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
24496 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
24497 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
24499 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
24500 if check_all_array_refs will be called.
24502 2018-01-23 David Malcolm <dmalcolm@redhat.com>
24504 * tree.c (selftest::test_location_wrappers): Add more test
24507 2018-01-23 David Malcolm <dmalcolm@redhat.com>
24509 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
24510 (selftest::test_bit_in_range): Likewise.
24512 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
24515 * doc/sourcebuild.texi (vect_float): Say that the selector
24516 only describes the situation when -funsafe-math-optimizations is on.
24517 (vect_float_strict): Document.
24519 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
24521 PR tree-optimization/83965
24522 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
24523 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
24524 instead of checking only for a reduction.
24525 (vect_recog_widen_sum_pattern): Likewise.
24527 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
24529 * predict.c (probably_never_executed): Only use precise profile info.
24530 (compute_function_frequency): Skip after inlining hack since we now
24531 have quality checking.
24533 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
24535 * profile-count.h (profile_probability::very_unlikely,
24536 profile_probability::unlikely, profile_probability::even): Set
24537 precision to guessed.
24539 2018-01-23 Richard Biener <rguenther@suse.de>
24541 PR tree-optimization/83963
24542 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
24543 Properly terminate dominator walk when crossing the exit edge not
24544 when visiting its source block.
24546 2018-01-23 Jakub Jelinek <jakub@redhat.com>
24549 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
24550 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
24552 2018-01-22 Jakub Jelinek <jakub@redhat.com>
24554 PR tree-optimization/83957
24555 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
24556 semicolon after for body surrounded by braces.
24558 PR tree-optimization/83081
24559 * profile-count.h (profile_probability::split): New method.
24560 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
24561 Use profile_probability::split.
24562 (do_compare_rtx_and_jump): Fix adjustment of probabilities
24563 when splitting a single conditional jump into 2.
24565 2018-01-22 David Malcolm <dmalcolm@redhat.com>
24567 PR tree-optimization/69452
24568 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
24571 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24573 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
24574 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
24575 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
24577 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24579 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
24581 * config/rl78/rl78.md (movdi): New define_expand.
24582 * config/rl78/rl78.c (rl78_split_movdi): New function.
24584 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
24587 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
24589 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
24590 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
24591 128-bit to produce an UNSPEC move to get the double word with the
24592 signbit and then a shift directly to do signbit.
24593 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
24594 implementation with a new version that just does either a direct
24595 move or a regular move. Move memory interface to separate insns.
24596 Move insns so they are next to the expander.
24597 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
24598 with signbit move. Split big and little endian case.
24599 (signbit<mode>2_dm_mem_le): Likewise.
24600 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
24601 (signbit<mode>2_dm2): Likewise.
24603 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24605 * config/rl78/rl78.md (anddi3): New define_expand.
24607 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24609 * config/rl78/rl78.md (umindi3): New define_expand.
24611 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24613 * config/rl78/rl78.md (smindi3): New define_expand.
24615 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24617 * config/rl78/rl78.md (smaxdi3): New define_expand.
24619 2018-01-22 Carl Love <cel@us.ibm.com>
24621 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
24622 LVX_V1TI): Add macro expansion.
24623 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
24624 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
24625 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
24626 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
24627 Change check to determine if the instruction is a byte reversing
24628 entry. Fix typo in comment.
24629 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
24630 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
24631 Add def_builtin calls for new builtins.
24632 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
24633 Add define_insn expansion.
24635 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24637 * config/rl78/rl78.md (umaxdi3): New define_expand.
24639 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24641 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
24642 for non-QImode registers.
24644 2018-01-22 Richard Biener <rguenther@suse.de>
24646 PR tree-optimization/83963
24647 * graphite-scop-detection.c (scop_detection::get_sese): Delay
24648 including the loop exit block.
24649 (scop_detection::merge_sese): Likewise.
24650 (scop_detection::add_scop): Do it here instead.
24652 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24654 * doc/sourcebuild.texi (arm_softfloat): Document.
24656 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
24659 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
24660 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
24661 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
24663 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24664 David Edelsohn <dje.gcc@gmail.com>
24667 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
24668 Change "crset eq" to "crset 2".
24669 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
24670 (*call_indirect_aix<mode>_nospec): Likewise.
24671 (*call_value_indirect_aix<mode>_nospec): Likewise.
24672 (*call_indirect_elfv2<mode>_nospec): Likewise.
24673 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
24674 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
24675 change assembly output from . to $.
24676 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
24677 (indirect_jump<mode>_nospec): Change assembly output from . to $.
24678 (*tablejump<mode>_internal1_nospec): Likewise.
24680 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
24683 * config/sh/sh_optimize_sett_clrt.cc:
24684 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
24686 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
24688 PR tree-optimization/83940
24689 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
24690 offset_dt to vect_constant_def rather than vect_unknown_def_type.
24691 (vect_check_load_store_mask): Add a mask_dt_out parameter and
24692 use it to pass back the definition type.
24693 (vect_check_store_rhs): Likewise rhs_dt_out.
24694 (vect_build_gather_load_calls): Add a mask_dt argument and use
24695 it instead of a call to vect_is_simple_use.
24696 (vectorizable_store): Update calls to vect_check_load_store_mask
24697 and vect_check_store_rhs. Use the dt returned by the latter instead
24698 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
24699 instead of calls to vect_is_simple_use. Pass the scalar rather
24700 than the vector operand to vect_is_simple_use when handling
24701 second and subsequent copies of an rhs value.
24702 (vectorizable_load): Update calls to vect_check_load_store_mask
24703 and vect_build_gather_load_calls. Use the cached mask_dt and
24704 gs_info.offset_dt instead of calls to vect_is_simple_use.
24706 2018-01-20 Jakub Jelinek <jakub@redhat.com>
24708 PR middle-end/83945
24709 * tree-emutls.c: Include gimplify.h.
24710 (lower_emutls_2): New function.
24711 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
24712 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
24713 it before further processing.
24716 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
24717 UINTVAL (trueop1) instead of INTVAL (op1).
24719 2018-01-19 Jakub Jelinek <jakub@redhat.com>
24723 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
24724 INCOMING_FRAME_SP_OFFSET if not defined.
24725 (scan_trace): Add ENTRY argument. If true and
24726 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
24727 emit a note to adjust the CFA offset.
24728 (create_cfi_notes): Adjust scan_trace callers.
24729 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
24730 INCOMING_FRAME_SP_OFFSET in the CIE.
24731 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
24732 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
24734 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
24735 * doc/tm.texi: Regenerated.
24737 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24739 PR rtl-optimization/83147
24740 * lra-constraints.c (remove_inheritance_pseudos): Use
24741 lra_substitute_pseudo_within_insn.
24743 2018-01-19 Tom de Vries <tom@codesourcery.com>
24744 Cesar Philippidis <cesar@codesourcery.com>
24747 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
24749 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
24752 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
24753 spaces for function labels.
24755 2018-01-19 Martin Liska <mliska@suse.cz>
24757 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
24758 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
24759 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
24760 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
24761 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
24762 (PRED_CONST_RETURN): Change from 69 to 65.
24763 (PRED_NULL_RETURN): Change from 91 to 71.
24764 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
24765 (PRED_LOOP_GUARD): Change from 66 to 73.
24767 2018-01-19 Martin Liska <mliska@suse.cz>
24769 * predict.c (predict_insn_def): Add new assert.
24770 (struct branch_predictor): Change type to signed integer.
24771 (test_prediction_value_range): Amend test to cover
24772 PROB_UNINITIALIZED.
24773 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
24774 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
24775 (PRED_LOOP_ITERATIONS_MAX): Likewise.
24776 (PRED_LOOP_IV_COMPARE): Likewise.
24777 * predict.h (PROB_UNINITIALIZED): Define new constant.
24779 2018-01-19 Martin Liska <mliska@suse.cz>
24781 * predict.c (dump_prediction): Add new format for
24782 analyze_brprob.py script which is enabled with -details
24784 * profile-count.h (precise_p): New function.
24786 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
24788 PR tree-optimization/83922
24789 * tree-vect-loop.c (vect_verify_full_masking): Return false if
24790 there are no statements that need masking.
24791 (vect_active_double_reduction_p): New function.
24792 (vect_analyze_loop_operations): Use it when handling phis that
24793 are not in the loop header.
24795 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
24797 PR tree-optimization/83914
24798 * tree-vect-loop.c (vectorizable_induction): Don't convert
24799 init_expr or apply the peeling adjustment for inductions
24800 that are nested within the vectorized loop.
24802 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24804 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
24807 2018-01-18 Jakub Jelinek <jakub@redhat.com>
24811 * function.h (gimplify_parameters): Add gimple_seq * argument.
24812 * function.c: Include gimple.h and options.h.
24813 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
24814 for the added local temporaries if needed.
24815 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
24816 if there are any parameter cleanups, wrap whole body into a
24817 try/finally with the cleanups.
24819 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
24822 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
24823 Use GET_MODE_CLASS for scalar floating point.
24825 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
24829 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
24830 Fix call of call_cgraph_insertion_hooks.
24832 2018-01-18 Martin Sebor <msebor@redhat.com>
24834 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
24836 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
24839 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
24842 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
24845 * common.opt: (-ffile-prefix-map): New option.
24846 * opts.c (common_handle_option): Defer it.
24847 * opts-global.c (handle_common_deferred_options): Handle it.
24848 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
24849 * file-prefix-map.h: New file.
24850 (remap_debug_filename, add_debug_prefix_map): ...here.
24851 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
24852 * final.c (debug_prefix_map, add_debug_prefix_map
24853 remap_debug_filename): Move to...
24854 * file-prefix-map.c: New file.
24855 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
24856 generalize, get rid of alloca(), use strrchr() instead of strchr().
24857 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
24858 Implement in terms of add_prefix_map().
24859 (remap_macro_filename, remap_debug_filename): Implement in term of
24861 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
24862 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
24863 * dbxout.c: Include file-prefix-map.h.
24864 * varasm.c: Likewise.
24865 * vmsdbgout.c: Likewise.
24866 * xcoffout.c: Likewise.
24867 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
24868 * doc/cppopts.texi (-fmacro-prefix-map): Document.
24869 * doc/invoke.texi (-ffile-prefix-map): Document.
24870 (-fdebug-prefix-map): Update description.
24872 2018-01-18 Martin Liska <mliska@suse.cz>
24874 * config/i386/i386.c (indirect_thunk_name): Document that also
24876 (output_indirect_thunk): Document why both instructions
24877 (pause and lfence) are generated.
24879 2018-01-18 Richard Biener <rguenther@suse.de>
24881 PR tree-optimization/83887
24882 * graphite-scop-detection.c
24883 (scop_detection::get_nearest_dom_with_single_entry): Remove.
24884 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
24885 (scop_detection::merge_sese): Re-implement with a flood-fill
24886 algorithm that properly finds a SESE region if it exists.
24888 2018-01-18 Jakub Jelinek <jakub@redhat.com>
24891 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
24892 pointer_diff optimizations use view_convert instead of convert.
24894 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24896 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
24897 Generate different code for -mno-speculate-indirect-jumps.
24898 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
24899 (*call_indirect_aix<mode>): Disable for
24900 -mno-speculate-indirect-jumps.
24901 (*call_indirect_aix<mode>_nospec): New define_insn.
24902 (*call_value_indirect_aix<mode>): Disable for
24903 -mno-speculate-indirect-jumps.
24904 (*call_value_indirect_aix<mode>_nospec): New define_insn.
24905 (*sibcall_nonlocal_sysv<mode>): Generate different code for
24906 -mno-speculate-indirect-jumps.
24907 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
24909 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
24911 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
24912 long double type, set the flags for noting the default long double
24913 type, even if we don't pass or return a long double type.
24915 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
24918 * ipa-inline.c (flatten_function): Do not overwrite final inlining
24921 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
24923 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
24924 support for merge[hl].
24925 (fold_mergehl_helper): New helper function.
24926 (tree-vector-builder.h): New #include for tree_vector_builder usage.
24927 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
24928 (altivec_vmrglw_direct): Add xxmrglw insn.
24930 2018-01-17 Andrew Waterman <andrew@sifive.com>
24932 * config/riscv/riscv.c (riscv_conditional_register_usage): If
24933 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
24935 2018-01-17 David Malcolm <dmalcolm@redhat.com>
24938 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
24939 call the lto_location_cache before reading the
24940 DECL_SOURCE_LOCATION of the types.
24942 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
24943 Richard Sandiford <richard.sandiford@linaro.org>
24945 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
24946 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
24947 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
24948 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
24949 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
24951 * config/aarch64/constraints.md (aarch64_movti_operand):
24953 * config/aarch64/predicates.md (Uti): Add new constraint.
24955 2018-01-17 Carl Love <cel@us.ibm.com>
24957 * config/rs6000/vsx.md (define_expand xl_len_r,
24958 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
24959 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
24961 (define_expand, define_insn): Move the shift left from the
24962 define_insn to the define_expand for lxvl and stxvl instructions.
24963 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
24964 and XL_LEN_R definitions to PURE.
24966 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
24968 * config/i386/i386.c (indirect_thunk_name): Declare regno
24969 as unsigned int. Compare regno with INVALID_REGNUM.
24970 (output_indirect_thunk): Ditto.
24971 (output_indirect_thunk_function): Ditto.
24972 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
24973 in the call to output_indirect_thunk_function.
24975 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
24977 PR middle-end/83884
24978 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
24979 rather than the size of inner_type to determine the stack slot size
24980 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
24982 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
24985 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
24988 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
24990 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
24991 endian Linux systems to optionally enable multilibs for selecting
24992 the long double type if the user configured an explicit type.
24993 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
24994 have no long double multilibs if not defined.
24995 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
24996 warn if the user used -mabi={ieee,ibm}longdouble and we built
24997 multilibs for long double.
24998 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
24999 appropriate multilib option.
25000 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
25002 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
25003 for building long double multilibs.
25004 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
25006 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
25008 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
25011 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
25013 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
25016 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
25019 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
25022 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
25024 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
25025 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
25027 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
25029 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
25030 different rtl trees depending on TARGET_64BIT.
25031 (rs6000_gen_lvx): Likewise.
25033 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
25035 * config/visium/visium.md (nop): Tweak comment.
25036 (hazard_nop): Likewise.
25038 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25040 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
25041 -mspeculate-indirect-jumps.
25042 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
25043 for -mno-speculate-indirect-jumps.
25044 (*call_indirect_elfv2<mode>_nospec): New define_insn.
25045 (*call_value_indirect_elfv2<mode>): Disable for
25046 -mno-speculate-indirect-jumps.
25047 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
25048 (indirect_jump): Emit different RTL for
25049 -mno-speculate-indirect-jumps.
25050 (*indirect_jump<mode>): Disable for
25051 -mno-speculate-indirect-jumps.
25052 (*indirect_jump<mode>_nospec): New define_insn.
25053 (tablejump): Emit different RTL for
25054 -mno-speculate-indirect-jumps.
25055 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
25056 (tablejumpsi_nospec): New define_expand.
25057 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
25058 (tablejumpdi_nospec): New define_expand.
25059 (*tablejump<mode>_internal1): Disable for
25060 -mno-speculate-indirect-jumps.
25061 (*tablejump<mode>_internal1_nospec): New define_insn.
25062 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
25065 2018-01-16 Artyom Skrobov tyomitch@gmail.com
25067 * caller-save.c (insert_save): Drop unnecessary parameter. All
25070 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25071 Richard Biener <rguenth@suse.de>
25074 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
25075 return early, inline manually is_gimple_sizepos. Make sure if we
25076 call gimplify_expr we don't end up with a gimple constant.
25077 * tree.c (variably_modified_type_p): Don't return true for
25078 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
25079 * gimplify.h (is_gimple_sizepos): Remove.
25081 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25083 PR tree-optimization/83857
25084 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
25085 vectorizable_live_operation for pure SLP statements.
25086 (vectorizable_live_operation): Handle PHIs.
25088 2018-01-16 Richard Biener <rguenther@suse.de>
25090 PR tree-optimization/83867
25091 * tree-vect-stmts.c (vect_transform_stmt): Precompute
25092 nested_in_vect_loop_p since the scalar stmt may get invalidated.
25094 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25097 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
25098 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
25099 If off is not INTEGER_CST, issue a may not be aligned warning
25100 rather than isn't aligned. Use isn%'t rather than isn't.
25101 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
25103 <case MULT_EXPR>: Improve the case when bottom and one of the
25104 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
25105 operand, in that case check if the other operand is multiple of
25106 bottom divided by the INTEGER_CST operand.
25108 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25111 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
25112 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
25113 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
25114 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
25115 * config/pa/pa.c (pa_function_arg_advance): Likewise.
25116 (pa_function_arg, pa_arg_partial_bytes): Likewise.
25117 (pa_function_arg_size): New function.
25119 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25121 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
25122 in a separate statement.
25124 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25126 PR tree-optimization/83847
25127 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
25128 group gathers and scatters.
25130 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25132 PR rtl-optimization/86620
25133 * params.def (max-sched-ready-insns): Bump minimum value to 1.
25135 PR rtl-optimization/83213
25136 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
25137 to last if both are JUMP_INSNs.
25139 PR tree-optimization/83843
25140 * gimple-ssa-store-merging.c
25141 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
25142 store_immediate_info for bswap/nop orig_stores.
25144 2018-01-15 Andrew Waterman <andrew@sifive.com>
25146 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
25148 <UDIV>: Increase cost if !TARGET_DIV.
25150 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25152 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
25153 (define_attr "cr_logical_3op"): New.
25154 (cceq_ior_compare): Adjust.
25155 (cceq_ior_compare_complement): Adjust.
25156 (*cceq_rev_compare): Adjust.
25157 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
25158 (is_cracked_insn): Adjust.
25159 (insn_must_be_first_in_group): Adjust.
25160 * config/rs6000/40x.md: Adjust.
25161 * config/rs6000/440.md: Adjust.
25162 * config/rs6000/476.md: Adjust.
25163 * config/rs6000/601.md: Adjust.
25164 * config/rs6000/603.md: Adjust.
25165 * config/rs6000/6xx.md: Adjust.
25166 * config/rs6000/7450.md: Adjust.
25167 * config/rs6000/7xx.md: Adjust.
25168 * config/rs6000/8540.md: Adjust.
25169 * config/rs6000/cell.md: Adjust.
25170 * config/rs6000/e300c2c3.md: Adjust.
25171 * config/rs6000/e500mc.md: Adjust.
25172 * config/rs6000/e500mc64.md: Adjust.
25173 * config/rs6000/e5500.md: Adjust.
25174 * config/rs6000/e6500.md: Adjust.
25175 * config/rs6000/mpc.md: Adjust.
25176 * config/rs6000/power4.md: Adjust.
25177 * config/rs6000/power5.md: Adjust.
25178 * config/rs6000/power6.md: Adjust.
25179 * config/rs6000/power7.md: Adjust.
25180 * config/rs6000/power8.md: Adjust.
25181 * config/rs6000/power9.md: Adjust.
25182 * config/rs6000/rs64.md: Adjust.
25183 * config/rs6000/titan.md: Adjust.
25185 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25187 * config/i386/predicates.md (indirect_branch_operand): Rewrite
25188 ix86_indirect_branch_register logic.
25190 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25192 * config/i386/constraints.md (Bs): Update
25193 ix86_indirect_branch_register check. Don't check
25194 ix86_indirect_branch_register with GOT_memory_operand.
25196 * config/i386/predicates.md (GOT_memory_operand): Don't check
25197 ix86_indirect_branch_register here.
25198 (GOT32_symbol_operand): Likewise.
25200 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25202 * config/i386/predicates.md (constant_call_address_operand):
25203 Rewrite ix86_indirect_branch_register logic.
25204 (sibcall_insn_operand): Likewise.
25206 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25208 * config/i386/constraints.md (Bs): Replace
25209 ix86_indirect_branch_thunk_register with
25210 ix86_indirect_branch_register.
25212 * config/i386/i386.md (indirect_jump): Likewise.
25213 (tablejump): Likewise.
25214 (*sibcall_memory): Likewise.
25215 (*sibcall_value_memory): Likewise.
25216 Peepholes of indirect call and jump via memory: Likewise.
25217 * config/i386/i386.opt: Likewise.
25218 * config/i386/predicates.md (indirect_branch_operand): Likewise.
25219 (GOT_memory_operand): Likewise.
25220 (call_insn_operand): Likewise.
25221 (sibcall_insn_operand): Likewise.
25222 (GOT32_symbol_operand): Likewise.
25224 2018-01-15 Jakub Jelinek <jakub@redhat.com>
25226 PR middle-end/83837
25227 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
25228 type rather than type addr's type points to.
25229 (expand_omp_atomic_mutex): Likewise.
25230 (expand_omp_atomic): Likewise.
25232 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25235 * config/i386/i386.c (output_indirect_thunk_function): Use
25236 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
25237 for __x86_return_thunk.
25239 2018-01-15 Richard Biener <rguenther@suse.de>
25241 PR middle-end/83850
25242 * expmed.c (extract_bit_field_1): Fix typo.
25244 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25247 * config/arm/iterators.md (VF): New mode iterator.
25248 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
25249 Remove integer-related logic from pattern.
25250 (neon_vabd<mode>_3): Likewise.
25252 2018-01-15 Jakub Jelinek <jakub@redhat.com>
25254 PR middle-end/82694
25255 * common.opt (fstrict-overflow): No longer an alias.
25256 (fwrapv-pointer): New option.
25257 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
25258 also for pointer types based on flag_wrapv_pointer.
25259 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
25260 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
25261 opts->x_flag_wrapv got set.
25262 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
25263 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
25264 POINTER_TYPE_OVERFLOW_UNDEFINED.
25265 * match.pd: Likewise in address comparison pattern.
25266 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
25268 2018-01-15 Richard Biener <rguenther@suse.de>
25271 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
25272 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
25273 Reset type names to their identifier if their TYPE_DECL doesn't
25274 have linkage (and thus is used for ODR and devirt).
25275 (save_debug_info_for_decl): Remove.
25276 (save_debug_info_for_type): Likewise.
25277 (add_tree_to_fld_list): Adjust.
25278 * tree-pretty-print.c (dump_generic_node): Make dumping of
25279 type names more robust.
25281 2018-01-15 Richard Biener <rguenther@suse.de>
25283 * BASE-VER: Bump to 8.0.1.
25285 2018-01-14 Martin Sebor <msebor@redhat.com>
25288 * builtins.c (check_access): Avoid warning when the no-warning bit
25291 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
25293 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
25294 * ira-color (allocno_hard_regs_compare): Likewise.
25296 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
25299 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
25300 Use .pushsection/.popsection.
25302 2018-01-14 Martin Sebor <msebor@redhat.com>
25305 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
25307 2018-01-14 Jakub Jelinek <jakub@redhat.com>
25309 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
25310 entry from extra_headers.
25311 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
25312 extra_headers, make the list bitwise identical to the i?86-*-* one.
25314 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25316 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
25317 -mcmodel=large with -mindirect-branch=thunk,
25318 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
25319 -mfunction-return=thunk-extern.
25320 * doc/invoke.texi: Document -mcmodel=large is incompatible with
25321 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
25322 -mfunction-return=thunk and -mfunction-return=thunk-extern.
25324 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25326 * config/i386/i386.c (print_reg): Print the name of the full
25327 integer register without '%'.
25328 (ix86_print_operand): Handle 'V'.
25329 * doc/extend.texi: Document 'V' modifier.
25331 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25333 * config/i386/constraints.md (Bs): Disallow memory operand for
25334 -mindirect-branch-register.
25336 * config/i386/predicates.md (indirect_branch_operand): Likewise.
25337 (GOT_memory_operand): Likewise.
25338 (call_insn_operand): Likewise.
25339 (sibcall_insn_operand): Likewise.
25340 (GOT32_symbol_operand): Likewise.
25341 * config/i386/i386.md (indirect_jump): Call convert_memory_address
25342 for -mindirect-branch-register.
25343 (tablejump): Likewise.
25344 (*sibcall_memory): Likewise.
25345 (*sibcall_value_memory): Likewise.
25346 Disallow peepholes of indirect call and jump via memory for
25347 -mindirect-branch-register.
25348 (*call_pop): Replace m with Bw.
25349 (*call_value_pop): Likewise.
25350 (*sibcall_pop_memory): Replace m with Bs.
25351 * config/i386/i386.opt (mindirect-branch-register): New option.
25352 * doc/invoke.texi: Document -mindirect-branch-register option.
25354 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25356 * config/i386/i386-protos.h (ix86_output_function_return): New.
25357 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
25358 set function_return_type.
25359 (indirect_thunk_name): Add ret_p to indicate thunk for function
25361 (output_indirect_thunk_function): Pass false to
25362 indirect_thunk_name.
25363 (ix86_output_indirect_branch_via_reg): Likewise.
25364 (ix86_output_indirect_branch_via_push): Likewise.
25365 (output_indirect_thunk_function): Create alias for function
25366 return thunk if regno < 0.
25367 (ix86_output_function_return): New function.
25368 (ix86_handle_fndecl_attribute): Handle function_return.
25369 (ix86_attribute_table): Add function_return.
25370 * config/i386/i386.h (machine_function): Add
25371 function_return_type.
25372 * config/i386/i386.md (simple_return_internal): Use
25373 ix86_output_function_return.
25374 (simple_return_internal_long): Likewise.
25375 * config/i386/i386.opt (mfunction-return=): New option.
25376 (indirect_branch): Mention -mfunction-return=.
25377 * doc/extend.texi: Document function_return function attribute.
25378 * doc/invoke.texi: Document -mfunction-return= option.
25380 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25382 * config/i386/i386-opts.h (indirect_branch): New.
25383 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
25384 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
25385 with local indirect jump when converting indirect call and jump.
25386 (ix86_set_indirect_branch_type): New.
25387 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
25388 (indirectlabelno): New.
25389 (indirect_thunk_needed): Likewise.
25390 (indirect_thunk_bnd_needed): Likewise.
25391 (indirect_thunks_used): Likewise.
25392 (indirect_thunks_bnd_used): Likewise.
25393 (INDIRECT_LABEL): Likewise.
25394 (indirect_thunk_name): Likewise.
25395 (output_indirect_thunk): Likewise.
25396 (output_indirect_thunk_function): Likewise.
25397 (ix86_output_indirect_branch_via_reg): Likewise.
25398 (ix86_output_indirect_branch_via_push): Likewise.
25399 (ix86_output_indirect_branch): Likewise.
25400 (ix86_output_indirect_jmp): Likewise.
25401 (ix86_code_end): Call output_indirect_thunk_function if needed.
25402 (ix86_output_call_insn): Call ix86_output_indirect_branch if
25404 (ix86_handle_fndecl_attribute): Handle indirect_branch.
25405 (ix86_attribute_table): Add indirect_branch.
25406 * config/i386/i386.h (machine_function): Add indirect_branch_type
25407 and has_local_indirect_jump.
25408 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
25410 (tablejump): Likewise.
25411 (*indirect_jump): Use ix86_output_indirect_jmp.
25412 (*tablejump_1): Likewise.
25413 (simple_return_indirect_internal): Likewise.
25414 * config/i386/i386.opt (mindirect-branch=): New option.
25415 (indirect_branch): New.
25418 (thunk-inline): Likewise.
25419 (thunk-extern): Likewise.
25420 * doc/extend.texi: Document indirect_branch function attribute.
25421 * doc/invoke.texi: Document -mindirect-branch= option.
25423 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
25426 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
25428 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
25430 * ipa-inline.c (want_inline_small_function_p): Return false if
25431 inlining has already failed with CIF_FINAL_ERROR.
25432 (update_caller_keys): Call want_inline_small_function_p before
25434 (update_callee_keys): Likewise.
25436 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25438 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
25440 (rs6000_quadword_masked_address_p): Likewise.
25441 (quad_aligned_load_p): Likewise.
25442 (quad_aligned_store_p): Likewise.
25443 (const_load_sequence_p): Add comment to describe the outer-most loop.
25444 (mimic_memory_attributes_and_flags): New function.
25445 (rs6000_gen_stvx): Likewise.
25446 (replace_swapped_aligned_store): Likewise.
25447 (rs6000_gen_lvx): Likewise.
25448 (replace_swapped_aligned_load): Likewise.
25449 (replace_swapped_load_constant): Capitalize argument name in
25450 comment describing this function.
25451 (rs6000_analyze_swaps): Add a third pass to search for vector loads
25452 and stores that access quad-word aligned addresses and replace
25453 with stvx or lvx instructions when appropriate.
25454 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
25455 New function prototype.
25456 (rs6000_quadword_masked_address_p): Likewise.
25457 (rs6000_gen_lvx): Likewise.
25458 (rs6000_gen_stvx): Likewise.
25459 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
25460 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
25461 when memory address is aligned.
25462 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
25463 this split to select lvx instruction when memory address is aligned.
25464 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
25465 instruction when memory address is aligned.
25466 (*vsx_le_perm_load_v16qi): Likewise.
25467 (four unnamed splitters): Modify to select the stvx instruction
25468 when memory is aligned.
25470 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
25472 * predict.c (determine_unlikely_bbs): Handle correctly BBs
25473 which appears in the queue multiple times.
25475 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25476 Alan Hayward <alan.hayward@arm.com>
25477 David Sherwood <david.sherwood@arm.com>
25479 * tree-vectorizer.h (vec_lower_bound): New structure.
25480 (_loop_vec_info): Add check_nonzero and lower_bounds.
25481 (LOOP_VINFO_CHECK_NONZERO): New macro.
25482 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
25483 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
25484 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
25485 fields. Make seg_len the distance travelled, not including the
25487 (dr_direction_indicator): Declare.
25488 (dr_zero_step_indicator): Likewise.
25489 (dr_known_forward_stride_p): Likewise.
25490 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
25492 (runtime_alias_check_p): Allow runtime alias checks with
25494 (operator ==): Compare access_size and align.
25495 (prune_runtime_alias_test_list): Rework for new distinction between
25496 the access_size and seg_len.
25497 (create_intersect_range_checks_index): Likewise. Cope with polynomial
25499 (get_segment_min_max): New function.
25500 (create_intersect_range_checks): Use it.
25501 (dr_step_indicator): New function.
25502 (dr_direction_indicator): Likewise.
25503 (dr_zero_step_indicator): Likewise.
25504 (dr_known_forward_stride_p): Likewise.
25505 * tree-loop-distribution.c (data_ref_segment_size): Return
25506 DR_STEP * (niters - 1).
25507 (compute_alias_check_pairs): Update call to the dr_with_seg_len
25509 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
25510 (vect_preserves_scalar_order_p): New function, split out from...
25511 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
25512 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
25513 (vect_vfa_access_size): New function.
25514 (vect_vfa_align): Likewise.
25515 (vect_compile_time_alias): Take access_size_a and access_b arguments.
25516 (dump_lower_bound): New function.
25517 (vect_check_lower_bound): Likewise.
25518 (vect_small_gap_p): Likewise.
25519 (vectorizable_with_step_bound_p): Likewise.
25520 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
25521 depencies if the vectorization factor is 1. Convert the checks
25522 for nonzero steps into checks on the bounds of DR_STEP. Try using
25523 a bunds check for variable steps if the minimum required step is
25524 relatively small. Update calls to the dr_with_seg_len
25525 constructor and to vect_compile_time_alias.
25526 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
25528 (vect_loop_versioning): Call it.
25529 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
25531 (vect_estimate_min_profitable_iters): Account for any bounds checks.
25533 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25534 Alan Hayward <alan.hayward@arm.com>
25535 David Sherwood <david.sherwood@arm.com>
25537 * doc/sourcebuild.texi (vect_scatter_store): Document.
25538 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
25540 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
25542 * genopinit.c (main): Add supports_vec_scatter_store and
25543 supports_vec_scatter_store_cached to target_optabs.
25544 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
25545 IFN_MASK_SCATTER_STORE.
25546 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
25548 * internal-fn.h (internal_store_fn_p): Declare.
25549 (internal_fn_stored_value_index): Likewise.
25550 * internal-fn.c (scatter_store_direct): New macro.
25551 (expand_scatter_store_optab_fn): New function.
25552 (direct_scatter_store_optab_supported_p): New macro.
25553 (internal_store_fn_p): New function.
25554 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
25555 IFN_MASK_SCATTER_STORE.
25556 (internal_fn_mask_index): Likewise.
25557 (internal_fn_stored_value_index): New function.
25558 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
25559 for scatter stores.
25560 * optabs-query.h (supports_vec_scatter_store_p): Declare.
25561 * optabs-query.c (supports_vec_scatter_store_p): New function.
25562 * tree-vectorizer.h (vect_get_store_rhs): Declare.
25563 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
25564 true for scatter stores.
25565 (vect_gather_scatter_fn_p): Handle scatter stores too.
25566 (vect_check_gather_scatter): Consider using scatter stores if
25567 supports_vec_scatter_store_p.
25568 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
25569 scatter stores too.
25570 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
25571 internal_fn_stored_value_index.
25572 (check_load_store_masking): Handle scatter stores too.
25573 (vect_get_store_rhs): Make public.
25574 (vectorizable_call): Use internal_store_fn_p.
25575 (vectorizable_store): Handle scatter store internal functions.
25576 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
25577 when deciding whether the end of the group has been reached.
25578 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
25579 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
25580 (mask_scatter_store<mode>): New insns.
25582 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25583 Alan Hayward <alan.hayward@arm.com>
25584 David Sherwood <david.sherwood@arm.com>
25586 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
25587 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
25588 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
25590 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
25591 Use vect_truncate_gather_scatter_offset if we can't treat the
25592 operation as a normal gather load or scatter store.
25593 (get_group_load_store_type): Take the gather_scatter_info
25594 as argument. Try using a gather load or scatter store for
25595 single-element groups.
25596 (get_load_store_type): Update calls to get_group_load_store_type
25597 and vect_use_strided_gather_scatters_p.
25599 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25600 Alan Hayward <alan.hayward@arm.com>
25601 David Sherwood <david.sherwood@arm.com>
25603 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
25604 optional tree argument.
25605 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
25607 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
25608 but continue to use the current value as a fallback.
25609 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
25610 to compare the updates.
25611 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
25612 (get_load_store_type): Use it when handling a strided access.
25613 (vect_get_strided_load_store_ops): New function.
25614 (vect_get_data_ptr_increment): Likewise.
25615 (vectorizable_load): Handle strided gather loads. Always pass
25616 a step to vect_create_data_ref_ptr and bump_vector_ptr.
25618 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25619 Alan Hayward <alan.hayward@arm.com>
25620 David Sherwood <david.sherwood@arm.com>
25622 * doc/md.texi (gather_load@var{m}): Document.
25623 (mask_gather_load@var{m}): Likewise.
25624 * genopinit.c (main): Add supports_vec_gather_load and
25625 supports_vec_gather_load_cached to target_optabs.
25626 * optabs-tree.c (init_tree_optimization_optabs): Use
25627 ggc_cleared_alloc to allocate target_optabs.
25628 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
25629 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
25631 * internal-fn.h (internal_load_fn_p): Declare.
25632 (internal_gather_scatter_fn_p): Likewise.
25633 (internal_fn_mask_index): Likewise.
25634 (internal_gather_scatter_fn_supported_p): Likewise.
25635 * internal-fn.c (gather_load_direct): New macro.
25636 (expand_gather_load_optab_fn): New function.
25637 (direct_gather_load_optab_supported_p): New macro.
25638 (direct_internal_fn_optab): New function.
25639 (internal_load_fn_p): Likewise.
25640 (internal_gather_scatter_fn_p): Likewise.
25641 (internal_fn_mask_index): Likewise.
25642 (internal_gather_scatter_fn_supported_p): Likewise.
25643 * optabs-query.c (supports_at_least_one_mode_p): New function.
25644 (supports_vec_gather_load_p): Likewise.
25645 * optabs-query.h (supports_vec_gather_load_p): Declare.
25646 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
25647 and memory_type field.
25648 (NUM_PATTERNS): Bump to 15.
25649 * tree-vect-data-refs.c: Include internal-fn.h.
25650 (vect_gather_scatter_fn_p): New function.
25651 (vect_describe_gather_scatter_call): Likewise.
25652 (vect_check_gather_scatter): Try using internal functions for
25653 gather loads. Recognize existing calls to a gather load function.
25654 (vect_analyze_data_refs): Consider using gather loads if
25655 supports_vec_gather_load_p.
25656 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
25657 (vect_get_gather_scatter_offset_type): Likewise.
25658 (vect_convert_mask_for_vectype): Likewise.
25659 (vect_add_conversion_to_patterm): Likewise.
25660 (vect_try_gather_scatter_pattern): Likewise.
25661 (vect_recog_gather_scatter_pattern): New pattern recognizer.
25662 (vect_vect_recog_func_ptrs): Add it.
25663 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
25664 internal_fn_mask_index and internal_gather_scatter_fn_p.
25665 (check_load_store_masking): Take the gather_scatter_info as an
25666 argument and handle gather loads.
25667 (vect_get_gather_scatter_ops): New function.
25668 (vectorizable_call): Check internal_load_fn_p.
25669 (vectorizable_load): Likewise. Handle gather load internal
25671 (vectorizable_store): Update call to check_load_store_masking.
25672 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
25673 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
25674 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
25675 (aarch64_gather_scale_operand_d): New predicates.
25676 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
25677 (mask_gather_load<mode>): New insns.
25679 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25680 Alan Hayward <alan.hayward@arm.com>
25681 David Sherwood <david.sherwood@arm.com>
25683 * optabs.def (fold_left_plus_optab): New optab.
25684 * doc/md.texi (fold_left_plus_@var{m}): Document.
25685 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
25686 * internal-fn.c (fold_left_direct): Define.
25687 (expand_fold_left_optab_fn): Likewise.
25688 (direct_fold_left_optab_supported_p): Likewise.
25689 * fold-const-call.c (fold_const_fold_left): New function.
25690 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
25691 * tree-parloops.c (valid_reduction_p): New function.
25692 (gather_scalar_reductions): Use it.
25693 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
25694 (vect_finish_replace_stmt): Declare.
25695 * tree-vect-loop.c (fold_left_reduction_fn): New function.
25696 (needs_fold_left_reduction_p): New function, split out from...
25697 (vect_is_simple_reduction): ...here. Accept reductions that
25698 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
25699 (vect_force_simple_reduction): Also store the reduction type in
25700 the assignment's STMT_VINFO_REDUC_TYPE.
25701 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
25702 (merge_with_identity): New function.
25703 (vect_expand_fold_left): Likewise.
25704 (vectorize_fold_left_reduction): Likewise.
25705 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
25706 scalar phi in place for it. Check for target support and reject
25707 cases that would reassociate the operation. Defer the transform
25708 phase to vectorize_fold_left_reduction.
25709 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
25710 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
25711 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
25713 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25715 * tree-if-conv.c (predicate_mem_writes): Remove redundant
25716 call to ifc_temp_var.
25718 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25719 Alan Hayward <alan.hayward@arm.com>
25720 David Sherwood <david.sherwood@arm.com>
25722 * target.def (legitimize_address_displacement): Take the original
25723 offset as a poly_int.
25724 * targhooks.h (default_legitimize_address_displacement): Update
25726 * targhooks.c (default_legitimize_address_displacement): Likewise.
25727 * doc/tm.texi: Regenerate.
25728 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
25729 as an argument, moving assert of ad->disp == ad->disp_term to...
25730 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
25731 Try calling targetm.legitimize_address_displacement before expanding
25732 the address rather than afterwards, and adjust for the new interface.
25733 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
25734 Match the new hook interface. Handle SVE addresses.
25735 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
25736 new hook interface.
25738 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25740 * Makefile.in (OBJS): Add early-remat.o.
25741 * target.def (select_early_remat_modes): New hook.
25742 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
25743 * doc/tm.texi: Regenerate.
25744 * targhooks.h (default_select_early_remat_modes): Declare.
25745 * targhooks.c (default_select_early_remat_modes): New function.
25746 * timevar.def (TV_EARLY_REMAT): New timevar.
25747 * passes.def (pass_early_remat): New pass.
25748 * tree-pass.h (make_pass_early_remat): Declare.
25749 * early-remat.c: New file.
25750 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
25752 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
25754 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25755 Alan Hayward <alan.hayward@arm.com>
25756 David Sherwood <david.sherwood@arm.com>
25758 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
25759 vfm1 with a bound_epilog parameter.
25760 (vect_do_peeling): Update calls accordingly, and move the prologue
25761 call earlier in the function. Treat the base bound_epilog as 0 for
25762 fully-masked loops and retain vf - 1 for other loops. Add 1 to
25763 this base when peeling for gaps.
25764 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
25765 with fully-masked loops.
25766 (vect_estimate_min_profitable_iters): Handle the single peeled
25767 iteration in that case.
25769 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25770 Alan Hayward <alan.hayward@arm.com>
25771 David Sherwood <david.sherwood@arm.com>
25773 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
25774 single-element interleaving even if the size is not a power of 2.
25775 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
25776 accesses for single-element interleaving if the group size is
25779 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25780 Alan Hayward <alan.hayward@arm.com>
25781 David Sherwood <david.sherwood@arm.com>
25783 * doc/md.texi (fold_extract_last_@var{m}): Document.
25784 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
25785 * optabs.def (fold_extract_last_optab): New optab.
25786 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
25787 * internal-fn.c (fold_extract_direct): New macro.
25788 (expand_fold_extract_optab_fn): Likewise.
25789 (direct_fold_extract_optab_supported_p): Likewise.
25790 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
25791 * tree-vect-loop.c (vect_model_reduction_cost): Handle
25792 EXTRACT_LAST_REDUCTION.
25793 (get_initial_def_for_reduction): Do not create an initial vector
25794 for EXTRACT_LAST_REDUCTION reductions.
25795 (vectorizable_reduction): Leave the scalar phi in place for
25796 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
25797 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
25798 epilogue code for EXTRACT_LAST_REDUCTION and defer the
25799 transform phase to vectorizable_condition.
25800 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
25802 (vect_finish_stmt_generation): ...here.
25803 (vect_finish_replace_stmt): New function.
25804 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
25805 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
25807 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
25809 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25810 Alan Hayward <alan.hayward@arm.com>
25811 David Sherwood <david.sherwood@arm.com>
25813 * doc/md.texi (extract_last_@var{m}): Document.
25814 * optabs.def (extract_last_optab): New optab.
25815 * internal-fn.def (EXTRACT_LAST): New internal function.
25816 * internal-fn.c (cond_unary_direct): New macro.
25817 (expand_cond_unary_optab_fn): Likewise.
25818 (direct_cond_unary_optab_supported_p): Likewise.
25819 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
25820 loops using EXTRACT_LAST.
25821 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
25822 (extract_last_<mode>): ...this optab.
25823 (vec_extract<mode><Vel>): Update accordingly.
25825 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25826 Alan Hayward <alan.hayward@arm.com>
25827 David Sherwood <david.sherwood@arm.com>
25829 * target.def (empty_mask_is_expensive): New hook.
25830 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
25831 * doc/tm.texi: Regenerate.
25832 * targhooks.h (default_empty_mask_is_expensive): Declare.
25833 * targhooks.c (default_empty_mask_is_expensive): New function.
25834 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
25835 if the target says that empty masks are expensive.
25836 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
25838 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
25840 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25841 Alan Hayward <alan.hayward@arm.com>
25842 David Sherwood <david.sherwood@arm.com>
25844 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
25845 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
25846 (vect_use_loop_mask_for_alignment_p): New function.
25847 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
25848 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
25849 niters_skip argument. Make sure that the first niters_skip elements
25850 of the first iteration are inactive.
25851 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
25852 Update call to vect_set_loop_masks_directly.
25853 (get_misalign_in_elems): New function, split out from...
25854 (vect_gen_prolog_loop_niters): ...here.
25855 (vect_update_init_of_dr): Take a code argument that specifies whether
25856 the adjustment should be added or subtracted.
25857 (vect_update_init_of_drs): Likewise.
25858 (vect_prepare_for_masked_peels): New function.
25859 (vect_do_peeling): Skip prologue peeling if we're using a mask
25860 instead. Update call to vect_update_inits_of_drs.
25861 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
25863 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
25864 alignment. Do not include the number of peeled iterations in
25865 the minimum threshold in that case.
25866 (vectorizable_induction): Adjust the start value down by
25867 LOOP_VINFO_MASK_SKIP_NITERS iterations.
25868 (vect_transform_loop): Call vect_prepare_for_masked_peels.
25869 Take the number of skipped iterations into account when calculating
25871 * tree-vect-stmts.c (vect_gen_while_not): New function.
25873 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25874 Alan Hayward <alan.hayward@arm.com>
25875 David Sherwood <david.sherwood@arm.com>
25877 * doc/sourcebuild.texi (vect_fully_masked): Document.
25878 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
25879 default value to 0.
25880 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
25882 (vect_analyze_loop_2): ...here. Don't check the vectorization
25883 factor against the number of loop iterations if the loop is
25886 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25887 Alan Hayward <alan.hayward@arm.com>
25888 David Sherwood <david.sherwood@arm.com>
25890 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
25891 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
25892 (dump_groups): Update accordingly.
25893 (iv_use::mem_type): New member variable.
25894 (address_p): New function.
25895 (record_use): Add a mem_type argument and initialize the new
25897 (record_group_use): Add a mem_type argument. Use address_p.
25898 Remove obsolete null checks of base_object. Update call to record_use.
25899 (find_interesting_uses_op): Update call to record_group_use.
25900 (find_interesting_uses_cond): Likewise.
25901 (find_interesting_uses_address): Likewise.
25902 (get_mem_type_for_internal_fn): New function.
25903 (find_address_like_use): Likewise.
25904 (find_interesting_uses_stmt): Try find_address_like_use before
25905 calling find_interesting_uses_op.
25906 (addr_offset_valid_p): Use the iv mem_type field as the type
25907 of the addressed memory.
25908 (add_autoinc_candidates): Likewise.
25909 (get_address_cost): Likewise.
25910 (split_small_address_groups_p): Use address_p.
25911 (split_address_groups): Likewise.
25912 (add_iv_candidate_for_use): Likewise.
25913 (autoinc_possible_for_pair): Likewise.
25914 (rewrite_groups): Likewise.
25915 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
25916 (determine_group_iv_cost): Update after split of USE_ADDRESS.
25917 (get_alias_ptr_type_for_ptr_address): New function.
25918 (rewrite_use_address): Rewrite address uses in calls that were
25919 identified by find_address_like_use.
25921 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25922 Alan Hayward <alan.hayward@arm.com>
25923 David Sherwood <david.sherwood@arm.com>
25925 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
25927 * gimple-expr.h (is_gimple_addressable: Likewise.
25928 * gimple-expr.c (is_gimple_address): Likewise.
25929 * internal-fn.c (expand_call_mem_ref): New function.
25930 (expand_mask_load_optab_fn): Use it.
25931 (expand_mask_store_optab_fn): Likewise.
25933 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25934 Alan Hayward <alan.hayward@arm.com>
25935 David Sherwood <david.sherwood@arm.com>
25937 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
25938 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
25939 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
25940 (cond_umax@var{mode}): Document.
25941 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
25942 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
25943 (cond_umin_optab, cond_umax_optab): New optabs.
25944 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
25945 (COND_IOR, COND_XOR): New internal functions.
25946 * internal-fn.h (get_conditional_internal_fn): Declare.
25947 * internal-fn.c (cond_binary_direct): New macro.
25948 (expand_cond_binary_optab_fn): Likewise.
25949 (direct_cond_binary_optab_supported_p): Likewise.
25950 (get_conditional_internal_fn): New function.
25951 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
25952 Cope with reduction statements that are vectorized as calls rather
25954 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
25955 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
25956 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
25957 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
25958 (UNSPEC_COND_EOR): New unspecs.
25959 (optab): Add mappings for them.
25960 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
25961 (sve_int_op, sve_fp_op): New int attributes.
25963 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25964 Alan Hayward <alan.hayward@arm.com>
25965 David Sherwood <david.sherwood@arm.com>
25967 * optabs.def (while_ult_optab): New optab.
25968 * doc/md.texi (while_ult@var{m}@var{n}): Document.
25969 * internal-fn.def (WHILE_ULT): New internal function.
25970 * internal-fn.h (direct_internal_fn_supported_p): New override
25971 that takes two types as argument.
25972 * internal-fn.c (while_direct): New macro.
25973 (expand_while_optab_fn): New function.
25974 (convert_optab_supported_p): Likewise.
25975 (direct_while_optab_supported_p): New macro.
25976 * wide-int.h (wi::udiv_ceil): New function.
25977 * tree-vectorizer.h (rgroup_masks): New structure.
25978 (vec_loop_masks): New typedef.
25979 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
25980 and fully_masked_p.
25981 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
25982 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
25983 (vect_max_vf): New function.
25984 (slpeel_make_loop_iterate_ntimes): Delete.
25985 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
25986 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
25987 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
25988 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
25989 internal-fn.h, stor-layout.h and optabs-query.h.
25990 (vect_set_loop_mask): New function.
25991 (add_preheader_seq): Likewise.
25992 (add_header_seq): Likewise.
25993 (interleave_supported_p): Likewise.
25994 (vect_maybe_permute_loop_masks): Likewise.
25995 (vect_set_loop_masks_directly): Likewise.
25996 (vect_set_loop_condition_masked): Likewise.
25997 (vect_set_loop_condition_unmasked): New function, split out from
25998 slpeel_make_loop_iterate_ntimes.
25999 (slpeel_make_loop_iterate_ntimes): Rename to..
26000 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
26001 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
26002 (vect_do_peeling): Update call accordingly.
26003 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
26005 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
26006 mask_compare_type, can_fully_mask_p and fully_masked_p.
26007 (release_vec_loop_masks): New function.
26008 (_loop_vec_info): Use it to free the loop masks.
26009 (can_produce_all_loop_masks_p): New function.
26010 (vect_get_max_nscalars_per_iter): Likewise.
26011 (vect_verify_full_masking): Likewise.
26012 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
26013 retries, and free the mask rgroups before retrying. Check loop-wide
26014 reasons for disallowing fully-masked loops. Make the final decision
26015 about whether use a fully-masked loop or not.
26016 (vect_estimate_min_profitable_iters): Do not assume that peeling
26017 for the number of iterations will be needed for fully-masked loops.
26018 (vectorizable_reduction): Disable fully-masked loops.
26019 (vectorizable_live_operation): Likewise.
26020 (vect_halve_mask_nunits): New function.
26021 (vect_double_mask_nunits): Likewise.
26022 (vect_record_loop_mask): Likewise.
26023 (vect_get_loop_mask): Likewise.
26024 (vect_transform_loop): Handle the case in which the final loop
26025 iteration might handle a partial vector. Call vect_set_loop_condition
26026 instead of slpeel_make_loop_iterate_ntimes.
26027 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
26028 (check_load_store_masking): New function.
26029 (prepare_load_store_mask): Likewise.
26030 (vectorizable_store): Handle fully-masked loops.
26031 (vectorizable_load): Likewise.
26032 (supportable_widening_operation): Use vect_halve_mask_nunits for
26034 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
26035 (vect_gen_while): New function.
26036 * config/aarch64/aarch64.md (umax<mode>3): New expander.
26037 (aarch64_uqdec<mode>): New insn.
26039 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26040 Alan Hayward <alan.hayward@arm.com>
26041 David Sherwood <david.sherwood@arm.com>
26043 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
26044 (reduc_xor_scal_optab): New optabs.
26045 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
26046 (reduc_xor_scal_@var{m}): Document.
26047 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
26048 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
26049 internal functions.
26050 * fold-const-call.c (fold_const_call): Handle them.
26051 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
26052 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
26053 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
26054 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
26055 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
26056 (UNSPEC_XORV): New unspecs.
26057 (optab): Add entries for them.
26058 (BITWISEV): New int iterator.
26059 (bit_reduc_op): New int attributes.
26061 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26062 Alan Hayward <alan.hayward@arm.com>
26063 David Sherwood <david.sherwood@arm.com>
26065 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
26066 * internal-fn.def (VEC_SHL_INSERT): New internal function.
26067 * optabs.def (vec_shl_insert_optab): New optab.
26068 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
26069 (duplicate_and_interleave): Likewise.
26070 * tree-vect-loop.c: Include internal-fn.h.
26071 (neutral_op_for_slp_reduction): New function, split out from
26072 get_initial_defs_for_reduction.
26073 (get_initial_def_for_reduction): Handle option 2 for variable-length
26074 vectors by loading the neutral value into a vector and then shifting
26075 the initial value into element 0.
26076 (get_initial_defs_for_reduction): Replace the code argument with
26077 the neutral value calculated by neutral_op_for_slp_reduction.
26078 Use gimple_build_vector for constant-length vectors.
26079 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
26080 but the first group_size elements have a neutral value.
26081 Use duplicate_and_interleave otherwise.
26082 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
26083 Update call to get_initial_defs_for_reduction. Handle SLP
26084 reductions for variable-length vectors by creating one vector
26085 result for each scalar result, with the elements associated
26086 with other scalar results stubbed out with the neutral value.
26087 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
26088 Require IFN_VEC_SHL_INSERT for double reductions on
26089 variable-length vectors, or SLP reductions that have
26090 a neutral value. Require can_duplicate_and_interleave_p
26091 support for variable-length unchained SLP reductions if there
26092 is no neutral value, such as for MIN/MAX reductions. Also require
26093 the number of vector elements to be a multiple of the number of
26094 SLP statements when doing variable-length unchained SLP reductions.
26095 Update call to vect_create_epilog_for_reduction.
26096 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
26097 and remove initial values.
26098 (duplicate_and_interleave): Make public.
26099 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
26100 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
26102 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26103 Alan Hayward <alan.hayward@arm.com>
26104 David Sherwood <david.sherwood@arm.com>
26106 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
26107 (can_duplicate_and_interleave_p): New function.
26108 (vect_get_and_check_slp_defs): Take the vector of statements
26109 rather than just the current one. Remove excess parentheses.
26110 Restriction rejectinon of vect_constant_def and vect_external_def
26111 for variable-length vectors to boolean types, or types for which
26112 can_duplicate_and_interleave_p is false.
26113 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
26114 (duplicate_and_interleave): New function.
26115 (vect_get_constant_vectors): Use gimple_build_vector for
26116 constant-length vectors and suitable variable-length constant
26117 vectors. Use duplicate_and_interleave for other variable-length
26118 vectors. Don't defer the update when inserting new statements.
26120 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26121 Alan Hayward <alan.hayward@arm.com>
26122 David Sherwood <david.sherwood@arm.com>
26124 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
26125 min_profitable_iters doesn't go negative.
26127 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26128 Alan Hayward <alan.hayward@arm.com>
26129 David Sherwood <david.sherwood@arm.com>
26131 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
26132 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
26133 * optabs.def (vec_mask_load_lanes_optab): New optab.
26134 (vec_mask_store_lanes_optab): Likewise.
26135 * internal-fn.def (MASK_LOAD_LANES): New internal function.
26136 (MASK_STORE_LANES): Likewise.
26137 * internal-fn.c (mask_load_lanes_direct): New macro.
26138 (mask_store_lanes_direct): Likewise.
26139 (expand_mask_load_optab_fn): Handle masked operations.
26140 (expand_mask_load_lanes_optab_fn): New macro.
26141 (expand_mask_store_optab_fn): Handle masked operations.
26142 (expand_mask_store_lanes_optab_fn): New macro.
26143 (direct_mask_load_lanes_optab_supported_p): Likewise.
26144 (direct_mask_store_lanes_optab_supported_p): Likewise.
26145 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
26147 (vect_load_lanes_supported): Likewise.
26148 * tree-vect-data-refs.c (strip_conversion): New function.
26149 (can_group_stmts_p): Likewise.
26150 (vect_analyze_data_ref_accesses): Use it instead of checking
26151 for a pair of assignments.
26152 (vect_store_lanes_supported): Take a masked_p parameter.
26153 (vect_load_lanes_supported): Likewise.
26154 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
26155 vect_store_lanes_supported and vect_load_lanes_supported.
26156 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
26157 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
26158 parameter. Don't allow gaps for masked accesses.
26159 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
26160 and vect_load_lanes_supported.
26161 (get_load_store_type): Take a masked_p parameter and update
26162 call to get_group_load_store_type.
26163 (vectorizable_store): Update call to get_load_store_type.
26164 Handle IFN_MASK_STORE_LANES.
26165 (vectorizable_load): Update call to get_load_store_type.
26166 Handle IFN_MASK_LOAD_LANES.
26168 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26169 Alan Hayward <alan.hayward@arm.com>
26170 David Sherwood <david.sherwood@arm.com>
26172 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
26174 * config/aarch64/aarch64-protos.h
26175 (aarch64_sve_struct_memory_operand_p): Declare.
26176 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
26177 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
26178 (VPRED, vpred): Handle SVE structure modes.
26179 * config/aarch64/constraints.md (Utx): New constraint.
26180 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
26181 (aarch64_sve_struct_nonimmediate_operand): New predicates.
26182 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
26183 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
26184 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
26185 structure modes. Split into pieces after RA.
26186 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
26187 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
26189 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
26190 SVE structure modes.
26191 (aarch64_classify_address): Likewise.
26192 (sizetochar): Move earlier in file.
26193 (aarch64_print_operand): Handle SVE register lists.
26194 (aarch64_array_mode): New function.
26195 (aarch64_sve_struct_memory_operand_p): Likewise.
26196 (TARGET_ARRAY_MODE): Redefine.
26198 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26199 Alan Hayward <alan.hayward@arm.com>
26200 David Sherwood <david.sherwood@arm.com>
26202 * target.def (array_mode): New target hook.
26203 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
26204 * doc/tm.texi: Regenerate.
26205 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
26206 * hooks.c (hook_optmode_mode_uhwi_none): New function.
26207 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
26208 targetm.array_mode.
26209 * stor-layout.c (mode_for_array): Likewise. Support polynomial
26212 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26213 Alan Hayward <alan.hayward@arm.com>
26214 David Sherwood <david.sherwood@arm.com>
26216 * fold-const.c (fold_binary_loc): Check the argument types
26217 rather than the result type when testing for a vector operation.
26219 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26221 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
26222 * doc/tm.texi: Regenerate.
26224 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26225 Alan Hayward <alan.hayward@arm.com>
26226 David Sherwood <david.sherwood@arm.com>
26228 * doc/invoke.texi (-msve-vector-bits=): Document new option.
26229 (sve): Document new AArch64 extension.
26230 * doc/md.texi (w): Extend the description of the AArch64
26231 constraint to include SVE vectors.
26232 (Upl, Upa): Document new AArch64 predicate constraints.
26233 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
26235 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
26236 (msve-vector-bits=): New option.
26237 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
26238 SVE when these are disabled.
26239 (sve): New extension.
26240 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
26241 modes. Adjust their number of units based on aarch64_sve_vg.
26242 (MAX_BITSIZE_MODE_ANY_MODE): Define.
26243 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
26244 aarch64_addr_query_type.
26245 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
26246 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
26247 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
26248 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
26249 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
26250 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
26251 (aarch64_simd_imm_zero_p): Delete.
26252 (aarch64_check_zero_based_sve_index_immediate): Declare.
26253 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
26254 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
26255 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
26256 (aarch64_sve_float_mul_immediate_p): Likewise.
26257 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
26258 rather than an rtx.
26259 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
26260 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
26261 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
26262 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
26263 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
26264 (aarch64_regmode_natural_size): Likewise.
26265 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
26266 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
26268 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
26269 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
26270 for VG and the SVE predicate registers.
26271 (V_ALIASES): Add a "z"-prefixed alias.
26272 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
26273 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
26274 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
26275 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
26276 (REG_CLASS_NAMES): Add entries for them.
26277 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
26278 and the predicate registers.
26279 (aarch64_sve_vg): Declare.
26280 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
26281 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
26282 (REGMODE_NATURAL_SIZE): Define.
26283 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
26285 * config/aarch64/aarch64.c: Include cfgrtl.h.
26286 (simd_immediate_info): Add a constructor for series vectors,
26287 and an associated step field.
26288 (aarch64_sve_vg): New variable.
26289 (aarch64_dbx_register_number): Handle VG and the predicate registers.
26290 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
26291 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
26292 (VEC_ANY_DATA, VEC_STRUCT): New constants.
26293 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
26294 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
26295 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
26296 (aarch64_get_mask_mode): New functions.
26297 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
26298 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
26299 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
26300 predicate modes and predicate registers. Explicitly restrict
26301 GPRs to modes of 16 bytes or smaller. Only allow FP registers
26302 to store a vector mode if it is recognized by
26303 aarch64_classify_vector_mode.
26304 (aarch64_regmode_natural_size): New function.
26305 (aarch64_hard_regno_caller_save_mode): Return the original mode
26307 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
26308 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
26309 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
26310 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
26312 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
26313 does not overlap dest if the function is frame-related. Handle
26315 (aarch64_split_add_offset): New function.
26316 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
26317 them aarch64_add_offset.
26318 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
26319 and update call to aarch64_sub_sp.
26320 (aarch64_add_cfa_expression): New function.
26321 (aarch64_expand_prologue): Pass extra temporary registers to the
26322 functions above. Handle the case in which we need to emit new
26323 DW_CFA_expressions for registers that were originally saved
26324 relative to the stack pointer, but now have to be expressed
26325 relative to the frame pointer.
26326 (aarch64_output_mi_thunk): Pass extra temporary registers to the
26328 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
26329 IP0 and IP1 values for SVE frames.
26330 (aarch64_expand_vec_series): New function.
26331 (aarch64_expand_sve_widened_duplicate): Likewise.
26332 (aarch64_expand_sve_const_vector): Likewise.
26333 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
26334 Handle SVE constants. Use emit_move_insn to move a force_const_mem
26335 into the register, rather than emitting a SET directly.
26336 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
26337 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
26338 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
26339 (offset_9bit_signed_scaled_p): New functions.
26340 (aarch64_replicate_bitmask_imm): New function.
26341 (aarch64_bitmask_imm): Use it.
26342 (aarch64_cannot_force_const_mem): Reject expressions involving
26343 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
26344 (aarch64_classify_index): Handle SVE indices, by requiring
26345 a plain register index with a scale that matches the element size.
26346 (aarch64_classify_address): Handle SVE addresses. Assert that
26347 the mode of the address is VOIDmode or an integer mode.
26348 Update call to aarch64_classify_symbol.
26349 (aarch64_classify_symbolic_expression): Update call to
26350 aarch64_classify_symbol.
26351 (aarch64_const_vec_all_in_range_p): New function.
26352 (aarch64_print_vector_float_operand): Likewise.
26353 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
26354 "vN" for FP registers with SVE modes. Handle (const ...) vectors
26355 and the FP immediates 1.0 and 0.5.
26356 (aarch64_print_address_internal): Handle SVE addresses.
26357 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
26358 (aarch64_regno_regclass): Handle predicate registers.
26359 (aarch64_secondary_reload): Handle big-endian reloads of SVE
26361 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
26362 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
26363 (aarch64_convert_sve_vector_bits): New function.
26364 (aarch64_override_options): Use it to handle -msve-vector-bits=.
26365 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
26366 rather than an rtx.
26367 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
26368 Handle SVE vector and predicate modes. Accept VL-based constants
26369 that need only one temporary register, and VL offsets that require
26370 no temporary registers.
26371 (aarch64_conditional_register_usage): Mark the predicate registers
26372 as fixed if SVE isn't available.
26373 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
26374 Return true for SVE vector and predicate modes.
26375 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
26376 rather than an unsigned int. Handle SVE modes.
26377 (aarch64_preferred_simd_mode): Update call accordingly. Handle
26379 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
26381 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
26382 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
26383 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
26384 (aarch64_sve_float_mul_immediate_p): New functions.
26385 (aarch64_sve_valid_immediate): New function.
26386 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
26387 Explicitly reject structure modes. Check for INDEX constants.
26388 Handle PTRUE and PFALSE constants.
26389 (aarch64_check_zero_based_sve_index_immediate): New function.
26390 (aarch64_simd_imm_zero_p): Delete.
26391 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
26392 vector modes. Accept constants in the range of CNT[BHWD].
26393 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
26394 ask for an Advanced SIMD mode.
26395 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
26396 (aarch64_simd_vector_alignment): Handle SVE predicates.
26397 (aarch64_vectorize_preferred_vector_alignment): New function.
26398 (aarch64_simd_vector_alignment_reachable): Use it instead of
26400 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
26401 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
26403 (MAX_VECT_LEN): Delete.
26404 (expand_vec_perm_d): Add a vec_flags field.
26405 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
26406 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
26407 (aarch64_evpc_ext): Don't apply a big-endian lane correction
26409 (aarch64_evpc_rev): Rename to...
26410 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
26411 (aarch64_evpc_rev_global): New function.
26412 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
26413 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
26415 (aarch64_evpc_sve_tbl): New function.
26416 (aarch64_expand_vec_perm_const_1): Update after rename of
26417 aarch64_evpc_rev. Handle SVE permutes too, trying
26418 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
26419 than aarch64_evpc_tbl.
26420 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
26421 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
26422 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
26423 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
26424 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
26425 (aarch64_expand_sve_vcond): New functions.
26426 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
26427 of aarch64_vector_mode_p.
26428 (aarch64_dwarf_poly_indeterminate_value): New function.
26429 (aarch64_compute_pressure_classes): Likewise.
26430 (aarch64_can_change_mode_class): Likewise.
26431 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
26432 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
26433 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
26434 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
26435 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
26436 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
26437 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
26438 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
26440 (Dn, Dl, Dr): Accept const as well as const_vector.
26441 (Dz): Likewise. Compare against CONST0_RTX.
26442 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
26443 of "vector" where appropriate.
26444 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
26445 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
26446 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
26447 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
26448 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
26449 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
26450 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
26451 (v_int_equiv): Extend to SVE modes.
26452 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
26454 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
26455 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
26456 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
26457 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
26458 (SVE_COND_FP_CMP): New int iterators.
26459 (perm_hilo): Handle the new unpack unspecs.
26460 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
26462 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
26463 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
26464 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
26465 (aarch64_equality_operator, aarch64_constant_vector_operand)
26466 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
26467 (aarch64_sve_nonimmediate_operand): Likewise.
26468 (aarch64_sve_general_operand): Likewise.
26469 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
26470 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
26471 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
26472 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
26473 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
26474 (aarch64_sve_float_arith_immediate): Likewise.
26475 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
26476 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
26477 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
26478 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
26479 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
26480 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
26481 (aarch64_sve_float_arith_operand): Likewise.
26482 (aarch64_sve_float_arith_with_sub_operand): Likewise.
26483 (aarch64_sve_float_mul_operand): Likewise.
26484 (aarch64_sve_vec_perm_operand): Likewise.
26485 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
26486 (aarch64_mov_operand): Accept const_poly_int and const_vector.
26487 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
26488 as well as const_vector.
26489 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
26490 in file. Use CONST0_RTX and CONSTM1_RTX.
26491 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
26492 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
26493 Use aarch64_simd_imm_zero.
26494 * config/aarch64/aarch64-sve.md: New file.
26495 * config/aarch64/aarch64.md: Include it.
26496 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
26497 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
26498 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
26499 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
26500 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
26501 (sve): New attribute.
26502 (enabled): Disable instructions with the sve attribute unless
26504 (movqi, movhi): Pass CONST_POLY_INT operaneds through
26505 aarch64_expand_mov_immediate.
26506 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
26507 CNT[BHSD] immediates.
26508 (movti): Split CONST_POLY_INT moves into two halves.
26509 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
26510 Split additions that need a temporary here if the destination
26511 is the stack pointer.
26512 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
26513 (*add<mode>3_poly_1): New instruction.
26514 (set_clobber_cc): New expander.
26516 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26518 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
26519 parameter and use it instead of GET_MODE_SIZE (innermode). Use
26520 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
26521 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
26522 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
26523 Change innermode from fixed_mode_size to machine_mode.
26524 (simplify_subreg): Update call accordingly. Handle a constant-sized
26525 subreg of a variable-length CONST_VECTOR.
26527 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26528 Alan Hayward <alan.hayward@arm.com>
26529 David Sherwood <david.sherwood@arm.com>
26531 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
26532 (add_offset_to_base): New function, split out from...
26533 (create_mem_ref): ...here. When handling a scale other than 1,
26534 check first whether the address is valid without the offset.
26535 Add it into the base if so, leaving the index and scale as-is.
26537 2018-01-12 Jakub Jelinek <jakub@redhat.com>
26540 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
26541 fold_for_warn before checking if arg2 is INTEGER_CST.
26543 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
26545 * config/rs6000/predicates.md (load_multiple_operation): Delete.
26546 (store_multiple_operation): Delete.
26547 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
26548 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
26549 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
26550 guarded by TARGET_STRING.
26551 (rs6000_output_load_multiple): Delete.
26552 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26553 OPTION_MASK_STRING / TARGET_STRING handling.
26554 (print_operand) <'N', 'O'>: Add comment that these are unused now.
26555 (const rs6000_opt_masks) <"string">: Change mask to 0.
26556 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
26557 (MASK_STRING): Delete.
26558 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
26560 (load_multiple): Delete.
26567 (store_multiple): Delete.
26574 (movmemsi_8reg): Delete.
26575 (corresponding unnamed define_insn): Delete.
26576 (movmemsi_6reg): Delete.
26577 (corresponding unnamed define_insn): Delete.
26578 (movmemsi_4reg): Delete.
26579 (corresponding unnamed define_insn): Delete.
26580 (movmemsi_2reg): Delete.
26581 (corresponding unnamed define_insn): Delete.
26582 (movmemsi_1reg): Delete.
26583 (corresponding unnamed define_insn): Delete.
26584 * config/rs6000/rs6000.opt (mno-string): New.
26585 (mstring): Replace by deprecation warning stub.
26586 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
26588 2018-01-12 Jakub Jelinek <jakub@redhat.com>
26590 * regrename.c (regrename_do_replace): If replacing the same
26591 reg multiple times, try to reuse last created gen_raw_REG.
26594 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
26595 main to workaround a bug in GDB.
26597 2018-01-12 Tom de Vries <tom@codesourcery.com>
26600 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
26602 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
26604 PR rtl-optimization/80481
26605 * ira-color.c (get_cap_member): New function.
26606 (allocnos_conflict_by_live_ranges_p): Use it.
26607 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
26608 (setup_slot_coalesced_allocno_live_ranges): Ditto.
26610 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
26613 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
26614 (*saddl_se_1): Ditto.
26615 (*ssubsi_1): Ditto.
26616 (*ssubl_se_1): Ditto.
26618 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
26620 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
26621 rather than wi::to_widest for DR_INITs.
26622 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
26623 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
26624 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
26626 (vect_analyze_group_access_1): Note that here.
26628 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
26630 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
26631 polynomial type sizes.
26633 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
26635 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
26636 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
26637 (gimple_add_tmp_var): Likewise.
26639 2018-01-12 Martin Liska <mliska@suse.cz>
26641 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
26642 (gimple_alloc_sizes): Likewise.
26643 (dump_gimple_statistics): Use PRIu64 in printf format.
26644 * gimple.h: Change uint64_t to int.
26646 2018-01-12 Martin Liska <mliska@suse.cz>
26648 * tree-core.h: Use uint64_t instead of int.
26649 * tree.c (tree_node_counts): Likewise.
26650 (tree_node_sizes): Likewise.
26651 (dump_tree_statistics): Use PRIu64 in printf format.
26653 2018-01-12 Martin Liska <mliska@suse.cz>
26655 * Makefile.in: As qsort_chk is implemented in vec.c, add
26656 vec.o to linkage of gencfn-macros.
26657 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
26658 passing the info to record_node_allocation_statistics.
26659 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
26661 * ggc-common.c (struct ggc_usage): Add operator== and use
26662 it in operator< and compare function.
26663 * mem-stats.h (struct mem_usage): Likewise.
26664 * vec.c (struct vec_usage): Remove operator< and compare
26665 function. Can be simply inherited.
26667 2018-01-12 Martin Jambor <mjambor@suse.cz>
26670 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
26671 * tree-ssa-math-opts.c: Include domwalk.h.
26672 (convert_mult_to_fma_1): New function.
26673 (fma_transformation_info): New type.
26674 (fma_deferring_state): Likewise.
26675 (cancel_fma_deferring): New function.
26676 (result_of_phi): Likewise.
26677 (last_fma_candidate_feeds_initial_phi): Likewise.
26678 (convert_mult_to_fma): Added deferring logic, split actual
26679 transformation to convert_mult_to_fma_1.
26680 (math_opts_dom_walker): New type.
26681 (math_opts_dom_walker::after_dom_children): New method, body moved
26682 here from pass_optimize_widening_mul::execute, added deferring logic
26684 (pass_optimize_widening_mul::execute): Moved most of code to
26685 math_opts_dom_walker::after_dom_children.
26686 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
26687 * config/i386/i386.c (ix86_option_override_internal): Added
26688 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
26690 2018-01-12 Richard Biener <rguenther@suse.de>
26693 * dwarf2out.c (gen_variable_die): Do not reset old_die for
26694 inline instance vars.
26696 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
26699 * config/rx/rx.c (rx_is_restricted_memory_address):
26700 Handle SUBREG case.
26702 2018-01-12 Richard Biener <rguenther@suse.de>
26704 PR tree-optimization/80846
26705 * target.def (split_reduction): New target hook.
26706 * targhooks.c (default_split_reduction): New function.
26707 * targhooks.h (default_split_reduction): Declare.
26708 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
26709 target requests first reduce vectors by combining low and high
26711 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
26712 (get_vectype_for_scalar_type_and_size): Export.
26713 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
26714 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
26715 * doc/tm.texi: Regenerate.
26716 * config/i386/i386.c (ix86_split_reduction): Implement
26717 TARGET_VECTORIZE_SPLIT_REDUCTION.
26719 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
26722 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
26723 in PIC mode except for TARGET_VXWORKS_RTP.
26724 * config/sparc/sparc.c: Include cfgrtl.h.
26725 (TARGET_INIT_PIC_REG): Define.
26726 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
26727 (sparc_pic_register_p): New predicate.
26728 (sparc_legitimate_address_p): Use it.
26729 (sparc_legitimize_pic_address): Likewise.
26730 (sparc_delegitimize_address): Likewise.
26731 (sparc_mode_dependent_address_p): Likewise.
26732 (gen_load_pcrel_sym): Remove 4th parameter.
26733 (load_got_register): Adjust call to above. Remove obsolete stuff.
26734 (sparc_expand_prologue): Do not call load_got_register here.
26735 (sparc_flat_expand_prologue): Likewise.
26736 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
26737 (sparc_use_pseudo_pic_reg): New function.
26738 (sparc_init_pic_reg): Likewise.
26739 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
26740 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
26742 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
26744 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26745 Add item for branch_cost.
26747 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
26749 PR rtl-optimization/83565
26750 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
26751 not extend the result to a larger mode for rotate operations.
26752 (num_sign_bit_copies1): Likewise.
26754 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26757 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
26759 Use values-Xc.o for -pedantic.
26760 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
26762 2018-01-12 Martin Liska <mliska@suse.cz>
26765 * ipa-devirt.c (final_warning_record::grow_type_warnings):
26767 (possible_polymorphic_call_targets): Use it.
26768 (ipa_devirt): Likewise.
26770 2018-01-12 Martin Liska <mliska@suse.cz>
26772 * profile-count.h (enum profile_quality): Use 0 as invalid
26773 enum value of profile_quality.
26775 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
26777 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
26778 -mext-string options.
26780 2018-01-12 Richard Biener <rguenther@suse.de>
26782 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
26783 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
26784 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
26786 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
26788 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
26790 * configure.ac (--with-long-double-format): Add support for the
26791 configuration option to change the default long double format on
26793 * config.gcc (powerpc*-linux*-*): Likewise.
26794 * configure: Regenerate.
26795 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
26796 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
26797 used without modification.
26799 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26801 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
26802 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
26803 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
26804 MISC_BUILTIN_SPEC_BARRIER.
26805 (rs6000_init_builtins): Likewise.
26806 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
26808 (speculation_barrier): New define_insn.
26809 * doc/extend.texi: Document __builtin_speculation_barrier.
26811 2018-01-11 Jakub Jelinek <jakub@redhat.com>
26814 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
26815 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
26816 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
26818 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
26819 integral modes instead of "ss" and "sd".
26820 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
26821 vectors with 32-bit and 64-bit elements.
26822 (vecdupssescalarmodesuffix): New mode attribute.
26823 (vec_dup<mode>): Use it.
26825 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
26828 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
26829 frame if argument is passed on stack.
26831 2018-01-11 Jakub Jelinek <jakub@redhat.com>
26834 * ree.c (combine_reaching_defs): Optimize also
26835 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
26836 reg2=any_extend(exp); reg1=reg2;, formatting fix.
26838 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
26840 PR middle-end/83189
26841 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
26843 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
26845 PR middle-end/83718
26846 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
26847 after they are computed.
26849 2018-01-11 Bin Cheng <bin.cheng@arm.com>
26851 PR tree-optimization/83695
26852 * gimple-loop-linterchange.cc
26853 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
26854 reset cached scev information after interchange.
26855 (pass_linterchange::execute): Remove call to scev_reset_htab.
26857 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26859 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
26860 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
26861 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
26862 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
26863 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
26864 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
26865 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
26866 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
26867 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
26868 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
26869 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
26870 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
26871 (V_lane_reg): Likewise.
26872 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
26874 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
26875 (vfmal_lane_low<mode>_intrinsic,
26876 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
26877 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
26878 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
26879 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
26880 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
26881 vfmsl_lane_high<mode>_intrinsic): New define_insns.
26883 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26885 * config/arm/arm-cpus.in (fp16fml): New feature.
26886 (ALL_SIMD): Add fp16fml.
26887 (armv8.2-a): Add fp16fml as an option.
26888 (armv8.3-a): Likewise.
26889 (armv8.4-a): Add fp16fml as part of fp16.
26890 * config/arm/arm.h (TARGET_FP16FML): Define.
26891 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
26893 * config/arm/arm-modes.def (V2HF): Define.
26894 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
26895 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
26896 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
26897 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
26898 vfmsl_low, vfmsl_high): New set of builtins.
26899 * config/arm/iterators.md (PLUSMINUS): New code iterator.
26900 (vfml_op): New code attribute.
26901 (VFMLHALVES): New int iterator.
26902 (VFML, VFMLSEL): New mode attributes.
26903 (V_reg): Define mapping for V2HF.
26904 (V_hi, V_lo): New mode attributes.
26905 (VF_constraint): Likewise.
26906 (vfml_half, vfml_half_selector): New int attributes.
26907 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
26909 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
26910 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
26912 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
26913 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
26914 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
26915 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
26917 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
26918 Document new effective target and option set.
26920 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26922 * config/arm/arm-cpus.in (armv8_4): New feature.
26923 (ARMv8_4a): New fgroup.
26924 (armv8.4-a): New arch.
26925 * config/arm/arm-tables.opt: Regenerate.
26926 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
26927 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
26928 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
26929 Add matching rules for -march=armv8.4-a and extensions.
26930 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
26932 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
26935 * config/rx/rx.md (BW): New mode attribute.
26936 (sync_lock_test_and_setsi): Add mode suffix to insn output.
26938 2018-01-11 Richard Biener <rguenther@suse.de>
26940 PR tree-optimization/83435
26941 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
26942 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
26943 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
26945 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
26946 Alan Hayward <alan.hayward@arm.com>
26947 David Sherwood <david.sherwood@arm.com>
26949 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
26951 (aarch64_classify_address): Initialize it. Track polynomial offsets.
26952 (aarch64_print_address_internal): Use it to check for a zero offset.
26954 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
26955 Alan Hayward <alan.hayward@arm.com>
26956 David Sherwood <david.sherwood@arm.com>
26958 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
26959 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
26960 Return a poly_int64 rather than a HOST_WIDE_INT.
26961 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
26962 rather than a HOST_WIDE_INT.
26963 * config/aarch64/aarch64.h (aarch64_frame): Protect with
26964 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
26965 hard_fp_offset, frame_size, initial_adjust, callee_offset and
26966 final_offset from HOST_WIDE_INT to poly_int64.
26967 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
26968 to_constant when getting the number of units in an Advanced SIMD
26970 (aarch64_builtin_vectorized_function): Check for a constant number
26972 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
26974 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
26975 attribute instead of GET_MODE_NUNITS.
26976 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
26977 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
26978 GET_MODE_SIZE for fixed-size registers.
26979 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
26980 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
26981 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
26982 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
26983 (aarch64_print_operand, aarch64_print_address_internal)
26984 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
26985 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
26986 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
26987 Handle polynomial GET_MODE_SIZE.
26988 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
26989 wider than SImode without modification.
26990 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
26991 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
26992 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
26993 passing and returning SVE modes.
26994 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
26995 rather than GEN_INT.
26996 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
26997 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
26998 (aarch64_allocate_and_probe_stack_space): Likewise.
26999 (aarch64_layout_frame): Cope with polynomial offsets.
27000 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
27001 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
27002 polynomial offsets.
27003 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
27004 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
27005 poly_int64 rather than a HOST_WIDE_INT.
27006 (aarch64_get_separate_components, aarch64_process_components)
27007 (aarch64_expand_prologue, aarch64_expand_epilogue)
27008 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
27009 (aarch64_anchor_offset): New function, split out from...
27010 (aarch64_legitimize_address): ...here.
27011 (aarch64_builtin_vectorization_cost): Handle polynomial
27012 TYPE_VECTOR_SUBPARTS.
27013 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
27015 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
27016 number of elements from the PARALLEL rather than the mode.
27017 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
27018 rather than GET_MODE_BITSIZE.
27019 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
27020 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
27021 (aarch64_expand_vec_perm_const_1): Handle polynomial
27022 d->perm.length () and d->perm elements.
27023 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
27024 Apply to_constant to d->perm elements.
27025 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
27026 polynomial CONST_VECTOR_NUNITS.
27027 (aarch64_move_pointer): Take amount as a poly_int64 rather
27029 (aarch64_progress_pointer): Avoid temporary variable.
27030 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
27031 the mode attribute instead of GET_MODE.
27033 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27034 Alan Hayward <alan.hayward@arm.com>
27035 David Sherwood <david.sherwood@arm.com>
27037 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
27038 x exists before using it.
27039 (aarch64_add_constant_internal): Rename to...
27040 (aarch64_add_offset_1): ...this. Replace regnum with separate
27041 src and dest rtxes. Handle the case in which they're different,
27042 including when the offset is zero. Replace scratchreg with an rtx.
27043 Use 2 additions if there is no spare register into which we can
27044 move a 16-bit constant.
27045 (aarch64_add_constant): Delete.
27046 (aarch64_add_offset): Replace reg with separate src and dest
27047 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
27048 Use aarch64_add_offset_1.
27049 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
27050 an rtx rather than an int. Take the delta as a poly_int64
27051 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
27052 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
27053 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
27054 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
27055 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
27056 and aarch64_add_sp.
27057 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
27058 aarch64_add_constant.
27060 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27062 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
27063 Use scalar_float_mode.
27065 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27067 * config/aarch64/aarch64-simd.md
27068 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
27069 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
27070 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
27071 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
27072 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
27073 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
27074 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
27075 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
27076 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
27077 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
27079 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27082 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
27083 targ_options->x_arm_arch_string is non NULL.
27085 2018-01-11 Tamar Christina <tamar.christina@arm.com>
27087 * config/aarch64/aarch64.h
27088 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
27090 2018-01-11 Sudakshina Das <sudi.das@arm.com>
27093 * expmed.c (emit_store_flag_force): Swap if const op0
27094 and change VOIDmode to mode of op0.
27096 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27098 PR rtl-optimization/83761
27099 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
27100 than bytes to mode_for_size.
27102 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
27104 PR middle-end/83189
27105 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
27106 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
27109 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
27111 PR middle-end/83575
27112 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
27113 when in layout mode.
27114 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
27115 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
27118 2018-01-10 Michael Collison <michael.collison@arm.com>
27120 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
27121 * config/aarch64/aarch64-option-extension.def: Add
27122 AARCH64_OPT_EXTENSION of 'fp16fml'.
27123 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27124 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
27125 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
27126 * config/aarch64/constraints.md (Ui7): New constraint.
27127 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
27128 (VFMLA_SEL_W): Ditto.
27131 (VFMLA16_LOW): New int iterator.
27132 (VFMLA16_HIGH): Ditto.
27133 (UNSPEC_FMLAL): New unspec.
27134 (UNSPEC_FMLSL): Ditto.
27135 (UNSPEC_FMLAL2): Ditto.
27136 (UNSPEC_FMLSL2): Ditto.
27137 (f16mac): New code attribute.
27138 * config/aarch64/aarch64-simd-builtins.def
27139 (aarch64_fmlal_lowv2sf): Ditto.
27140 (aarch64_fmlsl_lowv2sf): Ditto.
27141 (aarch64_fmlalq_lowv4sf): Ditto.
27142 (aarch64_fmlslq_lowv4sf): Ditto.
27143 (aarch64_fmlal_highv2sf): Ditto.
27144 (aarch64_fmlsl_highv2sf): Ditto.
27145 (aarch64_fmlalq_highv4sf): Ditto.
27146 (aarch64_fmlslq_highv4sf): Ditto.
27147 (aarch64_fmlal_lane_lowv2sf): Ditto.
27148 (aarch64_fmlsl_lane_lowv2sf): Ditto.
27149 (aarch64_fmlal_laneq_lowv2sf): Ditto.
27150 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
27151 (aarch64_fmlalq_lane_lowv4sf): Ditto.
27152 (aarch64_fmlsl_lane_lowv4sf): Ditto.
27153 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
27154 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
27155 (aarch64_fmlal_lane_highv2sf): Ditto.
27156 (aarch64_fmlsl_lane_highv2sf): Ditto.
27157 (aarch64_fmlal_laneq_highv2sf): Ditto.
27158 (aarch64_fmlsl_laneq_highv2sf): Ditto.
27159 (aarch64_fmlalq_lane_highv4sf): Ditto.
27160 (aarch64_fmlsl_lane_highv4sf): Ditto.
27161 (aarch64_fmlalq_laneq_highv4sf): Ditto.
27162 (aarch64_fmlsl_laneq_highv4sf): Ditto.
27163 * config/aarch64/aarch64-simd.md:
27164 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
27165 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
27166 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
27167 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
27168 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
27169 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
27170 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
27171 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
27172 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
27173 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
27174 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
27175 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
27176 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
27177 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
27178 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
27179 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
27180 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
27181 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
27182 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
27183 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
27184 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
27185 (vfmlsl_low_u32): Ditto.
27186 (vfmlalq_low_u32): Ditto.
27187 (vfmlslq_low_u32): Ditto.
27188 (vfmlal_high_u32): Ditto.
27189 (vfmlsl_high_u32): Ditto.
27190 (vfmlalq_high_u32): Ditto.
27191 (vfmlslq_high_u32): Ditto.
27192 (vfmlal_lane_low_u32): Ditto.
27193 (vfmlsl_lane_low_u32): Ditto.
27194 (vfmlal_laneq_low_u32): Ditto.
27195 (vfmlsl_laneq_low_u32): Ditto.
27196 (vfmlalq_lane_low_u32): Ditto.
27197 (vfmlslq_lane_low_u32): Ditto.
27198 (vfmlalq_laneq_low_u32): Ditto.
27199 (vfmlslq_laneq_low_u32): Ditto.
27200 (vfmlal_lane_high_u32): Ditto.
27201 (vfmlsl_lane_high_u32): Ditto.
27202 (vfmlal_laneq_high_u32): Ditto.
27203 (vfmlsl_laneq_high_u32): Ditto.
27204 (vfmlalq_lane_high_u32): Ditto.
27205 (vfmlslq_lane_high_u32): Ditto.
27206 (vfmlalq_laneq_high_u32): Ditto.
27207 (vfmlslq_laneq_high_u32): Ditto.
27208 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
27209 (AARCH64_FL_FOR_ARCH8_4): New.
27210 (AARCH64_ISA_F16FML): New ISA flag.
27211 (TARGET_F16FML): New feature flag for fp16fml.
27212 (doc/invoke.texi): Document new fp16fml option.
27214 2018-01-10 Michael Collison <michael.collison@arm.com>
27216 * config/aarch64/aarch64-builtins.c:
27217 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
27218 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27219 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
27220 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
27221 (AARCH64_ISA_SHA3): New ISA flag.
27222 (TARGET_SHA3): New feature flag for sha3.
27223 * config/aarch64/iterators.md (sha512_op): New int attribute.
27224 (CRYPTO_SHA512): New int iterator.
27225 (UNSPEC_SHA512H): New unspec.
27226 (UNSPEC_SHA512H2): Ditto.
27227 (UNSPEC_SHA512SU0): Ditto.
27228 (UNSPEC_SHA512SU1): Ditto.
27229 * config/aarch64/aarch64-simd-builtins.def
27230 (aarch64_crypto_sha512hqv2di): New builtin.
27231 (aarch64_crypto_sha512h2qv2di): Ditto.
27232 (aarch64_crypto_sha512su0qv2di): Ditto.
27233 (aarch64_crypto_sha512su1qv2di): Ditto.
27234 (aarch64_eor3qv8hi): Ditto.
27235 (aarch64_rax1qv2di): Ditto.
27236 (aarch64_xarqv2di): Ditto.
27237 (aarch64_bcaxqv8hi): Ditto.
27238 * config/aarch64/aarch64-simd.md:
27239 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
27240 (aarch64_crypto_sha512su0qv2di): Ditto.
27241 (aarch64_crypto_sha512su1qv2di): Ditto.
27242 (aarch64_eor3qv8hi): Ditto.
27243 (aarch64_rax1qv2di): Ditto.
27244 (aarch64_xarqv2di): Ditto.
27245 (aarch64_bcaxqv8hi): Ditto.
27246 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
27247 (vsha512h2q_u64): Ditto.
27248 (vsha512su0q_u64): Ditto.
27249 (vsha512su1q_u64): Ditto.
27250 (veor3q_u16): Ditto.
27251 (vrax1q_u64): Ditto.
27252 (vxarq_u64): Ditto.
27253 (vbcaxq_u16): Ditto.
27254 * config/arm/types.md (crypto_sha512): New type attribute.
27255 (crypto_sha3): Ditto.
27256 (doc/invoke.texi): Document new sha3 option.
27258 2018-01-10 Michael Collison <michael.collison@arm.com>
27260 * config/aarch64/aarch64-builtins.c:
27261 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
27262 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27263 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
27264 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
27265 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
27266 (AARCH64_ISA_SM4): New ISA flag.
27267 (TARGET_SM4): New feature flag for sm4.
27268 * config/aarch64/aarch64-simd-builtins.def
27269 (aarch64_sm3ss1qv4si): Ditto.
27270 (aarch64_sm3tt1aq4si): Ditto.
27271 (aarch64_sm3tt1bq4si): Ditto.
27272 (aarch64_sm3tt2aq4si): Ditto.
27273 (aarch64_sm3tt2bq4si): Ditto.
27274 (aarch64_sm3partw1qv4si): Ditto.
27275 (aarch64_sm3partw2qv4si): Ditto.
27276 (aarch64_sm4eqv4si): Ditto.
27277 (aarch64_sm4ekeyqv4si): Ditto.
27278 * config/aarch64/aarch64-simd.md:
27279 (aarch64_sm3ss1qv4si): Ditto.
27280 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
27281 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
27282 (aarch64_sm4eqv4si): Ditto.
27283 (aarch64_sm4ekeyqv4si): Ditto.
27284 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
27285 (sm3part_op): Ditto.
27286 (CRYPTO_SM3TT): Ditto.
27287 (CRYPTO_SM3PART): Ditto.
27288 (UNSPEC_SM3SS1): New unspec.
27289 (UNSPEC_SM3TT1A): Ditto.
27290 (UNSPEC_SM3TT1B): Ditto.
27291 (UNSPEC_SM3TT2A): Ditto.
27292 (UNSPEC_SM3TT2B): Ditto.
27293 (UNSPEC_SM3PARTW1): Ditto.
27294 (UNSPEC_SM3PARTW2): Ditto.
27295 (UNSPEC_SM4E): Ditto.
27296 (UNSPEC_SM4EKEY): Ditto.
27297 * config/aarch64/constraints.md (Ui2): New constraint.
27298 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
27299 * config/arm/types.md (crypto_sm3): New type attribute.
27300 (crypto_sm4): Ditto.
27301 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
27302 (vsm3tt1aq_u32): Ditto.
27303 (vsm3tt1bq_u32): Ditto.
27304 (vsm3tt2aq_u32): Ditto.
27305 (vsm3tt2bq_u32): Ditto.
27306 (vsm3partw1q_u32): Ditto.
27307 (vsm3partw2q_u32): Ditto.
27308 (vsm4eq_u32): Ditto.
27309 (vsm4ekeyq_u32): Ditto.
27310 (doc/invoke.texi): Document new sm4 option.
27312 2018-01-10 Michael Collison <michael.collison@arm.com>
27314 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
27315 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
27316 (AARCH64_FL_FOR_ARCH8_4): New.
27317 (AARCH64_FL_V8_4): New flag.
27318 (doc/invoke.texi): Document new armv8.4-a option.
27320 2018-01-10 Michael Collison <michael.collison@arm.com>
27322 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27323 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
27324 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
27325 * config/aarch64/aarch64-option-extension.def: Add
27326 AARCH64_OPT_EXTENSION of 'sha2'.
27327 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
27328 (crypto): Disable sha2 and aes if crypto disabled.
27329 (crypto): Enable aes and sha2 if enabled.
27330 (simd): Disable sha2 and aes if simd disabled.
27331 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
27333 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
27334 (TARGET_SHA2): New feature flag for sha2.
27335 (TARGET_AES): New feature flag for aes.
27336 * config/aarch64/aarch64-simd.md:
27337 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
27338 conditional on TARGET_AES.
27339 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
27340 (aarch64_crypto_sha1hsi): Make pattern conditional
27342 (aarch64_crypto_sha1hv4si): Ditto.
27343 (aarch64_be_crypto_sha1hv4si): Ditto.
27344 (aarch64_crypto_sha1su1v4si): Ditto.
27345 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
27346 (aarch64_crypto_sha1su0v4si): Ditto.
27347 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
27348 (aarch64_crypto_sha256su0v4si): Ditto.
27349 (aarch64_crypto_sha256su1v4si): Ditto.
27350 (doc/invoke.texi): Document new aes and sha2 options.
27352 2018-01-10 Martin Sebor <msebor@redhat.com>
27354 PR tree-optimization/83781
27355 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
27358 2018-01-11 Martin Sebor <msebor@gmail.com>
27359 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27361 PR tree-optimization/83501
27362 PR tree-optimization/81703
27364 * tree-ssa-strlen.c (get_string_cst): Rename...
27365 (get_string_len): ...to this. Handle global constants.
27366 (handle_char_store): Adjust.
27368 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
27369 Jim Wilson <jimw@sifive.com>
27371 * config/riscv/riscv-protos.h (riscv_output_return): New.
27372 * config/riscv/riscv.c (struct machine_function): New naked_p field.
27373 (riscv_attribute_table, riscv_output_return),
27374 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
27375 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
27376 (riscv_compute_frame_info): Only compute frame->mask if not a naked
27378 (riscv_expand_prologue): Add early return for naked function.
27379 (riscv_expand_epilogue): Likewise.
27380 (riscv_function_ok_for_sibcall): Return false for naked function.
27381 (riscv_set_current_function): New.
27382 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
27383 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
27384 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
27385 * doc/extend.texi (RISC-V Function Attributes): New.
27387 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
27389 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
27390 check for 128-bit long double before checking TCmode.
27391 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
27392 128-bit long doubles before checking TFmode or TCmode.
27393 (FLOAT128_IBM_P): Likewise.
27395 2018-01-10 Martin Sebor <msebor@redhat.com>
27397 PR tree-optimization/83671
27398 * builtins.c (c_strlen): Unconditionally return zero for the empty
27400 Use -Warray-bounds for warnings.
27401 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
27402 for non-constant array indices with COMPONENT_REF, arrays of
27403 arrays, and pointers to arrays.
27404 (gimple_fold_builtin_strlen): Determine and set length range for
27405 non-constant character arrays.
27407 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
27409 PR middle-end/81897
27410 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
27413 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
27415 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
27417 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
27420 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
27421 VECTOR_MEM_ALTIVEC_OR_VSX_P.
27422 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
27423 indexed_or_indirect_operand predicate.
27424 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
27425 (*vsx_le_perm_load_v8hi): Likewise.
27426 (*vsx_le_perm_load_v16qi): Likewise.
27427 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
27428 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
27429 (*vsx_le_perm_store_v8hi): Likewise.
27430 (*vsx_le_perm_store_v16qi): Likewise.
27431 (eight unnamed splitters): Likewise.
27433 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
27435 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
27436 * config/rs6000/emmintrin.h: Likewise.
27437 * config/rs6000/mmintrin.h: Likewise.
27438 * config/rs6000/xmmintrin.h: Likewise.
27440 2018-01-10 David Malcolm <dmalcolm@redhat.com>
27443 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
27445 * tree.c (tree_nop_conversion): Return true for location wrapper
27447 (maybe_wrap_with_location): New function.
27448 (selftest::check_strip_nops): New function.
27449 (selftest::test_location_wrappers): New function.
27450 (selftest::tree_c_tests): Call it.
27451 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
27452 (maybe_wrap_with_location): New decl.
27453 (EXPR_LOCATION_WRAPPER_P): New macro.
27454 (location_wrapper_p): New inline function.
27455 (tree_strip_any_location_wrapper): New inline function.
27457 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
27460 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
27461 stack_realign_offset for the largest alignment of stack slot
27463 (ix86_find_max_used_stack_alignment): New function.
27464 (ix86_finalize_stack_frame_flags): Use it. Set
27465 max_used_stack_alignment if we don't realign stack.
27466 * config/i386/i386.h (machine_function): Add
27467 max_used_stack_alignment.
27469 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
27471 * config/arm/arm.opt (-mbranch-cost): New option.
27472 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
27475 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
27478 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
27479 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
27481 2018-01-10 Richard Biener <rguenther@suse.de>
27484 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
27485 early out so it also covers the case where we have a non-NULL
27488 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
27490 PR tree-optimization/83753
27491 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
27492 for non-strided grouped accesses if the number of elements is 1.
27494 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
27497 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
27498 * i386.h (TARGET_USE_GATHER): Define.
27499 * x86-tune.def (X86_TUNE_USE_GATHER): New.
27501 2018-01-10 Martin Liska <mliska@suse.cz>
27504 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
27505 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
27507 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
27508 CLEANUP_NO_PARTITIONING is not set.
27510 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
27512 * doc/rtl.texi: Remove documentation of (const ...) wrappers
27513 for vectors, as a partial revert of r254296.
27514 * rtl.h (const_vec_p): Delete.
27515 (const_vec_duplicate_p): Don't test for vector CONSTs.
27516 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
27517 * expmed.c (make_tree): Likewise.
27520 * common.md (E, F): Use CONSTANT_P instead of checking for
27522 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
27523 checking for CONST_VECTOR.
27525 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
27527 PR middle-end/83575
27528 * predict.c (force_edge_cold): Handle in more sane way edges
27529 with no prediction.
27531 2018-01-09 Carl Love <cel@us.ibm.com>
27533 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
27535 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
27536 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
27537 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
27538 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
27539 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
27540 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
27541 * config/rs6000/rs6000-protos.h: Add extern defition for
27542 rs6000_generate_float2_double_code.
27543 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
27545 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
27546 (float2_v2df): Add define_expand.
27548 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
27551 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
27552 op_mode in the force_to_mode call.
27554 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
27556 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
27557 instead of checking each element individually.
27558 (aarch64_evpc_uzp): Likewise.
27559 (aarch64_evpc_zip): Likewise.
27560 (aarch64_evpc_ext): Likewise.
27561 (aarch64_evpc_rev): Likewise.
27562 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
27563 instead of checking each element individually. Return true without
27565 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
27566 whether all selected elements come from the same input, instead of
27567 checking each element individually. Remove calls to gen_rtx_REG,
27568 start_sequence and end_sequence and instead assert that no rtl is
27571 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
27573 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
27574 order of HIGH and CONST checks.
27576 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
27578 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
27579 if the destination isn't an SSA_NAME.
27581 2018-01-09 Richard Biener <rguenther@suse.de>
27583 PR tree-optimization/83668
27584 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
27586 (canonicalize_loop_form): ... here, renamed from ...
27587 (canonicalize_loop_closed_ssa_form): ... this and amended to
27588 swap successor edges for loop exit blocks to make us use
27589 the RPO order we need for initial schedule generation.
27591 2018-01-09 Joseph Myers <joseph@codesourcery.com>
27593 PR tree-optimization/64811
27594 * match.pd: When optimizing comparisons with Inf, avoid
27595 introducing or losing exceptions from comparisons with NaN.
27597 2018-01-09 Martin Liska <mliska@suse.cz>
27600 * asan.c (shadow_mem_size): Add gcc_assert.
27602 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
27604 Don't save registers in main().
27607 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
27608 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
27609 * config/avr/avr.c (avr_set_current_function): Don't error if
27610 naked, OS_task or OS_main are specified at the same time.
27611 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
27613 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
27615 * common/config/avr/avr-common.c (avr_option_optimization_table):
27616 Switch on -mmain-is-OS_task for optimizing compilations.
27618 2018-01-09 Richard Biener <rguenther@suse.de>
27620 PR tree-optimization/83572
27621 * graphite.c: Include cfganal.h.
27622 (graphite_transform_loops): Connect infinite loops to exit
27623 and remove fake edges at the end.
27625 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
27627 * ipa-inline.c (edge_badness): Revert accidental checkin.
27629 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
27632 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
27633 symbols; not inline clones.
27635 2018-01-09 Jakub Jelinek <jakub@redhat.com>
27638 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
27639 hard registers. Formatting fixes.
27641 PR preprocessor/83722
27642 * gcc.c (try_generate_repro): Pass
27643 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
27644 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
27647 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
27648 Kito Cheng <kito.cheng@gmail.com>
27650 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
27651 (riscv_leaf_function_p): Delete.
27652 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
27654 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27656 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
27658 (do_ifelse): New function.
27659 (do_isel): New function.
27660 (do_sub3): New function.
27661 (do_add3): New function.
27662 (do_load_mask_compare): New function.
27663 (do_overlap_load_compare): New function.
27664 (expand_compare_loop): New function.
27665 (expand_block_compare): Call expand_compare_loop() when appropriate.
27666 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
27667 option description.
27668 (-mblock-compare-inline-loop-limit): New option.
27670 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27673 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
27674 Reverse order of second and third operands in first alternative.
27675 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
27676 of first and second elements in UNSPEC_VPERMR vector.
27677 (altivec_expand_vec_perm_le): Likewise.
27679 2018-01-08 Jeff Law <law@redhat.com>
27681 PR rtl-optimizatin/81308
27682 * tree-switch-conversion.c (cfg_altered): New file scoped static.
27683 (process_switch): If group_case_labels makes a change, then set
27685 (pass_convert_switch::execute): If a switch is converted, then
27686 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
27688 PR rtl-optimization/81308
27689 * recog.c (split_all_insns): Conditionally cleanup the CFG after
27692 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
27694 PR target/83663 - Revert r255946
27695 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
27696 generation for cases where splatting a value is not useful.
27697 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
27698 across a vec_duplicate and a paradoxical subreg forming a vector
27699 mode to a vec_concat.
27701 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27703 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
27704 -march=armv8.3-a variants.
27705 * config/arm/t-multilib: Likewise.
27706 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
27708 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27710 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
27712 (cceq_ior_compare_complement): Give it a name so I can use it, and
27713 change boolean_or_operator predicate to boolean_operator so it can
27714 be used to generate a crand.
27715 (eqne): New code iterator.
27716 (bd/bd_neg): New code_attrs.
27717 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
27718 a single define_insn.
27719 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
27720 decrement (bdnzt/bdnzf/bdzt/bdzf).
27721 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
27722 with the new names of the branch decrement patterns, and added the
27723 names of the branch decrement conditional patterns.
27725 2018-01-08 Richard Biener <rguenther@suse.de>
27727 PR tree-optimization/83563
27728 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
27731 2018-01-08 Richard Biener <rguenther@suse.de>
27733 PR middle-end/83713
27734 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
27736 2018-01-08 Richard Biener <rguenther@suse.de>
27738 PR tree-optimization/83685
27739 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
27740 references to abnormals.
27742 2018-01-08 Richard Biener <rguenther@suse.de>
27745 * dwarf2out.c (output_indirect_strings): Handle empty
27746 skeleton_debug_str_hash.
27747 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
27749 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
27751 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
27752 (emit_store_direct): Likewise.
27753 (arc_trampoline_adjust_address): Likewise.
27754 (arc_asm_trampoline_template): New function.
27755 (arc_initialize_trampoline): Use asm_trampoline_template.
27756 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
27757 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
27758 * config/arc/arc.md (flush_icache): Delete pattern.
27760 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
27762 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
27763 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
27766 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
27769 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
27770 by not USED_FOR_TARGET.
27771 (make_pass_resolve_sw_modes): Likewise.
27773 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
27775 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
27778 2018-01-08 Richard Biener <rguenther@suse.de>
27780 PR middle-end/83580
27781 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
27783 2018-01-08 Richard Biener <rguenther@suse.de>
27785 PR middle-end/83517
27786 * match.pd ((t * 2) / 2) -> t): Add missing :c.
27788 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
27790 PR middle-end/81897
27791 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
27792 basic blocks with a small number of successors.
27793 (convert_control_dep_chain_into_preds): Improve handling of
27795 (dump_predicates): Split apart into...
27796 (dump_pred_chain): ...here...
27797 (dump_pred_info): ...and here.
27798 (can_one_predicate_be_invalidated_p): Add debugging printfs.
27799 (can_chain_union_be_invalidated_p): Improve check for invalidation
27801 (uninit_uses_cannot_happen): Avoid unnecessary if
27802 convert_control_dep_chain_into_preds yielded nothing.
27804 2018-01-06 Martin Sebor <msebor@redhat.com>
27806 PR tree-optimization/83640
27807 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
27808 subtracting negative offset from size.
27809 (builtin_access::overlap): Adjust offset bounds of the access to fall
27810 within the size of the object if possible.
27812 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
27814 PR rtl-optimization/83699
27815 * expmed.c (extract_bit_field_1): Restrict the vector usage of
27816 extract_bit_field_as_subreg to cases in which the extracted
27817 value is also a vector.
27819 * lra-constraints.c (process_alt_operands): Test for the equivalence
27820 substitutions when detecting a possible reload cycle.
27822 2018-01-06 Jakub Jelinek <jakub@redhat.com>
27825 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
27826 by default if flag_selective_schedling{,2}. Formatting fixes.
27828 PR rtl-optimization/83682
27829 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
27830 if it has non-VECTOR_MODE element mode.
27831 (vec_duplicate_p): Likewise.
27833 PR middle-end/83694
27834 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
27835 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
27837 2018-01-05 Jakub Jelinek <jakub@redhat.com>
27840 * config/i386/i386-builtin.def
27841 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
27842 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
27843 Require also OPTION_MASK_ISA_AVX512F in addition to
27844 OPTION_MASK_ISA_GFNI.
27845 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
27846 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
27847 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
27848 to OPTION_MASK_ISA_GFNI.
27849 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
27850 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
27851 OPTION_MASK_ISA_AVX512BW.
27852 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
27853 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
27854 addition to OPTION_MASK_ISA_GFNI.
27855 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
27856 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
27857 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
27858 to OPTION_MASK_ISA_GFNI.
27859 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
27860 a requirement for all ISAs rather than any of them with a few
27862 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
27864 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
27865 bitmasks to be enabled with 3 exceptions, instead of requiring any
27866 enabled ISA with lots of exceptions.
27867 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
27868 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
27869 Change avx512bw in isa attribute to avx512f.
27870 * config/i386/sgxintrin.h: Add license boilerplate.
27871 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
27872 to __AVX512F__ and __AVX512VL to __AVX512VL__.
27873 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
27874 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
27876 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
27877 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
27878 temporarily sse2 rather than sse if not enabled already.
27881 * config/i386/sse.md (VI248_VLBW): Rename to ...
27882 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
27883 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
27884 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
27885 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
27886 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
27887 mode iterator instead of VI248_VLBW.
27889 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
27891 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
27892 (record_modified): Skip clobbers; add debug output.
27893 (param_change_prob): Use sreal frequencies.
27895 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
27897 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
27898 punt for user-aligned variables.
27900 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
27902 * tree-chrec.c (chrec_contains_symbols): Return true for
27905 2018-01-05 Sudakshina Das <sudi.das@arm.com>
27908 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
27909 of (x|y) == x for BICS pattern.
27911 2018-01-05 Jakub Jelinek <jakub@redhat.com>
27913 PR tree-optimization/83605
27914 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
27915 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
27918 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
27920 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
27921 * config/epiphany/rtems.h: New file.
27923 2018-01-04 Jakub Jelinek <jakub@redhat.com>
27924 Uros Bizjak <ubizjak@gmail.com>
27927 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
27928 QIreg_operand instead of register_operand predicate.
27929 * config/i386/i386.c (ix86_rop_should_change_byte_p,
27930 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
27931 comments instead of -fmitigate[-_]rop.
27933 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27936 * cgraphunit.c (symbol_table::compile): Switch to text_section
27937 before calling assembly_start debug hook.
27938 * run-rtl-passes.c (run_rtl_passes): Likewise.
27941 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27943 * tree-vrp.c (extract_range_from_binary_expr_1): Check
27944 range_int_cst_p rather than !symbolic_range_p before calling
27945 extract_range_from_multiplicative_op_1.
27947 2018-01-04 Jeff Law <law@redhat.com>
27949 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
27950 redundant test in assertion.
27952 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27954 * doc/rtl.texi: Document machine_mode wrapper classes.
27956 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27958 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
27959 using tree_to_uhwi.
27961 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27963 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
27964 the VEC_PERM_EXPR fold to fail.
27966 2018-01-04 Jakub Jelinek <jakub@redhat.com>
27969 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
27970 to switched_sections.
27972 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27975 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
27976 test for d.testing.
27978 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
27981 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
27982 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
27984 2018-01-04 Jakub Jelinek <jakub@redhat.com>
27987 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
27988 is BLKmode and bitpos not zero or mode change is needed.
27990 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27993 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
27996 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
27999 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
28000 instead of MULT rtx. Update all corresponding splitters.
28001 (*saddl_se): Ditto.
28002 (*ssub<modesuffix>): Ditto.
28003 (*ssubl_se): Ditto.
28004 (*cmp_sadd_di): Update split patterns.
28005 (*cmp_sadd_si): Ditto.
28006 (*cmp_sadd_sidi): Ditto.
28007 (*cmp_ssub_di): Ditto.
28008 (*cmp_ssub_si): Ditto.
28009 (*cmp_ssub_sidi): Ditto.
28010 * config/alpha/predicates.md (const23_operand): New predicate.
28011 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
28012 Look for ASHIFT, not MULT inner operand.
28013 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
28015 2018-01-04 Martin Liska <mliska@suse.cz>
28017 PR gcov-profile/83669
28018 * gcov.c (output_intermediate_file): Add version to intermediate
28020 * doc/gcov.texi: Document new field 'version' in intermediate
28021 file format. Fix location of '-k' option of gcov command.
28023 2018-01-04 Martin Liska <mliska@suse.cz>
28026 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
28028 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28030 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
28032 2018-01-03 Martin Sebor <msebor@redhat.com>
28034 PR tree-optimization/83655
28035 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
28036 checking calls with invalid arguments.
28038 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28040 * tree-vect-stmts.c (vect_get_store_rhs): New function.
28041 (vectorizable_mask_load_store): Delete.
28042 (vectorizable_call): Return false for masked loads and stores.
28043 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
28044 instead of gimple_assign_rhs1.
28045 (vectorizable_load): Handle IFN_MASK_LOAD.
28046 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
28048 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28050 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
28052 (vectorizable_mask_load_store): ...here.
28053 (vectorizable_load): ...and here.
28055 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28057 * tree-vect-stmts.c (vect_build_all_ones_mask)
28058 (vect_build_zero_merge_argument): New functions, split out from...
28059 (vectorizable_load): ...here.
28061 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28063 * tree-vect-stmts.c (vect_check_store_rhs): New function,
28065 (vectorizable_mask_load_store): ...here.
28066 (vectorizable_store): ...and here.
28068 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28070 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
28072 (vectorizable_mask_load_store): ...here.
28074 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28076 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
28077 (vect_model_store_cost): Take a vec_load_store_type instead of a
28079 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
28080 (vect_model_store_cost): Take a vec_load_store_type instead of a
28082 (vectorizable_mask_load_store): Update accordingly.
28083 (vectorizable_store): Likewise.
28084 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
28086 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28088 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
28089 IFN_MASK_LOAD calls here rather than...
28090 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
28092 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28093 Alan Hayward <alan.hayward@arm.com>
28094 David Sherwood <david.sherwood@arm.com>
28096 * expmed.c (extract_bit_field_1): For vector extracts,
28097 fall back to extract_bit_field_as_subreg if vec_extract
28100 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28101 Alan Hayward <alan.hayward@arm.com>
28102 David Sherwood <david.sherwood@arm.com>
28104 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
28105 they are variable or constant sized.
28106 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
28107 slots for constant-sized data.
28109 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28110 Alan Hayward <alan.hayward@arm.com>
28111 David Sherwood <david.sherwood@arm.com>
28113 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
28114 handling COND_EXPRs with boolean comparisons, try to find a better
28115 basis for the mask type than the boolean itself.
28117 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28119 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
28120 is calculated and how it can be overridden.
28121 * genmodes.c (max_bitsize_mode_any_mode): New variable.
28122 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
28124 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
28127 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28128 Alan Hayward <alan.hayward@arm.com>
28129 David Sherwood <david.sherwood@arm.com>
28131 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
28132 Remove the mode argument.
28133 (aarch64_simd_valid_immediate): Remove the mode and inverse
28135 * config/aarch64/iterators.md (bitsize): New iterator.
28136 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
28137 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
28138 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
28139 aarch64_simd_valid_immediate.
28140 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
28141 (aarch64_reg_or_bic_imm): Likewise.
28142 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
28143 with an insn_type enum and msl with a modifier_type enum.
28144 Replace element_width with a scalar_mode. Change the shift
28145 to unsigned int. Add constructors for scalar_float_mode and
28146 scalar_int_mode elements.
28147 (aarch64_vect_float_const_representable_p): Delete.
28148 (aarch64_can_const_movi_rtx_p)
28149 (aarch64_simd_scalar_immediate_valid_for_move)
28150 (aarch64_simd_make_constant): Update call to
28151 aarch64_simd_valid_immediate.
28152 (aarch64_advsimd_valid_immediate_hs): New function.
28153 (aarch64_advsimd_valid_immediate): Likewise.
28154 (aarch64_simd_valid_immediate): Remove mode and inverse
28155 arguments. Rewrite to use the above. Use const_vec_duplicate_p
28156 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
28157 and aarch64_float_const_representable_p on the result.
28158 (aarch64_output_simd_mov_immediate): Remove mode argument.
28159 Update call to aarch64_simd_valid_immediate and use of
28160 simd_immediate_info.
28161 (aarch64_output_scalar_simd_mov_immediate): Update call
28164 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28165 Alan Hayward <alan.hayward@arm.com>
28166 David Sherwood <david.sherwood@arm.com>
28168 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
28169 (mode_nunits): Likewise CONST_MODE_NUNITS.
28170 * machmode.def (ADJUST_NUNITS): Document.
28171 * genmodes.c (mode_data::need_nunits_adj): New field.
28172 (blank_mode): Update accordingly.
28173 (adj_nunits): New variable.
28174 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
28176 (emit_mode_size_inline): Set need_bytesize_adj for all modes
28177 listed in adj_nunits.
28178 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
28179 listed in adj_nunits. Don't emit case statements for such modes.
28180 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
28181 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
28182 nothing if adj_nunits is nonnull.
28183 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
28184 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
28185 (emit_mode_fbit): Update use of print_maybe_const_decl.
28186 (emit_move_size): Likewise. Treat the array as non-const
28188 (emit_mode_adjustments): Handle adj_nunits.
28190 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28192 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
28193 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
28194 (VECTOR_MODES): Use it.
28195 (make_vector_modes): Take the prefix as an argument.
28197 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28198 Alan Hayward <alan.hayward@arm.com>
28199 David Sherwood <david.sherwood@arm.com>
28201 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
28202 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
28203 for MODE_VECTOR_BOOL.
28204 * machmode.def (VECTOR_BOOL_MODE): Document.
28205 * genmodes.c (VECTOR_BOOL_MODE): New macro.
28206 (make_vector_bool_mode): New function.
28207 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
28209 * lto-streamer-in.c (lto_input_mode_table): Likewise.
28210 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
28212 * stor-layout.c (int_mode_for_mode): Likewise.
28213 * tree.c (build_vector_type_for_mode): Likewise.
28214 * varasm.c (output_constant_pool_2): Likewise.
28215 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
28216 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
28217 for MODE_VECTOR_BOOL.
28218 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
28219 of mode class checks.
28220 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
28221 instead of a list of mode class checks.
28222 (expand_vector_scalar_condition): Likewise.
28223 (type_for_widest_vector_mode): Handle BImode as an inner mode.
28225 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28226 Alan Hayward <alan.hayward@arm.com>
28227 David Sherwood <david.sherwood@arm.com>
28229 * machmode.h (mode_size): Change from unsigned short to
28231 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
28232 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
28233 or if measurement_type is not polynomial.
28234 (fixed_size_mode::includes_p): Check for constant-sized modes.
28235 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
28236 return a poly_uint16 rather than an unsigned short.
28237 (emit_mode_size): Change the type of mode_size from unsigned short
28238 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
28239 (emit_mode_adjustments): Cope with polynomial vector sizes.
28240 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
28242 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
28244 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
28245 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
28246 * caller-save.c (setup_save_areas): Likewise.
28247 (replace_reg_with_saved_mem): Likewise.
28248 * calls.c (emit_library_call_value_1): Likewise.
28249 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
28250 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
28251 (gen_lowpart_for_combine): Likewise.
28252 * convert.c (convert_to_integer_1): Likewise.
28253 * cse.c (equiv_constant, cse_insn): Likewise.
28254 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
28255 (cselib_subst_to_values): Likewise.
28256 * dce.c (word_dce_process_block): Likewise.
28257 * df-problems.c (df_word_lr_mark_ref): Likewise.
28258 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
28259 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
28260 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
28261 (rtl_for_decl_location): Likewise.
28262 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
28263 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
28264 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
28265 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
28266 (expand_expr_real_1): Likewise.
28267 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
28268 (pad_below): Likewise.
28269 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
28270 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
28271 * ira.c (get_subreg_tracking_sizes): Likewise.
28272 * ira-build.c (ira_create_allocno_objects): Likewise.
28273 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
28274 (ira_sort_regnos_for_alter_reg): Likewise.
28275 * ira-costs.c (record_operand_costs): Likewise.
28276 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
28277 (resolve_simple_move): Likewise.
28278 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
28279 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
28280 (lra_constraints): Likewise.
28281 (CONST_POOL_OK_P): Reject variable-sized modes.
28282 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
28283 (add_pseudo_to_slot, lra_spill): Likewise.
28284 * omp-low.c (omp_clause_aligned_alignment): Likewise.
28285 * optabs-query.c (get_best_extraction_insn): Likewise.
28286 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
28287 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
28288 (expand_mult_highpart, valid_multiword_target_p): Likewise.
28289 * recog.c (offsettable_address_addr_space_p): Likewise.
28290 * regcprop.c (maybe_mode_change): Likewise.
28291 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
28292 * regrename.c (build_def_use): Likewise.
28293 * regstat.c (dump_reg_info): Likewise.
28294 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
28295 (find_reloads, find_reloads_subreg_address): Likewise.
28296 * reload1.c (eliminate_regs_1): Likewise.
28297 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
28298 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
28299 (simplify_binary_operation_1, simplify_subreg): Likewise.
28300 * targhooks.c (default_function_arg_padding): Likewise.
28301 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
28302 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
28303 (verify_gimple_assign_ternary): Likewise.
28304 * tree-inline.c (estimate_move_cost): Likewise.
28305 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
28306 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
28307 (get_address_cost_ainc): Likewise.
28308 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28309 (vect_supportable_dr_alignment): Likewise.
28310 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28311 (vectorizable_reduction): Likewise.
28312 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
28313 (vectorizable_operation, vectorizable_load): Likewise.
28314 * tree.c (build_same_sized_truth_vector_type): Likewise.
28315 * valtrack.c (cleanup_auto_inc_dec): Likewise.
28316 * var-tracking.c (emit_note_insn_var_location): Likewise.
28317 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
28318 (ADDR_VEC_ALIGN): Likewise.
28320 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28321 Alan Hayward <alan.hayward@arm.com>
28322 David Sherwood <david.sherwood@arm.com>
28324 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
28326 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
28327 or if measurement_type is polynomial.
28328 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
28329 * combine.c (make_extraction): Likewise.
28330 * dse.c (find_shift_sequence): Likewise.
28331 * dwarf2out.c (mem_loc_descriptor): Likewise.
28332 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
28333 (extract_bit_field, extract_low_bits): Likewise.
28334 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
28335 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
28336 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
28337 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
28338 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
28339 * reload.c (find_reloads): Likewise.
28340 * reload1.c (alter_reg): Likewise.
28341 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
28342 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
28343 * tree-if-conv.c (predicate_mem_writes): Likewise.
28344 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
28345 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
28346 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
28347 * valtrack.c (dead_debug_insert_temp): Likewise.
28348 * varasm.c (mergeable_constant_section): Likewise.
28349 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
28351 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28352 Alan Hayward <alan.hayward@arm.com>
28353 David Sherwood <david.sherwood@arm.com>
28355 * expr.c (expand_assignment): Cope with polynomial mode sizes
28356 when assigning to a CONCAT.
28358 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28359 Alan Hayward <alan.hayward@arm.com>
28360 David Sherwood <david.sherwood@arm.com>
28362 * machmode.h (mode_precision): Change from unsigned short to
28364 (mode_to_precision): Return a poly_uint16 rather than an unsigned
28366 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
28367 or if measurement_type is not polynomial.
28368 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
28369 in which the mode is already known to be a scalar_int_mode.
28370 * genmodes.c (emit_mode_precision): Change the type of mode_precision
28371 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
28373 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
28374 for GET_MODE_PRECISION.
28375 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
28376 for GET_MODE_PRECISION.
28377 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
28379 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
28380 (expand_field_assignment, make_extraction): Likewise.
28381 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
28382 (get_last_value): Likewise.
28383 * convert.c (convert_to_integer_1): Likewise.
28384 * cse.c (cse_insn): Likewise.
28385 * expr.c (expand_expr_real_1): Likewise.
28386 * lra-constraints.c (simplify_operand_subreg): Likewise.
28387 * optabs-query.c (can_atomic_load_p): Likewise.
28388 * optabs.c (expand_atomic_load): Likewise.
28389 (expand_atomic_store): Likewise.
28390 * ree.c (combine_reaching_defs): Likewise.
28391 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
28392 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
28393 * tree.h (type_has_mode_precision_p): Likewise.
28394 * ubsan.c (instrument_si_overflow): Likewise.
28396 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28397 Alan Hayward <alan.hayward@arm.com>
28398 David Sherwood <david.sherwood@arm.com>
28400 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
28401 polynomial numbers of units.
28402 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
28403 (valid_vector_subparts_p): New function.
28404 (build_vector_type): Remove temporary shim and take the number
28405 of units as a poly_uint64 rather than an int.
28406 (build_opaque_vector_type): Take the number of units as a
28407 poly_uint64 rather than an int.
28408 * tree.c (build_vector_from_ctor): Handle polynomial
28409 TYPE_VECTOR_SUBPARTS.
28410 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
28411 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
28412 (build_vector_from_val): If the number of units is variable,
28413 use build_vec_duplicate_cst for constant operands and
28414 VEC_DUPLICATE_EXPR otherwise.
28415 (make_vector_type): Remove temporary is_constant ().
28416 (build_vector_type, build_opaque_vector_type): Take the number of
28417 units as a poly_uint64 rather than an int.
28418 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
28420 * cfgexpand.c (expand_debug_expr): Likewise.
28421 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
28422 (store_constructor, expand_expr_real_1): Likewise.
28423 (const_scalar_mask_from_tree): Likewise.
28424 * fold-const-call.c (fold_const_reduction): Likewise.
28425 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
28426 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
28427 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
28428 (fold_relational_const): Likewise.
28429 (native_interpret_vector): Likewise. Change the size from an
28430 int to an unsigned int.
28431 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
28432 TYPE_VECTOR_SUBPARTS.
28433 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
28434 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
28435 duplicating a non-constant operand into a variable-length vector.
28436 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
28437 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
28438 * ipa-icf.c (sem_variable::equals): Likewise.
28439 * match.pd: Likewise.
28440 * omp-simd-clone.c (simd_clone_subparts): Likewise.
28441 * print-tree.c (print_node): Likewise.
28442 * stor-layout.c (layout_type): Likewise.
28443 * targhooks.c (default_builtin_vectorization_cost): Likewise.
28444 * tree-cfg.c (verify_gimple_comparison): Likewise.
28445 (verify_gimple_assign_binary): Likewise.
28446 (verify_gimple_assign_ternary): Likewise.
28447 (verify_gimple_assign_single): Likewise.
28448 * tree-pretty-print.c (dump_generic_node): Likewise.
28449 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
28450 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
28451 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
28452 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
28453 (vect_shift_permute_load_chain): Likewise.
28454 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
28455 (expand_vector_condition, optimize_vector_constructor): Likewise.
28456 (lower_vec_perm, get_compute_type): Likewise.
28457 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28458 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
28459 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
28460 (vect_recog_mask_conversion_pattern): Likewise.
28461 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
28462 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
28463 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
28464 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
28465 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
28466 (vectorizable_shift, vectorizable_operation, vectorizable_store)
28467 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
28468 (supportable_widening_operation): Likewise.
28469 (supportable_narrowing_operation): Likewise.
28470 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
28472 * varasm.c (output_constant): Likewise.
28474 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28475 Alan Hayward <alan.hayward@arm.com>
28476 David Sherwood <david.sherwood@arm.com>
28478 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
28479 so that both the length == 3 and length != 3 cases set up their
28480 own permute vectors. Add comments explaining why we know the
28481 number of elements is constant.
28482 (vect_permute_load_chain): Likewise.
28484 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28485 Alan Hayward <alan.hayward@arm.com>
28486 David Sherwood <david.sherwood@arm.com>
28488 * machmode.h (mode_nunits): Change from unsigned char to
28490 (ONLY_FIXED_SIZE_MODES): New macro.
28491 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
28492 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
28493 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
28495 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
28496 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
28497 or if measurement_type is not polynomial.
28498 * genmodes.c (ZERO_COEFFS): New macro.
28499 (emit_mode_nunits_inline): Make mode_nunits_inline return a
28501 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
28502 Use ZERO_COEFFS when emitting initializers.
28503 * data-streamer.h (bp_pack_poly_value): New function.
28504 (bp_unpack_poly_value): Likewise.
28505 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
28506 for GET_MODE_NUNITS.
28507 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
28508 for GET_MODE_NUNITS.
28509 * tree.c (make_vector_type): Remove temporary shim and make
28510 the real function take the number of units as a poly_uint64
28511 rather than an int.
28512 (build_vector_type_for_mode): Handle polynomial nunits.
28513 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
28514 * emit-rtl.c (const_vec_series_p_1): Likewise.
28515 (gen_rtx_CONST_VECTOR): Likewise.
28516 * fold-const.c (test_vec_duplicate_folding): Likewise.
28517 * genrecog.c (validate_pattern): Likewise.
28518 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
28519 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
28520 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
28521 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
28522 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
28523 * rtlanal.c (subreg_get_info): Likewise.
28524 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
28525 (vect_grouped_load_supported): Likewise.
28526 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
28527 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
28528 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
28529 (simplify_const_unary_operation, simplify_binary_operation_1)
28530 (simplify_const_binary_operation, simplify_ternary_operation)
28531 (test_vector_ops_duplicate, test_vector_ops): Likewise.
28532 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
28533 instead of CONST_VECTOR_NUNITS.
28534 * varasm.c (output_constant_pool_2): Likewise.
28535 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
28536 explicit-encoded elements in the XVEC for variable-length vectors.
28538 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28540 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
28542 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28543 Alan Hayward <alan.hayward@arm.com>
28544 David Sherwood <david.sherwood@arm.com>
28546 * coretypes.h (fixed_size_mode): Declare.
28547 (fixed_size_mode_pod): New typedef.
28548 * builtins.h (target_builtins::x_apply_args_mode)
28549 (target_builtins::x_apply_result_mode): Change type to
28550 fixed_size_mode_pod.
28551 * builtins.c (apply_args_size, apply_result_size, result_vector)
28552 (expand_builtin_apply_args_1, expand_builtin_apply)
28553 (expand_builtin_return): Update accordingly.
28555 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28557 * cse.c (hash_rtx_cb): Hash only the encoded elements.
28558 * cselib.c (cselib_hash_rtx): Likewise.
28559 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
28560 CONST_VECTOR encoding.
28562 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28563 Jeff Law <law@redhat.com>
28566 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
28567 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
28568 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
28569 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
28572 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
28573 explicitly probe *sp in a noreturn function if there were any callee
28574 register saves or frame pointer is needed.
28576 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28579 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
28580 BLKmode for ternary, binary or unary expressions.
28583 * var-tracking.c (delete_vta_debug_insn): New inline function.
28584 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
28585 insns from get_insns () to NULL instead of each bb separately.
28586 Use delete_vta_debug_insn. No longer static.
28587 (vt_debug_insns_local, variable_tracking_main_1): Adjust
28588 delete_vta_debug_insns callers.
28589 * rtl.h (delete_vta_debug_insns): Declare.
28590 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
28591 instead of variable_tracking_main.
28593 2018-01-03 Martin Sebor <msebor@redhat.com>
28595 PR tree-optimization/83603
28596 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
28597 arguments past the endof the argument list in functions declared
28598 without a prototype.
28599 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
28600 Avoid checking when arguments are null.
28602 2018-01-03 Martin Sebor <msebor@redhat.com>
28605 * doc/extend.texi (attribute const): Fix a typo.
28606 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
28607 issuing -Wsuggest-attribute for void functions.
28609 2018-01-03 Martin Sebor <msebor@redhat.com>
28611 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
28612 offset_int::from instead of wide_int::to_shwi.
28613 (maybe_diag_overlap): Remove assertion.
28614 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
28615 * gimple-ssa-sprintf.c (format_directive): Same.
28616 (parse_directive): Same.
28617 (sprintf_dom_walker::compute_format_length): Same.
28618 (try_substitute_return_value): Same.
28620 2018-01-03 Jeff Law <law@redhat.com>
28622 PR middle-end/83654
28623 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
28624 non-constant residual for zero at runtime and avoid probing in
28625 that case. Reorganize code for trailing problem to mirror handling
28628 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28630 PR tree-optimization/83501
28631 * tree-ssa-strlen.c (get_string_cst): New.
28632 (handle_char_store): Call get_string_cst.
28634 2018-01-03 Martin Liska <mliska@suse.cz>
28636 PR tree-optimization/83593
28637 * tree-ssa-strlen.c: Include tree-cfg.h.
28638 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
28639 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
28640 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
28642 (strlen_dom_walker::before_dom_children): Call
28643 gimple_purge_dead_eh_edges. Dump tranformation with details
28645 (strlen_dom_walker::before_dom_children): Update call by adding
28646 new argument cleanup_eh.
28647 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
28649 2018-01-03 Martin Liska <mliska@suse.cz>
28652 * cif-code.def (VARIADIC_THUNK): New enum value.
28653 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
28656 2018-01-03 Jan Beulich <jbeulich@suse.com>
28658 * sse.md (mov<mode>_internal): Tighten condition for when to use
28659 vmovdqu<ssescalarsize> for TI and OI modes.
28661 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28663 Update copyright years.
28665 2018-01-03 Martin Liska <mliska@suse.cz>
28668 * ipa-visibility.c (function_and_variable_visibility): Skip
28669 functions with noipa attribure.
28671 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28673 * gcc.c (process_command): Update copyright notice dates.
28674 * gcov-dump.c (print_version): Ditto.
28675 * gcov.c (print_version): Ditto.
28676 * gcov-tool.c (print_version): Ditto.
28677 * gengtype.c (create_file): Ditto.
28678 * doc/cpp.texi: Bump @copying's copyright year.
28679 * doc/cppinternals.texi: Ditto.
28680 * doc/gcc.texi: Ditto.
28681 * doc/gccint.texi: Ditto.
28682 * doc/gcov.texi: Ditto.
28683 * doc/install.texi: Ditto.
28684 * doc/invoke.texi: Ditto.
28686 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28688 * vector-builder.h (vector_builder::m_full_nelts): Change from
28689 unsigned int to poly_uint64.
28690 (vector_builder::full_nelts): Update prototype accordingly.
28691 (vector_builder::new_vector): Likewise.
28692 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
28693 (vector_builder::operator ==): Likewise.
28694 (vector_builder::finalize): Likewise.
28695 * int-vector-builder.h (int_vector_builder::int_vector_builder):
28696 Take the number of elements as a poly_uint64 rather than an
28698 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
28699 from unsigned int to poly_uint64.
28700 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
28701 (vec_perm_indices::new_vector): Likewise.
28702 (vec_perm_indices::length): Likewise.
28703 (vec_perm_indices::nelts_per_input): Likewise.
28704 (vec_perm_indices::input_nelts): Likewise.
28705 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
28706 number of elements per input as a poly_uint64 rather than an
28707 unsigned int. Use the original encoding for variable-length
28708 vectors, rather than clamping each individual element.
28709 For the second and subsequent elements in each pattern,
28710 clamp the step and base before clamping their sum.
28711 (vec_perm_indices::series_p): Handle polynomial element counts.
28712 (vec_perm_indices::all_in_range_p): Likewise.
28713 (vec_perm_indices_to_tree): Likewise.
28714 (vec_perm_indices_to_rtx): Likewise.
28715 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
28716 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
28717 (tree_vector_builder::new_binary_operation): Handle polynomial
28718 element counts. Return false if we need to know the number
28719 of elements at compile time.
28720 * fold-const.c (fold_vec_perm): Punt if the number of elements
28721 isn't known at compile time.
28723 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28725 * vec-perm-indices.h (vec_perm_builder): Change element type
28726 from HOST_WIDE_INT to poly_int64.
28727 (vec_perm_indices::element_type): Update accordingly.
28728 (vec_perm_indices::clamp): Handle polynomial element_types.
28729 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
28730 (vec_perm_indices::all_in_range_p): Likewise.
28731 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
28733 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
28734 polynomial vec_perm_indices element types.
28735 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
28736 * fold-const.c (fold_vec_perm): Likewise.
28737 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
28738 * tree-vect-generic.c (lower_vec_perm): Likewise.
28739 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
28740 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
28741 element type to HOST_WIDE_INT.
28743 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28744 Alan Hayward <alan.hayward@arm.com>
28745 David Sherwood <david.sherwood@arm.com>
28747 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
28748 rather than an int. Use plus_constant.
28749 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
28750 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
28752 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28753 Alan Hayward <alan.hayward@arm.com>
28754 David Sherwood <david.sherwood@arm.com>
28756 * calls.c (emit_call_1, expand_call): Change struct_value_size from
28757 a HOST_WIDE_INT to a poly_int64.
28759 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28760 Alan Hayward <alan.hayward@arm.com>
28761 David Sherwood <david.sherwood@arm.com>
28763 * calls.c (load_register_parameters): Cope with polynomial
28764 mode sizes. Require a constant size for BLKmode parameters
28765 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
28766 forces a parameter to be padded at the lsb end in order to
28767 fill a complete number of words, require the parameter size
28768 to be ordered wrt UNITS_PER_WORD.
28770 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28771 Alan Hayward <alan.hayward@arm.com>
28772 David Sherwood <david.sherwood@arm.com>
28774 * reload1.c (spill_stack_slot_width): Change element type
28775 from unsigned int to poly_uint64_pod.
28776 (alter_reg): Treat mode sizes as polynomial.
28778 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28779 Alan Hayward <alan.hayward@arm.com>
28780 David Sherwood <david.sherwood@arm.com>
28782 * reload.c (complex_word_subreg_p): New function.
28783 (reload_inner_reg_of_subreg, push_reload): Use it.
28785 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28786 Alan Hayward <alan.hayward@arm.com>
28787 David Sherwood <david.sherwood@arm.com>
28789 * lra-constraints.c (process_alt_operands): Reject matched
28790 operands whose sizes aren't ordered.
28791 (match_reload): Refer to this check here.
28793 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28794 Alan Hayward <alan.hayward@arm.com>
28795 David Sherwood <david.sherwood@arm.com>
28797 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
28798 that the mode size is in the set {1, 2, 4, 8, 16}.
28800 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28801 Alan Hayward <alan.hayward@arm.com>
28802 David Sherwood <david.sherwood@arm.com>
28804 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
28805 Use plus_constant instead of gen_rtx_PLUS.
28807 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28808 Alan Hayward <alan.hayward@arm.com>
28809 David Sherwood <david.sherwood@arm.com>
28811 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
28812 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
28813 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
28814 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
28815 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
28816 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
28817 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
28818 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
28819 * config/i386/i386.c (ix86_push_rounding): ...this new function.
28820 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
28822 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
28823 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
28824 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
28825 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
28826 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
28827 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
28828 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
28829 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
28830 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
28831 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
28833 * expr.c (emit_move_resolve_push): Treat the input and result
28834 of PUSH_ROUNDING as a poly_int64.
28835 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
28836 (emit_push_insn): Likewise.
28837 * lra-eliminations.c (mark_not_eliminable): Likewise.
28838 * recog.c (push_operand): Likewise.
28839 * reload1.c (elimination_effects): Likewise.
28840 * rtlanal.c (nonzero_bits1): Likewise.
28841 * calls.c (store_one_arg): Likewise. Require the padding to be
28842 known at compile time.
28844 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28845 Alan Hayward <alan.hayward@arm.com>
28846 David Sherwood <david.sherwood@arm.com>
28848 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
28849 Use plus_constant instead of gen_rtx_PLUS.
28851 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28852 Alan Hayward <alan.hayward@arm.com>
28853 David Sherwood <david.sherwood@arm.com>
28855 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
28856 rather than an int.
28858 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28859 Alan Hayward <alan.hayward@arm.com>
28860 David Sherwood <david.sherwood@arm.com>
28862 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
28863 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
28864 via stack temporaries. Treat the mode size as polynomial too.
28866 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28867 Alan Hayward <alan.hayward@arm.com>
28868 David Sherwood <david.sherwood@arm.com>
28870 * expr.c (expand_expr_real_2): When handling conversions involving
28871 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
28872 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
28873 as a poly_uint64 too.
28875 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28876 Alan Hayward <alan.hayward@arm.com>
28877 David Sherwood <david.sherwood@arm.com>
28879 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
28881 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28882 Alan Hayward <alan.hayward@arm.com>
28883 David Sherwood <david.sherwood@arm.com>
28885 * combine.c (can_change_dest_mode): Handle polynomial
28886 REGMODE_NATURAL_SIZE.
28887 * expmed.c (store_bit_field_1): Likewise.
28888 * expr.c (store_constructor): Likewise.
28889 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
28890 and polynomial REGMODE_NATURAL_SIZE.
28891 (gen_lowpart_common): Likewise.
28892 * reginfo.c (record_subregs_of_mode): Likewise.
28893 * rtlanal.c (read_modify_subreg_p): Likewise.
28895 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28896 Alan Hayward <alan.hayward@arm.com>
28897 David Sherwood <david.sherwood@arm.com>
28899 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
28900 numbers of elements.
28902 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28903 Alan Hayward <alan.hayward@arm.com>
28904 David Sherwood <david.sherwood@arm.com>
28906 * match.pd: Cope with polynomial numbers of vector elements.
28908 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28909 Alan Hayward <alan.hayward@arm.com>
28910 David Sherwood <david.sherwood@arm.com>
28912 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
28913 in a POINTER_PLUS_EXPR.
28915 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28916 Alan Hayward <alan.hayward@arm.com>
28917 David Sherwood <david.sherwood@arm.com>
28919 * omp-simd-clone.c (simd_clone_subparts): New function.
28920 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
28921 (ipa_simd_modify_function_body): Likewise.
28923 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28924 Alan Hayward <alan.hayward@arm.com>
28925 David Sherwood <david.sherwood@arm.com>
28927 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
28928 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
28929 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
28930 (expand_vector_condition, vector_element): Likewise.
28931 (subparts_gt): New function.
28932 (get_compute_type): Use subparts_gt.
28933 (count_type_subparts): Delete.
28934 (expand_vector_operations_1): Use subparts_gt instead of
28935 count_type_subparts.
28937 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28938 Alan Hayward <alan.hayward@arm.com>
28939 David Sherwood <david.sherwood@arm.com>
28941 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
28942 (vect_compile_time_alias): ...this new function. Do the calculation
28943 on poly_ints rather than trees.
28944 (vect_prune_runtime_alias_test_list): Update call accordingly.
28946 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28947 Alan Hayward <alan.hayward@arm.com>
28948 David Sherwood <david.sherwood@arm.com>
28950 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
28952 (vect_schedule_slp_instance): Likewise.
28954 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28955 Alan Hayward <alan.hayward@arm.com>
28956 David Sherwood <david.sherwood@arm.com>
28958 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
28959 constant and extern definitions for variable-length vectors.
28960 (vect_get_constant_vectors): Note that the number of units
28961 is known to be constant.
28963 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28964 Alan Hayward <alan.hayward@arm.com>
28965 David Sherwood <david.sherwood@arm.com>
28967 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
28968 of units as polynomial. Choose between WIDE and NARROW based
28971 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28972 Alan Hayward <alan.hayward@arm.com>
28973 David Sherwood <david.sherwood@arm.com>
28975 * tree-vect-stmts.c (simd_clone_subparts): New function.
28976 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
28978 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28979 Alan Hayward <alan.hayward@arm.com>
28980 David Sherwood <david.sherwood@arm.com>
28982 * tree-vect-stmts.c (vectorizable_call): Treat the number of
28983 vectors as polynomial. Use build_index_vector for
28984 IFN_GOMP_SIMD_LANE.
28986 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28987 Alan Hayward <alan.hayward@arm.com>
28988 David Sherwood <david.sherwood@arm.com>
28990 * tree-vect-stmts.c (get_load_store_type): Treat the number of
28991 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
28992 for variable-length vectors.
28993 (vectorizable_mask_load_store): Treat the number of units as
28994 polynomial, asserting that it is constant if the condition has
28995 already been enforced.
28996 (vectorizable_store, vectorizable_load): Likewise.
28998 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28999 Alan Hayward <alan.hayward@arm.com>
29000 David Sherwood <david.sherwood@arm.com>
29002 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
29003 of units as polynomial. Punt if we can't tell at compile time
29004 which vector contains the final result.
29006 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29007 Alan Hayward <alan.hayward@arm.com>
29008 David Sherwood <david.sherwood@arm.com>
29010 * tree-vect-loop.c (vectorizable_induction): Treat the number
29011 of units as polynomial. Punt on SLP inductions. Use an integer
29012 VEC_SERIES_EXPR for variable-length integer reductions. Use a
29013 cast of such a series for variable-length floating-point
29016 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29017 Alan Hayward <alan.hayward@arm.com>
29018 David Sherwood <david.sherwood@arm.com>
29020 * tree.h (build_index_vector): Declare.
29021 * tree.c (build_index_vector): New function.
29022 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
29023 of units as polynomial, forcibly converting it to a constant if
29024 vectorizable_reduction has already enforced the condition.
29025 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
29026 to create a {1,2,3,...} vector.
29027 (vectorizable_reduction): Treat the number of units as polynomial.
29028 Choose vectype_in based on the largest scalar element size rather
29029 than the smallest number of units. Enforce the restrictions
29032 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29033 Alan Hayward <alan.hayward@arm.com>
29034 David Sherwood <david.sherwood@arm.com>
29036 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
29037 number of units as polynomial.
29039 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29040 Alan Hayward <alan.hayward@arm.com>
29041 David Sherwood <david.sherwood@arm.com>
29043 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
29044 * target.def (autovectorize_vector_sizes): Return the vector sizes
29045 by pointer, using vector_sizes rather than a bitmask.
29046 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
29047 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
29048 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
29050 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
29051 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
29052 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
29053 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
29054 * omp-general.c (omp_max_vf): Likewise.
29055 * omp-low.c (omp_clause_aligned_alignment): Likewise.
29056 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
29057 * tree-vect-loop.c (vect_analyze_loop): Likewise.
29058 * tree-vect-slp.c (vect_slp_bb): Likewise.
29059 * doc/tm.texi: Regenerate.
29060 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
29062 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
29063 the vector size as a poly_uint64 rather than an unsigned int.
29064 (current_vector_size): Change from an unsigned int to a poly_uint64.
29065 (get_vectype_for_scalar_type): Update accordingly.
29066 * tree.h (build_truth_vector_type): Take the size and number of
29067 units as a poly_uint64 rather than an unsigned int.
29068 (build_vector_type): Add a temporary overload that takes
29069 the number of units as a poly_uint64 rather than an unsigned int.
29070 * tree.c (make_vector_type): Likewise.
29071 (build_truth_vector_type): Take the number of units as a poly_uint64
29072 rather than an unsigned int.
29074 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29075 Alan Hayward <alan.hayward@arm.com>
29076 David Sherwood <david.sherwood@arm.com>
29078 * target.def (get_mask_mode): Take the number of units and length
29079 as poly_uint64s rather than unsigned ints.
29080 * targhooks.h (default_get_mask_mode): Update accordingly.
29081 * targhooks.c (default_get_mask_mode): Likewise.
29082 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
29083 * doc/tm.texi: Regenerate.
29085 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29086 Alan Hayward <alan.hayward@arm.com>
29087 David Sherwood <david.sherwood@arm.com>
29089 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
29090 * omp-general.c (omp_max_vf): Likewise.
29091 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
29092 (expand_omp_simd): Handle polynomial safelen.
29093 * omp-low.c (omplow_simd_context): Add a default constructor.
29094 (omplow_simd_context::max_vf): Change from int to poly_uint64.
29095 (lower_rec_simd_input_clauses): Update accordingly.
29096 (lower_rec_input_clauses): Likewise.
29098 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29099 Alan Hayward <alan.hayward@arm.com>
29100 David Sherwood <david.sherwood@arm.com>
29102 * tree-vectorizer.h (vect_nunits_for_cost): New function.
29103 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
29104 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
29105 (vect_analyze_slp_cost): Likewise.
29106 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
29107 (vect_model_load_cost): Likewise.
29109 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29110 Alan Hayward <alan.hayward@arm.com>
29111 David Sherwood <david.sherwood@arm.com>
29113 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
29114 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
29115 from an unsigned int * to a poly_uint64_pod *.
29116 (calculate_unrolling_factor): New function.
29117 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
29119 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29120 Alan Hayward <alan.hayward@arm.com>
29121 David Sherwood <david.sherwood@arm.com>
29123 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
29124 from an unsigned int to a poly_uint64.
29125 (_loop_vec_info::slp_unrolling_factor): Likewise.
29126 (_loop_vec_info::vectorization_factor): Change from an int
29128 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
29129 (vect_get_num_vectors): New function.
29130 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
29131 (vect_get_num_copies): Use vect_get_num_vectors.
29132 (vect_analyze_data_ref_dependences): Change max_vf from an int *
29133 to an unsigned int *.
29134 (vect_analyze_data_refs): Change min_vf from an int * to a
29136 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
29137 than an unsigned HOST_WIDE_INT.
29138 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
29139 (vect_analyze_data_ref_dependence): Change max_vf from an int *
29140 to an unsigned int *.
29141 (vect_analyze_data_ref_dependences): Likewise.
29142 (vect_compute_data_ref_alignment): Handle polynomial vf.
29143 (vect_enhance_data_refs_alignment): Likewise.
29144 (vect_prune_runtime_alias_test_list): Likewise.
29145 (vect_shift_permute_load_chain): Likewise.
29146 (vect_supportable_dr_alignment): Likewise.
29147 (dependence_distance_ge_vf): Take the vectorization factor as a
29148 poly_uint64 rather than an unsigned HOST_WIDE_INT.
29149 (vect_analyze_data_refs): Change min_vf from an int * to a
29151 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
29152 vfm1 as a poly_uint64 rather than an int. Make the same change
29153 for the returned bound_scalar.
29154 (vect_gen_vector_loop_niters): Handle polynomial vf.
29155 (vect_do_peeling): Likewise. Update call to
29156 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
29157 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
29159 * tree-vect-loop.c (vect_determine_vectorization_factor)
29160 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
29161 (vect_get_known_peeling_cost): Likewise.
29162 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
29163 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
29164 (vect_transform_loop): Likewise. Use the lowest possible VF when
29165 updating the upper bounds of the loop.
29166 (vect_min_worthwhile_factor): Make static. Return an unsigned int
29167 rather than an int.
29168 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
29169 polynomial unroll factors.
29170 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
29171 (vect_make_slp_decision): Likewise.
29172 (vect_supported_load_permutation_p): Likewise, and polynomial
29174 (vect_analyze_slp_cost): Handle polynomial vf.
29175 (vect_slp_analyze_node_operations): Likewise.
29176 (vect_slp_analyze_bb_1): Likewise.
29177 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
29178 than an unsigned HOST_WIDE_INT.
29179 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
29180 (vectorizable_load): Handle polynomial vf.
29181 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
29183 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
29185 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29186 Alan Hayward <alan.hayward@arm.com>
29187 David Sherwood <david.sherwood@arm.com>
29189 * match.pd: Handle bit operations involving three constants
29190 and try to fold one pair.
29192 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29194 * tree-vect-loop-manip.c: Include gimple-fold.h.
29195 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
29196 niters_maybe_zero parameters. Handle other cases besides a step of 1.
29197 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
29198 Add a path that uses a step of VF instead of 1, but disable it
29200 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
29201 and niters_no_overflow parameters. Update calls to
29202 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
29203 Create a new SSA name if the latter choses to use a ste other
29204 than zero, and return it via niters_vector_mult_vf_var.
29205 * tree-vect-loop.c (vect_transform_loop): Update calls to
29206 vect_do_peeling, vect_gen_vector_loop_niters and
29207 slpeel_make_loop_iterate_ntimes.
29208 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
29209 (vect_gen_vector_loop_niters): Update declarations after above changes.
29211 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
29213 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
29214 128-bit round to integer instructions.
29215 (ceil<mode>2): Likewise.
29216 (btrunc<mode>2): Likewise.
29217 (round<mode>2): Likewise.
29219 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29221 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
29222 unaligned VSX load/store on P8/P9.
29223 (expand_block_clear): Allow the use of unaligned VSX
29224 load/store on P8/P9.
29226 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29228 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
29230 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
29231 swap associated with both a load and a store.
29233 2018-01-02 Andrew Waterman <andrew@sifive.com>
29235 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
29236 * config/riscv/riscv.md (clear_cache): Use it.
29238 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
29240 * web.c: Remove out-of-date comment.
29242 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29244 * expr.c (fixup_args_size_notes): Check that any existing
29245 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
29246 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
29247 (emit_single_push_insn): ...here.
29249 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29251 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
29252 (const_vector_encoded_nelts): New function.
29253 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
29254 (const_vector_int_elt, const_vector_elt): Declare.
29255 * emit-rtl.c (const_vector_int_elt_1): New function.
29256 (const_vector_elt): Likewise.
29257 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
29258 of CONST_VECTOR_ELT.
29260 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29262 * expr.c: Include rtx-vector-builder.h.
29263 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
29264 directly on the tree encoding.
29265 (const_vector_from_tree): Likewise.
29266 * optabs.c: Include rtx-vector-builder.h.
29267 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
29268 sequence of "u" values.
29269 * vec-perm-indices.c: Include rtx-vector-builder.h.
29270 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
29271 directly on the vec_perm_indices encoding.
29273 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29275 * doc/rtl.texi (const_vector): Describe new encoding scheme.
29276 * Makefile.in (OBJS): Add rtx-vector-builder.o.
29277 * rtx-vector-builder.h: New file.
29278 * rtx-vector-builder.c: Likewise.
29279 * rtl.h (rtx_def::u2): Add a const_vector field.
29280 (CONST_VECTOR_NPATTERNS): New macro.
29281 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
29282 (CONST_VECTOR_DUPLICATE_P): Likewise.
29283 (CONST_VECTOR_STEPPED_P): Likewise.
29284 (CONST_VECTOR_ENCODED_ELT): Likewise.
29285 (const_vec_duplicate_p): Check for a duplicated vector encoding.
29286 (unwrap_const_vec_duplicate): Likewise.
29287 (const_vec_series_p): Check for a non-duplicated vector encoding.
29288 Say that the function only returns true for integer vectors.
29289 * emit-rtl.c: Include rtx-vector-builder.h.
29290 (gen_const_vec_duplicate_1): Delete.
29291 (gen_const_vector): Call gen_const_vec_duplicate instead of
29292 gen_const_vec_duplicate_1.
29293 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
29294 (gen_const_vec_duplicate): Use rtx_vector_builder.
29295 (gen_const_vec_series): Likewise.
29296 (gen_rtx_CONST_VECTOR): Likewise.
29297 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
29298 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
29299 Build a new vector rather than modifying a CONST_VECTOR in-place.
29300 (handle_special_swappables): Update call accordingly.
29301 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
29302 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
29303 Build a new vector rather than modifying a CONST_VECTOR in-place.
29304 (handle_special_swappables): Update call accordingly.
29306 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29308 * simplify-rtx.c (simplify_const_binary_operation): Use
29309 CONST_VECTOR_ELT instead of XVECEXP.
29311 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29313 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
29314 the selector elements to be different from the data elements
29315 if the selector is a VECTOR_CST.
29316 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
29317 ssizetype for the selector.
29319 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29321 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
29322 before testing each element individually.
29323 * tree-vect-generic.c (lower_vec_perm): Likewise.
29325 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29327 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
29328 * selftest-run-tests.c (selftest::run_tests): Call it.
29329 * vector-builder.h (vector_builder::operator ==): New function.
29330 (vector_builder::operator !=): Likewise.
29331 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
29332 (vec_perm_indices::all_from_input_p): New function.
29333 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
29334 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
29335 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
29336 instead of reading the VECTOR_CST directly. Detect whether both
29337 vector inputs are the same before constructing the vec_perm_indices,
29338 and update the number of inputs argument accordingly. Use the
29339 utility functions added above. Only construct sel2 if we need to.
29341 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29343 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
29344 the broadcast of the low byte.
29345 (expand_mult_highpart): Use an explicit encoding for the permutes.
29346 * optabs-query.c (can_mult_highpart_p): Likewise.
29347 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
29348 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29349 (vectorizable_bswap): Likewise.
29350 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
29351 explicit encoding for the power-of-2 permutes.
29352 (vect_permute_store_chain): Likewise.
29353 (vect_grouped_load_supported): Likewise.
29354 (vect_permute_load_chain): Likewise.
29356 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29358 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
29359 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
29360 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
29361 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
29362 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
29363 (vect_gen_perm_mask_any): Likewise.
29365 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29367 * int-vector-builder.h: New file.
29368 * vec-perm-indices.h: Include int-vector-builder.h.
29369 (vec_perm_indices): Redefine as an int_vector_builder.
29370 (auto_vec_perm_indices): Delete.
29371 (vec_perm_builder): Redefine as a stand-alone class.
29372 (vec_perm_indices::vec_perm_indices): New function.
29373 (vec_perm_indices::clamp): Likewise.
29374 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
29375 (vec_perm_indices::new_vector): New function.
29376 (vec_perm_indices::new_expanded_vector): Update for new
29377 vec_perm_indices class.
29378 (vec_perm_indices::rotate_inputs): New function.
29379 (vec_perm_indices::all_in_range_p): Operate directly on the
29380 encoded form, without computing elided elements.
29381 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
29382 encoding. Update for new vec_perm_indices class.
29383 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
29384 the given vec_perm_builder.
29385 (expand_vec_perm_var): Update vec_perm_builder constructor.
29386 (expand_mult_highpart): Use vec_perm_builder instead of
29387 auto_vec_perm_indices.
29388 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
29389 vec_perm_indices instead of auto_vec_perm_indices. Use a single
29390 or double series encoding as appropriate.
29391 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
29392 vec_perm_indices instead of auto_vec_perm_indices.
29393 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
29394 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
29395 (vect_permute_store_chain): Likewise.
29396 (vect_grouped_load_supported): Likewise.
29397 (vect_permute_load_chain): Likewise.
29398 (vect_shift_permute_load_chain): Likewise.
29399 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
29400 (vect_transform_slp_perm_load): Likewise.
29401 (vect_schedule_slp_instance): Likewise.
29402 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29403 (vectorizable_mask_load_store): Likewise.
29404 (vectorizable_bswap): Likewise.
29405 (vectorizable_store): Likewise.
29406 (vectorizable_load): Likewise.
29407 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
29408 vec_perm_indices instead of auto_vec_perm_indices. Use
29409 tree_to_vec_perm_builder to read the vector from a tree.
29410 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
29411 vec_perm_builder instead of a vec_perm_indices.
29412 (have_whole_vector_shift): Use vec_perm_builder and
29413 vec_perm_indices instead of auto_vec_perm_indices. Leave the
29414 truncation to calc_vec_perm_mask_for_shift.
29415 (vect_create_epilog_for_reduction): Likewise.
29416 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
29417 from auto_vec_perm_indices to vec_perm_indices.
29418 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
29419 instead of changing individual elements.
29420 (aarch64_vectorize_vec_perm_const): Use new_vector to install
29421 the vector in d.perm.
29422 * config/arm/arm.c (expand_vec_perm_d::perm): Change
29423 from auto_vec_perm_indices to vec_perm_indices.
29424 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
29425 instead of changing individual elements.
29426 (arm_vectorize_vec_perm_const): Use new_vector to install
29427 the vector in d.perm.
29428 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
29429 Update vec_perm_builder constructor.
29430 (rs6000_expand_interleave): Likewise.
29431 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
29432 (rs6000_expand_interleave): Likewise.
29434 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29436 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
29437 to qimode could truncate the indices.
29438 * optabs.c (expand_vec_perm_var): Likewise.
29440 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29442 * Makefile.in (OBJS): Add vec-perm-indices.o.
29443 * vec-perm-indices.h: New file.
29444 * vec-perm-indices.c: Likewise.
29445 * target.h (vec_perm_indices): Replace with a forward class
29447 (auto_vec_perm_indices): Move to vec-perm-indices.h.
29448 * optabs.h: Include vec-perm-indices.h.
29449 (expand_vec_perm): Delete.
29450 (selector_fits_mode_p, expand_vec_perm_var): Declare.
29451 (expand_vec_perm_const): Declare.
29452 * target.def (vec_perm_const_ok): Replace with...
29453 (vec_perm_const): ...this new hook.
29454 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
29455 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
29456 * doc/tm.texi: Regenerate.
29457 * optabs.def (vec_perm_const): Delete.
29458 * doc/md.texi (vec_perm_const): Likewise.
29459 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
29460 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
29461 expand_vec_perm for constant permutation vectors. Assert that
29462 the mode of variable permutation vectors is the integer equivalent
29463 of the mode that is being permuted.
29464 * optabs-query.h (selector_fits_mode_p): Declare.
29465 * optabs-query.c: Include vec-perm-indices.h.
29466 (selector_fits_mode_p): New function.
29467 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
29468 is defined, instead of checking whether the vec_perm_const_optab
29469 exists. Use targetm.vectorize.vec_perm_const instead of
29470 targetm.vectorize.vec_perm_const_ok. Check whether the indices
29471 fit in the vector mode before using a variable permute.
29472 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
29473 vec_perm_indices instead of an rtx.
29474 (expand_vec_perm): Replace with...
29475 (expand_vec_perm_const): ...this new function. Take the selector
29476 as a vec_perm_indices rather than an rtx. Also take the mode of
29477 the selector. Update call to shift_amt_for_vec_perm_mask.
29478 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
29479 Use vec_perm_indices::new_expanded_vector to expand the original
29480 selector into bytes. Check whether the indices fit in the vector
29481 mode before using a variable permute.
29482 (expand_vec_perm_var): Make global.
29483 (expand_mult_highpart): Use expand_vec_perm_const.
29484 * fold-const.c: Includes vec-perm-indices.h.
29485 * tree-ssa-forwprop.c: Likewise.
29486 * tree-vect-data-refs.c: Likewise.
29487 * tree-vect-generic.c: Likewise.
29488 * tree-vect-loop.c: Likewise.
29489 * tree-vect-slp.c: Likewise.
29490 * tree-vect-stmts.c: Likewise.
29491 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
29493 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
29494 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
29495 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
29496 (aarch64_vectorize_vec_perm_const): ...this new function.
29497 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
29498 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29499 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
29500 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
29501 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
29502 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29503 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
29505 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
29506 check for NEON modes.
29507 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
29508 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
29509 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
29510 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
29512 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
29513 the old VEC_PERM_CONST conditions.
29514 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
29515 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
29516 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
29517 (ia64_vectorize_vec_perm_const_ok): Merge into...
29518 (ia64_vectorize_vec_perm_const): ...this new function.
29519 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
29520 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
29521 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
29522 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
29523 * config/mips/mips.c (mips_expand_vec_perm_const)
29524 (mips_vectorize_vec_perm_const_ok): Merge into...
29525 (mips_vectorize_vec_perm_const): ...this new function.
29526 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
29527 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
29528 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
29529 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
29530 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
29531 (rs6000_expand_vec_perm_const): Delete.
29532 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
29534 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29535 (altivec_expand_vec_perm_const_le): Take each operand individually.
29536 Operate on constant selectors rather than rtxes.
29537 (altivec_expand_vec_perm_const): Likewise. Update call to
29538 altivec_expand_vec_perm_const_le.
29539 (rs6000_expand_vec_perm_const): Delete.
29540 (rs6000_vectorize_vec_perm_const_ok): Delete.
29541 (rs6000_vectorize_vec_perm_const): New function.
29542 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
29543 an element count and rtx array.
29544 (rs6000_expand_extract_even): Update call accordingly.
29545 (rs6000_expand_interleave): Likewise.
29546 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
29547 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
29548 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
29549 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
29550 (rs6000_expand_vec_perm_const): Delete.
29551 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
29552 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29553 (altivec_expand_vec_perm_const_le): Take each operand individually.
29554 Operate on constant selectors rather than rtxes.
29555 (altivec_expand_vec_perm_const): Likewise. Update call to
29556 altivec_expand_vec_perm_const_le.
29557 (rs6000_expand_vec_perm_const): Delete.
29558 (rs6000_vectorize_vec_perm_const_ok): Delete.
29559 (rs6000_vectorize_vec_perm_const): New function. Remove stray
29560 reference to the SPE evmerge intructions.
29561 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
29562 an element count and rtx array.
29563 (rs6000_expand_extract_even): Update call accordingly.
29564 (rs6000_expand_interleave): Likewise.
29565 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
29566 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
29568 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29570 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29572 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
29573 vector mode and that that mode matches the mode of the data
29575 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
29576 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
29577 directly using expand_vec_perm_1 when forcing selectors into
29579 (expand_vec_perm_var): New function, split out from expand_vec_perm.
29581 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29583 * optabs-query.h (can_vec_perm_p): Delete.
29584 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
29585 * optabs-query.c (can_vec_perm_p): Split into...
29586 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
29587 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
29588 particular selector is valid.
29589 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
29590 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
29591 (vect_grouped_load_supported): Likewise.
29592 (vect_shift_permute_load_chain): Likewise.
29593 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
29594 (vect_transform_slp_perm_load): Likewise.
29595 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29596 (vectorizable_bswap): Likewise.
29597 (vect_gen_perm_mask_checked): Likewise.
29598 * fold-const.c (fold_ternary_loc): Likewise. Don't take
29599 implementations of variable permutation vectors into account
29600 when deciding which selector to use.
29601 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
29602 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
29603 with a false third argument.
29604 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
29605 to test whether the constant selector is valid and can_vec_perm_var_p
29606 to test whether a variable selector is valid.
29608 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29610 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
29611 * optabs-query.c (can_vec_perm_p): Likewise.
29612 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
29613 instead of vec_perm_indices.
29614 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
29615 (vect_gen_perm_mask_checked): Likewise,
29616 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
29617 (vect_gen_perm_mask_checked): Likewise,
29619 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29621 * optabs-query.h (qimode_for_vec_perm): Declare.
29622 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
29623 (qimode_for_vec_perm): ...this new function.
29624 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
29626 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29628 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
29629 does not have a conditional at the top.
29631 2018-01-02 Richard Biener <rguenther@suse.de>
29633 * ipa-inline.c (big_speedup_p): Fix expression.
29635 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
29638 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
29641 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
29645 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
29646 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
29647 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
29648 cond_taken_branch_cost 3->4.
29650 2018-01-01 Jakub Jelinek <jakub@redhat.com>
29652 PR tree-optimization/83581
29653 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
29654 TODO_cleanup_cfg if any changes have been made.
29656 PR middle-end/83608
29657 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
29658 convert_modes if target mode has the right side, but different mode
29661 PR middle-end/83609
29662 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
29663 last argument when extracting from CONCAT. If either from_real or
29664 from_imag is NULL, use expansion through memory. If result is not
29665 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
29666 the parts directly to inner mode, if even that fails, use expansion
29669 PR middle-end/83623
29670 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
29671 check for bswap in mode rather than HImode and use that in expand_unop
29674 Copyright (C) 2018 Free Software Foundation, Inc.
29676 Copying and distribution of this file, with or without modification,
29677 are permitted in any medium without royalty provided the copyright
29678 notice and this notice are preserved.