1 2018-10-30 Martin Sebor <msebor@redhat.com>
3 * doc/extend.texi (optimize): Clarify/expand attribute documentation.
4 (target, pragma GCC optimize, pragma GCC target): Ditto.
6 2018-10-30 Jonathan Wakely <jwakely@redhat.com>
8 * doc/extend.texi: Fix prototype and description of
9 __builtin_expect_with_probability.
11 2018-10-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
13 * cgraph.h (clone_function_name_1): Replaced by new
14 clone_function_name_numbered that takes name as string; for
15 privatize_symbol_name_1 use only.
16 (clone_function_name): Renamed to
17 clone_function_name_numbered to be explicit about numbering.
18 (clone_function_name): New two-argument function that does
19 not number its output.
20 (clone_function_name): New three-argument function that
21 takes a number to append to its output.
22 * cgraphclones.c (duplicate_thunk_for_node):
23 (clone_function_name_1): Renamed.
24 (clone_function_name_numbered): Two new functions.
25 (clone_function_name): Improved documentation.
26 (cgraph_node::create_virtual_clone): Use clone_function_name_numbered.
27 * config/rs6000/rs6000.c (make_resolver_func): Ditto.
28 * final.c (final_scan_insn_1): Use the new clone_function_name
30 * multiple_target.c (create_dispatcher_calls): Ditto.
31 (create_target_clone): Ditto.
32 * omp-expand.c (grid_expand_target_grid_body): Ditto.
33 * omp-low.c (create_omp_child_function_name): Ditto.
34 * omp-simd-clone.c (simd_clone_create): Ditto.
35 * symtab.c (simd_symtab_node::noninterposable_alias): Use the
36 new clone_function_name without numbering.
38 2018-10-30 Richard Earnshaw <rearnsha@arm.com>
40 * alloc-pool.h (base_pool_allocator <TBlockAllocator>::initialize):
41 Assert that the allocation size is not zero.
43 2018-10-30 Richard Biener <rguenther@suse.de>
45 PR tree-optimization/87800
46 * tree-vect-slp.c (vect_build_slp_tree_2): Reject any
47 non-induction or reduction PHIs.
49 2018-10-30 Sameera Deshpande <sameera.deshpande@linaro.org
51 * config/aarch64/falkor-tag-collision-avoidance.c
52 (execute_tag_collision_avoidance): Call df_note_add_problem.
54 2018-10-30 Martin Liska <mliska@suse.cz>
56 * doc/extend.texi: Fix typo in documentation
57 of __builtin_expect_with_probability.
59 2018-10-29 David Malcolm <dmalcolm@redhat.com>
62 * input.c (get_substring_ranges_for_loc): Detect if
63 linemap_resolve_location gives us a NULL map, and reject
66 2018-10-29 Iain Buclaw <ibuclaw@gdcproject.org>
68 * config.gcc (xstormy16-*-elf): Set tm_d_file.
70 2018-10-29 David Malcolm <dmalcolm@redhat.com>
71 Martin Sebor <msebor@redhat.com>
72 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
74 * doc/ux.texi (Quoting): New subsection, adapted from material at
75 https://gcc.gnu.org/wiki/DiagnosticsGuidelines written by
76 MartinSebor and ManuelLopezIbanez.
77 (Fix-it hints): Note that fix-it hints shouldn't be marked for
80 2018-10-29 Kugan Vivekanandarajah <kuganv@linaro.org>
83 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Fix niter
86 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
88 * gcc.target/powerpc/ssse3-check.h: Remove inaccurate comment.
90 2018-10-29 Paul Koning <ni1d@arrl.net>
92 * doc/rtl.texi (CONST_WIDE_INT_ELT): Give correct macro name.
94 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
96 * gcc/config/rs6000/mmintrin.h (_mm_packs_pi16, _mm_packs_pi32,
97 _mm_packs_pu16, _mm_unpackhi_pi8, _mm_unpacklo_pi8, _mm_add_pi8,
98 _mm_add_pi16, _mm_add_pi32, _mm_sub_pi8, _mm_sub_pi16, _mm_sub_pi32,
99 _mm_cmpgt_pi8, _mm_cmpeq_pi16, _mm_cmpgt_pi16, _mm_cmpeq_pi32,
100 _mm_cmpgt_pi32, _mm_adds_pi8, _mm_adds_pi16, _mm_adds_pu8,
101 _mm_adds_pu16, _mm_subs_pi8, _mm_subs_pi16, _mm_subs_pu8,
102 _mm_subs_pu16, _mm_madd_pi16, _mm_mulhi_pi16, _mm_mullo_pi16,
103 _mm_sll_pi16, _mm_sra_pi16, _mm_srl_pi16, _mm_set1_pi16, _mm_set1_pi8):
104 Change 'vector' to '__vector'.
105 * gcc/config/rs6000/xmmintrin.h (_mm_cvtps_pi32, _mm_cvttps_pi32,
106 _mm_cvtps_pi16, _mm_cvtps_pi8, _mm_max_pi16, _mm_max_pu8, _mm_min_pi16,
107 _mm_min_pu8, _mm_mulhi_pu16, _mm_shuffle_pi16, _mm_avg_pu8,
108 _mm_avg_pu16): Likewise. And, whitespace corrections.
110 2018-10-29 Richard Biener <rguenther@suse.de>
112 PR tree-optimization/87785
113 * tree-vect-slp.c (vect_gather_slp_loads): Only gather
116 2018-10-29 Olivier Hainque <hainque@adacore.com>
118 * VXWORKS_NET_LIBS_RTP: New macro, network part of VXWORKS_LIBS_RTP.
120 2018-10-29 Olivier Hainque <hainque@adacore.com>
122 * config/vxworks.h (VXWORKS_PERSONALITY): New VxWorks
123 ports configuration macro, defaults to "gnu".
124 (VXWORKS_OS_CPP_BUILTINS): Use it. Feed TOOL and
125 TOOL_FAMILY instead of _VX_TOOL/_VXTOOL_FAMILY.
127 2018-10-29 Olivier Hainque <hainque@adacore.com>
129 * config/vxworks.h (CLEAR_INSN_CACHE): #undef before #define.
131 2018-10-29 Richard Biener <rguenther@suse.de>
133 PR tree-optimization/87790
134 * tree-vect-slp.c (vect_mark_slp_stmts): Simplify.
135 (vect_make_slp_decision): Adjust.
136 (vect_slp_analyze_bb_1): Likewise.
137 (vect_detect_hybrid_slp_stmts): Properly union SLP type over
140 2018-10-29 Richard Biener <rguenther@suse.de>
142 PR tree-optimization/87785
143 * tree-vect-slp.c (vect_build_slp_tree_2): Remove loads argument
145 (vect_build_slp_tree): Likewise.
146 (vect_gather_slp_loads): New function.
147 (vect_analyze_slp_instance): Gather loads separately from the
150 2018-10-29 Martin Liska <mliska@suse.cz>
152 * Makefile.in: Make dependency to json.o.
153 * doc/gcov.texi: Document new JSON format, remove
154 old intermediate format documentation.
155 * gcov.c (struct function_info): Come up with m_name and
157 (function_info::function_info): Initialize it.
158 (function_info::~function_info): Release it.
159 (main): Rename flag_intermediate_format to flag_json_format.
160 (print_usage): Describe --json-format.
161 (process_args): Set flag_json_format.
162 (output_intermediate_line): Remove.
163 (output_intermediate_json_line): Likewise.
164 (get_gcov_intermediate_filename): Return new extension
166 (output_intermediate_file): Implement JSON emission.
167 (output_json_intermediate_file): Implement JSON emission.
168 (generate_results): Use ::get_name for function name.
169 Handle JSON output file.
170 (read_graph_file): Use ::get_name instead of cplus_demangle.
171 (read_count_file): Likewise.
172 (solve_flow_graph): Likewise.
173 (add_line_counts): Likewise.
174 (accumulate_line_counts): Use new flag_json_format.
175 (output_function_details): Use ::get_name instead of cplus_demangle.
176 (output_lines): Likewise.
177 * json.cc (test_writing_literals): Add new tests.
178 * json.h (class literal): Add new boolean constructor.
180 2018-10-29 Segher Boessenkool <segher@kernel.crashing.org>
182 PR rtl-optimization/87701
183 PR rtl-optimization/87780
184 * combine.c (make_more_copies): Rewrite.
186 2018-10-28 Kugan Vivekanandarajah <kuganv@linaro.org>
188 * doc/generic.texi (ABSU_EXPR): Document.
189 * match.pd (absu(x)*absu(x) -> x*x): Handle.
190 (absu(absu(X)) -> absu(X)): Likewise.
191 (absu(-X) -> absu(X)): Likewise.
192 (absu(X) where X is nonnegative -> X): Likewise.
194 2018-10-28 Iain Buclaw <ibuclaw@gdcproject.org>
196 * Makefile.in (tm_d_file_list, tm_d_include_list): New variables.
197 (TM_D_H, D_TARGET_DEF, D_TARGET_H, D_TARGET_OBJS): New variables.
198 (tm_d.h, cs-tm_d.h, default-d.o): New rules.
199 (d/d-target-hooks-def.h, s-d-target-hooks-def-h): New rules.
200 (s-tm-texi): Also check timestamp on d-target.def.
201 (generated_files): Add TM_D_H and d-target-hooks-def.h.
202 (build/genhooks.o): Also depend on D_TARGET_DEF.
203 * config.gcc (tm_d_file, d_target_objs, target_has_targetdm): New
205 * config/aarch64/aarch64-d.c: New file.
206 * config/aarch64/aarch64-linux.h (GNU_USER_TARGET_D_CRITSEC_SIZE):
208 * config/aarch64/aarch64-protos.h (aarch64_d_target_versions): New
210 * config/aarch64/aarch64.h (TARGET_D_CPU_VERSIONS): Define.
211 * config/aarch64/t-aarch64 (aarch64-d.o): New rule.
212 * config/arm/arm-d.c: New file.
213 * config/arm/arm-protos.h (arm_d_target_versions): New prototype.
214 * config/arm/arm.h (TARGET_D_CPU_VERSIONS): Define.
215 * config/arm/linux-eabi.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
216 * config/arm/t-arm (arm-d.o): New rule.
217 * config/default-d.c: New file.
218 * config/glibc-d.c: New file.
219 * config/gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
220 * config/i386/i386-d.c: New file.
221 * config/i386/i386-protos.h (ix86_d_target_versions): New prototype.
222 * config/i386/i386.h (TARGET_D_CPU_VERSIONS): Define.
223 * config/i386/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
224 (GNU_USER_TARGET_D_CRITSEC_SIZE): Define.
225 * config/i386/t-i386 (i386-d.o): New rule.
226 * config/kfreebsd-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
227 * config/kopensolaris-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
228 * config/linux-android.h (ANDROID_TARGET_D_OS_VERSIONS): Define.
229 * config/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
230 * config/mips/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
231 * config/mips/mips-d.c: New file.
232 * config/mips/mips-protos.h (mips_d_target_versions): New prototype.
233 * config/mips/mips.h (TARGET_D_CPU_VERSIONS): Define.
234 * config/mips/t-mips (mips-d.o): New rule.
235 * config/powerpcspe/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
236 * config/powerpcspe/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
237 * config/powerpcspe/powerpcspe-d.c: New file.
238 * config/powerpcspe/powerpcspe-protos.h (rs6000_d_target_versions):
240 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
241 Support GNU D by using 0 as the language type.
242 * config/powerpcspe/powerpcspe.h (TARGET_D_CPU_VERSIONS): Define.
243 * config/powerpcspe/t-powerpcspe (powerpcspe-d.o): New rule.
244 * config/riscv/riscv-d.c: New file.
245 * config/riscv/riscv-protos.h (riscv_d_target_versions): New
247 * config/riscv/riscv.h (TARGET_D_CPU_VERSIONS): Define.
248 * config/riscv/t-riscv (riscv-d.o): New rule.
249 * config/rs6000/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
250 * config/rs6000/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
251 * config/rs6000/rs6000-d.c: New file.
252 * config/rs6000/rs6000-protos.h (rs6000_d_target_versions): New
254 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
255 Support GNU D by using 0 as the language type.
256 * config/rs6000/rs6000.h (TARGET_D_CPU_VERSIONS): Define.
257 * config/rs6000/t-rs6000 (rs6000-d.o): New rule.
258 * config/s390/s390-d.c: New file.
259 * config/s390/s390-protos.h (s390_d_target_versions): New prototype.
260 * config/s390/s390.h (TARGET_D_CPU_VERSIONS): Define.
261 * config/s390/t-s390 (s390-d.o): New rule.
262 * config/sparc/sparc-d.c: New file.
263 * config/sparc/sparc-protos.h (sparc_d_target_versions): New
265 * config/sparc/sparc.h (TARGET_D_CPU_VERSIONS): Define.
266 * config/sparc/t-sparc (sparc-d.o): New rule.
267 * config/t-glibc (glibc-d.o): New rule.
268 * configure: Regenerated.
269 * configure.ac (tm_d_file): New variable.
270 (tm_d_file_list, tm_d_include_list, d_target_objs): Add substitutes.
271 * doc/contrib.texi (Contributors): Add self for the D frontend.
272 * doc/frontends.texi (G++ and GCC): Mention D as a supported language.
273 * doc/install.texi (Configuration): Mention libphobos as an option for
274 --enable-shared. Mention d as an option for --enable-languages.
275 (Testing): Mention check-d as a target.
276 * doc/invoke.texi (Overall Options): Mention .d, .dd, and .di as file
277 name suffixes. Mention d as a -x option.
278 * doc/sourcebuild.texi (Top Level): Mention libphobos.
279 * doc/standards.texi (Standards): Add section on D language.
280 * doc/tm.texi: Regenerated.
281 * doc/tm.texi.in: Add @node for D language and ABI, and @hook for
282 TARGET_CPU_VERSIONS, TARGET_D_OS_VERSIONS, and TARGET_D_CRITSEC_SIZE.
283 * dwarf2out.c (is_dlang): New function.
284 (gen_compile_unit_die): Use DW_LANG_D for D.
285 (declare_in_namespace): Return module die for D, instead of adding
286 extra declarations into the namespace.
287 (gen_namespace_die): Generate DW_TAG_module for D.
288 (gen_decl_die): Handle CONST_DECLSs for D.
289 (dwarf2out_decl): Likewise.
290 (prune_unused_types_walk_local_classes): Handle DW_tag_interface_type.
291 (prune_unused_types_walk): Handle DW_tag_interface_type same as other
293 * gcc.c (default_compilers): Add entries for .d, .dd and .di.
294 * genhooks.c: Include d/d-target.def.
296 2018-10-28 Iain Sandoe <iain@sandoe.co.uk>
299 * config/rs6000/darwin.h (STACK_BOUNDARY): New.
300 (RS6000_STARTING_FRAME_OFFSET): Adjust to preserve 16byte alignment.
301 (STACK_DYNAMIC_OFFSET): Likewise.
303 2018-10-27 Sandra Loosemore <sandra@codesourcery.com>
306 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Fix
309 2018-10-26 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
311 * doc/extend.texi (PowerPC builtins): Fix __builtin_unpack_ibm128
312 return type and other typos.
314 2018-10-26 Aaron Sawdey <acsawdey@linux.ibm.com>
316 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Change to
317 a shorter sequence with fewer branches.
318 (emit_final_str_compare_gpr): Ditto.
320 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
322 * config/rs6000/tmmintrin.h: New file.
323 * config.gcc (powerpc*-*-*): Add tmmintrin.h to extra_headers.
325 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
327 * config/rs6000/mmintrin.h: Enable 32bit compilation.
328 * config/rs6000/xmmintrin.h: Likewise.
330 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
332 * config/rs6000/xmmintrin.h (_mm_extract_pi16): Fix for big-endian.
334 2018-10-26 Richard Biener <rguenther@suse.de>
336 * tree-vect-slp.c (vect_mark_slp_stmts): Add visited hash_set
338 (vect_mark_slp_stmts_relevant): Likewise.
339 (vect_detect_hybrid_slp_stmts): Likewise.
340 (vect_bb_slp_scalar_cost): Likewise.
341 (vect_remove_slp_scalar_calls): Likewise.
343 2018-10-26 Jan Hubicka <jh@suse.cz>
345 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix recursion.
346 (warn_types_mismatch): Fix walk of DECL_NAME.
347 (odr_types_equivalent_p): Fix overactive assert.
349 2018-10-26 Richard Biener <rguenther@suse.de>
351 PR tree-optimization/87105
352 * tree-vectorizer.h (_slp_tree::refcnt): New member.
353 * tree-vect-slp.c (vect_free_slp_tree): Decrement and honor
355 (vect_create_new_slp_node): Initialize refcnt to one.
357 (scalar_stmts_set_t, bst_fail): Remove.
358 (vect_build_slp_tree_2): Add bst_map argument and adjust calls.
359 (vect_build_slp_tree): Add bst_map argument and lookup
360 already created SLP nodes.
361 (vect_print_slp_tree): Handle a SLP graph, print SLP node
363 (vect_slp_rearrange_stmts): Handle a SLP graph.
364 (vect_analyze_slp_instance): Adjust and free SLP nodes from
365 the CSE map. Fix indenting.
366 (vect_schedule_slp_instance): Add short-cut.
368 2018-10-26 Martin Liska <mliska@suse.cz>
371 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Hash with
372 addr_expr and not with pointers.
374 2018-10-26 Jan Hubicka <jh@suse.cz>
376 * tree.c (free_lang_data_in_type): Only check main variants.
377 * ipa-devirt.c (warn_odr): Make static.
378 (types_same_for_odr): Drop strict variant.
379 (types_odr_comparable): Likewise.
380 (odr_or_derived_type_p): Look for main variants.
381 (odr_name_hasher::equal): Cleanup comment.
382 (odr_subtypes_equivalent): Add warn and warned arguments; check main
384 (type_variants_equivalent_p): break out from ...
385 (odr_types_equivalent): ... here; go for main variants where needed.
386 (warn_odr): ... here; turn static.
387 (warn_types_mismatch): Compare mangled names of main variants.
388 * ipa-utils.h (types_odr_comparable): Drop strict parameter.
389 (type_with_linkage_p): Sanity check that we look at main variant.
390 * lto.c (lto_read_decls): Only consider main variant to be ODR type.
391 * tree.h (types_same_for_odr): Drop strict argument.
393 2018-10-26 Richard Biener <rguenther@suse.de>
395 PR tree-optimization/87746
396 * tree-vect-data-refs.c (vect_update_misalignment_for_peel):
397 Simplify and fix WRT strided store groups with size not
398 equal to step in element count.
399 (vect_analyze_group_access_1): Dump the whole group.
401 2018-10-25 Carl Love <cel@us.ibm.com>
403 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_VSCEDPGT,
404 P9V_BUILTIN_VEC_VSCEDPLT, P9V_BUILTIN_VEC_VSCEDPEQ,
405 P9V_BUILTIN_VEC_VSCEDPUO): Rename base overloaded name. Add quad
406 precicion entry for each overloaded builtin.
407 * config/rs6000/rs6000-builtin.def (VSCEDPGT, VSCEDPLT, VSCEDPEQ,
408 VSCEDPUO): Rename overloaded name.
409 (VSCEDPGT, VSCEQPGT, VSCEDPLT, VSCEQPLT, VSCEDPEQ, VSCEQPEQ,
410 VSCEDPUO, VSCEQPUO): Add defitions for overloaded builtins.
411 * config/rs6000/vsx.md (xscmpexpqp_<code>_<mode>): Add
412 define_expand for xscmpexqp instruction.
413 (*xscmpexpqp): Add define_insn for the xscmpexqp instruction.
415 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
416 Jinsong Ji <jji@us.ibm.com>
418 * config/rs6000/emmintrin.h (_mm_slli_epi16): Replace deprecated
419 function with vec_sl.
420 (_mm_slli_epi32): Likewise.
421 (_mm_slli_epi64): Likewise.
422 (_mm_srai_epi16): Replace deprecated function with vec_sra.
423 (_mm_srai_epi32): Likewise.
424 (_mm_srli_epi16): Replace deprecated function with vec_sr.
425 (_mm_srli_epi32): Likewise.
426 (_mm_srli_epi64): Likewise.
427 (_mm_sll_epi16): Replace deprecated function with vec_sl.
428 (_mm_sll_epi32): Likewise.
429 (_mm_sll_epi64): Likewise.
430 (_mm_sra_epi16): Replace deprecated function with vec_sra.
431 (_mm_sra_epi32): Likewise.
432 (_mm_srl_epi16): Replace deprecated function with vec_sr.
433 (_mm_srl_epi32): Likewise.
434 (_mm_srl_epi64): Likewise.
436 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
437 Jinsong Ji <jji@us.ibm.com>
439 * gcc/config/rs6000/emmintrin.h (_mm_sll_epi16): Replace
440 comparison operators with vec_cmp* for compatibility due to
441 unfortunate history; clean up formatting and use types more
443 (_mm_sll_epi32): Likewise.
444 (_mm_sll_epi64): Likewise.
445 (_mm_srl_epi16): Likewise.
446 (_mm_srl_epi32): Likewise.
447 (_mm_srl_epi64): Likewise.
449 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
450 Jinsong Ji <jji@us.ibm.com>
452 * config/rs6000/emmintrin.h (_mm_sll_epi64): Remove wrong cast.
453 * config/rs6000/xmmintrin.h (_mm_min_ps): Change m's type to
454 __vector __bool int. Use vec_cmpgt in preference to deprecated
455 function vec_vcmpgtfp.
456 (_mm_max_ps): Likewise.
458 2018-10-25 Jeff Law <law@redhat.com>
460 * config/rl78/rl78.c (insn_ok_now): Always re-recognize the insn
463 2018-10-25 Martin Sebor <msebor@redhat.com>
465 * doc/extend.texi (aligned): Expand attribute description.
466 (Alignment): Rename section. Discuss function arguments.
468 2018-10-25 Jan Hubicka <jh@suse.cz>
470 * ipa-devirt.c (main_odr_variant): Remove.
471 (hash_odr_name, types_same_for_odr, types_odr_comparable,
472 odr_name_hasher::equal, odr_subtypes_equivalent_p):
473 Drop use of main_odr_variant.
474 (add_type_duplicate): Silence confused warnings on integer types.
475 (get_odr_type): Always look for main variant.
476 (register_odr_type): Simplify.
478 2018-10-25 Richard Biener <rguenther@suse.de>
480 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses):
481 Initialize ng to silence error with release checking bootstrap.
483 2018-10-25 Richard Biener <rguenther@suse.de>
485 * tree-if-conv.c: Include tree-ssa-sccvn.h.
486 (tree_if_conversion): Run CSE on the if-converted loop body.
488 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
490 * config/s390/constraints.md (ZL): New constraint.
491 * config/s390/s390.c (legitimate_pic_operand_p): Accept LARL
493 * config/s390/s390.md (movdi_larl): Remove.
494 (movdi_64): Add the LARL alternative.
496 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
499 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
500 (RTX_CODE_HWINT_P): New macro.
501 (rtx_code_size): Use RTX_CODE_HWINT_P ().
503 2018-10-25 Jan Hubicka <jh@suse.cz>
505 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
508 2018-10-25 Richard Biener <rguenther@suse.de>
510 PR tree-optimization/87665
511 PR tree-optimization/87745
512 * tree-vectorizer.h (get_earlier_stmt): Remove.
513 (get_later_stmt): Pick up UID from the original non-pattern stmt.
515 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
517 * options.texi (Deprecated): Move list to Var section.
519 2018-10-24 Bill Schmidt <wschmidt@linux.ibm.com>
520 Jinsong Ji <jji@us.ibm.com>
522 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Change deprecated
523 __vector long to __vector long long.
524 (_mm_cvtpd_ps): Likewise.
525 (_mm_cvttpd_epi32): Likewise.
526 (_mm_cvtpi32_pd): Likewise.
527 (_mm_unpackhi_epi64): Likewise.
528 (_mm_unpacklo_epi64): Likewise.
530 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
532 PR rtl-optimization/87720
533 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
535 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
537 * gimple-ssa-isolate-paths.c
538 (find_implicit_erroneous_behavior): Do not change code if the
539 pass is running for warnings only.
540 (find_explicit_erroneous_behavior): Likewise.
542 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
544 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
545 Define as rs6000_mangle_decl_assembler_name.
546 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
547 long double to IEEE long double, switch the names of the long
548 double built-in functions to be <func>f128 instead of <func>l.
550 2018-10-24 Martin Sebor <msebor@redhat.com>
552 * doc/extend.texi (nonnull): List no-argument form. Reference
553 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
555 2018-10-24 Richard Biener <rguenther@suse.de>
557 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
559 2018-10-24 Martin Liska <mliska@suse.cz>
561 PR tree-optimization/84436
562 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
564 (switch_conversion::contains_linear_function_p): New.
565 (switch_conversion::build_one_array): Support linear
566 transformation on input.
567 * tree-switch-conversion.h (struct switch_conversion): Add
568 contains_linear_function_p declaration.
570 2018-10-24 Richard Biener <rguenther@suse.de>
572 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
573 if its argument is CONSTANT_CLASS_P.
575 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
577 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
578 it is wrong for forward declarations.
580 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
582 * config/s390/s390.c (s390_check_qrst_address): Add the missing
583 SYMBOL_REF_P () check.
585 2018-10-24 Richard Biener <rguenther@suse.de>
587 PR tree-optimization/87105
588 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
590 (vect_analyze_data_ref_accesses): Handle duplicate loads and
591 stores by splitting the affected group after the fact.
592 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
593 fail the SLP build because of size constraints.
595 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
597 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
598 * configure: Regenerate.
599 * config.in: Regenerate.
600 * varasm.c (mergeable_string_section): Use readonly_data_section
601 if linker doesn't support SHF_MERGE with alignment > 8.
602 (mergeable_constant_section): Likewise.
604 2018-10-24 Richard Biener <rguenther@suse.de>
606 PR tree-optimization/84013
607 * tree-ssa-structalias.c (struct msdi_data): New struct for
608 marshalling data to walk_stmt_load_store_ops.
609 (maybe_set_dependence_info): Refactor as callback for
610 walk_stmt_load_store_ops.
611 (compute_dependence_clique): Set restrict info on all stmt kinds.
613 2018-10-24 Martin Liska <mliska@suse.cz>
615 * cgraph.c (cgraph_node::dump):
616 Remove reduntant dumps and make tp_first_run dump more compact.
618 2018-10-24 Richard Biener <rguenther@suse.de>
620 PR tree-optimization/87665
621 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
624 2018-10-12 Jeff Law <law@redhat.com>
626 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
629 2018-10-23 Richard Biener <rguenther@suse.de>
631 * tree-vrp.c (add_assert_info): Guard dump_printf with
633 * gimple-ssa-evrp-analyze.c
634 (evrp_range_analyzer::record_ranges_from_incoming_edge):
635 Use value_range::ignore_equivs_equal_p.
637 2018-10-23 Richard Biener <rguenther@suse.de>
639 PR tree-optimization/87105
640 PR tree-optimization/87608
641 * passes.def (pass_all_early_optimizations): Add early phi-opt
643 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
644 addition to debug stmts.
645 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
646 and abs replacement early.
647 * tree-cfg.c (gimple_empty_block_p): Likewise.
649 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
652 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
653 specified to configure.
654 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
656 2018-10-23 Richard Biener <rguenther@suse.de>
658 PR tree-optimization/87700
659 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
661 2018-10-23 Jakub Jelinek <jakub@redhat.com>
664 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
665 second argument from __mmask16 to __mmask8.
666 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
667 _mm_mask_packs_epi32): Likewise.
668 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
670 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
672 2018-10-23 Richard Biener <rguenther@suse.de>
674 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
676 2018-10-23 Richard Biener <rguenther@suse.de>
678 PR tree-optimization/86144
679 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
682 2018-10-23 Richard Biener <rguenther@suse.de>
684 PR tree-optimization/87693
685 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
686 the case we do not find the taken edge.
688 2018-10-22 Bill Schmidt <wschmidt@linux.ibm.com>
689 Jinsong Ji <jji@us.ibm.com>
691 * config/rs6000/emmintrin.h (_MM_SHUFFLE2): Comment cleanup.
692 (_mm_store_pd): Use unaligned vector type for pointer cast.
693 (_mm_maskmoveu_si128): Likewise.
694 * config/rs6000/xmmintrin.h (__m128_u): New typedef.
695 (_mm_store_ps): Use unaligned vector type for pointer cast.
697 2018-10-22 Paul Koning <ni1d@arrl.net>
699 * symtab.c (symtab_node::increase_alignment): Correct max
702 2018-10-22 Yury Gribov <tetra2005@gmail.com>
704 PR tree-optimization/87633
705 * match.pd: Do not generate unordered integer comparisons.
707 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
709 PR rtl-optimization/87600
710 * combine.c: Add include of expr.h.
711 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
712 register to a pseudo.
713 (make_more_copies): New function, add a copy to a new pseudo after
714 the moves from hard registers into pseudos.
715 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
716 later. Call make_more_copies.
718 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
720 * lra-constraints.c (process_alt_operands): New local array,
721 matching_early_clobber. Check matching_early_clobber before
722 decrementing reject, and set matching_early_clobber after.
724 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
727 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
728 call output_addr_const and hope for the best.
730 2018-10-22 Richard Biener <rguenther@suse.de>
732 * gimple-ssa-evrp-analyze.c
733 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
734 smarter about what ranges to use.
735 * tree-vrp.c (add_assert_info): Dump here.
736 (register_edge_assert_for_2): Instead of here at multiple but
739 * gcc.dg/tree-ssa/evrp12.c: New testcase.
740 * gcc.dg/predict-6.c: Adjust.
741 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
742 * gcc.dg/tree-ssa/vrp02.c: Likewise.
743 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
745 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
746 Richard Biener <rguenther@suse.de>
748 * bitmap.h: Update data structure documentation, including a
749 description of bitmap views as either linked-lists or splay trees.
750 (struct bitmap_element_def): Update comments for splay tree bitmaps.
751 (struct bitmap_head_def): Likewise.
752 (bitmap_list_view, bitmap_tree_view): New prototypes.
753 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
755 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
756 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
757 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
758 released bitmap element here.
759 (bitmap_element_free): Remove.
760 (bitmap_elt_clear_from): Work on splay tree bitmaps.
761 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
762 this function similar ones such that linked-list bitmap implementation
763 functions are grouped.
764 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
765 and moved for grouping.
766 (bitmap_list_insert_element_after): Renamed from
767 bitmap_elt_insert_after, and moved for grouping.
768 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
769 (bitmap_tree_link_left, bitmap_tree_link_right,
770 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
771 bitmap_tree_link_element, bitmap_tree_unlink_element,
772 bitmap_tree_find_element): New functions for splay-tree bitmap
774 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
775 Renamed and moved, see above entries.
776 (bitmap_tree_listify_from): New function to convert part of a splay
777 tree bitmap to a linked-list bitmap.
778 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
779 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
780 (bitmap_find_bit): Remove.
781 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
782 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
783 Handle splay tree bitmaps.
784 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
785 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
786 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
787 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
788 bitmap_intersect_compl_p, bitmap_ior_and_compl,
789 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
790 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
791 corresponding changes to use linked-list specific bitmap_element
792 manipulation functions as applicable for efficiency.
793 (bitmap_tree_to_vec): New function.
794 (debug_bitmap_elt_file): New function split out from ...
795 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
796 (bitmap_print): Likewise.
798 PR tree-optimization/63155
799 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
801 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
804 2018-10-22 Martin Liska <mliska@suse.cz>
806 PR tree-optimization/87686
808 2018-08-29 Martin Liska <mliska@suse.cz>
810 * tree-switch-conversion.c (switch_conversion::expand):
811 Strenghten assumption about gswitch statements.
813 2018-10-22 Martin Liska <mliska@suse.cz>
815 * ipa-icf.c (sem_item::compare_attributes): Remove.
816 (sem_item::compare_referenced_symbol_properties): Use
817 attribute_list_equal instead.
818 (sem_function::equals_wpa): Likewise.
819 * ipa-icf.h: Remove compare_attributes.
821 2018-10-22 Richard Biener <rguenther@suse.de>
824 * mem-stats.h (mem_usage::operator==): Fix pasto.
826 2018-10-22 Richard Biener <rguenther@suse.de>
828 PR tree-optimization/87640
829 * tree-vrp.c (set_value_range_with_overflow): Decompose
831 (extract_range_from_binary_expr_1): Adjust.
833 2018-10-22 Martin Jambor <mjambor@suse.cz>
835 * tree-eh.h (stmt_could_throw_p): Add function parameter.
836 (stmt_can_throw_external): Likewise.
837 (stmt_can_throw_internal): Likewise.
838 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
839 (lower_eh_constructs_2): Likewise.
840 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
841 (stmt_can_throw_external): Likewise.
842 (stmt_can_throw_internal): Likewise.
843 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
844 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
845 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
846 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
847 (pass_lower_eh_dispatch::execute): Pass cfun to
848 stmt_can_throw_external.
849 (cleanup_empty_eh): Likewise.
850 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
851 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
852 stmt_can_throw_external instead of pushing it to cfun.
853 (symbol_table::create_edge): Likewise.
854 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
855 stmt_can_throw_internal.
856 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
857 to stmt_could_throw_p.
858 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
859 stmt_can_throw_internal.
860 (pass_store_merging::execute): Likewise.
861 * gimple-ssa-strength-reduction.c
862 (find_candidates_dom_walker::before_dom_children): Pass cfun to
864 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
865 stmt_can_throw_internal.
866 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
867 to stmt_can_throw_external.
868 (check_stmt): Pass cfun to stmt_could_throw_p.
869 (check_stmt): Pass cfun to stmt_can_throw_external.
870 (pass_nothrow::execute): Likewise.
871 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
872 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
873 stmt_can_throw_internal.
874 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
875 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
876 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
877 * tree-complex.c (expand_complex_libcall): Pass cfun to
878 stmt_could_throw_p and to stmt_can_throw_internal.
879 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
880 * tree-inline.c (copy_edges_for_bb): Likewise.
881 (maybe_move_debug_stmts_to_successors): Likewise.
882 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
884 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
885 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
886 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
887 stmt_can_throw_internal.
888 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
889 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
891 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
892 stmt_can_throw_internal.
893 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
894 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
896 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
897 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
898 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
899 (convert_mult_to_fma_1): Likewise.
900 (convert_to_divmod): Likewise.
901 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
902 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
903 * tree-ssa-propagate.c
904 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
905 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
906 (maybe_optimize_range_tests): Likewise.
907 (linearize_expr_tree): Likewise.
908 (reassociate_bb): Likewise.
909 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
910 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
911 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
912 (handle_char_store): Likewise.
913 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
914 stmt_can_throw_internal.
915 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
917 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
918 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
919 (vectorizable_simd_clone_call): Likewise.
920 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
921 (gimple_stringop_fixed_value): Likewise.
923 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
925 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
926 literal pool references.
927 (s390_check_qrst_address): Adapt to the new behavior of
928 s390_loadrelative_operand_p ().
930 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
933 * config/i386/sse.md (*andnot<mode>3_bcst): New.
935 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
938 * config/i386/sse.md (*<code><mode>3_bcst): New.
940 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
943 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
944 V4DI, V16SI and V8DI.
945 (*sub<mode>3<mask_name>_bcst): New.
946 (*add<mode>3<mask_name>_bcst): Likewise.
948 2018-10-21 Bill Schmidt <wschmidt@linux.ibm.com>
949 Jinsong Ji <jji@us.ibm.com>
951 * config/rs6000/emmintrin.h (_mm_movemask_pd): Replace __vector
952 __m64 with __vector unsigned long long for compatibility.
953 (_mm_movemask_epi8): Likewise.
954 * config/rs6000/xmmintrin.h (_mm_cvtps_pi32): Likewise.
955 (_mm_cvttps_pi32): Likewise.
956 (_mm_cvtpi32_ps): Likewise.
957 (_mm_cvtps_pi16): Likewise.
958 (_mm_loadh_pi): Likewise.
959 (_mm_storeh_pi): Likewise.
960 (_mm_movehl_ps): Likewise.
961 (_mm_movelh_ps): Likewise.
962 (_mm_loadl_pi): Likewise.
963 (_mm_storel_pi): Likewise.
964 (_mm_movemask_ps): Likewise.
965 (_mm_shuffle_pi16): Likewise.
967 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
970 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
971 __builtin_ia32_vfnmsubpd512_mask.
972 (_mm512_mask_fnmsub_round_pd): Likewise.
973 (_mm512_fnmsub_pd): Likewise.
974 (_mm512_mask_fnmsub_pd): Likewise.
975 (_mm512_maskz_fnmsub_round_pd): Use
976 __builtin_ia32_vfnmsubpd512_maskz.
977 (_mm512_maskz_fnmsub_pd): Likewise.
978 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
979 (_mm512_mask_fnmsub_round_ps): Likewise.
980 (_mm512_fnmsub_ps): Likewise.
981 (_mm512_mask_fnmsub_ps): Likewise.
982 (_mm512_maskz_fnmsub_round_ps): Use
983 __builtin_ia32_vfnmsubps512_maskz.
984 (_mm512_maskz_fnmsub_ps): Likewise.
985 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
986 __builtin_ia32_vfnmsubpd256_mask.
987 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
988 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
989 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
990 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
991 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
992 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
993 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
994 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
995 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
996 __builtin_ia32_vfnmsubpd.
997 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
998 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
999 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
1000 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
1001 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
1002 * config/i386/i386-builtin.def: Add
1003 __builtin_ia32_vfnmsubpd256_mask,
1004 __builtin_ia32_vfnmsubpd256_maskz,
1005 __builtin_ia32_vfnmsubpd128_mask,
1006 __builtin_ia32_vfnmsubpd128_maskz,
1007 __builtin_ia32_vfnmsubps256_mask,
1008 __builtin_ia32_vfnmsubps256_maskz,
1009 __builtin_ia32_vfnmsubps128_mask,
1010 __builtin_ia32_vfnmsubps128_maskz,
1011 __builtin_ia32_vfnmsubpd512_mask,
1012 __builtin_ia32_vfnmsubpd512_maskz,
1013 __builtin_ia32_vfnmsubps512_mask,
1014 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
1015 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
1016 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
1017 __builtin_ia32_vfnmsubpd256.
1018 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
1019 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
1020 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
1022 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
1024 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
1026 (fmai_vmfnmsub_<mode><round_name>): Likewise.
1028 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1031 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
1032 __builtin_ia32_vfnmaddpd512_mask.
1033 (_mm512_mask_fnmadd_round_pd): Likewise.
1034 (_mm512_fnmadd_pd): Likewise.
1035 (_mm512_mask_fnmadd_pd): Likewise.
1036 (_mm512_maskz_fnmadd_round_pd): Use
1037 __builtin_ia32_vfnmaddpd512_maskz.
1038 (_mm512_maskz_fnmadd_pd): Likewise.
1039 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
1040 (_mm512_mask_fnmadd_round_ps): Likewise.
1041 (_mm512_fnmadd_ps): Likewise.
1042 (_mm512_mask_fnmadd_ps): Likewise.
1043 (_mm512_maskz_fnmadd_round_ps): Use
1044 __builtin_ia32_vfnmaddps512_maskz.
1045 (_mm512_maskz_fnmadd_ps): Likewise.
1046 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
1047 __builtin_ia32_vfnmaddpd256_mask.
1048 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
1049 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
1050 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
1051 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
1052 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
1053 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
1054 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
1055 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
1056 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
1057 __builtin_ia32_vfnmaddpd.
1058 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
1059 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
1060 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
1061 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
1062 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
1063 * config/i386/i386-builtin.def: Add
1064 __builtin_ia32_vfnmaddpd256_mask,
1065 __builtin_ia32_vfnmaddpd256_maskz,
1066 __builtin_ia32_vfnmaddpd128_mask,
1067 __builtin_ia32_vfnmaddpd128_maskz,
1068 __builtin_ia32_vfnmaddps256_mask,
1069 __builtin_ia32_vfnmaddps256_maskz,
1070 __builtin_ia32_vfnmaddps128_mask,
1071 __builtin_ia32_vfnmaddps128_maskz,
1072 __builtin_ia32_vfnmaddpd512_mask,
1073 __builtin_ia32_vfnmaddpd512_maskz,
1074 __builtin_ia32_vfnmaddps512_mask,
1075 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
1076 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
1077 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
1078 __builtin_ia32_vfnmaddpd256.
1079 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
1080 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
1081 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
1083 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
1085 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
1087 (fmai_vmfnmadd_<mode><round_name>): Likewise.
1089 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1092 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
1093 __builtin_ia32_vfmsubpd512_mask.
1094 (_mm512_mask_fmsub_round_pd): Likewise.
1095 (_mm512_fmsub_pd): Likewise.
1096 (_mm512_mask_fmsub_pd): Likewise.
1097 (_mm512_maskz_fmsub_round_pd): Use
1098 __builtin_ia32_vfmsubpd512_maskz.
1099 (_mm512_maskz_fmsub_pd): Likewise.
1100 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
1101 (_mm512_mask_fmsub_round_ps): Likewise.
1102 (_mm512_fmsub_ps): Likewise.
1103 (_mm512_mask_fmsub_ps): Likewise.
1104 (_mm512_maskz_fmsub_round_ps): Use
1105 __builtin_ia32_vfmsubps512_maskz.
1106 (_mm512_maskz_fmsub_ps): Likewise.
1107 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
1108 __builtin_ia32_vfmsubpd256_mask.
1109 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
1110 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
1111 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
1112 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
1113 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
1114 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
1115 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
1116 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
1117 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
1118 __builtin_ia32_vfmsubpd.
1119 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
1120 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
1121 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
1122 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
1123 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
1124 * config/i386/i386-builtin.def: Add
1125 __builtin_ia32_vfmsubpd256_mask,
1126 __builtin_ia32_vfmsubpd256_maskz,
1127 __builtin_ia32_vfmsubpd128_mask,
1128 __builtin_ia32_vfmsubpd128_maskz,
1129 __builtin_ia32_vfmsubps256_mask,
1130 __builtin_ia32_vfmsubps256_maskz,
1131 __builtin_ia32_vfmsubps128_mask,
1132 __builtin_ia32_vfmsubps128_maskz,
1133 __builtin_ia32_vfmsubpd512_mask,
1134 __builtin_ia32_vfmsubpd512_maskz,
1135 __builtin_ia32_vfmsubps512_mask,
1136 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
1137 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
1138 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
1139 __builtin_ia32_vfmsubpd256.
1140 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
1141 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
1142 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
1144 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
1146 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
1148 (fmai_vmfmsub_<mode><round_name>): Likewise.
1150 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1152 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
1153 Remove plus. Renamed to ...
1154 (*sub<mode>3<mask_name>_bcst): This.
1155 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
1156 (*add<mode>3<mask_name>_bcst): This.
1158 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1161 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
1163 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1166 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
1167 (_mm_or_epi32): Likewise.
1168 (_mm256_xor_epi32): Likewise.
1169 (_mm_xor_epi32): Likewise.
1170 (_mm256_or_epi64): Likewise.
1171 (_mm_or_epi64): Likewise.
1172 (_mm256_xor_epi64): Likewise.
1173 (_mm_xor_epi64): Likewise.
1175 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
1178 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
1180 2018-10-20 Jakub Jelinek <jakub@redhat.com>
1183 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
1185 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
1187 * doc/ux.texi: Move @section directly after @node.
1189 2018-10-19 Jakub Jelinek <jakub@redhat.com>
1193 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
1194 depend closely nested inside of loop with ordered clause with
1197 2018-10-19 David Malcolm <dmalcolm@redhat.com>
1199 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
1200 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
1201 * doc/ux.texi: New file.
1203 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
1205 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
1206 be the first CR field allocated.
1208 2018-10-19 Richard Biener <rguenther@suse.de>
1211 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
1212 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
1214 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
1217 * config/i386/sse.md
1218 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
1219 (*add<mode>3<mask_name>_bcst_2): Likewise.
1221 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
1223 * config/i386/sse.md
1224 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
1225 Replace nonimmediate_operand with register_operand.
1226 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
1228 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
1231 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
1233 PR rtl-optimization/87596
1234 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
1235 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
1236 for instructions in FROM..TO range.
1238 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
1240 * cfgexpand.c (expand_one_var): Use specific wording in error message
1241 for non-local frame variables.
1242 * stor-layout.c (layout_decl): Do not issue a warning for them.
1244 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
1246 * haifa-sched.c (priority): Add force_recompute parameter.
1247 (apply_replacement): Call priority () with force_recompute = true.
1248 (restore_pattern): Likewise.
1250 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1252 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
1253 HOST_BITS_PER_WIDE_INT.
1254 (test_vector_ops_duplicate): Likewise.
1256 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1259 * config/i386/sse.md (VF_AVX512): New.
1260 (avx512bcst): Likewise.
1261 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
1263 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
1265 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
1268 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
1270 * doc/invoke.texi (-dumpversion): Improve grammar.
1271 (-dumpfullversion): Make more consistent with -dumpversion.
1273 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
1275 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
1276 Set JUMP_LABEL to the jump insn.
1277 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
1278 Predict emitted jump and add label to jump insn.
1280 2018-10-18 David Malcolm <dmalcolm@redhat.com>
1282 PR tree-optimization/87562
1283 * input.c (get_substring_ranges_for_loc): Use
1284 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
1285 getting the linemap for the endpoint. Verify that it's either
1286 in the same linemap as the start point's spelling location, or
1287 at least in the same file.
1289 2018-10-18 Richard Biener <rguenther@suse.de>
1291 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
1292 feed width-specific load/store costs through ix86_vec_cost.
1293 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
1294 (k8_cost): Likewise.
1295 (bdver_cost): Likewise.
1296 (znver1_cost): Likewise.
1297 (btver1_cost): Likewise.
1298 (btver2_cost): Likewise.
1300 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1302 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
1303 to simplify subreg of vec_merge.
1305 2018-10-18 Richard Biener <rguenther@suse.de>
1307 * config/i386/i386.c: Fix costing of vector FMA.
1309 2018-10-18 Richard Biener <rguenther@suse.de>
1311 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
1313 (ix86_builtin_vectorization_cost): For vec_construct properly
1314 cost insertion into SSE regs.
1315 (...): Adjust calls to ix86_vec_cost.
1317 2018-10-18 Richard Biener <rguenther@suse.de>
1321 2018-02-07 Richard Biener <rguenther@suse.de>
1323 PR tree-optimization/84204
1324 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
1327 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1330 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
1332 (test_vector_ops_duplicate): Add test for a scalar subreg of a
1333 VEC_MERGE of a VEC_DUPLICATE.
1335 2018-10-17 Joseph Myers <joseph@codesourcery.com>
1337 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
1338 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
1339 * doc/standards.texi (C Language): Document C2X.
1340 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
1341 (rl78_option_override): Handle "GNU C2X" language name.
1343 2018-10-17 Joseph Myers <joseph@codesourcery.com>
1345 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
1346 Document C17 as published in 2018.
1348 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
1351 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
1352 bail out if both sides do not have the same storage order.
1354 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
1356 * bitmap.c (bitmap_head::dump): New.
1357 * bitmap.h (bitmap_head): Add dump().
1358 * gimple-ssa-evrp-analyze.c
1359 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
1360 (evrp_range_analyzer::set_ssa_range_info): Same.
1361 (evrp_range_analyzer::record_ranges_from_phis): Same.
1362 (evrp_range_analyzer::record_ranges_from_stmt): Same.
1363 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
1364 * gimple-ssa-sprintf.c (get_int_range): Same.
1365 (format_integer): Same.
1366 (sprintf_dom_walker::handle_gimple_call): Same.
1367 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
1368 (ipcp_vr_lattice::top_p): Same.
1369 (ipcp_vr_lattice::bottom_p): Same.
1370 (ipcp_vr_lattice::set_to_bottom): Same.
1371 (ipa_vr_operation_and_type_effects): Same.
1372 (propagate_vr_across_jump_function): Same.
1373 (ipcp_store_vr_results): Same.
1374 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
1375 (ipa_print_node_jump_functions_for_edge): Same.
1376 (ipa_get_value_range): Same.
1377 (ipa_compute_jump_functions_for_edge): Same.
1378 (ipa_write_jump_function): Same.
1379 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
1380 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
1382 * vr-values.c (set_value_range_to_nonnegative): Same.
1383 (set_value_range_to_truthvalue): Same.
1384 (vr_values::get_value_range): Same.
1385 (vr_values::set_defs_to_varying): Same.
1386 (vr_values::update_value_range): Same.
1387 (symbolic_range_based_on_p): Same.
1388 (vr_values::op_with_boolean_value_range_p): Same.
1389 (vr_values::extract_range_for_var_from_comparison_expr): Same.
1390 (vr_values::extract_range_from_ssa_name): Same.
1391 (vr_values::extract_range_from_binary_expr): Same.
1392 (vr_values::extract_range_from_unary_expr): Same.
1393 (vr_values::extract_range_from_cond_expr): Same.
1394 (vr_values::extract_range_from_comparison): Same.
1395 (vr_values::check_for_binary_op_overflow): Same.
1396 (vr_values::extract_range_basic): Same.
1397 (vr_values::extract_range_from_assignment): Same.
1398 (compare_ranges): Same.
1399 (compare_range_with_value): Same.
1400 (vr_values::adjust_range_with_scev): Same.
1401 (vrp_valueize): Same.
1402 (vrp_valueize_1): Same.
1403 (vr_values::get_vr_for_comparison): Same.
1404 (vr_values::compare_name_with_value): Same.
1405 (vr_values::compare_names): Same.
1406 (vr_values::vrp_evaluate_conditional): Same.
1407 (find_case_label_ranges): Same.
1408 (vr_values::vrp_visit_switch_stmt): Same.
1409 (vr_values::extract_range_from_phi_node): Same.
1410 (vr_values::simplify_div_or_mod_using_ranges): Same.
1411 (vr_values::simplify_bit_ops_using_ranges): Same.
1412 (test_for_singularity): Same.
1413 (range_fits_type_p): Same.
1414 (vr_values::simplify_cond_using_ranges_1): Same.
1415 (vr_values::simplify_switch_using_ranges): Same.
1416 (vr_values::simplify_float_conversion_using_ranges): Same.
1417 (vr_values::two_valued_val_range_p): Same.
1418 (vr_values::add_equivalence): Move to value_range::equiv_add.
1419 * vr-values.h (vr_values::add_equivalence): Remove.
1420 (VR_INITIALIZER): Remove.
1421 * tree-vrp.c (value_range::set): New.
1422 (value_range::equiv_add): New.
1423 (value_range::value_range): New.
1424 (value_range::deep_copy): New.
1425 (value_range::check): New.
1426 (value_range::equal_p): New.
1427 (value_range::ignore_equivs_equal_p): New.
1428 (value_range::operator==): New.
1429 (value_range::operator!=): New.
1430 (value_range::symbolic_p): New.
1431 (value_range::numeric_p): New.
1432 (value_range::set_undefined): New.
1433 (value_range::set_varying): New.
1434 (value_range::may_contain_p): New.
1435 (value_range::equiv_clear): New.
1436 (value_range::singleton_p): New.
1437 (value_range::intersect): New.
1438 (value_range::dump): New.
1439 (value_range::set_and_canonicalize): New.
1440 (set_value_range): Adjust for value_range API.
1441 (set_value_range_to_undefined): Same.
1442 (set_value_range_to_varying): Same.
1443 (set_and_canonicalize_value_range): Same.
1444 (set_value_range_to_nonnull): Same.
1445 (set_value_range_to_null): Same.
1446 (range_is_null): Same.
1447 (range_is_nonnull): Same.
1448 (range_int_cst_p): Same.
1449 (range_int_cst_singleton_p): Same.
1450 (symbolic_range_p): Same.
1451 (range_includes_zero_p): Same.
1452 (value_range_constant_singleton): Same.
1453 (vrp_set_zero_nonzero_bits): Same.
1454 (ranges_from_anti_range): Same.
1455 (extract_range_into_wide_ints): Same.
1456 (extract_range_from_multiplicative_op): Same.
1457 (set_value_range_with_overflow): Same.
1458 (extract_range_from_binary_expr_1): Same.
1459 (extract_range_from_unary_expr): Same.
1460 (dump_value_range): Same.
1461 (debug_value_range): Same.
1462 (vrp_prop::check_array_ref): Same.
1463 (vrp_prop::check_mem_ref): Same.
1464 (vrp_prop::vrp_initialize): Same.
1465 (vrp_prop::visit_stmt): Same.
1466 (intersect_ranges): Same.
1467 (vrp_prop::visit_phi): Same.
1468 (vrp_prop::vrp_finalize): Same.
1469 (determine_value_range_1): Same.
1470 (determine_value_range): Same.
1471 (vrp_intersect_ranges_1): Rename to...
1472 (vrp_intersect_1): this.
1473 (vrp_intersect_ranges): Rename to...
1474 (value_range::intersect_helper): ...this.
1475 (vrp_meet_1): Rename to...
1476 (value_range::union_helper): ...this.
1477 (vrp_meet): Rename to...
1478 (value_range::union_): ...this.
1479 (copy_value_range): Remove.
1480 * tree-vrp.h (struct value_range): Rewrite into a proper class.
1481 (value_range::vrtype): New.
1482 (value_range::type): New.
1483 (value_range::equiv): New.
1484 (value_range::min): New.
1485 (value_range::max): New.
1486 (value_range::varying_p): New.
1487 (value_range::undefined_p): New.
1488 (value_range::null_p): New.
1489 (value_range::equiv_add): New.
1490 (copy_value_range): Remove.
1492 2018-10-17 David Malcolm <dmalcolm@redhat.com>
1494 * Makefile.in (SELFTEST_TARGETS): New.
1495 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
1496 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
1497 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
1499 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
1500 (selftest-c++-gdb, selftest-c++-valgrind): Move to
1502 * configure: Regenerate.
1503 * configure.ac (selftest_languages): New.
1505 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
1507 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
1508 overflow wraps argument.
1509 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
1510 wide_int_range_multiplicative_op.
1511 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
1512 overflow wraps argument.
1513 (wide_int_range_multiplicative_op): Same.
1514 (wide_int_range_lshift): Same.
1515 (wide_int_range_div): Same.
1516 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
1517 (wide_int_range_lshift): Same.
1518 (wide_int_range_div): Same.
1520 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
1522 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
1523 use sign as argument.
1524 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
1525 wide_int_range_shift_undefined_p.
1527 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1529 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
1531 (@despeculate_copy<ALLI_TI:mode>): ... This.
1532 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
1535 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1537 * config.gcc: Obsolete *-*-solaris2.10*.
1538 * doc/install.texi (Specific, *-*-solaris2*): Document it.
1540 2018-10-12 Jeff Law <law@redhat.com>
1542 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
1543 reg + sym +- const_int addressing modes.
1545 2018-10-15 David Malcolm <dmalcolm@redhat.com>
1547 * common.opt (fdiagnostics-minimum-margin-width=): New option.
1548 * diagnostic-show-locus.c (layout::layout): Apply the minimum
1550 (layout::start_annotation_line): Only print up to 3 of the
1551 margin character, to avoid touching the left-hand side.
1552 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
1553 minimum margin width, as set by test_diagnostic_context's ctor.
1554 (selftest::test_fixit_insert_containing_newline): Likewise.
1555 (selftest::test_fixit_insert_containing_newline_2): Likewise.
1556 (selftest::test_line_numbers_multiline_range): Clear
1557 dc.min_margin_width.
1558 * diagnostic.c (diagnostic_initialize): Initialize
1560 * diagnostic.h (struct diagnostic_context): Add field
1562 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
1563 * opts.c (common_handle_option): Handle
1564 OPT_fdiagnostics_minimum_margin_width_.
1565 * selftest-diagnostic.c
1566 (selftest::test_diagnostic_context::test_diagnostic_context):
1567 Initialize min_margin_width to 6.
1568 * toplev.c (general_init): Initialize global_dc->min_margin_width.
1570 2018-10-15 David Malcolm <dmalcolm@redhat.com>
1572 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1573 Fix usage of "error_at_rich_loc" in the comment.
1575 2018-10-15 Renlin Li <renlin.li@arm.com>
1578 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
1579 if-conversioned loop when it contains ifn with types not
1580 supported by backend.
1581 * internal-fn.c (expand_direct_optab_fn): Add an assert.
1582 (direct_internal_fn_supported_p): New helper function.
1583 * internal-fn.h (direct_internal_fn_supported_p): Declare.
1585 2018-10-15 Jakub Jelinek <jakub@redhat.com>
1588 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
1589 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
1590 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
1592 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
1594 PR tree-optimization/87022
1595 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
1596 bits in dist vector rather than the first one.
1598 2018-10-15 Richard Biener <rguenther@suse.de>
1601 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
1602 (visit_loadstore): When a used restrict tag escaped verify that
1603 the points-to solution of "other" pointers do not include
1605 (compute_dependence_clique): If a used restrict tag escaped
1606 communicated that down to visit_loadstore.
1608 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
1610 * config/s390/s390.c (s390_expand_vec_init): Force vector element
1611 into reg if it isn't a general operand.
1613 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
1616 * config/i386/sse.md (*vec_dupv2di): Add register source to
1619 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
1622 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
1623 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
1624 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
1625 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
1627 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
1629 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
1630 (notice_args_size): Set it in the current trace if no insn that can
1631 throw internally has been seen yet.
1632 (connect_traces): When connecting args_size between traces, allow the
1633 incoming values not to match if there is an insn setting it before the
1634 first insn that can throw internally; in that case, force the creation
1635 of a CFI note on this latter insn.
1637 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
1639 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
1640 the base class, because using the injected-class-name was not clearly
1641 specified until DR 176.
1643 2018-10-12 Paul Koning <ni1d@arrl.net>
1645 * config/pdp11/pdp11.md (doloop_end): New expander.
1646 (doloop_end_insn): renamed from "doloop_end".
1647 (addqi3): New pattern.
1648 (subqi3): New pattern.
1649 * config/pdp11/predicates.md (incdec_operand): New predicate.
1651 2018-10-12 Yury Gribov <tetra2005@gmail.com>
1654 * real.c (format_helper::can_represent_integral_type_p): New function
1655 * real.h (format_helper::can_represent_integral_type_p): Ditto.
1656 * match.pd: New pattern.
1658 2018-10-12 Alexandre Oliva <oliva@adacore.com>
1660 * configure.ac: Introduce --enable-large-address-aware
1661 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
1662 * doc/install.texi: Document it.
1663 * configure, config.in: Rebuilt.
1664 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
1665 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
1666 (LINK_SPEC): Insert it.
1667 * config/i386/mingw-264.h: Likewise.
1669 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
1671 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
1673 PR rtl-optimization/87600
1674 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
1676 2018-10-12 Paul Koning <ni1d@arrl.net>
1678 * doc/md.texi (doloop_end): Document that the pattern code may
1679 need to check operand mode.
1681 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
1683 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
1684 to zero-extend between int and floating-point registers.
1685 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
1686 ldp into floating-point registers. Add type and arch attributes.
1687 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
1688 Use f_loads for type attribute.
1690 2018-10-11 Martin Sebor <msebor@redhat.com>
1692 * doc/extend.texi (attribute packed): Correct typos.
1694 2018-10-11 Martin Sebor <msebor@redhat.com>
1696 * doc/extend.texi (attribute flatten): Mention interaction with
1699 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
1702 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
1703 Set new_decl virtual flag to zero.
1705 2018-10-11 Martin Sebor <msebor@redhat.com>
1708 * doc/extend.texi (attribute format_arg): Discuss using multiple
1709 attributes on a single function.
1711 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
1713 PR tree-optimization/86829
1714 * match.pd (sin (atan (x))): New simplification rules.
1715 (cos (atan (x))): Likewise.
1716 * real.c (build_sinatan_real): New function.
1717 * real.h (build_sinatan_real): Prototype.
1719 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
1721 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
1723 (fold_mergeeo_helper): New helper function.
1724 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
1725 intrinsics. Correct some whitespace indentation issues.
1727 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
1730 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
1731 Use HOST_WIDE_INT_1U for shift.
1733 2018-10-11 Doug Rupp <rupp@adacore.com>
1734 Olivier Hainque <hainque@adacore.com>
1736 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
1737 Pass --relax to the linker for RTPs.
1738 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
1740 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
1741 Jan Hubicka <jh@suse.cz>
1742 Martin Jambor <mjambor@suse.cz>
1744 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
1745 the same elements are repeated rather than printing all of them.
1746 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
1747 "repeated" elements.
1748 * read-rtl-function.c (test_loading_repeat): New function.
1749 (read_rtl_function_c_tests): Call test_loading_repeat.
1750 * rtl-tests.c (test_dumping_repeat): New function.
1751 (rtl_tests_c_tests): Call test_dumping_repeat.
1753 2018-10-11 Richard Biener <rguenther@suse.de>
1755 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
1756 bdver?_cost): Unify to ...
1757 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
1758 * config/i386/i386.c (processor_cost_table): Adjust.
1760 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
1763 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
1764 the thunk when expanding to GIMPLE.
1766 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1768 * varasm.c (mergeable_string_section): Don't try to move zero-length
1769 strings to the merge section.
1771 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
1774 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
1776 2018-10-10 Jakub Jelinek <jakub@redhat.com>
1779 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
1780 to special_args set.
1782 2018-10-10 Richard Biener <rguenther@suse.de>
1784 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
1785 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
1786 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
1787 and recursing and pattern terminating the recursion on SSE
1788 vector width using ix86_expand_reduc.
1789 (reduc_sminmax_scal_<mode>): Split into part reducing to half
1790 width and recursing and SSE2 vector variant doing the final
1791 reduction with ix86_expand_reduc.
1792 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
1793 with terminating the recursion at AVX level, splitting that
1796 2018-10-09 David Malcolm <dmalcolm@redhat.com>
1798 * genmatch.c (error_cb): Rename to...
1799 (diagnostic_cb): ...this, converting int params to enums.
1800 (fatal_at): Update for renaming.
1801 (warning_at): Likewise.
1803 * input.c (selftest::ebcdic_execution_charset::apply):
1804 Update for renaming of...
1805 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
1807 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
1808 converting level and reason to enums.
1809 (class selftest::lexer_error_sink): Rename to...
1810 (class selftest::lexer_test_options): ...this, renaming field
1811 "m_errors" to "m_diagnostics".
1812 (selftest::lexer_test_options::apply): Update for renaming of...
1813 (selftest::lexer_test_options::on_error): ...this, renaming to...
1814 (selftest::lexer_test_options::on_diagnostic): ...this
1815 converting level and reason to enums.
1816 (selftest::test_lexer_string_locations_raw_string_unterminated):
1817 Update for renamings.
1818 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
1821 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
1823 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
1824 * config/rs6000/pmmintrin.h: New file.
1826 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
1828 PR tree-optimization/86659
1829 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
1831 2018-10-09 Richard Biener <rguenther@suse.de>
1833 PR tree-optimization/63155
1834 * tree-ssa-structalias.c: Include tree-ssa.h.
1835 (get_constraint_for_ssa_var): For undefs return nothing_id.
1836 (find_func_aliases): Cleanup PHI handling.
1838 2018-10-09 Richard Biener <rguenther@suse.de>
1840 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
1843 2018-10-09 Martin Liska <mliska@suse.cz>
1845 * asan.c (asan_emit_stack_protection): If a stack variable
1846 is located in a same file as current function, then emit
1847 line info into variable definition string.
1849 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
1851 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
1854 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
1856 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
1859 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
1861 PR tree-optimization/86659
1862 * gimple-match.h (struct gimple_match_op): Add reverse field.
1863 (gimple_match_op::set_op): New overloaded method.
1864 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
1865 the REF_REVERSE_STORAGE_ORDER flag on the value.
1866 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
1867 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
1869 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
1872 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
1873 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
1875 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
1878 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
1879 Defined with __builtin_ia32_vfmaddsubpd512_mask.
1881 2018-10-08 Richard Biener <rguenther@suse.de>
1883 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
1884 cost the same as AVX128 ones.
1886 2018-10-08 Paul Koning <ni1d@arrl.net>
1888 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
1889 (expand_block_move): New function.
1890 * config/pdp11/pdp11.c (output_block_move): Remove.
1891 (expand_block_move): New function.
1892 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
1893 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
1894 (*movmemhi1): Remove.
1896 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
1898 * config/s390/2827.md: Increase latencies for some FP instructions.
1900 2018-10-08 Richard Biener <rguenther@suse.de>
1902 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
1904 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
1905 * tree-vectorizer.h (dump_stmt_cost): Adjust.
1906 (add_stmt_cost): Dump return value of the hook.
1908 2018-10-08 Richard Biener <rguenther@suse.de>
1910 PR tree-optimization/63155
1911 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
1912 (ssa_propagation_engine::ssa_propagate): Remove redundant
1913 bitmap bit clearing.
1915 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
1917 PR rtl-optimization/86939
1918 PR rtl-optimization/87479
1919 * ira.h (non_conflicting_reg_copy_p): New prototype.
1920 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
1921 (make_hard_regno_dead): Don't add conflicts for register
1922 ignore_reg_for_conflicts.
1923 (make_object_dead): Likewise.
1924 (non_conflicting_reg_copy_p): New function.
1925 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
1926 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
1927 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
1928 (make_hard_regno_dead): Don't add conflicts for register
1929 ignore_reg_for_conflicts. Remove special conflict handling of
1930 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
1931 check_pic_pseudo_p and update callers.
1932 (mark_pseudo_dead): Don't add conflicts for register
1933 ignore_reg_for_conflicts.
1934 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
1936 2018-10-05 Andrew Waterman <andrew@sifive.com>
1937 Jim Wilson <jimw@sifive.com>
1939 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
1940 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
1941 new pattern using HONOR_SNANS that emits one extra instruction.
1943 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
1945 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
1946 patterns): Merge SI and DI patterns to a GPR pattern.
1947 (unnamed define_insn and define_split for record form of that): Merge
1948 to a single define_insn_and_split pattern.
1950 2018-10-05 David Malcolm <dmalcolm@redhat.com>
1953 * input.c (expand_location_to_spelling_point): Add param "aspect"
1954 and use rather than hardcoding LOCATION_ASPECT_CARET.
1955 (get_substring_ranges_for_loc): Handle the case of a single token
1956 within a macro expansion.
1957 * input.h (expand_location_to_spelling_point): Add "aspect" param,
1958 defaulting to LOCATION_ASPECT_CARET.
1960 2018-10-05 Paul Koning <ni1d@arrl.net>
1962 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
1963 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
1964 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
1965 (pdp11_guard_type): New function.
1967 2018-10-05 Paul Koning <ni1d@arrl.net>
1969 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
1970 * config/pdp11/pdp11.opt (mfloat32): Remove.
1972 * doc/invoke.texi (pdp11 -mfloat32): Remove:
1973 (pdp11 -mfloat64): Remove.
1975 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
1977 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
1978 (*cmp<mode>_cc_i387): Ditto.
1979 (*cmpu<mode>_cc_i387): Ditto.
1980 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
1981 * config/i386/i386.c (ix86_expand_fp_compare): Remove
1983 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
1984 Emit x86_sahf_1 pattern.
1985 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
1986 (ix86_expand_carry_flag_compare): Ditto.
1988 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
1990 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
1991 to reg_or_0_operand. Add "C" constraint.
1992 (*cmpxf_cc_i387): Ditto.
1993 (*cmp<mode>_i387): Change operand 2 predicate
1994 to nonimm_or_0_operand. Add "C" constraint.
1995 (*cmp<mode>_cc_i387): Ditto.
1996 (*cmp<mode>_0_i387): Remove insn pattern.
1997 (*cmp<mode>_0_cc_i387): Ditto.
1999 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
2001 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
2002 * config/i386/predicates.md (nonimm_or_0_operand): Rename
2003 from vector_move_operand. Update all uses.
2005 2018-10-05 Martin Sebor <msebor@redhat.com>
2007 PR tree-optimization/87490
2008 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
2011 2018-10-05 Richard Biener <rguenther@suse.de>
2013 PR tree-optimization/63155
2014 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
2015 vertical space in dumpfiles.
2016 * tree-ssa-propagate.h
2017 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
2018 * tree-ssa-propagate.c (cfg_blocks_back): New global.
2019 (ssa_edge_worklist_back): Likewise.
2020 (curr_order): Likewise.
2021 (cfg_blocks_get): Remove abstraction.
2022 (cfg_blocks_add): Likewise.
2023 (cfg_blocks_empty_p): Likewise.
2024 (add_ssa_edge): Add to current or next worklist based on
2026 (add_control_edge): Likewise.
2027 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
2029 (ssa_propagation_engine::ssa_propagate): ... here. Unify
2030 iteration from CFG and SSA edge worklist so we process
2031 everything in RPO order, prioritizing forward progress
2033 (ssa_prop_init): Allocate new worklists, do not dump
2035 (ssa_prop_fini): Free new worklists.
2037 2018-10-05 Richard Biener <rguenther@suse.de>
2039 * tree-core.h (tree_block::abstract_flag): Remove.
2040 (tree_block::block_num): Make full 32bits.
2041 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
2042 * tree.h (BLOCK_ABSTRACT): Remove.
2043 * dwarf2out.c (gen_lexical_block_die): Remove dead code
2044 resulting from BLOCK_ABSTRACT being always false.
2045 (gen_inlined_subroutine_die): Likewise.
2046 (gen_block_die): Likewise.
2047 * tree.c (block_ultimate_origin): Likewise.
2048 * tree-pretty-print.c (dump_block_node): Remove code dealing
2049 with BLOCK_ABSTRACT.
2050 * tree-ssa-live.c (dump_scope_block): Likewise.
2051 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
2052 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
2054 2018-10-05 Richard Biener <rguenther@suse.de>
2056 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
2057 is asked for initialize mode to the component mode of the
2060 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
2063 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
2064 assembler for -mavx.
2065 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
2067 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2070 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
2072 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
2073 to be DImode. When using mffscrn, force the operand to a register.
2075 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
2077 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
2078 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
2079 X87MODEF mode iterator.
2080 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
2081 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
2082 X87MODEF mode iterator.
2084 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
2086 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
2087 -Wno-prio-ctor-dtor.
2089 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2091 * Makefile.in (OBJS): Add opt-problem.o.
2092 * dump-context.h: Include "selftest.h.
2093 (selftest::temp_dump_context): New forward decl.
2094 (class dump_context): Make friend of class
2095 selftest::temp_dump_context.
2096 (dump_context::dump_loc_immediate): New decl.
2097 (class dump_pretty_printer): Move here from dumpfile.c.
2098 (class temp_dump_context): Move to namespace selftest.
2099 (temp_dump_context::temp_dump_context): Add param
2100 "forcibly_enable_dumping".
2101 (selftest::verify_dumped_text):
2102 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
2103 (selftest::verify_item):
2104 (ASSERT_IS_TEXT): Move here from dumpfile.c.
2105 (ASSERT_IS_TREE): Likewise.
2106 (ASSERT_IS_GIMPLE): Likewise.
2107 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
2109 (dump_context::dump_loc_immediate): ...this new function.
2110 (class dump_pretty_printer): Move to dump-context.h.
2111 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
2112 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
2113 (temp_dump_context::temp_dump_context): Move to "selftest"
2114 namespace. Add param "forcibly_enable_dumping", and use it to
2115 conditionalize the use of m_pp;
2116 (selftest::verify_dumped_text): Make non-static.
2117 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
2118 (selftest::verify_item): Make non-static.
2119 (ASSERT_IS_TEXT): Move to dump-context.h.
2120 (ASSERT_IS_TREE): Likewise.
2121 (ASSERT_IS_GIMPLE): Likewise.
2122 (selftest::test_capture_of_dump_calls): Pass "true" for new
2123 param of temp_dump_context.
2124 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
2125 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
2127 * opt-problem.cc: New file.
2128 * opt-problem.h: New file.
2129 * optinfo-emit-json.cc
2130 (selftest::test_building_json_from_dump_calls): Pass "true" for
2131 new param of temp_dump_context.
2132 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
2133 (optinfo::emit_for_opt_problem): New function.
2134 (optinfo::emit): Clarity which emit_item is used.
2135 * optinfo.h (optinfo::get_dump_location): New accessor.
2136 (optinfo::emit_for_opt_problem): New decl.
2137 (optinfo::emit): Make const.
2138 * selftest-run-tests.c (selftest::run_tests): Call
2139 selftest::opt_problem_cc_tests.
2140 * selftest.h (selftest::opt_problem_cc_tests): New decl.
2141 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
2142 bool to opt_result, converting fprintf messages to
2143 opt_result::failure_at calls. Add "stmt" param for use by the
2145 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
2146 (runtime_alias_check_p): Convert return type from bool to
2147 opt_result, converting dump_printf calls to
2148 opt_result::failure_at, using the statement DDR_A for their
2150 (find_data_references_in_stmt): Convert return type from bool to
2151 opt_result, converting "return false" to opt_result::failure_at
2153 * tree-data-ref.h: Include "opt-problem.h".
2154 (dr_analyze_innermost): Convert return type from bool to opt_result,
2155 and add a const gimple * param.
2156 (find_data_references_in_stmt): Convert return type from bool to
2158 (runtime_alias_check_p): Likewise.
2159 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
2160 dr_analyze_innermost.
2161 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
2162 Convert return type from bool to opt_result, adding a message for
2163 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
2164 (vect_analyze_data_ref_dependence): Convert return type from bool
2165 to opt_result. Change sense of return type from "false"
2166 effectively meaning "no problems" to "false" meaning a problem,
2167 so that "return false" becomes "return opt_result::success".
2168 Convert "return true" calls to opt_result::failure_at, using
2169 the location of statement A rather than vect_location.
2170 (vect_analyze_data_ref_dependences): Convert return type from bool
2172 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
2173 calls to opt_result::failure_at, using the stmt location rather
2175 (vect_verify_datarefs_alignment): Convert return type from bool
2177 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
2178 into multiple more-tightly-scoped copies.
2179 (vect_analyze_data_refs_alignment): Convert return type from bool
2181 (vect_analyze_data_ref_accesses): Likewise, converting a
2182 "return false" to a "return opt_result::failure_at", adding a
2184 (vect_prune_runtime_alias_test_list): Convert return type from
2185 bool to opt_result, converting dump_printf_loc to
2186 opt_result::failure_at. Add a %G to show the pertinent statement,
2187 and use the stmt's location rather than vect_location.
2188 (vect_find_stmt_data_reference): Convert return type from
2189 bool to opt_result, converting dump_printf_loc to
2190 opt_result::failure_at, using stmt's location.
2191 (vect_analyze_data_refs): Convert return type from bool to
2192 opt_result. Convert "return false" to "return
2193 opt_result::failure_at", adding messages as needed.
2194 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
2195 type from bool to opt_result.
2196 (vect_determine_vf_for_stmt): Likewise.
2197 (vect_determine_vectorization_factor): Likewise, converting
2198 dump_printf_loc to opt_result::failure_at, using location of phi
2199 rather than vect_location.
2200 (vect_analyze_loop_form_1): Convert return type from bool to
2201 opt_result, converting dump_printf_loc calls, retaining the use of
2203 (vect_analyze_loop_form): Convert return type from loop_vec_info
2204 to opt_loop_vec_info.
2205 (vect_analyze_loop_operations): Convert return type from bool to
2206 opt_result, converting dump_printf_loc calls, using the location
2207 of phi/stmt rather than vect_location where available. Convert
2208 various "return false" to "return opt_result::failure_at" with
2209 "unsupported phi" messages.
2210 (vect_get_datarefs_in_loop): Convert return type from bool to
2211 opt_result. Add a message for the
2212 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
2213 (vect_analyze_loop_2): Convert return type from bool to
2214 opt_result. Ensure "ok" is set to a opt_result::failure_at before
2215 each "goto again;", adding new messages where needed.
2216 Add "unsupported grouped {store|load}" messages.
2217 (vect_analyze_loop): Convert return type from loop_vec_info to
2219 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
2221 * tree-vect-stmts.c (process_use): Likewise, converting
2222 dump_printf_loc call and using stmt location, rather than
2224 (vect_mark_stmts_to_be_vectorized): Likeise.
2225 (vect_analyze_stmt): Likewise, adding a %G.
2226 (vect_get_vector_types_for_stmt): Convert return type from bool to
2227 opt_result, converting dump_printf_loc calls and using stmt
2228 location, rather than vect_location.
2229 (vect_get_mask_type_for_stmt): Convert return type from tree to
2230 opt_tree, converting dump_printf_loc calls and using stmt location.
2231 * tree-vectorizer.c: Include "opt-problem.h.
2232 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
2233 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
2234 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
2235 enabled, use it to report at the top level "couldn't vectorize
2236 loop" followed by the problem.
2237 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
2238 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
2240 (vect_analyze_stmt): Likewise.
2241 (vect_get_vector_types_for_stmt): Likewise.
2242 (tree vect_get_mask_type_for_stmt): Likewise.
2243 (vect_analyze_data_ref_dependences): Likewise.
2244 (vect_enhance_data_refs_alignment): Likewise.
2245 (vect_analyze_data_refs_alignment): Likewise.
2246 (vect_verify_datarefs_alignment): Likewise.
2247 (vect_analyze_data_ref_accesses): Likewise.
2248 (vect_prune_runtime_alias_test_list): Likewise.
2249 (vect_find_stmt_data_reference): Likewise.
2250 (vect_analyze_data_refs): Likewise.
2251 (vect_analyze_loop): Convert return type from loop_vec_info to
2253 (vect_analyze_loop_form): Likewise.
2254 (vect_analyze_slp): Convert return type from bool to opt_result.
2256 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2258 * doc/invoke.texi (-fopt-info): Document new "internals"
2260 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
2261 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
2263 (optinfo_verbosity_options): Add "internals".
2264 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
2265 (dump_context::apply_dump_filter_p): New member function.
2266 (dump_context::dump_loc): Use apply_dump_filter_p rather than
2267 explicitly masking the dump_kind.
2268 (dump_context::begin_scope): Increment the scope depth first. Use
2269 apply_dump_filter_p rather than explicitly masking the dump_kind.
2270 (dump_context::emit_item): Use apply_dump_filter_p rather than
2271 explicitly masking the dump_kind.
2272 (dump_dec): Likewise.
2273 (dump_hex): Likewise.
2274 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
2275 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
2276 (opt_info_switch_p): Update handling of default
2277 MSG_OPTIMIZED_LOCATIONS to cope with default of
2278 MSG_PRIORITY_USER_FACING.
2279 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
2280 masking the dump_kind.
2281 (selftest::test_capture_of_dump_calls): Update test_dump_context
2282 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
2283 than MSG_ALL. Generalize scope test to be run at all four
2284 combinations of with/without MSG_PRIORITY_USER_FACING and
2285 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
2286 for each of the two values.
2287 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
2288 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
2289 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
2290 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
2291 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
2292 with MSG_PRIORITY_*.
2293 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
2294 dump messages as MSG_PRIORITY_USER_FACING.
2295 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
2296 about the interaction with MSG_PRIORITY_*.
2298 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2300 * varasm.c (output_constant): Add new parameter merge_strings.
2301 Make strings properly zero terminated in merge string sections.
2302 (mergeable_string_section): Don't fail if the last char is non-zero.
2303 (assemble_variable_contents): Handle merge string sections.
2304 (assemble_variable): Likewise.
2305 (assemble_constant_contents): Likewise.
2306 (output_constant_def_contents): Likewise.
2307 (output_constructor_array_range,
2308 output_constructor_regular_field): Adjust call to output_constant.
2309 (output_object_block): Adjust call to assemble_constant_contents
2310 and assemble_variable_contents.
2312 2018-10-04 Martin Liska <mliska@suse.cz>
2315 * cgraphunit.c (process_function_and_variable_attributes):
2316 Warn about a function with alias attribute and a body.
2318 2018-10-04 Martin Liska <mliska@suse.cz>
2321 * multiple_target.c (redirect_to_specific_clone): New function.
2322 (ipa_target_clone): Use it.
2323 * tree-inline.c: Fix comment.
2325 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2327 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
2329 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
2330 (gcc::dump_manager::register_pass): New member function, adapted
2331 from loop body in gcc::pass_manager::register_pass, adding a
2332 call to update_dfi_for_opt_info.
2333 (gcc::dump_manager::opt_info_enable_passes): Store the
2334 -fopt-info options into the new fields. Move the loop
2336 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
2338 * dumpfile.h (struct opt_pass): New forward decl.
2339 (gcc::dump_manager::register_pass): New decl.
2340 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
2341 (class gcc::dump_manager): Add fields "m_optgroup_flags",
2342 "m_optinfo_flags", and "m_optinfo_filename".
2343 * passes.c (gcc::pass_manager::register_pass): Move all of the
2344 dump-handling code to gcc::dump_manager::register_pass.
2346 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
2348 PR rtl-optimization/87466
2349 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
2350 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
2351 * doc/tm.texi: Regenerate.
2352 * ira-lives.c (process_bb_node_lives): Use the new target hook.
2353 * lra-lives.c (process_bb_lives): Likewise.
2354 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
2357 2018-10-04 Tamar Christina <tamar.christina@arm.com>
2359 * params.c (add_params): Fix initialization.
2361 2018-10-04 Martin Liska <mliska@suse.cz>
2363 PR gcov-profile/84107
2364 * tree-profile.c (init_ic_make_global_vars):
2365 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
2366 Come up with new ic_tuple* variables. Emit
2367 __gcov_indirect_call{,_topn} variables.
2368 (gimple_gen_ic_profiler): Access the variable
2370 (gimple_gen_ic_func_profiler): Access
2371 __gcov_indirect_call.callee field.
2372 (gimple_init_gcov_profiler): Use ptr_type_node.
2373 * value-prof.c (gimple_ic): Use ptr_type_node.
2375 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2377 PR tree-optimization/85787
2378 * ipa-pure-const.c (malloc_candidate_p_1): Move most of malloc_candidate_p
2379 into this function and add support for detecting multiple phis.
2380 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
2382 2018-10-04 Martin Liska <mliska@suse.cz>
2385 * ipa-inline.c (inline_to_all_callers_1):
2386 Call ultimate_alias_target for node being inlined.
2388 2018-10-03 Jeff Law <law@redhat.com>
2390 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
2392 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
2393 * tree.h (get_typenode_from_name): Prototype.
2395 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
2397 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
2398 Change operand 2 predicate to nonimmediate_operand.
2399 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2401 2018-10-03 Martin Sebor <msebor@redhat.com>
2402 Jeff Law <law@redhat.com>
2404 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
2406 (get_string_length): Detect unterminated arrays.
2407 (format_string): Same.
2408 (format_directive): Warn about unterminated arrays.
2409 (handle_gimple_call): Mark statements with no_warning as needed.
2411 2018-10-03 Jim Wilson <jimw@sifive.com>
2413 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
2414 also define __riscv_abi_rve. Delete trailing white space.
2416 2018-10-03 Paul Koning <ni1d@arrl.net>
2418 Enable LRA register allocator for PDP11.
2419 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
2422 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
2423 * config/pdp11/pdp11.opt (-mlra): New option.
2424 * doc/invoke.texi (PDP-11 Options): Document -mlra.
2426 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
2428 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
2429 (*<absneg:code>extend<mode>xf2): Ditto.
2431 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
2433 PR tree-optimization/87415
2434 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
2437 2018-10-02 Jeff Law <law@redhat.com>
2439 * gimple-fold.c (get_range_strlen): Only set *nonstr when
2440 an unterminated string is discovered. Bubble up range
2441 even for unterminated strings.
2442 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
2443 indicates the string was not terminated via NONSTR.
2445 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
2447 * tree-vrp.c (extract_range_from_unary_expr): Special case all
2448 pointer conversions.
2449 Do not do anything special for anti-ranges.
2451 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
2453 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
2454 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
2456 2018-10-03 Martin Liska <mliska@suse.cz>
2458 PR gcov-profile/86109
2459 * coverage.c (coverage_begin_function): Do not
2460 mark lambdas as artificial.
2461 * tree-core.h (struct GTY): Remove tm_clone_flag
2462 and introduce new lambda_function.
2463 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
2465 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
2468 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
2469 P8_VECTOR and VSX are enabled.
2471 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
2473 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
2474 0x3907 as CPU model number.
2476 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
2478 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
2479 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
2480 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
2481 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
2482 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
2483 * config/s390/s390.md: Likewise. Rename also the cpu attribute
2484 value from arch12 to z14.
2486 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
2488 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
2490 (isinf<mode>2): Ditto.
2492 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
2494 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
2495 before emitting fxam. Perform calculations in XFmode.
2497 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
2499 * match.pd (((X /[ex] A) +- B) * A): New transformation.
2501 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
2504 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
2505 * tree.c (signed_or_unsigned_type_for): Handle complex.
2507 2018-10-02 Jeff Law <law@redhat.com>
2509 * gimple-fold.c (get_range_strlen): Remove dead code.
2511 2018-10-02 Martin Sebor <msebor@redhat.com>
2512 Jeff Law <law@redhat.com>
2514 * builtins.c (unterminated_array): Add new arguments.
2515 If argument is not terminated, bubble up size and exact
2517 (expand_builtin_strnlen): Detect, avoid expanding
2518 and diagnose unterminated arrays.
2519 (c_strlen): Fill in offset of start of unterminated strings.
2520 * builtins.h (unterminated_array): Update prototype.
2522 2018-10-02 Richard Biener <rguenther@suse.de>
2524 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
2525 of haddv4df, first reduce to SSE width and exploit the fact
2526 that we only need element zero with the reduction result.
2527 (reduc_plus_scal_v2df): Likewise.
2529 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
2531 * dojump.h (do_jump): Delete.
2532 (do_jump_1): Likewise.
2533 (split_comparison): Move around.
2534 * dojump.c (do_jump): Make static.
2535 (do_jump_1): Likewise.
2536 (jumpifnot): Move around.
2537 (jumpifnot_1): Likewise.
2539 (jumpif_1): Likewise.
2540 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
2542 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
2544 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
2545 insns in the delay slot and add_insn_after for the jump insn.
2547 2018-10-02 Richard Biener <rguenther@suse.de>
2549 * tree-inline.c (expand_call_inline): Use the location of
2550 the callee declaration for the inline-entry marker.
2551 * final.c (notice_source_line): Remove special-casing of
2552 NOTE_INSN_INLINE_ENTRY.
2554 2018-10-01 Carl Love <cel@us.ibm.com>
2557 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
2558 (__builtin_mtfsb0): New.
2559 (__builtin_mtfsb1): New.
2560 ( __builtin_set_fpscr_rn): New.
2561 (__builtin_set_fpscr_drn): New.
2562 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
2563 (rs6000_expand_set_fpscr_rn_builtin): Add.
2564 (rs6000_expand_set_fpscr_drn_builtin): Add.
2565 (rs6000_expand_builtin): Add case statement entries for
2566 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
2567 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
2568 RS6000_BUILTIN_MFFSL.
2569 (rs6000_init_builtins): Add ftype initialization and def_builtin
2570 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
2571 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
2572 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
2573 rs6000_mffscdrn): Add define_insn.
2574 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
2575 * doc/extend.texi: Add documentation for the builtins.
2577 2018-10-01 Richard Biener <rguenther@suse.de>
2579 PR tree-optimization/87465
2580 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
2581 causing branch miscounts.
2583 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2585 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
2586 aarch64_option_default_param): New.
2587 (params.h): Include.
2588 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
2589 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
2590 stack-clash protection validation code.
2592 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2594 * params.c (validate_param): New.
2595 (add_params): Use it.
2596 (set_param_value): Refactor param validation into validate_param.
2597 (diagnostic.h): Include.
2598 * diagnostic.h (diagnostic_ready_p): New.
2600 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2602 * params.c (set_param_value):
2603 Add index of parameter being validated.
2604 * common/common-target.def (option_validate_param): New.
2605 * common/common-targhooks.h (default_option_validate_param): New.
2606 * common/common-targhooks.c (default_option_validate_param): New.
2607 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
2608 * doc/tm.texi: Regenerate.
2610 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2613 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2614 Add validation for stack-clash parameters and set defaults.
2616 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2619 * configure.ac: Add stack-clash-protection-guard-size.
2620 * doc/install.texi: Document it.
2621 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
2622 * params.def: Update comment for guard-size.
2623 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
2624 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
2625 * configure: Regenerate.
2627 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2630 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
2631 STACK_DYNAMIC_OFFSET): New.
2632 * config/aarch64/aarch64.c (aarch64_layout_frame):
2633 Update outgoing args size.
2634 (aarch64_stack_clash_protection_alloca_probe_range,
2635 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
2637 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2640 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
2642 * target.def (stack_clash_protection_alloca_probe_range): New.
2643 (stack_clash_protection_final_dynamic_probe): Remove.
2644 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
2645 (default_stack_clash_protection_final_dynamic_probe): Remove.
2646 * targhooks.c: Likewise.
2647 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
2648 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
2649 * doc/tm.texi: Regenerate.
2651 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2654 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
2655 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
2656 aarch64_clamp_to_uimm12_shift): New.
2657 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
2658 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
2660 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2663 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
2665 2018-10-01 Jeff Law <law@redhat.com>
2666 Richard Sandiford <richard.sandiford@linaro.org>
2667 Tamar Christina <tamar.christina@arm.com>
2670 * config/aarch64/aarch64.md
2671 (probe_stack_range): Add k (SP) constraint.
2672 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
2673 STACK_CLASH_MAX_UNROLL_PAGES): New.
2674 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
2675 stack probes for stack clash.
2676 (aarch64_allocate_and_probe_stack_space): New.
2677 (aarch64_expand_prologue): Use it.
2678 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
2679 (aarch64_sub_sp): Add emit_move_imm optional param.
2681 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
2683 PR tree-optimization/87261
2684 * match.pd: Remove trailing whitespace.
2685 Add (x & y) | ~(x | y) -> ~(x ^ y),
2686 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
2688 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
2690 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
2692 (ashlsi3_insn): Update instruction constraints.
2693 (ashrsi3_insn): Likewise.
2694 (rotrsi3): Likewise.
2695 (add_shift): Likewise.
2696 * config/arc/constraints.md (Csz): New 32 bit constraint. It
2697 avoids placing in the limm field small constants which, otherwise,
2698 could end into a small instruction.
2700 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
2702 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
2703 destination register is not odd-even.
2704 (umaddsidi4_split): Likewise.
2706 2018-10-01 Richard Biener <rguenther@suse.de>
2708 * tree-inline.c (expand_call_inline): Store origin of fn
2709 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
2710 * tree.c (block_ultimate_origin): Simplify and do some
2713 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
2715 * config/i386/mmx.md (EMMS): New int iterator.
2716 (emms): New int attribute.
2717 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
2718 EMMS int iterator. Explicitly declare clobbers.
2719 (mmx_emms): Remove expander.
2721 * config/i386/predicates.md (emms_operation): Remove predicate.
2722 (vzeroall_pattern): New predicate.
2723 (vzeroupper_pattern): Rename from vzeroupper_operation.
2724 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
2725 vzeroupper_pattern and vzeroall_pattern predicates.
2727 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
2729 PR rtl-optimization/86939
2730 * ira-lives.c (make_hard_regno_born): Rename from this...
2731 (make_hard_regno_live): ... to this. Remove update to conflict
2732 information. Update function comment.
2733 (make_hard_regno_dead): Add conflict information update. Update
2735 (make_object_born): Rename from this...
2736 (make_object_live): ... to this. Remove update to conflict information.
2737 Update function comment.
2738 (make_object_dead): Add conflict information update. Update function
2740 (mark_pseudo_regno_live): Call make_object_live.
2741 (mark_pseudo_regno_subword_live): Likewise.
2742 (mark_hard_reg_dead): Update function comment.
2743 (mark_hard_reg_live): Call make_hard_regno_live.
2744 (process_bb_node_lives): Likewise.
2745 * lra-lives.c (make_hard_regno_born): Rename from this...
2746 (make_hard_regno_live): ... to this. Remove update to conflict
2747 information. Remove now uneeded check_pic_pseudo_p argument.
2748 Update function comment.
2749 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
2750 to conflict information. Update function comment.
2751 (mark_pseudo_live): Remove update to conflict information. Update
2753 (mark_pseudo_dead): Add conflict information update.
2754 (mark_regno_live): Call make_hard_regno_live.
2755 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
2756 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
2758 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
2761 * config/i386/i386.c (construct_container): Use TImode for
2762 BLKmode values in 2 integer registers.
2764 2018-09-29 Jeff Law <law@redhat.com>
2766 * builtins.c (unterminated_array): Pass in c_strlen_data * to
2767 c_strlen rather than just a tree *.
2768 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
2769 Update recursive calls appropriately. If caller did not provide a
2770 suitable data pointer, create a local one. When a non-terminated
2771 string is discovered, bubble up information about the string via the
2772 c_strlen_data object.
2773 * builtins.h (c_strlen): Update prototype.
2774 (c_strlen_data): New structure.
2775 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
2776 For a type 2 call, if c_strlen indicates a non-terminated string
2777 use the length of the non-terminated string.
2778 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
2780 2018-09-29 Jakub Jelinek <jakub@redhat.com>
2783 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
2784 __m512d type for __A argument rather than __m512.
2786 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
2788 * match.pd (simple_comparison): Don't optimize if either operand is
2789 a function pointer when target needs function pointer canonicalization.
2791 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
2793 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
2794 power5 .. power9 to remove indirection.
2795 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
2796 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
2797 ASM_CPU_476_SPEC): Delete.
2798 (ASM_CPU_SPEC): Adjust.
2799 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
2800 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
2802 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
2804 * config.in: Delete HAVE_AS_DCI.
2805 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
2806 * config/rs6000/rs6000.h: Ditto.
2807 * configure.ac: Delete HAVE_AS_DCI.
2808 * configure: Regenerate.
2810 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
2812 * config.in (HAVE_AS_LWSYNC): Delete.
2813 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
2814 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
2816 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
2817 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
2819 * configure.ac: Delete HAVE_AS_LWSYNC.
2820 * configure: Regenerate.
2822 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
2823 Pierre-Marie de Rodat <derodat@adacore.com>
2825 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
2826 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
2827 (cgraph_node::create_thunk): Add indirect_offset parameter.
2828 (thunk_adjust): Likewise.
2829 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
2830 and initialize the corresponding field with it.
2831 (cgraph_node::dump): Dump indirect_offset field.
2832 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
2833 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
2834 (thunk_adjust): Add indirect_offset parameter and deal with it.
2835 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
2836 pass it to thunk_adjust. Do not call the target hook if it's non-zero
2837 or if the thunk is external or local. Fix formatting. Do not chain
2838 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
2839 if any, in the GIMPLE representation.
2840 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
2841 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
2842 (input_node): Read indirect_offset field.
2843 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
2844 call to thunk_adjust.
2845 * tree-nested.c (struct nesting_info): Add thunk_p field.
2846 (create_nesting_tree): Set it.
2847 (convert_all_function_calls): Copy static chain from targets to thunks.
2848 (finalize_nesting_tree_1): Return early for thunks.
2849 (unnest_nesting_tree_1): Do not finalize thunks.
2850 (gimplify_all_functions): Do not gimplify thunks.
2852 2018-09-28 David Malcolm <dmalcolm@redhat.com>
2854 * opt-suggestions.c (option_proposer::build_option_suggestions):
2855 Release "option_values".
2857 2018-09-28 David Malcolm <dmalcolm@redhat.com>
2859 * coverage.c (get_coverage_counts): Convert problem-reporting dump
2860 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
2861 * dumpfile.c (kind_as_string): New function.
2862 (dump_loc): Rather than a hardcoded prefix of "note: ", use
2863 kind_as_string to vary the prefix based on dump_kind.
2864 (selftest::test_capture_of_dump_calls): Update for above.
2866 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
2868 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
2869 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
2871 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
2873 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
2874 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
2875 INVALID_REGNUM instead of FPSR_REG.
2876 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
2877 * config/i386/i386.md: Update comment of FP compares.
2878 (fldenv): Do not clobber FPSR_REG.
2880 2018-09-28 Richard Biener <rguenther@suse.de>
2882 * tree.h (BLOCK_ORIGIN): New.
2883 * omp-expand.c (grid_expand_target_grid_body): Assign
2884 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
2885 * tree-inline.c (remap_block): Likewise.
2886 * auto-profile.c (get_function_decl_from_block): Simplify
2887 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
2888 * langhooks.c (lhd_print_error_function): Likewise.
2889 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
2891 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
2892 * tree.c (block_nonartificial_location): Likewise.
2893 (block_ultimate_origin): Likewise.
2894 * tree-pretty-print.c (percent_K_format): Likewise. Remove
2895 no longer needed LTO case.
2897 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
2898 Jan Hubicka <jh@suse.cz>
2899 Martin Jambor <mjambor@suse.cz>
2901 * simplify-rtx.c (simplify_merge_mask): New function.
2902 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
2903 same masks are used in op1 or op2.
2904 (test_vec_merge): New function.
2905 (test_vector_ops): Call test_vec_merge.
2907 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
2909 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
2910 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
2911 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
2912 (cypress_costs): Set it.
2913 (supersparc_costs): Likewise.
2914 (hypersparc_costs): Likewise.
2915 (leon_cost): Likewise.
2916 (leon3_costs): Likewise.
2917 (sparclet_costs): Likewise.
2918 (ultrasparc_costs): Likewise.
2919 (ultrasparc_costs): Likewise.
2920 (niagara_costs): Likewise.
2921 (niagara2_costs): Likewise.
2922 (niagara3_costs): Likewise.
2923 (niagara4_costs): Likewise.
2924 (niagara7_costs): Likewise.
2925 (m8_costs): Likewise.
2926 (TARGET_CAN_FOLLOW_JUMP): Define.
2927 (pass_work_around_errata::gate): Minor tweak.
2928 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
2929 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
2930 Automaitcally clear MASK_FSMULD mask for V7 processors.
2931 (sparc_can_follow_jump): New static function.
2932 (output_ubranch): Deal with CROSSING_JUMP_P.
2933 (sparc_use_sched_lookahead): Rewrite using switch statement.
2934 (sparc_issue_rate): Reorder.
2935 (sparc_branch_cost): New function.
2937 2018-09-27 Martin Sebor <msebor@redhat.com>
2939 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
2940 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
2941 (int_fits_type_p): Same.
2943 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
2945 * config/i386/i386.md (FPCR_REG): Remove.
2946 (UNSPEC_FLDCW): Remove.
2947 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
2948 (x86_fldcw_1): Remove insn pattern.
2949 (fnstenv): Do not clobber FPCR_REG.
2951 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
2952 (CALL_USED_REGISTERS): Ditto.
2953 (REG_ALLOC_ORDER): Ditto.
2954 (REG_CLASS_CONTENTS): Ditto.
2955 (HI_REGISTER_NAMES): Ditto.
2956 (ADDITIONAL_REGISTER_NAMES): Use defines instead
2957 of numerical constants.
2958 * config/i386/i386.c (regclass_map): Remove fpsr register.
2959 (dbx_register_map): Ditto.
2960 (dbx64_register_map): Ditto.
2961 (svr4_dbx_register_map): Ditto.
2962 (print_reg): Do not handle FPCR_REG.
2964 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
2967 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
2968 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
2969 Delete, always treat as true.
2970 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
2971 Ditto. Simplify remaining code.
2972 * config/powerpcspe/powerpcspe.h: Ditto.
2973 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
2974 Simplify remaining code.
2975 (rs6000_expand_builtin): Ditto.
2976 * config/rs6000/rs6000.h: Ditto.
2977 * configure.ac: Ditto.
2978 * configure: Regenerate.
2980 2018-09-27 Martin Liska <mliska@suse.cz>
2982 * coverage.c (get_coverage_counts): Revert the formatting
2983 of missing profile opt info.
2985 2018-09-27 Richard Biener <rguenther@suse.de>
2989 * dwarf2out.c (set_block_origin_self): Do not mark outermost
2990 block as we do not output that.
2991 (gen_inlined_subroutine_die): Elide the originally outermost
2992 block, matching what we do for concrete instances.
2993 (decls_for_scope): Add parameter specifying whether to recurse
2996 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
2997 Tom de Vries <tom@codesourcery.com>
3001 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
3002 STORE_FLAG_VALUE == 1.
3004 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
3006 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
3007 constant definitions.
3008 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
3009 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
3010 ("speculation_barrier"): New expander definition.
3012 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
3014 PR gcov-profile/86957
3015 * common.opt: New warning option -Wmissing-profile.
3016 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
3017 * doc/invoke.texi: Document -Wmissing-profile.
3019 2018-09-26 Jim Wilson <jimw@sifive.com>
3021 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
3022 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
3024 2018-09-26 Martin Sebor <msebor@redhat.com>
3026 * tree.c (zerop): Change return type to bool.
3027 (integer_zerop, integer_onep, integer_each_onep): Same.
3028 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
3029 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
3030 (real_onep, real_minus_onep, chain_index): Same.
3031 (print_type_hash_statistics, type_list_equal): Same.
3032 * tree.h (zerop): Same.
3033 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
3034 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
3035 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
3036 (real_onep, real_minus_onep, chain_index): Same.
3037 (print_type_hash_statistics, type_list_equal): Same.
3039 2018-09-26 Jim Wilson <jimw@sifive.com>
3041 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
3043 2018-09-26 Jakub Jelinek <jakub@redhat.com>
3046 * config/i386/i386.c: Include debug.h and dwarf2out.h.
3047 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
3050 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
3052 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
3054 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
3056 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
3057 and FP_SECOND_SSE_REGS.
3058 (REG_CLASS_NAMES): Ditto.
3059 (REG_CLASS_CONTENTS): Ditto.
3060 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
3061 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
3062 (ix86_preferred_output_reload_class): Ditto.
3063 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
3064 clobber constraint to "=&f".
3065 (fix_truncdi_i387): Ditto.
3066 (lrintxfdi2): Ditto.
3067 (fistdi2_<rounding>): Ditto.
3068 (fpremxf4_i387): Change "=u" constraint to "=f".
3069 (fprem1xf4_i387): Ditto.
3071 (fptanxf4_i387): Ditto.
3072 (fxtractxf3_i387): Ditto.
3073 (fscalexf4_i387): Ditto.
3074 (atan2xf3): Change "u" constraint to "f".
3075 (fyl2xxf3_i387): Ditto.
3076 (fyl2xp1xf3_i387): Ditto.
3078 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
3081 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
3082 for removed I387_MASK_PM entity.
3085 2018-09-26 Jeff Law <law@redhat.com>
3087 2018-09-26 Alexey Neyman <stilor@att.net>
3089 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
3090 headers are no longer pulled in by <isl/val.h>.
3092 2018-09-26 Richard Biener <rguenther@suse.de>
3095 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
3096 or concrete instance DIE to the tree. Create abstract origin
3097 attributes also for concrete instances.
3099 2018-09-26 Alexey Neyman <stilor@att.net>
3101 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
3102 headers are no longer pulled in by <isl/val.h>.
3104 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
3106 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
3107 Use new helper functions.
3108 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
3109 Use new helper functions.
3110 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
3111 aarch_mm_needs_release): New declarations.
3112 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
3113 aarch_mm_needs_release): New.
3115 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
3117 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
3118 (arm32_output_mi_thunk): Deal with long calls.
3120 2018-09-26 Richard Biener <rguenther@suse.de>
3124 * tree-inline.c (expand_call_inline): When the location
3125 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
3126 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
3127 the inserted BLOCK to make inlined_function_outer_scope_p
3129 * dwarf2out.c (add_call_src_coords_attributes): Do not add
3130 coords for reserved locations.
3132 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3134 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
3135 (*call_indirect_nonlocal_sysv<mode>): Ditto.
3136 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
3137 (*sibcall_nonlocal_sysv<mode>): Ditto.
3138 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
3139 (<bd>_<mode>): Ditto.
3140 (<bd>tf_<mode>): Ditto.
3142 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3144 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
3145 control string as a list of templates instead of as C code.
3146 (*altivec_movti): Ditto.
3147 * config/rs6000/darwin.md (movdf_low_di): Ditto.
3149 2018-09-25 Jim Wilson <jimw@sifive.com>
3151 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
3152 when target symbol is weak.
3154 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
3157 * builtins.c (unterminated_array): Simplify.
3158 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
3159 where pointer arithmetic is safe.
3161 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3164 * config/rs6000/altivec.md (altivec_vspltb): Use
3165 const_0_to_15_operand instead of u5bit_cint_operand.
3166 (*altivec_vspltb_internal): Ditto.
3167 (altivec_vspltb_direct): Ditto.
3168 (altivec_vsplth): Use const_0_to_7_operand instead of
3170 (*altivec_vsplth_internal): Ditto.
3171 (altivec_vsplth_direct): Ditto.
3172 (altivec_vspltw): Use const_0_to_3_operand instead of
3174 (*altivec_vspltw_internal): Ditto.
3175 (altivec_vspltw_direct): Ditto.
3176 (altivec_vspltsf): Ditto.
3177 (*altivec_vspltsf_internal): Ditto.
3178 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
3179 various splats with the proper size immediate. Reorder the various
3180 cases by ascending size of immediate, and put all such together.
3182 2018-09-25 Richard Biener <rguenther@suse.de>
3185 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
3187 (maybe_create_die_with_external_ref): Do not create
3188 DW_TAG_imported_unit here.
3189 (add_abstract_origin_attribute): Handle external BLOCK refs.
3190 (dwarf2out_abstract_function): Simplify LTO case.
3191 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
3192 rather than using maybe_create_die_with_external_ref.
3194 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
3197 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
3198 (frndintxf2_mask_pm_i387): Ditto.
3199 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
3200 Enable for !flag_trapping_math.
3201 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
3202 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
3203 Change operand 1 predicate to nonimmediate_operand.
3204 (attr "i387_cw"): Remove mask_pm.
3205 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
3206 (enum ix86_entity): Remove I387_MASK_PM.
3207 * config/i386/i386.c (ix86_i387_mode_needed): Do not
3208 handle I387_MASK_PM.
3209 (ix86_mode_needed): Ditto.
3210 (ix86_mode_after): Ditto.
3211 (ix86_mode_entry): Ditto.
3212 (ix86_mode_exit): Ditto.
3213 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
3215 2018-09-25 Jakub Jelinek <jakub@redhat.com>
3217 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
3218 to_update_switch_stmts to vNULL instead of calling create on them
3221 2018-09-25 Richard Biener <rguenther@suse.de>
3223 PR tree-optimization/87402
3224 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
3225 (visit_phi): Re-instantiate handling of supposed to be VARYING
3226 but non-VARYING backedge value.
3228 2018-09-25 Richard Biener <rguenther@suse.de>
3231 * dwarf2out.c (struct sym_off_pair): New.
3232 (external_die_map): New global.
3233 (lookup_decl_die): When in LTO create DIEs lazily from the
3235 (lookup_block_die): New function, create DIEs lazily in LTO.
3236 (equate_block_to_die): New function.
3237 (dwarf2out_die_ref_for_decl): During WPA get the association
3238 from the external DIE map.
3239 (dwarf2out_register_external_die): Record mapping into the
3241 (maybe_create_die_with_external_ref): New function split out from
3242 DIE generation part of old dwarf2out_register_external_die.
3243 (add_abstract_origin_attribute): Do not return the DIE. When
3244 in LTO reference externals directly.
3245 (dwarf2out_abstract_function): When in LTO ignore calls for
3246 decls with external DIEs (already present abstract instances).
3247 (gen_call_site_die): Adjust.
3248 (add_high_low_attributes): Likewise.
3249 (gen_lexical_block_die): Likewise.
3250 (gen_inlined_subroutine_die): Likewie.
3251 (gen_block_die): Likewise.
3252 (dwarf2out_inline_entry): Likewise.
3253 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
3256 2018-09-25 Martin Liska <mliska@suse.cz>
3258 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
3259 integers and not by a float value.
3261 2018-09-25 Martin Liska <mliska@suse.cz>
3264 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
3266 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
3267 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
3268 * godump.c (go_format_type): Remove extra parenthesis.
3270 2018-09-25 Martin Liska <mliska@suse.cz>
3272 * alias.c (set_dest_equal_p): Remove unused function.
3273 * config/i386/i386.c (def_builtin_pure2): Likewise.
3274 * diagnostic-show-locus.c (class layout): Remove
3276 (layout::layout): Likewise here.
3277 * dump-context.h (class temp_dump_context): Likewise.
3278 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
3279 (add_AT_loclistsptr): Likewise.
3280 (add_AT_offset): Likewise.
3281 (get_AT_hi_pc): Likewise.
3282 (is_comdat_die): Likewise.
3283 (type_is_enum): Likewise.
3284 (ceiling): Likewise.
3285 (add_AT_vms_delta): Likewise.
3286 (is_class_die): Likewise.
3287 * edit-context.c (class line_event): Remove unused field.
3288 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
3290 * ipa-cp.c (ipa_get_vr_lat): Likewise.
3291 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
3292 (ok_for_base_p_nonstrict): Likewise.
3293 * tree-chrec.c (is_not_constant_evolution): Likewise.
3294 (chrec_fold_poly_cst): Likewise.
3295 * tree-if-conv.c (has_pred_critical_p): Likewise.
3296 * tree-ssa-coalesce.c (print_exprs): Likewise.
3297 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
3298 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
3299 * tree-vrp.c (value_ranges_intersect_p): Likewise.
3300 (value_range_nonnegative_p): Likewise.
3302 2018-09-25 Martin Liska <mliska@suse.cz>
3304 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
3305 Do not handle "GNU Pascal".
3306 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
3308 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
3309 from documentation. Likewise.
3310 * dbxout.c (dbxout_range_type): Likewise.
3311 * doc/cpp.texi: Likewise.
3312 * doc/extend.texi: Likewise.
3313 * doc/frontends.texi: Likewise.
3314 * doc/invoke.texi: Remove Pascal entry.
3315 * tree.def (CLEANUP_POINT_EXPR): Likewise.
3316 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
3318 2018-09-25 Martin Liska <mliska@suse.cz>
3321 * doc/invoke.texi: Document all parameters and remove default
3324 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
3327 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
3328 contains HOST_WIDE_INTs when computing its size.
3330 2018-09-24 Jim Wilson <jimw@sifive.com>
3333 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
3335 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
3337 2018-09-24 Andrew Pinski <apinski@marvell.com>
3339 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
3340 access prev before checking it for NULLness in the
3341 AARCH64_FUSE_CMP_BRANCH case.
3343 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
3346 * config/i386/i386.c (rest_of_insert_endbranch): Set
3347 endbr_queued_at_entrance to true and don't insert ENDBR if
3348 x86_function_profiler will be called.
3349 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
3351 * config/i386/i386.h (machine_function): Add
3352 endbr_queued_at_entrance.
3354 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
3356 * genattrtab.c (mk_attr_alt): Use alternative_mask.
3357 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
3359 (check_attr_test): Use alternative_mask.
3360 (get_attr_value): Likewise.
3361 (compute_alternative_mask): Use alternative_mask and XWINT.
3362 (make_alternative_compare): Use alternative_mask.
3363 (attr_alt_subset_p): Use XWINT.
3364 (attr_alt_subset_of_compl_p): Likewise.
3365 (attr_alt_intersection): Use alternative_mask and XWINT.
3366 (attr_alt_union): Likewise.
3367 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
3368 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
3369 (simplify_test_exp): Use alternative_mask and XWINT.
3370 (write_test_expr): Use alternative_mask and XWINT, adjust bit
3371 number calculation to support 64 bits. Generate code that
3372 checks 64-bit masks.
3373 (main): Use alternative_mask.
3374 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
3376 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
3379 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
3380 RETURN+USE when returning via %r14.
3382 2018-09-24 Martin Liska <mliska@suse.cz>
3384 * gcov.c (output_lines): Print colorization legend
3385 for both flag_use_colors and flag_use_hotness_colors.
3388 2018-09-24 Martin Liska <mliska@suse.cz>
3390 * coverage.c (get_coverage_counts): Use warning_at
3391 with current_function_decl location. Use %qD in warning
3394 2018-09-24 Martin Liska <mliska@suse.cz>
3396 * memory-block.h (memory_block_pool::release): Annotate with
3397 valgrind that the memory is not accessible.
3399 2018-09-24 Martin Liska <mliska@suse.cz>
3402 * asan.c: Make asan_handled_variables extern.
3404 * cfgexpand.c (expand_stack_vars): Make sure
3405 a representative is unpoison if another
3406 variable in the partition is handled by
3407 use-after-scope sanitization.
3409 2018-09-24 Richard Biener <rguenther@suse.de>
3411 PR tree-optimization/63155
3412 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
3413 the worklist when the edge of the respective argument isn't
3416 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3418 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
3419 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
3420 (MASK_CLASS_P): Update for rename.
3421 (MAYBE_MASK_CLASS_P): Ditto.
3422 (REG_CLASS_NAMES): Update.
3423 (REG_CLASS_CONTENT): Update.
3424 * config/i386/i386.c (regclass_map): Update for MASK_REG
3425 and ALL_MASK_REGS rename.
3426 * config/i386/constraints.md (Yk): Update for rename.
3429 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3431 * config/i386/i386.h (enum reg_class): Remove
3432 EVEX_SSE_REGS and MOD4_SSE_REGS.
3433 (REG_CLASS_NAMES): Update.
3434 (REG_CLASS_CONTENT): Update.
3435 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
3436 registers as ALL_SSE_REGS.
3437 (ix86_additional_allocno_class_p): Remove.
3438 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
3439 (ix86_register_priority): Lower priority of EVEX SSE registers.
3440 Use IN_RANGE macro where appropriate.
3441 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
3442 AVX-5124VNNIW checks.
3443 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
3444 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
3445 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
3446 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
3447 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
3448 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
3449 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
3450 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
3451 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
3452 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
3453 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
3454 * config/i386/constraints.md (Yh): Remove.
3456 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3458 * config/i386/i386.c (regclass_map): Declare integer REX registers
3461 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
3463 * doc/service.texi (Service): Switch the fsf.org link to https.
3465 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
3468 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3469 Define to speculation_safe_value_not_needed.
3471 2018-09-21 Florian Weimer <fweimer@redhat.com>
3474 * doc/extend.texi (Common Function Attributes): Mention that
3475 noreturn suppresses tail call optimization.
3477 2018-09-21 Jeff Law <law@redhat.com>
3479 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
3480 vr_values::cleanup_edges_and_switches.
3481 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
3483 (identify_jump_threads): Remove EDGE_IGNORE handling.
3484 (execute_vrp): Move handling of to_remove_edges and
3485 to_update_switch_stmts into vr_values class member functions.
3486 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
3487 (to_update_switch_stmts): Likewise.
3488 * vr-values.c: Include cfghooks.h.
3489 (vr_values::vr_values): Initialize to_remove_edges and
3490 to_update_switch_stmts.
3491 (vr_values::~vr_values): Verify to_remove_edges and
3492 to_update_switch_stmts are empty.
3493 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
3494 (vr_values::cleanup_edges_and_switches): New member function.
3495 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
3496 function. Add new data members.
3498 2018-09-21 David Malcolm <dmalcolm@redhat.com>
3500 PR tree-optimization/87309
3501 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
3502 calls with pflags and alt_flags.
3503 (selftest::test_capture_of_dump_calls): Add test of interaction of
3504 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
3506 2018-09-21 Olivier Hainque <hainque@adacore.com>
3508 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
3510 2018-09-21 Olivier Hainque <hainque@adacore.com>
3512 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
3514 2018-09-21 Olivier Hainque <hainque@adacore.com>
3516 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
3517 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
3518 (VXWORKS_LIBS_RTP): Minor reordering.
3520 2018-09-21 Olivier Hainque <hainque@adacore.com>
3522 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
3523 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
3525 2018-09-21 Olivier Hainque <hainque@adacore.com>
3527 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
3528 (PTRDIFF_TYPE): Likewise.
3530 2018-09-21 Olivier Hainque <hainque@adacore.com>
3532 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
3533 triplet, similar to support for VxWorks7.
3534 * config/vxworks-dummy.h: Provide a default definition
3535 of TARGET_VXWORKS64 to 0.
3537 2018-09-21 Olivier Hainque <hainque@adacore.com>
3539 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
3540 * config/vxworks-dummy.h: here.
3542 2018-09-21 Olivier Hainque <hainque@adacore.com>
3544 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
3546 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
3547 Bo Zhou <zbo.zhou@hisilicon.com>
3549 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
3550 * config/aarch64/aarch64-tune.md: Regenerated.
3551 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
3552 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
3553 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
3555 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
3556 Julian Brown <julian@codesourcery.com>
3558 * builtins.c (get_builtin_sync_mem): Handle address spaces.
3560 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
3562 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
3563 if the call takes a static chain.
3565 2018-09-21 Martin Liska <mliska@suse.cz>
3567 * auto-profile.c (autofdo_source_profile::read): Do not
3569 (read_profile): Do not add working sets.
3570 (read_autofdo_file): Remove sum_all.
3571 (afdo_callsite_hot_enough_for_early_inline): Remove const
3573 * coverage.c (struct counts_entry): Remove gcov_summary.
3574 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
3575 do not support GCOV_TAG_PROGRAM_SUMMARY.
3576 (get_coverage_counts): Remove summary and expected
3578 * coverage.h (get_coverage_counts): Likewise.
3579 * doc/gcov-dump.texi: Remove -w option.
3580 * gcov-dump.c (dump_working_sets): Remove.
3581 (main): Do not support '-w' option.
3582 (print_usage): Likewise.
3583 (tag_summary): Likewise.
3584 * gcov-io.c (gcov_write_summary): Do not dump
3586 (gcov_read_summary): Likewise.
3587 (gcov_histo_index): Remove.
3588 (gcov_histogram_merge): Likewise.
3589 (compute_working_sets): Likewise.
3590 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
3592 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
3593 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
3594 (GCOV_HISTOGRAM_SIZE): Remove.
3595 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
3596 (struct gcov_summary): Simplify rapidly just
3597 to runs and sum_max fields.
3598 (gcov_histo_index): Remove.
3599 (NUM_GCOV_WORKING_SETS): Likewise.
3600 (compute_working_sets): Likewise.
3601 * gcov-tool.c (print_overlap_usage_message): Remove
3602 trailing empty line.
3603 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
3604 (output_lines): Remove program related line.
3605 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
3606 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
3608 (input_profile_summary): Do not read it.
3609 (merge_profile_summaries): And do not merge it.
3610 (input_symtab): Do not call removed function.
3611 * modulo-sched.c (sms_schedule): Do not print sum_max.
3612 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
3613 removed when histogram method was invented.
3614 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
3616 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
3618 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
3619 and dump selected value.
3620 * profile.c (add_working_set): Remove.
3621 (get_working_sets): Likewise.
3622 (find_working_set): Likewise.
3623 (get_exec_counts): Do not work with working sets.
3624 (read_profile_edge_counts): Do not inform as sum_max is removed.
3625 (compute_branch_probabilities): Likewise.
3626 (compute_value_histograms): Remove argument for call of
3627 get_coverage_counts.
3628 * profile.h: Do not make gcov_summary const.
3630 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
3632 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
3634 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
3636 PR tree-optimization/86990
3637 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
3638 Check that the entire merged store group is made of constants only for
3641 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
3643 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
3644 (VTABLE_VERIFICATION_SPEC): Likewise.
3645 (SANITIZER_EARLY_SPEC): Likewise.
3646 (SANITIZER_SPEC): Likewise.
3647 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
3648 * doc/invoke.texi (Link Options): Document -r.
3650 2018-09-20 Richard Biener <rguenther@suse.de>
3653 * gimplify.c (gimplify_expr): Retain alignment of
3654 addressable lvalue in dereference.
3656 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
3659 * configure.ac: Check for .loc is_stmt support.
3660 * configure, config.in: Rebuilt.
3661 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
3664 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3666 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
3667 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
3670 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3672 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
3674 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
3677 2018-09-20 Marek Polacek <polacek@redhat.com>
3679 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
3681 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
3683 PR tree-optimization/87288
3684 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
3685 into account when determining PEELING_FOR_NITERS.
3687 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
3689 PR tree-optimization/86877
3690 * tree-vect-loop.c (vect_analyze_loop_2): Call
3691 vect_verify_datarefs_alignment.
3693 2018-09-19 Marek Polacek <polacek@redhat.com>
3695 * doc/invoke.texi: Document -Wclass-conversion.
3697 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
3699 * config/pa/pa.c (pa_adjust_priority): Delete.
3700 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
3702 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
3703 (atomic_storehi): Likewise.
3704 (atomic_storesi): Likewise.
3705 (atomic_loaddi): Restore compare and swap exchange loop code.
3707 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
3709 PR rtl-optimization/86902
3710 * combine.c (try_combine): When changing the CC mode used, don't change
3711 an unrelated mode in other_insn to that new CC mode.
3713 2018-09-19 David Malcolm <dmalcolm@redhat.com>
3715 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
3716 with %T in place of calls to dump_generic_expr.
3717 (prune_runtime_alias_test_list): Likewise.
3718 (create_runtime_alias_checks): Likewise.
3719 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
3720 (vect_analyze_data_ref_dependence): Likewise.
3721 (vect_slp_analyze_data_ref_dependence): Likewise.
3722 (vect_record_base_alignment): Likewise. Use %G in place of call
3723 to dump_gimple_stmt.
3724 (vect_compute_data_ref_alignment): Likewise.
3725 (verify_data_ref_alignment): Likewise.
3726 (vect_find_same_alignment_drs): Likewise.
3727 (vect_analyze_group_access_1): Likewise.
3728 (vect_analyze_data_ref_accesses): Likewise.
3729 (dependence_distance_ge_vf): Likewise.
3730 (dump_lower_bound): Likewise.
3731 (vect_prune_runtime_alias_test_list): Likewise.
3732 (vect_find_stmt_data_reference): Likewise.
3733 (vect_analyze_data_refs): Likewise.
3734 (vect_create_addr_base_for_vector_ref): Likewise.
3735 (vect_create_data_ref_ptr): Likewise.
3736 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
3737 (vect_can_advance_ivs_p): Likewise.
3738 (vect_update_ivs_after_vectorizer): Likewise.
3739 (vect_gen_prolog_loop_niters): Likewise.
3740 (vect_prepare_for_masked_peels): Likewise.
3741 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
3742 (vect_determine_vectorization_factor): Likewise.
3743 (vect_is_simple_iv_evolution): Likewise.
3744 (vect_analyze_scalar_cycles_1): Likewise.
3745 (vect_analyze_loop_operations): Likewise.
3746 (report_vect_op): Likewise.
3747 (vect_is_slp_reduction): Likewise.
3748 (check_reduction_path): Likewise.
3749 (vect_is_simple_reduction): Likewise.
3750 (vect_create_epilog_for_reduction): Likewise.
3751 (vect_finalize_reduction:): Likewise.
3752 (vectorizable_induction): Likewise.
3753 (vect_transform_loop_stmt): Likewise.
3754 (vect_transform_loop): Likewise.
3755 (optimize_mask_stores): Likewise.
3756 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
3757 (vect_split_statement): Likewise.
3758 (vect_recog_over_widening_pattern): Likewise.
3759 (vect_recog_average_pattern): Likewise.
3760 (vect_determine_min_output_precision_1): Likewise.
3761 (vect_determine_precisions_from_range): Likewise.
3762 (vect_determine_precisions_from_users): Likewise.
3763 (vect_mark_pattern_stmts): Likewise.
3764 (vect_pattern_recog_1): Likewise.
3765 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
3766 (vect_record_max_nunits): Likewise.
3767 (vect_build_slp_tree_1): Likewise.
3768 (vect_build_slp_tree_2): Likewise.
3769 (vect_print_slp_tree): Likewise.
3770 (vect_analyze_slp_instance): Likewise.
3771 (vect_detect_hybrid_slp_stmts): Likewise.
3772 (vect_detect_hybrid_slp_1): Likewise.
3773 (vect_slp_analyze_operations): Likewise.
3774 (vect_slp_analyze_bb_1): Likewise.
3775 (vect_transform_slp_perm_load): Likewise.
3776 (vect_schedule_slp_instance): Likewise.
3777 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
3778 (vect_mark_stmts_to_be_vectorized): Likewise.
3779 (vect_init_vector_1): Likewise.
3780 (vect_get_vec_def_for_operand): Likewise.
3781 (vect_finish_stmt_generation_1): Likewise.
3782 (vect_check_load_store_mask): Likewise.
3783 (vectorizable_call): Likewise.
3784 (vectorizable_conversion): Likewise.
3785 (vectorizable_operation): Likewise.
3786 (vectorizable_load): Likewise.
3787 (vect_analyze_stmt): Likewise.
3788 (vect_is_simple_use): Likewise.
3789 (vect_get_vector_types_for_stmt): Likewise.
3790 (vect_get_mask_type_for_stmt): Likewise.
3791 * tree-vectorizer.c (increase_alignment): Likewise.
3793 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
3795 * doc/rtl.texi: Adjust vec_select description.
3796 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
3797 non-constant selectors.
3799 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
3801 * config/aarch64/aarch64-protos.h
3802 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
3803 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
3804 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
3805 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
3806 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
3807 (AARCH64_FL_PROFILE): Move index so flags are ordered.
3808 (AARCH64_ISA_RCPC8_4): New flag.
3809 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
3810 to aarch64_offset_9bit_signed_unscaled_p.
3811 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
3813 * config/aarch64/constraints.md (Ust): New constraint.
3814 * config/aarch64/predicates.md.
3815 (aarch64_9bit_offset_memory_operand): New predicate.
3816 (aarch64_rcpc_memory_operand): New predicate.
3818 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
3820 PR rtl-optimization/87361
3821 * rtlanal.c (nonzero_bits1): Revert accidental change.
3823 2018-09-19 Richard Biener <rguenther@suse.de>
3825 PR tree-optimization/87349
3826 PR tree-optimization/87342
3827 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
3829 2018-09-18 Marek Polacek <polacek@redhat.com>
3831 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
3832 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
3834 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
3836 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
3838 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
3840 PR rtl-optimization/86882
3841 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
3843 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
3845 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
3846 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
3848 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
3851 * doc/invoke.texi (Link Options): Fix formatting and grammar.
3853 2018-09-18 Richard Biener <rguenther@suse.de>
3856 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
3857 (compute_samebase_partition_bases): Likewise.
3858 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
3859 (gimple_can_coalesce_p): Simplify.
3861 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
3863 Handle a library implementation of ffs calling __builtin_ffs.
3864 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
3865 (mmix_init_libfuncs): New function: make __builtin_ffs expand
3868 2018-09-17 David Malcolm <dmalcolm@redhat.com>
3870 * diagnostic-show-locus.c (class layout_range): Add field
3872 (layout_range::layout_range): Add "original_idx" param and use it
3873 to initialize new field.
3874 (make_range): Use 0 for original_idx.
3875 (layout::layout): Pass in index to calls to
3876 maybe_add_location_range.
3877 (layout::maybe_add_location_range): Add param "original_idx" and
3878 pass it on to layout_range.
3879 (layout::print_any_labels): Pass on range->m_original_idx to
3881 (gcc_rich_location::add_location_if_nearby): Use 0 for
3883 * gcc-rich-location.h (text_range_label::get_text): Update for new
3885 (range_label_for_type_mismatch::get_text): Likewise.
3887 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
3889 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
3891 2018-09-17 David Malcolm <dmalcolm@redhat.com>
3893 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
3894 format_string_diagnostic_t.
3895 (fmtwarn_n): Likewise.
3896 * substring-locations.c
3897 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
3898 (format_warning_n_va): Convert to...
3899 (format_string_diagnostic_t::emit_warning_n_va): ...this.
3900 (format_warning_va): Convert to...
3901 (format_string_diagnostic_t::emit_warning_va): ...this.
3902 (format_warning_at_substring): Convert to...
3903 (format_string_diagnostic_t::emit_warning): ...this.
3904 (format_warning_at_substring_n): Convert to...
3905 (format_string_diagnostic_t::emit_warning_n): ...this.
3906 * substring-locations.h (class format_string_diagnostic_t): New
3908 (format_warning_va): Convert to
3909 format_string_diagnostic_t::emit_warning_va.
3910 (format_warning_n_va): Convert to
3911 format_string_diagnostic_t::emit_warning_n_va.
3912 (format_warning_at_substring): Convert to
3913 format_string_diagnostic_t::emit_warning.
3914 (format_warning_at_substring_n): Convert to
3915 format_string_diagnostic_t::emit_warning_n.
3917 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
3918 Bernd Schmidt <bernds_cb1@t-online.de>
3920 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
3923 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
3925 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
3926 operand 0 predicate to nonimmediate operand.
3927 (rint<mode>2_frndint): Remove insn pattern.
3928 (rint<mode>2): Change operand 1 predicate to general_operand.
3929 Extend operand 1 to XFmode and generate rintxf2 insn.
3930 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
3931 Do not use X87MODEF mode macro.
3932 (frndintxf2_<rounding>_i387): Rename from
3933 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
3934 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
3935 to XFmode and generate significandxf3 insn.
3937 2018-09-17 Richard Biener <rguenther@suse.de>
3939 PR tree-optimization/87328
3940 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
3941 visiting unexecutable backedges when not iterating.
3942 (do_rpo_vn): Mark all edges not executable even when not
3945 2018-09-17 Martin Jambor <mjambor@suse.cz>
3948 * doc/invoke.texi (Warning Options): Likewise.
3950 2018-09-17 Richard Biener <rguenther@suse.de>
3952 PR tree-optimization/87301
3953 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
3954 clean EH info from leftover copy assignments.
3956 2018-09-17 Martin Liska <mliska@suse.cz>
3958 PR gcov-profile/85871
3959 * gcov.c (output_intermediate_file): Fix out of bounds
3962 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
3964 * config/arc/arc.c: Object attributes for core4 not reflected
3966 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
3969 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
3971 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
3973 2018-09-17 Martin Liska <mliska@suse.cz>
3975 * doc/gcov.texi: Document new option --use-hotness-colors.
3976 * gcov.c (struct source_info): Declare new field.
3977 (source_info::source_info): Set default for maximum_count.
3978 (print_usage): Add new -q option.
3979 (process_args): Process it.
3980 (accumulate_line_info): Save src->maximum_count.
3981 (output_line_beginning): Make color line number if
3982 flag_use_hotness_colors is set.
3983 (output_line_details): Pass default argument value.
3984 (output_lines): Pass src->maximum_count.
3986 2018-09-17 Martin Liska <mliska@suse.cz>
3988 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
3989 Use processor_names table.
3990 * config/i386/i386.c (ix86_default_align): Use
3991 processor_cost_table for alignment values.
3992 (ix86_option_override_internal): Use processor_names.
3993 (ix86_function_specific_print): Likewise.
3994 * config/i386/i386.h (struct processor_costs):
3995 Add alignment values.
3996 (struct ptt): Remove and replace with const char *.
3997 * config/i386/x86-tune-costs.h (struct processor_costs):
3998 Declare default alignments for all costs.
4000 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
4002 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
4003 symbolics or VR_VARYING ranges for ABS_EXPR.
4004 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
4005 when range will wrap.
4007 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
4010 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
4011 before and after a JUMP_TABLE_DATA.
4013 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
4016 * dojump.c (do_compare_and_jump): Canonicalize function pointers
4017 when one operand is a function pointer. Use POINTER_TYPE_P and
4018 FUNC_OR_METHOD_TYPE_P.
4019 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
4020 * fold-const.c (build_range_check): Likewise.
4021 * match.pd (simple_comparison): Likewise.
4023 2018-09-14 David Malcolm <dmalcolm@redhat.com>
4026 * spellcheck.c (get_edit_distance_cutoff): New function.
4027 (selftest::test_edit_distance_unit_test_oneway): Rename to...
4028 (selftest::test_get_edit_distance_one_way): ...this.
4029 (selftest::test_get_edit_distance_unit): Rename to...
4030 (selftest::test_get_edit_distance_both_ways): ...this.
4031 (selftest::test_edit_distances): Move tests to this new function,
4032 and test some more pairs of strings. Update for above renaming.
4033 (selftest::get_old_cutoff): New function.
4034 (selftest::test_get_edit_distance_cutoff): New function.
4035 (selftest::assert_suggested_for): New function.
4036 (ASSERT_SUGGESTED_FOR): New macro.
4037 (selftest::assert_not_suggested_for): New function.
4038 (ASSERT_NOT_SUGGESTED_FOR): New macro.
4039 (selftest::test_suggestions): New function.
4040 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
4041 tests to selftest::test_edit_distances and call it. Add calls to
4042 selftest::test_get_edit_distance_cutoff and
4043 selftest::test_suggestions.
4044 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
4045 (best_match::consider): Replace hard-coded cutoff calculation with
4047 (best_match::get_cutoff): New declaration.
4048 (best_match::get_best_meaningful_candidate): Likewise.
4050 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
4052 * builtins.c (fold_builtin_strlen): Remove TODO comment.
4054 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
4057 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
4059 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
4060 terminated string literal.
4062 2018-09-14 Martin Sebor <msebor@redhat.com>
4064 * builtins.c (unterminated_array): Handle ARRAY_REF.
4065 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
4066 * builtins.h (unterminated_array): Declare extern.
4067 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
4069 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
4072 2018-09-14 Martin Sebor <msebor@redhat.com>
4073 Jeff Law <law@redhat.com>
4075 * builtins.c (unterminated_array): New.
4076 (expand_builtin_strcpy): Adjust.
4077 (expand_builtin_strcpy_args): Detect unterminated arrays.
4078 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
4079 unterminated arrays.
4080 * gimple-fold.h (get_maxval_strlen): Add argument.
4081 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
4083 * gimple-fold.c (get_range_strlen): Add argument.
4084 (get_maxval_strlen): Adjust.
4085 * gimple-fold.h (get_range_strlen): Add argument.
4087 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
4089 * config/i386/movdirintrin.h: Fix copyright year.
4091 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
4093 * reg-stack.c: Include regs.h.
4094 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
4095 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
4096 FIRST_STACK_REG, not DFmode.
4097 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
4098 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
4099 (change stack): Default register mode to the reg_raw_mode of
4100 FIRST_STACK_REG, not DFmode.
4101 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
4102 (*swapxf): Rename from swapxf.
4104 2018-09-14 Carl Love <cel@us.ibm.com>
4106 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
4107 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
4109 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
4112 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
4115 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
4118 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
4120 2018-09-14 Jason Merrill <jason@redhat.com>
4122 Fix --enable-gather-detailed-mem-stats.
4123 * hash-table.c (hash_table_usage): Change from variable to function.
4124 * hash-table.h: Adjust.
4125 * Makefile.in: Add missing dependencies on hash-table.h.
4127 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4129 PR tree-optimization/87259
4131 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
4132 execute_cse_reciprocals_1 has tried transforming.
4134 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
4136 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
4137 VR_VARYING for PLUS/MINUS_EXPR.
4139 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
4141 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
4144 2018-09-14 Richard Biener <rguenther@suse.de>
4147 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
4148 bits for the merged partition.
4150 2018-09-13 Martin Sebor <msebor@redhat.com>
4151 Bernd Edlinger <bernd.edlinger@hotmail.de>
4153 * builtins.h (c_srlen): Add argument.
4154 * builtins.c (warn_string_no_nul): New function.
4155 (c_strlen): Add argument and use it. Update recursive calls.
4156 Pass DECL argument to string_constant to get info on non
4157 terminated strings. Update *NONSTR as needed.
4158 (fold_builtin_strlen): Add argument to calls to c_strlen.
4159 Warn for unterminated arrays.
4160 (warn_string_no_null): Add prototype.
4161 * expr.c (string_constant): Update arguments. Update recursive
4162 calls appropriately. Detect missing NUL terminator and outermost
4163 declaration its missing in.
4164 Improve checks for arrays with nonzero lower bound or elements
4165 that are not a single byte. Simplify offset computation.
4166 Simplify checks for non-NUL terminated strings.
4167 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
4168 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
4170 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
4172 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
4174 * fold-const.c (c_getstr): Fix function comment. Remove unused third
4175 argument. Fix range checks.
4176 * fold-const.h (c_getstr): Adjust protoype.
4177 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
4178 string is constant but contains no NUL byte.
4180 * expr.c (string_constant): Adjust function comment.
4181 Remove bogus check for zero termination.
4183 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
4185 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
4186 (get_constant_size): Don't make STRING_CSTs larger than they are.
4187 (check_string_literal): New check function for STRING_CSTs.
4188 (output_constant): Use it.
4190 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
4193 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
4195 2018-09-13 Richard Biener <rguenther@suse.de>
4197 PR tree-optimization/87263
4198 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
4199 (struct unwind_state): Add max_rpo field.
4200 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
4201 Compute max_rpo, the max RPO number a block can be backwards reached
4202 from. Re-write non-iterating mode to a RPO ordered worklist approach,
4203 separating it from the iterating mode.
4205 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
4207 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
4208 (rfs_decision): New scheduling decision.
4210 2018-09-13 Richard Biener <rguenther@suse.de>
4213 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
4214 (vn_nary_op_insert_pieces_predicated): Do not write useless
4215 valid_dominated_by_p entry outside of the allocated storage.
4217 2018-09-13 Omar Sandoval <osandov@osandov.com>
4218 Tom de Vries <tdevries@suse.de>
4221 * dwarf2out.c (is_c): New function.
4222 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
4224 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
4227 * config/aarch64/aarch64.md (*aarch64_bfxil):
4229 * config/aarch64/constraints.md (Ulc): Define.
4230 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
4232 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
4235 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
4237 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
4238 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
4239 aarch64_layout_frame call.
4240 (aarch64_expand_epilogue): Likewise.
4241 (aarch64_initial_elimination_offset): Likewise.
4242 (aarch64_get_separate_components): Likewise.
4243 (aarch64_use_return_insn_p): Likewise.
4244 (aarch64_layout_frame): Remove unneeded check.
4246 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4248 * configure.ac: Only append
4249 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
4250 gcc_config_arguments if it was never reconfigured or last reconfigure
4251 was with different arguments.
4252 * configure: Regenerated.
4254 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4255 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4258 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
4259 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
4261 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4263 PR tree-optimization/87287
4264 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
4265 X % C == 0 to X % (unsigned) C == 0 optimization to ...
4266 * match.pd (X % C == 0): ... here. New optimization.
4268 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4271 * expr.h (maybe_optimize_mod_cmp): Declare.
4272 * expr.c (mod_inv): New function.
4273 (maybe_optimize_mod_cmp): New function.
4274 (do_store_flag): Use it.
4275 * cfgexpand.c (expand_gimple_cond): Likewise.
4277 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
4278 Julian Brown <julian@codesourcery.com>
4281 * gimplify.c (gimplify_scan_omp_clauses): Set
4282 target_firstprivatize_array_bases in OpenACC parallel and kernels
4283 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
4284 OpenACC data regions.
4286 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
4288 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
4289 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
4291 2018-09-12 Richard Biener <rguenther@suse.de>
4293 PR tree-optimization/87280
4294 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
4295 edge but unreachable target.
4296 (do_rpo_vn): For conservatively handling a PHI only mark
4297 the backedge executable but not the block reachable.
4299 2018-09-12 Richard Biener <rguenther@suse.de>
4301 PR tree-optimization/87266
4302 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
4305 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
4307 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
4309 ("trunc<BFP:mode><DFP_ALL:mode>2")
4310 ("trunc<DFP_ALL:mode><BFP:mode>2")
4311 ("extend<BFP:mode><DFP_ALL:mode>2")
4312 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
4313 according to the target operand type.
4315 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4316 Andreas Krebbel <krebbel@linux.ibm.com>
4318 PR tree-optimization/86844
4319 * gimple-ssa-store-merging.c
4320 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
4321 there are any overlapping stores in between them, make sure they are
4322 also coalesced or we give up completely.
4324 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4327 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
4328 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
4330 2018-09-12 Tom de Vries <tdevries@suse.de>
4332 * common.opt (gdescribe-dies): Add option.
4333 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
4334 attribute for artifical and nameless decls.
4335 (dwarf2out_register_external_die): Add description attribute to
4336 external reference die.
4337 (add_desc_attribute): New functions.
4338 (gen_subprogram_die): Add description attribute to
4339 DW_TAG_call_site_parameter.
4340 * tree-pretty-print.c (print_generic_expr_to_str): New function.
4341 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
4342 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
4344 (@item -gdescribe-dies): Add.
4346 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
4348 * tree-vrp.c (vrp_shift_undefined_p): Remove.
4349 (extract_range_from_binary_expr_1: Call
4350 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
4351 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
4354 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
4356 * gimple-ssa-warn-alloca.c
4357 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
4358 field for ALLOCA_BOUND_*_LARGE.
4360 2018-09-11 Nathan Sidwell <nathan@acm.org>
4362 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
4364 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4366 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
4367 for clobbers. Remove obsolete comment.
4369 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4371 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
4372 mpxchk, mpxld and mpxst types.
4373 (define_attr length_immediate): Remove all processing of mpx types.
4374 (define_attr prefix_0f): Ditto.
4375 (define_attr memory): Ditto.
4377 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4379 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
4380 (log<mode>2): Change operand 1 predicate to general_operand.
4381 Extend operand 1 to XFmode and generate logxf3 insn.
4382 (log10<mode>2): Change operand 1 predicate to general_operand.
4383 Extend operand 1 to XFmode and generate log10xf3 insn.
4384 (log2<mode>2): Change operand 1 predicate to general_operand.
4385 Extend operand 1 to XFmode and generate log2xf3 insn.
4386 (fyl2xp1_extend<mode>xf3_i387): Remove.
4387 (log1p<mode>2): Change operand 1 predicate to general_operand.
4388 Extend operand 1 to XFmode and generate log1pxf3 insn.
4389 (fxtract_extend<mode>xf3_i387): Remove.
4390 (logb<mode>2): Change operand 1 predicate to general_operand.
4391 Extend operand 1 to XFmode and generate logbxf3 insn.
4392 (ilogb<mode>2): Change operand 1 predicate to general_operand.
4393 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
4394 (significand<mode>2): Change operand 1 predicate to general_operand.
4395 Extend operand 1 to XFmode and generate significandxf3 insn.
4397 2018-09-11 Nathan Sidwell <nathan@acm.org>
4399 * gcc.c (perror_with_name, pfatal_with_name): Delete.
4400 (load_specs): Use fatal_error.
4401 (DELETE_IF_ORDINARY, process_command): Use error.
4402 (execute, run_attempt): Use fatal_error.
4404 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
4406 * diagnostic-core.h (sorry_at): New prototype.
4407 * diagnostic.c (sorry_at): New function.
4409 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
4411 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
4412 by zero as VR_UNDEFINED.
4414 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
4416 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
4417 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
4418 (<sincos>mode2): New expander.
4419 (sincos_extend<mode>xf3_i387): Remove insn pattern.
4420 (sincos -> sin, cos splitters): Remove splitter patterns.
4421 (sincos<mode>3): Change operand 2 predicate to general_operand.
4422 Extend operand 2 to XFmode and generate sincosxf3 insn.
4423 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
4424 Change operand 3 predicate to const1_operand.
4425 (fptan_extend<mode>xf4_i387): Remove insn pattern.
4426 (tanxf2): Update operands in the call to fptanxf4_i387.
4427 (tan<mode>2): Change operand 1 predicate to general_operand.
4428 Extend operand 1 to XFmode and generate tanxf3 insn.
4429 (atan2xf3): Rename from *fpatanxf3_i387.
4430 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
4431 (atan2xf3): Remove expander.
4432 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
4433 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
4434 (atan<mode>2): Change operand 1 predicate to general_operand.
4435 Extend operand 1 to XFmode and generate atanxf3 insn.
4437 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
4439 * config/i386/i386.md (x87/SSE constant load splitter): Use
4440 memory_operand instead of nonimmediate_operand for input operand
4443 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
4445 * config/i386/i386.md (float partial SSE register stall splitter): Move
4446 splitter near its instruction pattern.
4447 (float_extend partial SSE register stall splitter): Ditto.
4448 (float_truncate partial SSE register stall splitter): Ditto.
4450 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
4453 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
4454 to speculation_safe_value_not_needed.
4457 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
4458 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
4459 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
4460 leaf_function_p, instead use has_hard_reg_initial_val.
4462 2018-09-09 Nathan Sidwell <nathan@acm.org>
4464 * gcc.h (pfatal_with_name): Don't declare here.
4465 * gcc.c (pfatal_with_name): Make static.
4467 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
4469 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
4472 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
4474 PR rtl-optimization/85458
4475 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
4476 priority hook to reduce the priority of EXPR.
4478 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
4480 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
4481 DImode for x87 on 32bit targets. Conditionally disable x87 modes
4482 with X87_ENABLE_FLOAT. Remove preparation code.
4483 (*float<SWI48:mode><MODEF:mode>2): Rename from
4484 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
4485 math using "enabled" attribute.
4486 (*floatdi<MODEF:mode>2_i387): Rename from
4487 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
4488 enable for 32bit targets only.
4489 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
4491 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
4492 as operand 1 predicate. Rewrite as define_insn_and_split.
4493 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
4495 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
4497 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
4498 to fallthru to FLOAT case.
4500 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
4503 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
4504 around folding of vec_sl to handle out of range shift values.
4506 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
4508 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
4509 Update callers to gen_fix_trunc<mode>_i387_fisttp
4510 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
4511 nonimmediate_operand.
4512 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
4513 and corresponding splitters.
4514 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
4515 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
4516 (fix_truncdi_i387_with_temp): Remove insn pattern
4517 and corresponding splitters.
4518 (fix_trunc<mode>_i387): Change operand 0 predicate to
4519 nonimmediate_operand.
4520 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
4521 and corresponding splitters.
4522 (*fistdi2_1): Remove.
4524 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
4525 (lrintxfdi2): Remove expander. Reimplement as define_insn.
4526 (*fist<mode>2_1): Remove.
4527 (fist<mode>2): Ditto.
4528 (fist<mode>2_with_temp): Remove insn pattern and corresponding
4530 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
4531 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
4532 (fistdi2_<rounding>): Change operand 0 predicate to
4533 nonimmediate_operand.
4534 (fistdi2_<rounding>_with_temp): Remove insn pattern
4535 and corresponding splitters.
4536 (fist<mode>2_<rounding>): Change operand 0 predicate to
4537 nonimmediate_operand.
4538 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
4539 and corresponding splitters.
4541 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
4543 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
4545 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
4548 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
4550 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4551 early gimple folding of vec_splat().
4552 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
4553 * gimple-fold.h: Add an extern define for tree_vec_extract().
4555 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
4557 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
4558 wrappers around TREE_TYPE comparisons.
4560 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
4563 * config/s390/predicates.md: Add nonsym_memory_operand.
4564 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
4565 contains a SYMBOL_REF, load it into an intermediate pseudo.
4566 (s390_emit_compare_and_swap): Legitimize operand.
4567 * config/s390/s390.md: Use the new nonsym_memory_operand
4568 with UNSPECV_CAS patterns.
4570 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
4573 * config/s390/s390-passes.def: New file.
4574 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
4576 (make_pass_s390_early_mach): Add declaration.
4577 * config/s390/s390.c (make_pass_s390_early_mach):
4578 (s390_option_override): Remove dynamic registration.
4579 * config/s390/t-s390: Add s390-passes.def.
4581 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
4583 * config/s390/s390.c (s390_decompose_constant_pool_ref):
4584 Remove UNSPEC_LTREL_BASE check.
4585 (annotate_constant_pool_refs): Likewise.
4586 (find_constant_pool_ref): Likewise.
4587 (find_ltrel_base): Removed.
4588 (replace_ltrel_base): Removed.
4589 (s390_mainpool_finish): Remove replace_ltrel_base call.
4590 (s390_chunkify_start): Remove pending LTREL_BASE logic.
4591 (s390_chunkify_finish): Remove replace_ltrel_base call.
4592 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
4594 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
4597 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
4598 to speculation_safe_value_not_needed.
4600 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
4601 Bernd Schmidt <bernds_cb1@t-online.de>
4603 * config/nvptx/nvptx-opts.h: New file.
4604 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
4605 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
4607 (TARGET_SM35): New macro.
4608 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
4610 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
4612 (misa=): New option.
4613 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
4615 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
4617 * config/i386/i386.md (truncdfsf2): Remove expander.
4618 (truncdfsf2_with_temp): Ditto.
4619 (truncxf<mode>2): Ditto.
4620 (*truncdfsf_fast_mixed): Remove insn pattern.
4621 (*truncdfsf_fast_i387): Ditto.
4622 (*truncdfsf_mixed): Ditto.
4623 (*truncdfsf_i387): Ditto.
4624 (*truncdfsf2_i387_1): Ditto.
4625 (*truncxfsf2_mixed): Ditto.
4626 (*truncxfdf2_mixed): Ditto.
4627 (*truncxf<mode>2_i387_noop): Ditto. Update callers
4628 to call gen_truncxf<mode>2 instead.
4629 (*truncxf<mode>2_i387): Remove.
4630 (reg->reg splitters): Remove splitter pattern.
4631 (reg->mem splitters): Ditto.
4633 (truncdfsf2): New insn pattern.
4634 (truncxf<mode>2): Ditto.
4636 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4638 * tree-ssa-math-opts.c (is_mult_by): New function.
4639 (is_square_of): Use the above.
4640 (optimize_recip_sqrt): New function.
4641 (pass_cse_reciprocals::execute): Use the above.
4643 2018-09-05 Richard Biener <rguenther@suse.de>
4646 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
4647 to zero-init the emplaced vec.
4649 2018-09-05 Martin Liska <mliska@suse.cz>
4651 PR tree-optimization/87205
4652 * tree-switch-conversion.c (pass_lower_switch::execute):
4653 Group cases for switch statements.
4655 2018-09-05 Richard Biener <rguenther@suse.de>
4657 PR tree-optimization/87217
4658 * tree-ssa-sccvn.c (vuse_valueize): New.
4659 (vn_reference_lookup_pieces): Use it.
4660 (vn_reference_lookup): Likewise.
4662 2018-09-05 Nathan Sidwell <nathan@acm.org>
4665 * stor-layout.c (place_field): Scan forwards to check last
4666 bitfield when ms_bitfield_placement is in effect.
4668 2018-09-05 Richard Biener <rguenther@suse.de>
4671 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
4674 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
4675 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
4677 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
4678 * config.gcc (extra_objs): Build it.
4679 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
4681 * config/aarch64/aarch64-passes.def
4682 (pass_tag_collision_avoidance): New pass.
4683 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
4684 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
4685 (aarch64_classify_address): Remove static qualifier.
4686 (aarch64_address_info, aarch64_address_type): Move to...
4687 * config/aarch64/aarch64-protos.h: ... here.
4688 (make_pass_tag_collision_avoidance): New function.
4689 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
4692 2018-09-05 Martin Liska <mliska@suse.cz>
4694 * doc/gcov.texi: Update documentation of humar
4696 * gcov.c (format_count): Print one decimal place, it provides
4697 more fine number of situations like '1G' vs. '1.4G'.
4699 2018-09-05 Martin Liska <mliska@suse.cz>
4702 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
4703 * optc-gen.awk: Allow 'Var' for Deprecated options in order
4704 to generate a MASK value.
4706 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
4709 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
4710 if frame pointer isn't used.
4711 (compute_frame_pointer_to_fb_displacement): Likewise.
4713 2018-09-04 Jakub Jelinek <jakub@redhat.com>
4716 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
4717 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
4718 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
4719 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
4720 and OPTION_MASK_ISA_XSAVEC_UNSET.
4722 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
4724 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
4725 XOR operations in NAND case.
4727 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
4729 * wide-int-range.cc (wide_int_range_convert): New.
4730 * wide-int-range.h (wide_int_range_convert): New.
4731 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
4732 code into wide_int_range_convert.
4733 (extract_range_into_wide_ints): Do not munge anti range constants
4734 into the entire domain. Just return the range back.
4736 2018-09-04 Martin Liska <mliska@suse.cz>
4738 * genmatch.c (output_line_directive): Add new argument
4740 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
4742 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
4744 * doc/invoke.texi (Option Summary): Add whitespace.
4746 * doc/invoke.texi (Option Summary): Add -Waligned-new.
4748 2018-09-04 Richard Biener <rguenther@suse.de>
4750 PR tree-optimization/87211
4751 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
4752 backedge value we're supposed to treat as VARYING also number
4753 the PHI to VARYING in case it got a different value-number already.
4755 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
4757 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
4758 (extract_range_from_binary_expr_1): Do not call
4759 vrp_can_optimize_bit_op.
4760 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
4762 (wide_int_range_get_mask_and_bounds): New.
4763 (wide_int_range_optimize_bit_op): New.
4764 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
4765 (wide_int_range_bit_and): Same.
4766 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
4767 (wide_int_range_optimize_bit_op): New.
4768 (wide_int_range_get_mask_and_bounds): New.
4770 2018-09-04 Richard Biener <rguenther@suse.de>
4772 PR tree-optimization/87176
4773 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
4774 variable. When value-numbering a virtual PHI node make sure
4775 to not value-number to the backedge value.
4777 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
4779 * doc/extend.texi (Long Long, Hex Floats): Document support for
4780 long long and hex floats in more recent versions of ISO C++.
4782 2018-09-03 Richard Biener <rguenther@suse.de>
4784 PR tree-optimization/87177
4785 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
4788 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4790 * bb-reorder.c (edge_order): Convert to C-qsort-style
4791 tri-state comparator.
4792 (reorder_basic_blocks_simple): Change std::stable_sort to
4795 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4797 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
4798 tri-state comparator.
4799 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
4801 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4803 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
4804 (mergesort): ... here as maximum count for using netsort.
4805 (gcc_qsort): Set nlim to 3 if stable sort is requested.
4806 (gcc_stablesort): New.
4807 * system.h (gcc_stablesort): Declare.
4809 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4811 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
4812 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
4813 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
4815 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
4817 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
4818 lxsdx and stxsdx alternatives.
4819 (*mov<mode>_hardfloat64): Ditto.
4820 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
4822 2018-09-03 Richard Biener <rguenther@suse.de>
4824 PR tree-optimization/87200
4825 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
4828 2018-09-03 Martin Liska <mliska@suse.cz>
4830 PR tree-optimization/87201
4831 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
4832 Fix parenthesis in an expression.
4834 2018-09-03 Richard Biener <rguenther@suse.de>
4836 PR tree-optimization/87197
4837 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
4838 visited. CSE the VN_INFO hashtable lookup.
4840 PR tree-optimization/87169
4841 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
4842 iterating make sure there's no extra backedges from irreducible
4843 regions feeding the header. Mark the destination block
4846 2018-09-03 Martin Liska <mliska@suse.cz>
4849 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
4850 * common/common-targhooks.c (default_get_valid_option_values):
4852 * common/common-targhooks.h (default_get_valid_option_values):
4854 * common/config/i386/i386-common.c: Move processor_target_table
4856 (ix86_get_valid_option_values): New function.
4857 (TARGET_GET_VALID_OPTION_VALUES): New macro.
4858 * config/i386/i386.c (struct ptt): Move to i386-common.c.
4859 (PTA_*): Move all defined masks into i386-common.c.
4860 (ix86_function_specific_restore): Use new processor_cost_table.
4861 * config/i386/i386.h (struct ptt): Moved from i386.c.
4862 (struct pta): Likewise.
4863 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
4864 * doc/tm.texi.in: Likewise.
4865 * opt-suggestions.c (option_proposer::suggest_option):
4866 Pass prefix to build_option_suggestions.
4867 (option_proposer::get_completions): Likewise.
4868 (option_proposer::build_option_suggestions): Use the new target
4870 * opts.c (struct option_help_tuple): New struct.
4871 (print_filtered_help): Use the new target hook.
4873 2018-09-03 Martin Liska <mliska@suse.cz>
4876 * predict.c (set_even_probabilities): Add likely_edges
4877 argument and handle cases where we have precisely one
4879 (combine_predictions_for_bb): Catch also likely_edges.
4880 (tree_predict_by_opcode): Handle gswitch statements.
4881 * tree-cfg.h (find_case_label_for_value): New declaration.
4882 (find_taken_edge_switch_expr): Likewise.
4883 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
4884 Find pivot in decision tree based on probabily, not by number of
4887 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
4889 * doc/standards.texi (Standards): Update Objective-C reference.
4891 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
4893 * doc/install.texi (Prerequisites): Update link for MPC.
4895 2018-09-01 Michael Matz <matz@suse.de>
4897 PR tree-optimization/87074
4898 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
4899 PHIs for outer-loop uses.
4901 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
4903 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
4904 * doc/invoke.texi (C Dialect Options): Ditto.
4906 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
4908 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
4910 2018-08-31 Richard Biener <rguenther@suse.de>
4912 PR tree-optimization/87168
4913 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
4914 (rpo_elim::eliminate_avail): When OP was not visited it must
4917 2018-08-31 David Malcolm <dmalcolm@redhat.com>
4919 * tree-vrp.c (copy_value_range): Convert param "from" from
4920 "value_range *" to "const value_range *".
4921 (range_is_null): Likewise for param "vr".
4922 (range_int_cst_p): Likewise.
4923 (range_int_cst_singleton_p): Likewise.
4924 (symbolic_range_p): Likewise.
4925 (value_ranges_intersect_p): Likewise for both params.
4926 (value_range_nonnegative_p): Likewise for param "vr".
4927 (value_range_constant_singleton): Likewise.
4928 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
4929 (extract_range_into_wide_ints): Likewise for param "vr".
4930 (extract_range_from_multiplicative_op): Likewise for params "vr0"
4932 (vrp_can_optimize_bit_op): Likewise.
4933 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
4935 (extract_range_from_unary_expr): Likewise.
4936 (debug_value_range): Likewise for param "vr".
4937 (value_range::dump): Add "const" qualifier.
4938 (vrp_prop::check_array_ref): Convert local "vr" from
4939 "value_range *" to "const value_range *".
4940 (vrp_prop::check_mem_ref): Likewise.
4941 (vrp_prop::visit_stmt): Likewise for local "old_vr".
4942 (vrp_intersect_ranges_1): Likewise for param "vr_1".
4943 (vrp_intersect_ranges): Likewise.
4944 (simplify_stmt_for_jump_threading): Likewise for local "vr".
4945 (vrp_prop::vrp_finalize): Likewise.
4946 * tree-vrp.h (value_range::dump): Add "const" qualifier.
4947 (vrp_intersect_ranges): Add "const" qualifier to params as above.
4948 (extract_range_from_unary_expr): Likewise.
4949 (value_range_constant_singleton): Likewise.
4950 (symbolic_range_p): Likewise.
4951 (copy_value_range): Likewise.
4952 (extract_range_from_binary_expr_1): Likewise.
4953 (range_int_cst_p): Likewise.
4954 (vrp_set_zero_nonzero_bits): Likewise.
4955 (range_int_cst_singleton_p): Likewise.
4957 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
4959 * config/aarch64/arm_neon.h (vabsd_s64): New.
4960 (vnegd_s64): Likewise.
4962 2018-08-31 Martin Jambor <mjambor@suse.cz>
4964 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
4966 2018-08-31 Martin Liska <mliska@suse.cz>
4968 * ipa-icf.c (sem_item::add_type): Use
4969 sem_item::m_type_hash_cache.
4970 * ipa-icf.h: Move the cache from sem_item_optimizer
4973 2018-08-31 Nathan Sidwell <nathan@acm.org>
4975 * doc/extend.texi (Backwards Compatibility): Remove implicit
4976 extern C leeway of () being (...).
4978 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4980 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
4982 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
4986 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
4988 2018-08-31 Jakub Jelinek <jakub@redhat.com>
4991 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
4992 gen_int_mode. Formatting fixes.
4994 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
4996 * target.def (custom_function_descriptors): Improve documentation.
4997 * doc/tm.texi.in (Trampolines): Expand discussion of function
4998 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
4999 beginning of the section.
5000 * doc/tm.texi: Regenerated.
5002 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
5004 * cfg.h (class auto_edge_flag): Spell out the template-id of the
5005 base class in the initializer list. This is a workaround for
5006 building with older GCC.
5007 (class auto_bb_flag): Likewise.
5009 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
5011 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
5012 (altivec_vcmpequ<VI_char>_p): Remove star.
5013 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
5015 (expand_strncmp_vec_sequence): New function.
5016 (emit_final_str_compare_vec): New function.
5017 (expand_strn_compare): Add support for vector strncmp.
5018 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
5019 length specification to bytes.
5020 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
5021 (vcmpnezb_p): New pattern.
5022 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
5023 for option -mstring-compare-inline-limit.
5025 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
5027 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
5028 (PTA_SKYLAKE): Add PTA_AES.
5029 (PTA_GOLDMONT): Likewise.
5031 2018-08-29 Jan Hubicka <jh@suse.cz>
5034 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
5035 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
5037 2018-08-29 Jan Hubicka <jh@suse.cz>
5039 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
5041 (hash_tree): Do not visit TYPE_STUB_DECL.
5042 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
5043 stream TYPE_STUB_DECL.
5044 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
5045 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
5046 after free_lang_data.
5047 (type_in_anonymous_namespace_p): Likewise.
5049 2018-08-29 Jan Hubicka <jh@suse.cz>
5051 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
5052 comment that it has to be even number.
5053 (class sreal): Change m_sig type to int32_t.
5054 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
5055 int64_t for temporary calculations.
5056 (sreal_verify_basics): Drop one bit from minimum and maximum.
5058 2018-08-30 Richard Biener <rguenther@suse.de>
5060 PR tree-optimization/87147
5061 * tree-ssa-sccvn.c (SSA_VISITED): New function.
5062 (visit_phi): When the degenerate result is from the backedge and
5063 we didn't visit its definition yet drop to VARYING.
5064 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
5066 2018-08-29 Jan Hubicka <jh@suse.cz>
5068 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
5070 (hash_tree): Likewise.
5072 2018-08-29 Jan Hubicka <jh@suse.cz>
5074 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
5075 and TYPE_NEXT_REF_TO.
5077 2018-08-29 Jan Hubicka <jh@suse.cz>
5079 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
5080 comment that it has to be even number.
5081 (class sreal): Change m_sig type to int32_t.
5082 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
5083 int64_t for temporary calculations.
5084 (sreal_verify_basics): Drop one bit from minimum and maximum.
5086 2018-08-30 Tamar Christina <tamar.christina@arm.com>
5088 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
5090 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
5093 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
5094 if to_add is negative.
5096 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
5099 * builtins.c (c_strlen): Improve range checks.
5101 2018-08-29 Martin Sebor <msebor@redhat.com>
5102 Jeff Law <law@redhat.com>
5104 PR tree-optimization/86714
5105 PR tree-optimization/86711
5106 * builtins.c (c_strlen): Add arguments to call to string_constant.
5107 * expr.c (string_constant): Add argument. Detect missing nul
5108 terminator and outermost declaration it's missing in.
5109 * expr.h (string_constant): Add argument.
5110 * fold-const.c (read_from_constant_string): Add arguments to call to
5112 (c_getstr): Likewise.
5113 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
5115 * tree-ssa-strlen.c (get_stridx): Likewise.
5117 2018-08-29 Jan Hubicka <jh@suse.cz>
5119 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
5120 Do not stream DECL_VINDEX.
5121 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
5122 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
5123 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
5125 2018-08-29 Richard Biener <rguenther@suse.de>
5127 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
5128 virtual operands that are not default defs to honor region
5130 (rpo_vn_valueize): Remove ineffective code here.
5132 2018-08-29 Richard Biener <rguenther@suse.de>
5134 PR tree-optimization/87132
5135 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
5136 when skipping defs reachable over backedges.
5138 2018-08-29 Richard Biener <rguenther@suse.de>
5140 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
5141 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
5142 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
5143 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
5144 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
5145 (vn_reference_lookup_3): Remove use of const_parms.
5146 (free_rpo_vn): Do not free const_parms.
5147 (do_rpo_vn): Do not call init_const_parms.
5148 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
5149 SSA_NAME_POINTS_TO_READONLY_MEMORY.
5150 (call_may_clobber_ref_p_1): Likewise.
5152 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
5155 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
5157 (-ftree-scev-cprop): Document.
5159 2018-08-29 Jan Hubicka <jh@suse.cz>
5161 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
5163 (sreal constructor): Update.
5164 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
5165 sreal:operator/): Update.
5167 2018-08-29 Martin Liska <mliska@suse.cz>
5169 * tree-switch-conversion.c (switch_conversion::expand):
5170 Strenghten assumption about gswitch statements.
5172 2018-08-29 Richard Biener <rguenther@suse.de>
5174 PR tree-optimization/87117
5175 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
5176 re-value-number released SSA VDEFs.
5178 2018-08-29 Richard Biener <rguenther@suse.de>
5180 PR tree-optimization/87126
5181 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
5183 2018-08-28 Jim Wilson <jimw@sifive.com>
5185 * config/riscv/pic.md: Rewrite.
5186 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
5188 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
5189 (SOFTF, default_load, softload, softstore): New.
5191 2018-08-28 Jeff Law <law@redhat.com>
5193 * fold-const.c (fold_binary_loc): Remove recently added assert.
5195 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
5197 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
5198 to OP parmeter of generated function.
5200 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
5202 PR tree-optimization/87009
5203 * match.pd: Add boolean optimizations.
5205 2018-08-28 Martin Sebor <msebor@redhat.com>
5208 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
5209 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
5210 (pass_walloca::gate): Use it.
5211 (alloca_call_type): Same.
5212 (pass_walloca::execute): Same.
5213 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
5215 2018-08-28 David Malcolm <dmalcolm@redhat.com>
5217 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
5218 GCC_VERSION for usage of "__gcc_dump_printf__" format from
5221 2018-08-28 Richard Biener <rguenther@suse.de>
5223 PR tree-optimization/87124
5224 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
5225 constants before looking up avail.
5227 2018-08-28 Jakub Jelinek <jakub@redhat.com>
5230 * calls.c (maybe_warn_nonstring_arg): Punt early if
5231 warn_stringop_overflow is zero. Don't call get_range_strlen
5232 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
5233 Swap comparison operands to have constants on rhs. Only use
5234 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
5235 increment lenrng[0].
5237 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
5239 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
5240 use of tree_to_shwi. Remove duplicated test for the size being
5241 a whole number of bytes.
5243 2018-08-28 Richard Biener <rguenther@suse.de>
5245 PR tree-optimization/87117
5246 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
5247 Handle removed stmt without LHS (GIMPLE_NOP).
5249 2018-08-28 Richard Biener <rguenther@suse.de>
5251 PR tree-optimization/87117
5252 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
5253 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
5255 2018-08-28 Richard Biener <rguenther@suse.de>
5257 PR tree-optimization/87117
5258 * tree-ssa-pre.c (compute_avail): Do not make expressions
5259 with predicated values available.
5260 (get_expr_value_id): Assert we do not run into predicated value
5263 2018-08-28 Richard Biener <rguenther@suse.de>
5265 PR tree-optimization/87117
5266 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
5267 get virtual operands.
5268 (get_expr_operands): Handle STRING_CST like other decls.
5270 2018-08-28 Martin Liska <mliska@suse.cz>
5272 * tree.h: Update documentation of fndecl_built_in_p
5276 2018-08-27 Jeff Law <law@redhat.com>
5277 PR tree-optimization/87110
5278 * tree-ssa-dse.c (compute_trims): Handle non-constant
5281 2018-08-27 Martin Sebor <msebor@redhat.com>
5283 PR tree-optimization/86914
5284 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
5286 2018-08-27 Martin Sebor <msebor@redhat.com>
5288 PR tree-optimization/87112
5289 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
5290 the type of the bound argument.
5292 2018-08-27 Jeff Law <law@redhat.com>
5294 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
5295 type does not have a TYPE_SIZE_UNIT.
5297 2018-08-27 Steve Ellcey <sellcey@cavium.com>
5299 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
5300 with include of backend.h.
5302 2018-08-27 Richard Biener <rguenther@suse.de>
5304 PR tree-optimization/86927
5305 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
5306 use const cond reduction code.
5308 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
5310 PR tree-optimization/85758
5311 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
5313 2018-08-27 David Malcolm <dmalcolm@redhat.com>
5316 * diagnostic-show-locus.c (class layout_range): Update for
5317 conversion of show_caret_p to a tri-state.
5318 (layout_range::layout_range): Likewise.
5319 (make_range): Likewise.
5320 (layout::maybe_add_location_range): Likewise.
5321 (layout::should_print_annotation_line_p): Don't show annotation
5322 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
5323 (layout::get_state_at_point): Update for conversion of
5324 show_caret_p to a tri-state. Bail out early for
5325 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
5326 underlining or source colorization.
5327 (gcc_rich_location::add_location_if_nearby): Update for conversion
5328 of show_caret_p to a tri-state.
5329 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
5330 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
5332 (selftest::test_one_liner_labels): Likewise.
5333 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
5334 conversion of show_caret_p to a tri-state.
5335 * pretty-print.c (text_info::set_location): Likewise.
5336 * pretty-print.h (text_info::set_location): Likewise.
5337 * substring-locations.c (format_warning_n_va): Likewise.
5338 * tree-diagnostic.c (default_tree_printer): Likewise.
5339 * tree-pretty-print.c (newline_and_indent): Likewise.
5341 2018-08-27 David Malcolm <dmalcolm@redhat.com>
5344 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
5345 line above for line-insertion fix-it hints.
5346 (selftest::test_fixit_insert_containing_newline): Update the
5347 expected results, and add a test with line-numbering enabled.
5349 2018-08-27 Martin Liska <mliska@suse.cz>
5352 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
5353 params with DECL_HAS_VALUE_EXPR_P.
5355 2018-08-27 Martin Liska <mliska@suse.cz>
5357 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
5358 selected expansion strategy.
5360 2018-08-27 Martin Liska <mliska@suse.cz>
5362 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
5363 * builtins.c (is_builtin_fn): Likewise.
5364 * attribs.c (diag_attr_exclusions): Use new function
5365 fndecl_built_in_p and remove check for FUNCTION_DECL if
5367 (builtin_mathfn_code): Likewise.
5368 (fold_builtin_expect): Likewise.
5369 (fold_call_expr): Likewise.
5370 (fold_builtin_call_array): Likewise.
5371 (fold_call_stmt): Likewise.
5372 (set_builtin_user_assembler_name): Likewise.
5373 (is_simple_builtin): Likewise.
5374 * calls.c (gimple_alloca_call_p): Likewise.
5375 (maybe_warn_nonstring_arg): Likewise.
5376 * cfgexpand.c (expand_call_stmt): Likewise.
5377 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
5378 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
5379 (cgraph_node::verify_node): Likewise.
5380 * cgraphclones.c (build_function_decl_skip_args): Likewise.
5381 (cgraph_node::create_clone): Likewise.
5382 * config/arm/arm.c (arm_insert_attributes): Likewise.
5383 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
5384 * dse.c (scan_insn): Likewise.
5385 * expr.c (expand_expr_real_1): Likewise.
5386 * fold-const.c (operand_equal_p): Likewise.
5387 (fold_binary_loc): Likewise.
5388 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
5389 * gimple-low.c (lower_stmt): Likewise.
5390 * gimple-pretty-print.c (dump_gimple_call): Likewise.
5391 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
5393 * gimple.c (gimple_build_call_from_tree): Likewise.
5394 (gimple_call_builtin_p): Likewise.
5395 (gimple_call_combined_fn): Likewise.
5396 * gimplify.c (gimplify_call_expr): Likewise.
5397 (gimple_boolify): Likewise.
5398 (gimplify_modify_expr): Likewise.
5399 (gimplify_addr_expr): Likewise.
5400 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
5401 * ipa-cp.c (determine_versionability): Likewise.
5402 * ipa-fnsummary.c (compute_fn_summary): Likewise.
5403 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
5404 * ipa-split.c (visit_bb): Likewise.
5405 (split_function): Likewise.
5406 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
5407 * lto-cgraph.c (input_node): Likewise.
5408 * lto-streamer-out.c (write_symbol): Likewise.
5409 * omp-low.c (setjmp_or_longjmp_p): Likewise.
5410 (lower_omp_1): Likewise.
5411 * predict.c (strip_predict_hints): Likewise.
5412 * print-tree.c (print_node): Likewise.
5413 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
5414 * trans-mem.c (is_tm_irrevocable): Likewise.
5415 (is_tm_load): Likewise.
5416 (is_tm_simple_load): Likewise.
5417 (is_tm_store): Likewise.
5418 (is_tm_simple_store): Likewise.
5419 (is_tm_abort): Likewise.
5420 (tm_region_init_1): Likewise.
5421 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
5422 * tree-cfg.c (verify_gimple_call): Likewise.
5423 (move_stmt_r): Likewise.
5424 (stmt_can_terminate_bb_p): Likewise.
5425 * tree-eh.c (lower_eh_constructs_2): Likewise.
5426 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
5427 * tree-inline.c (remap_gimple_stmt): Likewise.
5428 (copy_bb): Likewise.
5429 (estimate_num_insns): Likewise.
5430 (fold_marked_statements): Likewise.
5431 * tree-sra.c (scan_function): Likewise.
5432 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
5433 (optimize_stack_restore): Likewise.
5434 (pass_fold_builtins::execute): Likewise.
5435 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5436 (mark_all_reaching_defs_necessary_1): Likewise.
5437 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
5438 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
5439 (pass_forwprop::execute): Likewise.
5440 * tree-ssa-loop-im.c (stmt_cost): Likewise.
5441 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5442 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
5443 * tree-ssa-strlen.c (get_string_length): Likewise.
5444 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
5445 (find_func_aliases_for_call): Likewise.
5446 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
5447 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
5448 * tree-tailcall.c (find_tail_calls): Likewise.
5449 * tree.c (need_assembler_name_p): Likewise.
5450 (free_lang_data_in_decl): Likewise.
5451 (get_call_combined_fn): Likewise.
5452 * ubsan.c (is_ubsan_builtin_p): Likewise.
5453 * varasm.c (incorporeal_function_p): Likewise.
5454 * tree.h (DECL_BUILT_IN): Remove and replace with
5456 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
5457 (fndecl_built_in_p): New.
5459 2018-08-27 Martin Liska <mliska@suse.cz>
5461 PR tree-optimization/86847
5462 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
5463 Dump also subtree probability.
5464 (switch_decision_tree::do_jump_if_equal): New function.
5465 (switch_decision_tree::emit_case_nodes): Handle special
5466 situations in balanced tree that can be emitted much simpler.
5467 Fix calculation of probabilities that happen in tree expansion.
5468 * tree-switch-conversion.h (struct cluster): Add
5470 (struct simple_cluster): Likewise.
5471 (struct case_tree_node): Add new function has_child.
5472 (do_jump_if_equal): New.
5474 2018-08-27 Martin Liska <mliska@suse.cz>
5476 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
5477 Add new argument to bit_test_cluster constructor.
5478 (bit_test_cluster::emit): Set bits really number of values
5480 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
5481 probability argument.
5482 * tree-switch-conversion.h (struct bit_test_cluster):
5483 Add m_handles_entire_switch.
5485 2018-08-27 Martin Liska <mliska@suse.cz>
5487 PR tree-optimization/86702
5488 * tree-switch-conversion.c (jump_table_cluster::emit):
5489 Make probabilities even for values in jump table
5490 according to number of cases handled.
5491 (switch_decision_tree::compute_cases_per_edge): Pass
5492 argument to reset_out_edges_aux function.
5493 (switch_decision_tree::analyze_switch_statement): Likewise.
5494 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
5497 2018-08-27 Martin Liska <mliska@suse.cz>
5499 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
5500 cfun argument explicitly.
5501 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
5502 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
5503 function gimple_switch_default_bb.
5504 (convert_switch_statements):
5506 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
5507 * stmt.c (label_to_block_fn): Use label_to_block and pass
5508 cfun argument explicitly and use gimple_switch_label_bb.
5509 (expand_case): Likewise.
5510 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
5511 cfun argument explicitly. Likewise.
5512 (make_edges_bb): Likewise.
5513 (make_cond_expr_edges): Likewise.
5514 (get_cases_for_edge): Likewise.
5515 (make_gimple_switch_edges): Likewise.
5516 (label_to_block_fn): Likewise.
5517 (label_to_block): Likewise.
5518 (make_goto_expr_edges): Likewise.
5519 (make_gimple_asm_edges): Likewise.
5520 (main_block_label): Likewise.
5521 (group_case_labels_stmt): Likewise.
5522 (find_taken_edge_computed_goto): Likewise.
5523 (find_taken_edge_switch_expr): Likewise.
5524 (gimple_verify_flow_info): Likewise.
5525 (gimple_redirect_edge_and_branch): Likewise.
5526 (gimple_switch_label_bb): New function.
5527 (gimple_switch_default_bb): Likewise.
5528 (gimple_switch_edge): Likewise.
5529 (gimple_switch_default_edge): Likewise.
5530 * tree-cfg.h (label_to_block_fn): Remove and replace ...
5531 (label_to_block): ... with this.
5532 (gimple_switch_label_bb): New.
5533 (gimple_switch_default_bb): Likewise.
5534 (gimple_switch_edge): Likewise.
5535 (gimple_switch_default_edge): Likewise.
5536 * tree-cfgcleanup.c (convert_single_case_switch): Use
5537 new gimple functions and pass new argument to label_to_block.
5538 (cleanup_control_flow_bb):
5539 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
5540 cfun argument explicitly.
5541 (make_eh_edges): Likewise.
5542 (redirect_eh_dispatch_edge): Likewise.
5543 (lower_resx): Likewise.
5544 (lower_eh_dispatch): Likewise.
5545 (maybe_remove_unreachable_handlers): Likewise.
5546 (unsplit_eh): Likewise.
5547 (cleanup_empty_eh): Likewise.
5548 (verify_eh_edges): Likewise.
5549 (verify_eh_dispatch_edge): Likewise.
5550 * tree-ssa-dom.c (record_edge_info): Likewise.
5551 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
5552 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
5553 (thread_through_normal_block): Likewise.
5554 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
5555 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
5556 * tree-switch-conversion.c (switch_conversion::collect): Use new
5558 (switch_conversion::check_final_bb): Likewise.
5559 (switch_conversion::gather_default_values): Pass new argument
5561 (switch_conversion::build_constructors): Likewise.
5562 (switch_decision_tree::compute_cases_per_edge): Use new
5563 gimple_switch_edge function.
5564 (switch_decision_tree::analyze_switch_statement): Pass new argument
5566 (switch_decision_tree::try_switch_expansion): Use
5567 gimple_switch_default_edge.
5568 * tree-vrp.c (find_switch_asserts): Pass new argument
5570 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
5571 (vr_values::simplify_switch_using_ranges): Likewise.
5573 2018-08-27 Richard Biener <rguenther@suse.de>
5575 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
5576 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
5578 * tree-ssa-sccvn.h (struct vn_pval): New structure.
5579 (struct vn_nary_op_s): Add unwind_to member. Add
5580 predicated_values flag and put result into a union together
5581 with a linked list of vn_pval.
5582 (struct vn_ssa_aux): Add name member to make maintaining
5583 a map of SSA name to vn_ssa_aux possible. Remove no longer
5584 needed info, dfsnum, low, visited, on_sccstack, use_processed
5585 and range_info_anti_range_p members.
5586 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
5587 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
5589 (vn_valueize): New global.
5590 (vn_context_bb): Likewise.
5591 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
5592 VN_INFO_PTR_INFO): Remove.
5593 * tree-ssa-sccvn.c: ... (rewrite)
5594 (pass_fre::execute): For -O2+ initialize loops and run
5595 RPO VN in optimistic mode (iterating). For -O1 and -Og
5596 run RPO VN in non-optimistic mode.
5597 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
5598 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
5599 * doc/invoke.texi (sccvn-max-scc-size): Remove.
5600 (rpo-vn-max-loop-depth): Document.
5601 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
5602 when valuezing the VUSE signals we walked out of the region.
5603 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
5604 (phi_translate): Set VN context block to use for availability
5606 (compute_avail): Likewise.
5607 (pre_valueize): New function.
5608 (pass_pre::execute): Adjust to the RPO VN API.
5610 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
5611 (propagate_constants_for_unrolling): Remove.
5612 (tree_unroll_loops_completely): Perform value-numbering
5613 on the unrolled bodies loop parent.
5615 2018-08-27 Richard Biener <rguenther@suse.de>
5617 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
5618 for partial antic compute.
5620 2018-08-27 Jakub Jelinek <jakub@redhat.com>
5622 PR rtl-optimization/87065
5623 * combine.c (simplify_if_then_else): Formatting fix.
5624 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
5626 (known_cond): Don't return const_true_rtx for vector modes. Use
5627 CONST0_RTX instead of const0_rtx. Formatting fixes.
5629 2018-08-27 Martin Liska <mliska@suse.cz>
5631 PR gcov-profile/87069
5632 * gcov.c (process_file): Record files already processed
5633 and warn about a file being processed multiple times.
5635 2018-08-27 Martin Liska <mliska@suse.cz>
5638 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5639 Set default values for x_aarch64_*_string strings.
5640 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
5641 prefix. For -mabi do not print '=ABI' in help and use
5642 <option_value> format for -msve-vector-bits and -moverride
5645 2018-08-26 Jeff Law <law@redhat.com>
5647 * config/mips/frame-header-opt.c: Include "backend.h" rather than
5650 2018-08-26 Marek Polacek <polacek@redhat.com>
5652 PR c++/87029, Implement -Wredundant-move.
5653 * doc/invoke.texi: Document -Wredundant-move.
5655 2018-08-25 Martin Sebor <msebor@redhat.com>
5657 PR tree-optimization/87059
5658 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
5659 to the same type as the other.
5660 * fold-const.c (fold_binary_loc): Assert expectation.
5662 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
5664 * config/darwin.c (machopic_legitimize_pic_address): Clean up
5665 extraneous parentheses, dead code section and formatting.
5667 2018-08-24 David Malcolm <dmalcolm@redhat.com>
5670 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
5671 wide enough for jumps in the line-numbering to be visible.
5672 (layout::print_gap_in_line_numbering): New member function.
5673 (layout::calculate_line_spans): When using line numbering, merge
5674 line spans that are only 1 line apart.
5675 (diagnostic_show_locus): When printing line numbers, show gaps in
5676 line numbering directly, rather than printing headers.
5677 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
5678 line-numbering with multiple line spans.
5679 (selftest::test_fixit_insert_containing_newline_2): Add test of
5680 line-numbering, in which the spans are close enough to be merged.
5682 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
5684 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
5685 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
5686 * tree-vrp.c (range_is_nonnull): Remove.
5687 (range_includes_zero_p): Accept value_range instead of min/max.
5688 (extract_range_from_binary_expr_1): Do not early bail on
5690 Use range_includes_zero_p instead of range_is_nonnull.
5691 (extract_range_from_unary_expr): Use range_includes_zero_p instead
5692 of range_is_nonnull.
5693 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
5694 special case VR_ANTI_RANGE.
5695 (vrp_finalize): Same.
5696 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
5698 (range_is_nonnull): Remove.
5699 * vr-values.c (vrp_stmt_computes_nonzero): Use
5700 range_includes_zero_p instead of range_is_nonnull.
5701 (extract_range_basic): Pass value_range to range_includes_zero_p
5702 instead of range_is_nonnull.
5704 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
5706 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
5707 * emit-rtl.h (rtl_data): Remove return_bnd.
5708 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
5709 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
5710 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
5711 (POINTER_BOUNDS_MODE): Remove definition.
5712 (make_pointer_bounds_mode): Remove.
5713 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
5714 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
5715 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
5716 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
5717 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
5718 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
5719 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
5721 * config/i386/i386-modes.def (BND32, BND64): Remove.
5722 * config/i386/i386.c (dbx_register_map): Remove bound registers.
5723 (dbx64_register_map): Ditto.
5724 (svr4_dbx_register_map): Ditto.
5725 (indirect_thunk_bnd_needed): Remove.
5726 (indirect_thunks_bnd_used): Ditto.
5727 (indirect_return_bnd_needed): Ditto.
5728 (indirect_return_via_cx_bnd): Ditto.
5729 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
5730 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
5731 (output_indirect_thunk): Ditto. Remove need_prefix argument.
5732 (output_indirect_thunk_function): Remove handling of
5733 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
5734 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
5735 (ix86_save_reg): Remove handling of crtl->return_bnd.
5736 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
5737 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
5738 and UNSPEC_BNDLX_ADDR.
5739 (ix86_output_indirect_branch_via_reg): Remove handling of
5740 indirect_thunk_prefix_bnd.
5741 (ix86_output_indirect_branch_via_push): Ditto.
5742 (ix86_output_function_return): Ditto.
5743 (ix86_output_indirect_function_return): Ditto.
5744 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
5745 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
5746 (CALL_USED_REGISTERS): Ditto.
5747 (REG_ALLOC_ORDER): Update for removal of bound registers.
5748 (HI_REGISTER_NAMES): Ditto.
5749 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
5750 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
5751 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
5752 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
5753 (FIRST_PSEUDO_REG): Update.
5754 (BND): Remove mode iterator.
5755 * config/i386/predicates.md (bnd_mem_operator): Remove.
5757 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
5759 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
5762 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
5764 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
5765 the case in which the permute needs only a single element and
5766 repeats for every vector of the result. Extend that case to
5767 handle variable-length vectors.
5768 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
5770 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
5773 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
5774 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
5776 2018-08-24 Richard Biener <rguenther@suse.de>
5778 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
5779 bb_flags_allocated members.
5780 (auto_flag): New RAII class for allocating flags.
5781 (auto_edge_flag): New RAII class for allocating edge flags.
5782 (auto_bb_flag): New RAII class for allocating bb flags.
5783 * cfgloop.c (verify_loop_structure): Allocate temporary edge
5785 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
5786 in favor of temporarily allocated BB flag.
5787 * hsa-brig.c: Re-order includes.
5788 * hsa-dump.c: Likewise.
5789 * hsa-regalloc.c: Likewise.
5790 * print-rtl.c: Likewise.
5791 * profile-count.c: Likewise.
5793 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
5796 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
5799 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
5802 * wide-int-range.cc (wide_int_range_div): Do not ignore result
5803 from wide_int_range_multiplicative_op.
5805 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5807 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
5810 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
5812 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
5815 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
5817 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
5818 full GENERIC expression used for replacement.
5820 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
5822 * tree-vrp.c (abs_extent_range): Remove.
5823 (extract_range_into_wide_ints): Pass wide ints by reference.
5824 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
5825 Pass wide ints by reference in all calls to
5826 extract_range_into_wide_ints.
5827 * wide-int-range.cc (wide_int_range_div): New.
5828 * wide-int-range.h (wide_int_range_div): New.
5829 (wide_int_range_includes_zero_p): New.
5830 (wide_int_range_zero_p): New.
5832 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
5834 * config/aarch64/aarch64.md (arches): New enum.
5835 (arch): New enum attr.
5836 (arch_enabled): New attr.
5837 (enabled): Now uses arch_enabled only.
5838 (simd, sve, fp16): Removed attribute.
5839 (fp): Attr now defined in terms of 'arch'.
5840 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
5841 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
5842 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
5843 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
5844 attributes into 'arch'.
5845 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
5846 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
5847 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
5848 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
5849 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
5850 'simd' attribute into 'arch'.
5851 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
5852 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
5853 Convert use of 'fp' attribute to 'arch'.
5854 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
5855 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
5857 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
5858 (different modes) Merge 'fp' and 'simd' into 'arch'.
5859 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
5862 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
5864 PR rtl-optimization/87026
5865 * expmed.c (canonicalize_comparison): If we can no longer create
5866 pseudoregisters, don't.
5868 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
5871 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
5873 * config/arm/arm.c (speculation_barrier_libfunc): New static
5875 (arm_init_libfuncs): Initialize it.
5876 (arm_emit_speculation_barrier): New function.
5877 * config/arm/arm.md (speculation_barrier): Call
5878 arm_emit_speculation_barrier for architectures that do not have
5880 (speculation_barrier_insn): Only match on Armv7 or later.
5882 2018-08-23 Richard Biener <rguenther@suse.de>
5885 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
5888 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
5890 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
5891 of single-vector TBLs.
5892 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
5895 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
5898 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
5899 aarch64_evpc_tbl guard.
5901 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
5903 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
5906 2018-08-22 Martin Sebor <msebor@redhat.com>
5909 * tree-pretty-print.c (pretty_print_string): Add argument.
5910 (dump_generic_node): Call to pretty_print_string with string size.
5912 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
5914 PR rtl-optimization/86771
5915 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
5916 of two SETs into those two SETs, one to be placed at i2, if that SETs
5917 destination is modified between i2 and i3.
5919 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
5921 PR tree-optimization/86725
5922 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
5924 (vect_analyze_scalar_cycles_1): Check it.
5926 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
5928 PR tree-optimization/86725
5929 * tree-vect-loop.c (vect_is_simple_reduction): When treating
5930 an outer loop phi as a double reduction, make sure that the
5931 single user of the phi result is an inner loop phi.
5933 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
5935 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
5936 grouped stores with gaps to a strided group.
5938 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
5940 * tree-vect-stmts.c (get_group_load_store_type)
5941 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
5942 first statement in a group.
5944 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5946 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
5947 the sequence used in gcc/gcc.c.
5949 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5952 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
5955 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5957 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
5958 Darwin10-specific unwinder-shim.
5959 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
5960 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
5961 New to cater for Darwin10 Rosetta.
5963 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5965 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
5966 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
5968 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
5973 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
5974 (dwarf2out_switch_text_section): Generate a local label for the
5975 second function sub-section and apply it as the second FDE start
5977 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
5978 second sub-section start.
5980 2018-08-22 Richard Biener <rguenther@suse.de>
5982 PR tree-optimization/86988
5983 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
5985 2018-08-22 Richard Biener <rguenther@suse.de>
5987 PR tree-optimization/86945
5988 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
5990 2018-08-22 Alexandre Oliva <oliva@adacore.com>
5992 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
5993 a comment about how uses of r2 for .sdata2 come about.
5995 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
5997 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
5999 2018-08-21 Marek Polacek <polacek@redhat.com>
6001 PR c++/86981, Implement -Wpessimizing-move.
6002 * doc/invoke.texi: Document -Wpessimizing-move.
6004 2018-08-21 Jan Hubicka <jh@suse.cz>
6006 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
6007 * tree.h (is_redundant_typedef): Remove.
6008 * dwarf2out.c (is_redundant_typedef): Turn into static function.
6010 2018-08-21 Jan Hubicka <jh@suse.cz>
6012 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
6015 2018-08-21 Tamar Christina <tamar.christina@arm.com>
6017 * expmed.c (extract_low_bits): Reject invalid subregs early.
6019 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
6022 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
6025 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6027 * config/vxworks.h: Guard vxworks_asm_out_constructor and
6028 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
6029 * config/vxworks.c: Likewise.
6031 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6033 * config/vxworks.c: Set targetm.have_ctors_dtors
6034 if HAVE_INITFINI_ARRAY_SUPPORT.
6035 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
6036 if HAVE_INITFINI_ARRAY_SUPPORT.
6038 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6040 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
6041 default search path for VxWorks < 7.
6043 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
6045 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
6046 (get_format_string): Refer to c_getstr.
6048 2018-08-21 Tom de Vries <tdevries@suse.de>
6050 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
6051 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
6052 (debuginfo_early_stop): Declare.
6053 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
6054 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
6055 (debuginfo_early_stop): New function.
6056 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
6057 and debuginfo_early_stop.
6058 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
6059 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
6060 (general_init): Call debuginfo_early_init.
6061 (finalize): Call debuginfo_fini.
6062 (do_compile): Call debuginfo_init.
6063 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
6065 (@item -fdump-debug, @item -fdump-earlydebug): Add.
6067 2018-08-21 Tom de Vries <tdevries@suse.de>
6069 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
6070 flag_dump_noaddr and flag_dump_unnumbered.
6072 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
6074 * wide-int-range.cc (wide_int_range_abs): New.
6075 (wide_int_range_order_set): Rename from wide_int_range_min_max.
6076 * wide-int-range.h (wide_int_range_abs): New.
6077 (wide_int_range_min_max): New.
6078 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
6079 case to call wide_int_range_abs.
6080 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
6081 (extract_range_from_abs_expr): Delete.
6083 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
6086 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
6087 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
6090 2018-08-20 Nathan Sidwell <nathan@acm.org>
6091 Jeff Law <law@redhat.com>
6093 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
6094 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
6096 2018-08-20 David Malcolm <dmalcolm@redhat.com>
6099 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
6100 (decl_attributes): Likewise.
6101 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
6103 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
6104 * diagnostic-core.h (class auto_diagnostic_group): New class.
6105 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
6106 (diagnostic_report_diagnostic): Handle the first diagnostics within
6108 (emit_diagnostic): Add auto_diagnostic_group instance.
6110 (inform_n): Likewise.
6111 (warning): Likewise.
6112 (warning_at): Likewise.
6113 (warning_n): Likewise.
6114 (pedwarn): Likewise.
6115 (permerror): Likewise.
6117 (error_n): Likewise.
6118 (error_at): Likewise.
6120 (fatal_error): Likewise.
6121 (internal_error): Likewise.
6122 (internal_error_no_backtrace): Likewise.
6123 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
6124 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
6125 * diagnostic.h (struct diagnostic_context): Add fields
6126 "diagnostic_group_nesting_depth",
6127 "diagnostic_group_emission_count", "begin_group_cb",
6129 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
6130 Add auto_diagnostic_group instance(s).
6131 (find_explicit_erroneous_behavior): Likewise.
6132 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
6133 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
6134 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
6135 (gimplify_va_arg_expr): Likewise.
6136 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
6137 (HSA_SORRY_AT): Likewise.
6138 * ipa-devirt.c (compare_virtual_tables): Likewise.
6139 (warn_odr): Likewise.
6140 * multiple_target.c (expand_target_clones): Likewise.
6141 * opts-common.c (cmdline_handle_error): Likewise.
6142 * reginfo.c (globalize_reg): Likewise.
6143 * substring-locations.c (format_warning_n_va): Likewise.
6144 * tree-inline.c (expand_call_inline): Likewise.
6145 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
6146 * tree-ssa-loop-niter.c
6147 (do_warn_aggressive_loop_optimizations): Likewise.
6148 * tree-ssa-uninit.c (warn_uninit): Likewise.
6149 * tree.c (warn_deprecated_use): Likewise.
6151 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
6154 * config/i386/i386.md (eh_return): Always update EH return
6155 address in word_mode.
6157 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
6159 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
6160 TARGET_SPLIT_COMPLEX_ARG is defined.
6162 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
6164 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
6166 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
6169 * expr.c (expand_assignment): Assert that bitpos is positive.
6170 (store_field): Likewise
6171 (expand_expr_real_1): Make sure that bitpos is positive.
6172 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
6175 2018-08-20 Nathan Sidwell <nathan@acm.org>
6177 * Makefile.in (CPP_ID_DATA_H): Delete.
6178 (CPP_INTERNAL_H): Don't add it.
6179 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
6180 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
6182 2018-08-20 Richard Biener <rguenther@suse.de>
6184 PR tree-optimization/78655
6185 * tree-vrp.c (extract_range_from_binary_expr_1): Make
6186 pointer + offset nonnull if either operand is nonnull work.
6188 2018-08-20 Tom de Vries <tdevries@suse.de>
6190 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
6191 unless the referenced die describes the added property using
6192 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
6193 Otherwise, add a DW_AT_location to the referenced die.
6195 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
6198 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
6199 register_operand when calling ix86_set_reg_reg_cost.
6200 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
6201 Set *total to 0 for operands that satisfy x86_64_immediate_operand
6202 predicate and to 1 otherwise.
6204 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
6206 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
6207 emit a diagnostic that it is not supported and reset the option.
6208 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
6209 supported and consume the option. (ASM_FINAL_SPEC): New.
6211 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
6213 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
6216 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
6218 C-SKY port: Documentation
6220 * doc/extend.texi (C-SKY Function Attributes): New section.
6221 * doc/invoke.texi (Option Summary): Add C-SKY options.
6222 (C-SKY Options): New section.
6223 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
6225 2018-08-17 Jojo <jijie_rong@c-sky.com>
6226 Huibin Wang <huibin_wang@c-sky.com>
6227 Sandra Loosemore <sandra@codesourcery.com>
6228 Chung-Lin Tang <cltang@codesourcery.com>
6230 C-SKY port: Backend implementation
6232 * config/csky/*: New.
6233 * common/config/csky/*: New.
6235 2018-08-17 Jojo <jijie_rong@c-sky.com>
6236 Huibin Wang <huibin_wang@c-sky.com>
6237 Sandra Loosemore <sandra@codesourcery.com>
6238 Chung-Lin Tang <cltang@codesourcery.com>
6239 Andrew Jenner <andrew@codesourcery.com>
6241 C-SKY port: Configury
6243 * config.gcc (csky-*-*): New.
6244 * configure.ac: Add csky to targets for dwarf2 debug_line support.
6245 * configure: Regenerated.
6247 2018-08-17 David Malcolm <dmalcolm@redhat.com>
6249 * dump-context.h: Include "dumpfile.h".
6250 (dump_context::dump_printf_va): Convert final param from va_list
6251 to va_list *. Convert from ATTRIBUTE_PRINTF to
6252 ATTRIBUTE_GCC_DUMP_PRINTF.
6253 (dump_context::dump_printf_loc_va): Likewise.
6254 * dumpfile.c: Include "stringpool.h".
6255 (make_item_for_dump_printf_va): Delete.
6256 (make_item_for_dump_printf): Delete.
6257 (class dump_pretty_printer): New class.
6258 (dump_pretty_printer::dump_pretty_printer): New ctor.
6259 (dump_pretty_printer::emit_items): New member function.
6260 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
6262 (dump_pretty_printer::emit_item): New member function.
6263 (dump_pretty_printer::stash_item): New member function.
6264 (dump_pretty_printer::format_decoder_cb): New member function.
6265 (dump_pretty_printer::decode_format): New member function.
6266 (dump_context::dump_printf_va): Reimplement in terms of
6267 dump_pretty_printer.
6268 (dump_context::dump_printf_loc_va): Convert final param from va_list
6270 (dump_context::begin_scope): Reimplement call to
6271 make_item_for_dump_printf.
6272 (dump_printf): Update for change to dump_printf_va.
6273 (dump_printf_loc): Likewise.
6274 (selftest::test_capture_of_dump_calls): Convert "stmt" from
6275 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
6276 with %T, %E, and %G.
6277 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
6278 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
6279 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
6280 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
6281 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
6282 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
6283 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
6284 within a dump_printf_loc call to "%wu".
6285 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
6286 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
6287 missing space after "=".
6288 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
6289 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
6290 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
6291 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
6292 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
6293 duplicate "vectorized" from message.
6295 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
6297 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
6298 polyNxK_t element's TYPE_STRING_FLAG.
6300 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
6302 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
6303 (they were unnamed before). Fix comments.
6305 2018-08-17 Nathan Sidwell <nathan@acm.org>
6307 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
6309 2018-08-17 Richard Biener <rguenther@suse.de>
6311 PR tree-optimization/86841
6312 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
6314 2018-08-17 Martin Liska <mliska@suse.cz>
6316 * common.opt: Remove Warn, Init and Report for options with
6317 Ignore/Deprecated flag. Warning is done automatically for
6319 * config/i386/i386.opt: Likewise.
6320 * config/ia64/ia64.opt: Likewise.
6321 * config/rs6000/rs6000.opt: Likewise.
6322 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
6323 Remove usage of flag_check_pointer_bounds.
6324 * lto-wrapper.c (merge_and_complain): Do not handle
6325 OPT_fcheck_pointer_bounds.
6326 (append_compiler_options): Likewise.
6327 * opt-functions.awk: Do not handle Deprecated.
6328 * optc-gen.awk: Check that Var, Report and Init are not
6329 used for an option with Ignore/Deprecated flag.
6330 * opts-common.c (decode_cmdline_option): Do not report
6332 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
6334 * opts.h (struct cl_option): Remove cl_deprecated flag.
6335 (CL_ERR_DEPRECATED): Remove error enum value.
6337 2018-08-17 Richard Biener <rguenther@suse.de>
6340 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
6341 across a va-arg-pack using call adjust its return value accordingly.
6343 2018-08-16 Martin Sebor <msebor@redhat.com>
6345 PR tree-optimization/86853
6346 * gimple-ssa-sprintf.c (struct format_result): Rename member.
6347 (struct fmtresult): Add member and initialize it in ctors.
6348 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
6349 (format_string): Handle %S the same as %ls. Set MAYFAIL.
6350 (format_directive): Set POSUNDER4K when MAYFAIL is set.
6351 (parse_directive): Handle %C same as %c.
6352 (sprintf_dom_walker::compute_format_length): Adjust.
6353 (is_call_safe): Adjust.
6355 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
6357 * builtins.c (c_strlen): Add new parameter eltsize. Use it
6358 for determining how to count the elements.
6359 * builtins.h (c_strlen): Adjust prototype.
6360 * expr.c (string_constant): Add new parameter mem_size.
6361 Set *mem_size appropriately.
6362 * expr.h (string_constant): Adjust protoype.
6363 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
6364 * gimple-fold.h (get_range_strlen): Adjust prototype.
6365 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
6366 (format_string): Call get_string_length with eltsize.
6368 2018-08-16 David Malcolm <dmalcolm@redhat.com>
6370 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
6371 to emit the span, rather than setting it as the prefix.
6373 2018-08-16 David Malcolm <dmalcolm@redhat.com>
6375 * diagnostic-show-locus.c (layout::start_annotation_line): Add
6376 "margin_char" parameter, defaulting to space. Use it in place
6377 of pp_space for the initial part of the margin.
6378 (layout::print_leading_fixits): Use '+' when filling the margin
6379 of line-insertion fix-it hints.
6381 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6383 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
6386 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6388 * config/rs6000/altivec.md: Don't set length attribute to the default
6390 * config/rs6000/darwin.md: Ditto.
6391 * config/rs6000/dfp.md: Ditto.
6392 * config/rs6000/htm.md: Ditto.
6393 * config/rs6000/rs6000.md: Ditto.
6394 * config/rs6000/sync.md: Ditto.
6395 * config/rs6000/vsx.md: Ditto.
6397 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6399 * config/rs6000/altivec.md: Don't set length attribute to the default
6400 value, for branch instructions.
6401 * config/rs6000/darwin.md: Ditto.
6402 * config/rs6000/rs6000.md: Ditto.
6404 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6406 * config/rs6000/rs6000.md (length): Always define as const_int 4.
6407 (unnamed conditional branch define_insn): Set length to 4 or 8
6408 depending on offset.
6409 (<bd>_<mode>): Similar, for alternative 0.
6410 (<bd>tf_<mode>): Ditto.
6412 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6414 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
6416 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
6418 * doc/rtl.texi: Replace old RTX class names with new names.
6421 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
6423 * expmed.h (canonicalize_comparison): New declaration.
6424 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
6425 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
6426 * optabs.c (prepare_cmp_insn): Likewise.
6427 * rtl.h (unsigned_condition_p): New function which checks if a
6428 comparison operator is unsigned.
6430 2018-08-16 Nathan Sidwell <nathan@acm.org>
6432 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
6433 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
6435 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6438 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
6439 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
6440 (mov<mov>): ..this and enable unconditionally.
6442 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6444 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
6446 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
6448 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
6449 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
6450 "Common" with "Target".
6452 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
6454 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
6455 * doc/invoke.texi (mmitigate-rop): Remove.
6456 * config/i386/i386.c: Do not include "regrename.h".
6457 (ix86_rop_should_change_byte_p, reg_encoded_number)
6458 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
6460 (ix86_reorg): Remove call to ix86_mitigate_rop.
6461 * config/i386/i386.md (attr "modrm_class"): Remove.
6462 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
6463 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
6464 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
6466 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6468 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
6469 allow folding of mergeh() and mergel() for the float and double types.
6470 (fold_mergehl_helper): Rework to handle building a permute tree
6473 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
6475 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
6478 2018-08-15 David Malcolm <dmalcolm@redhat.com>
6480 * common.opt (fdiagnostics-show-labels): New option.
6481 * diagnostic-show-locus.c (class layout_range): Add field
6483 (class layout): Add field "m_show_labels_p".
6484 (layout_range::layout_range): Add param "label" and use it to
6486 (make_range): Pass in NULL for new "label" param of layout_range's
6488 (layout::layout): Initialize m_show_labels_p.
6489 (layout::maybe_add_location_range): Pass in loc_range->m_label
6490 when constructing layout_range instances.
6491 (struct line_label): New struct.
6492 (layout::print_any_labels): New member function.
6493 (layout::print_line): Call it if label-printing is enabled.
6494 (selftest::test_one_liner_labels): New test.
6495 (selftest::test_diagnostic_show_locus_one_liner): Call it.
6496 * diagnostic.c (diagnostic_initialize): Initialize
6497 context->show_labels_p.
6498 * diagnostic.h (struct diagnostic_context): Add field
6500 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
6501 -fno-diagnostics-show-labels.
6502 * dwarf2out.c (gen_producer_string): Add
6503 OPT_fdiagnostics_show_labels to the ignored options.
6504 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
6506 (gcc_rich_location::maybe_add_expr): Likewise.
6507 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
6508 label" param, defaulting to NULL.
6509 (gcc_rich_location::add_expr): Add "label" param.
6510 (gcc_rich_location::maybe_add_expr): Likewise.
6511 (class text_range_label): New class.
6512 (class range_label_for_type_mismatch): New class.
6513 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
6514 of format_warning_va.
6515 (fmtwarn_n): Likewise for new params of format_warning_n_va.
6516 * lto-wrapper.c (merge_and_complain): Add
6517 OPT_fdiagnostics_show_labels to the "pick one setting" options.
6518 (append_compiler_options): Likewise to the dropped options.
6519 (append_diag_options): Likewise to the passed-on options.
6520 * opts.c (common_handle_option): Handle the new option.
6521 * selftest-diagnostic.c
6522 (test_diagnostic_context::test_diagnostic_context): Enable
6524 * substring-locations.c: Include "gcc-rich-location.h".
6525 (format_warning_n_va): Add "fmt_label" and "param_label" params
6526 and use them as appropriate.
6527 (format_warning_va): Add "fmt_label" and "param_label" params,
6528 passing them on to format_warning_n_va.
6529 (format_warning_at_substring): Likewise.
6530 (format_warning_at_substring_n): Likewise.
6531 * substring-locations.h (format_warning_va): Add "fmt_label" and
6532 "param_label" params.
6533 (format_warning_n_va): Likewise.
6534 (format_warning_at_substring): Likewise.
6535 (format_warning_at_substring_n): Likewise.
6536 * toplev.c (general_init): Initialize global_dc->show_labels_p.
6538 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
6541 * builtins.c (expand_builtin_memcmp): Do not expand the call
6542 when overflow is detected.
6544 2018-08-15 Martin Sebor <msebor@redhat.com>
6546 PR tree-optimization/71625
6547 * config/aarch64/aarch64-builtins.c
6548 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
6550 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
6552 * config/s390/s390.c (s390_reorg): Remove loop.
6554 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
6557 (darwin_function_switched_text_sections): Delete.
6558 * gcc/config/darwin.h
6559 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
6561 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
6564 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
6565 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
6566 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
6568 2018-08-15 Martin Liska <mliska@suse.cz>
6570 PR tree-optimization/86925
6571 * predict.c (expr_expected_value_1): When taking
6572 later predictor, assign also probability.
6573 Use fold_build2_initializer_loc in order to fold
6574 the expression in -frounding-math.
6576 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
6578 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
6580 (expand_vec_perm_1): Try the new method.
6582 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
6585 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
6586 Check whether lra_live_max_point is 0 before dividing.
6588 2018-08-14 Martin Sebor <msebor@redhat.com>
6590 PR tree-optimization/86650
6591 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
6592 (vrp_prop::check_mem_ref): Same.
6594 2018-08-13 Liu Hao <lh_mouse@126.com>
6596 * pretty-print.c (eat_esc_sequence): Swap the foreground and
6597 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
6598 and clear it thereafter, as it only works for DBCS.
6600 2018-08-13 Liu Hao <lh_mouse@126.com>
6602 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
6603 handle returned by _get_osf_handle().
6605 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6607 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
6608 for folding vec_perm.
6610 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6612 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
6613 Add support for gimple-folding of vec_pack() and vec_unpack()
6616 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6618 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
6619 vec_xst variants to the list.
6620 (rs6000_gimple_fold_builtin): Add support for folding unaligned
6621 vector loads and stores.
6623 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
6625 * config.gcc (rs6000-ibm-aix4.x): Delete.
6626 (rs6000-ibm-aix5.1): Delete.
6627 (rs6000-ibm-aix5.2): Delete.
6628 (rs6000-ibm-aix5.3): Delete.
6629 * config/rs6000/aix43.h: Delete.
6630 * config/rs6000/aix51.h: Delete.
6631 * config/rs6000/aix52.h: Delete.
6632 * config/rs6000/t-aix43: Delete.
6634 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
6636 * config/s390/s390.c (s390_decompose_constant_pool_ref):
6638 (s390_decompose_address): Factor out constant pool ref
6641 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6643 * config/nds32/nds32-predicates.c
6644 (nds32_can_use_bclr_p): Change return type as bool.
6645 (nds32_can_use_bset_p): Ditto.
6646 (nds32_can_use_btgl_p): Ditto.
6647 (nds32_can_use_bitci_p): Ditto.
6648 * config/nds32/nds32-protos.h
6649 (nds32_can_use_bclr_p): Change declaration.
6650 (nds32_can_use_bset_p): Ditto.
6651 (nds32_can_use_btgl_p): Ditto.
6652 (nds32_can_use_bitci_p): Ditto.
6654 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6656 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
6657 Support -msched-prolog-epilog option.
6658 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
6660 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6662 * common/config/nds32/nds32-common.c
6663 (nds32_option_optimization_table): Enalbe -malways-align.
6665 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6667 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
6669 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
6670 OPT_misr_secure_ case.
6671 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
6672 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
6673 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
6675 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
6676 (nds32_isr_info): New field security_level.
6677 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
6678 * config/nds32/nds32.md (return_internal): Consider critical attribute.
6679 * config/nds32/nds32.opt (misr-secure): New option.
6680 * config/nds32/nds32_init.inc: New file.
6681 * config/nds32/nds32_isr.h: New file.
6683 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
6685 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
6686 Update comment for atomic instructions.
6687 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
6688 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
6690 (atomic_loaddi): Revise fence expansion to only emit fence prior to
6691 load for __ATOMIC_SEQ_CST model.
6692 (atomic_loaddi_1): Remove float register target.
6693 (atomic_storedi): Handle CONST_INT values.
6694 (atomic_storedi_1): Remove float register source. Add special case
6696 (memory_barrier): New expander and insn.
6698 2018-08-11 Jakub Jelinek <jakub@redhat.com>
6700 PR tree-optimization/86835
6701 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
6702 new_stmt after def_gsi, make sure to insert new_square_stmt after
6703 that stmt, not 2 stmts before it.
6705 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
6708 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
6711 2018-08-10 Martin Liska <mliska@suse.cz>
6714 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
6716 * builtins.c (expand_builtin_expect_with_probability):
6718 (expand_builtin_expect_with_probability): New function.
6719 (build_builtin_expect_predicate): Add new argumnet probability
6720 for BUILT_IN_EXPECT_WITH_PROBABILITY.
6721 (fold_builtin_expect):
6724 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
6725 * builtins.h (fold_builtin_expect): Set new argument.
6726 * doc/extend.texi: Document __builtin_expect_with_probability.
6727 * doc/invoke.texi: Likewise.
6728 * gimple-fold.c (gimple_fold_call): Pass new argument.
6729 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
6730 also BUILT_IN_EXPECT_WITH_PROBABILITY.
6731 * predict.c (get_predictor_value): New function.
6732 (expr_expected_value): Add new argument probability. Assume
6733 that predictor and probability are always non-null.
6734 (expr_expected_value_1): Likewise. For __builtin_expect and
6735 __builtin_expect_with_probability set probability. Handle
6736 combination in binary expressions.
6737 (tree_predict_by_opcode): Simplify code by simply calling
6738 get_predictor_value.
6739 (pass_strip_predict_hints::execute): Add handling of
6740 BUILT_IN_EXPECT_WITH_PROBABILITY.
6741 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
6743 * tree.h (DECL_BUILT_IN_P): New function.
6745 2018-08-10 Martin Liska <mliska@suse.cz>
6747 PR tree-optimization/85799
6748 * passes.def: Add argument for pass_strip_predict_hints.
6749 * predict.c (class pass_strip_predict_hints): Add new argument
6751 (strip_predictor_early): New function.
6752 (pass_strip_predict_hints::execute): Call the function to
6754 (strip_predict_hints): New function.
6755 * predict.def: Fix comment.
6757 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
6759 * Makefile.in: Clarify which tm.texi to copy over to assert the
6760 right to grant a GFDL license for all.
6762 2018-08-09 Jeff Law <law@redhat.com>
6764 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
6767 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
6769 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
6772 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
6774 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
6775 reductions for variable-length vectors.
6777 2018-08-09 David Malcolm <dmalcolm@redhat.com>
6780 * common.opt (fdiagnostics-show-line-numbers): New option.
6781 * diagnostic-show-locus.c (class layout): Add fields
6782 "m_show_line_numbers_p" and "m_linenum_width";
6783 (num_digits): New function.
6784 (test_num_digits): New function.
6785 (layout::layout): Initialize new fields. Update m_x_offset
6786 logic to handle any left margin.
6787 (layout::print_source_line): Print line number when requested.
6788 (layout::start_annotation_line): New member function.
6789 (layout::print_annotation_line): Call it.
6790 (layout::print_leading_fixits): Likewise.
6791 (layout::print_trailing_fixits): Likewise. Update calls to
6792 move_to_column for new parameter.
6793 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
6794 it to potentially call start_annotation_line.
6795 (layout::show_ruler): Call start_annotation_line.
6796 (selftest::test_line_numbers_multiline_range): New selftest.
6797 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
6798 and selftest::test_line_numbers_multiline_range.
6799 * diagnostic.c (diagnostic_initialize): Initialize
6800 show_line_numbers_p.
6801 * diagnostic.h (struct diagnostic_context): Add field
6802 "show_line_numbers_p".
6803 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
6804 -fno-diagnostics-show-line-numbers.
6805 * dwarf2out.c (gen_producer_string): Add
6806 OPT_fdiagnostics_show_line_numbers to the ignored options.
6807 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
6808 one setting" options.
6809 (append_compiler_options): Likewise to the dropped options.
6810 (append_diag_options): Likewise to the passed-on options.
6811 * opts.c (common_handle_option): Handle the new option.
6812 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
6814 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
6816 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
6817 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
6818 third argument of type "const signed char" to descriptions of
6819 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
6820 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
6821 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
6822 __builtin_bcdsub_ov functions.
6824 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
6826 PR tree-optimization/86858
6827 * tree-vect-loop.c (vect_is_simple_reduction): Restore
6828 flow_bb_inside_loop_p calls.
6830 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
6832 PR tree-optimization/86871
6833 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
6834 instead of gimple_assign_lhs.
6836 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
6839 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
6840 register constraint to operand 0.
6841 (add<mode>3_carryinC): Likewise.
6842 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
6844 2018-08-09 Martin Liska <mliska@suse.cz>
6847 * common.opt: Remove extra line.
6849 2018-08-09 Martin Liska <mliska@suse.cz>
6851 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
6852 at the end of a line, make first letter capital and end up
6853 a sentence with a dot.
6854 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
6855 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
6856 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
6857 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
6858 (PARAM_MAX_ISL_OPERATIONS): Likewise.
6859 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
6860 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
6861 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
6862 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
6863 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
6864 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
6865 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
6866 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
6867 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
6868 (PARAM_TREE_REASSOC_WIDTH): Likewise.
6869 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
6870 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
6871 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
6873 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
6876 * config/s390/s390.c (s390_option_override_internal): Reduce the
6877 stack-clash-protection-probe-interval param if it would be too big
6880 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
6883 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
6884 * config/m68k/m68k.c (handle_move_double): Don't call
6885 m68k_final_prescan_insn.
6886 (m68k_adjust_decorated_operand): Renamed from
6887 m68k_final_prescan_insn, remove first and third operand and
6889 (print_operand): Call it.
6890 (print_operand_address): Call it.
6892 2018-08-08 Nathan Sidwell <nathan@acm.org>
6894 * diagnostic.c (diagnostic_report_current_module): Use
6895 linemap_included_from & linemap_included_from_linemap.
6897 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
6899 * config/aarch64/aarch64-cores.def: Add phecda core.
6900 * config/aarch64/aarch64-tune.md: Regenerate.
6901 * doc/invoke.texi: Add phecda core.
6903 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
6906 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
6908 * config/s390/s390.md ("movti"): Add more alternatives for
6909 constant to GPR copies.
6911 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
6913 * config/s390/s390.c: Fix whitespace damage throughout the file.
6914 * config/s390/s390.h: Likewise.
6915 * config/s390/tpf.h: Likewise.
6917 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
6919 * config/s390/s390.c (s390_loadrelative_operand_p):
6920 Remove TARGET_CPU_ZARCH usages.
6921 (s390_rtx_costs): Likewise.
6922 (s390_legitimate_constant_p): Likewise.
6923 (s390_cannot_force_const_mem): Likewise.
6924 (legitimate_reload_constant_p): Likewise.
6925 (s390_preferred_reload_class): Likewise.
6926 (legitimize_pic_address): Likewise.
6927 (legitimize_tls_address): Likewise.
6928 (s390_split_branches): Removed.
6929 (s390_add_execute): Removed.
6930 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
6931 (s390_mainpool_start): Likewise.
6932 (s390_mainpool_finish): Likewise.
6933 (s390_mainpool_cancel): Removed.
6934 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
6935 (s390_chunkify_cancel): Likewise.
6936 (s390_return_addr_rtx): Likewise.
6937 (s390_register_info): Remove split_branches_pending_p uages.
6938 (s390_optimize_register_info): Likewise.
6939 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
6940 split_branches_pending_p usages.
6941 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
6942 (s390_load_got): Likewise.
6943 (s390_expand_split_stack_prologue): Likewise.
6944 (output_asm_nops): Likewise.
6945 (s390_function_profiler): Likewise.
6946 (s390_emit_call): Likewise.
6947 (s390_conditional_register_usage): Likewise.
6948 (s390_optimize_prologue): Likewise.
6949 (s390_reorg): Remove TARGET_CPU_ZARCH and
6950 split_branches_pending_p usages.
6951 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
6953 (s390_output_indirect_thunk_function): Likewise.
6954 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
6955 (TARGET_CPU_ZARCH_P): Removed.
6956 (struct machine_function): Remove split_branches_pending_p.
6957 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
6959 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
6961 * common/config/s390/s390-common.c (processor_flags_table):
6963 * config.gcc: Remove with_arch/with_tune support.
6964 * config/s390/2064.md: Remove cpu attribute comparisons.
6965 * config/s390/driver-native.c (s390_host_detect_local_cpu):
6967 * config/s390/linux.h (ASM_SPEC):
6968 Remove -march support.
6969 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
6970 Use a table to get an arch level.
6971 * config/s390/s390-opts.h (enum processor_type):
6973 * config/s390/s390.c
6974 (processor_table): Remove entries, add arch_level values.
6975 (s390_issue_rate): Remove cases.
6976 (s390_option_override): Adjust
6977 s390_option_override_internal() call.
6978 (s390_option_override_internal): Remove deprecation warning.
6979 (s390_valid_target_attribute_tree): Adjust
6980 s390_option_override_internal() call.
6981 * config/s390/s390.h (struct s390_processor):
6982 Share with s390-c.c, add arch_level field.
6983 * config/s390/s390.md:
6984 Remove occurrences in cpu attribute.
6985 * config/s390/s390.opt: Remove -march/-mtune support.
6986 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
6987 * doc/invoke.texi: Remove deprecation warning.
6989 2018-08-08 Luis Machado <luis.machado@linaro.org>
6991 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
6993 (qdf24xx_tunings): Set vector cost structure to
6994 qdf24xx_vector_cost.
6996 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
6997 <register_sextend>: Set to 3.
6999 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
7002 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
7003 * config/aarch64/aarch64-simd.md
7004 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
7005 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
7006 (aarch64_frecpx<mode>): ...this new pattern.
7007 * config/aarch64/aarch64-simd-builtins.def: Remove comment
7008 about aarch64_frecp<FRECP:frecp_suffix><mode>.
7010 2018-08-07 Martin Liska <mliska@suse.cz>
7013 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
7014 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
7015 * predict.def (PRED_MALLOC_NONNULL): New predictor.
7016 * doc/extend.texi: Document that malloc attribute adds
7019 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
7022 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7023 Define to speculation_safe_value_not_needed.
7025 2018-08-06 Jeff Law <law@redhat.com>
7027 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
7028 the vr_values instance to cprop_into_stmt.
7029 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
7030 (cprop_operand): Also query EVRP to determine if OP is a constant.
7032 2018-08-06 Nathan Sidwell <nathan@acm.org>
7034 * diagnostic.c (diagnostic_report_current_module): Reroll
7035 included-at loop. Translate text.
7037 2018-08-06 David Malcolm <dmalcolm@redhat.com>
7039 * function-tests.c (selftest::test_expansion_to_rtl): Call
7040 free_after_compilation.
7042 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7044 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
7046 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
7048 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
7049 loops with memory block operations from getting unrolled.
7051 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
7054 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7055 Define to speculation_safe_value_not_needed.
7057 2018-08-06 Jeff Law <law@redhat.com>
7059 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
7062 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
7065 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
7066 with all enabled __intN types.
7068 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
7070 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7072 * alias.c (record_set): Check for clobber high.
7073 * cfgexpand.c (expand_gimple_stmt): Likewise.
7074 * combine-stack-adj.c (single_set_for_csa): Likewise.
7075 * combine.c (find_single_use_1): Likewise.
7076 (set_nonzero_bits_and_sign_copies): Likewise.
7077 (get_combine_src_dest): Likewise.
7078 (is_parallel_of_n_reg_sets): Likewise.
7079 (try_combine): Likewise.
7080 (record_dead_and_set_regs_1): Likewise.
7081 (reg_dead_at_p_1): Likewise.
7082 (reg_dead_at_p): Likewise.
7083 * dce.c (deletable_insn_p): Likewise.
7084 (mark_nonreg_stores_1): Likewise.
7085 (mark_nonreg_stores_2): Likewise.
7086 * df-scan.c (df_find_hard_reg_defs): Likewise.
7087 (df_uses_record): Likewise.
7088 (df_get_call_refs): Likewise.
7089 * dwarf2out.c (mem_loc_descriptor): Likewise.
7090 * haifa-sched.c (haifa_classify_rtx): Likewise.
7091 * ira-build.c (create_insn_allocnos): Likewise.
7092 * ira-costs.c (scan_one_insn): Likewise.
7093 * ira.c (equiv_init_movable_p): Likewise.
7094 (rtx_moveable_p): Likewise.
7095 (interesting_dest_for_shprep): Likewise.
7096 * jump.c (mark_jump_label_1): Likewise.
7097 * postreload-gcse.c (record_opr_changes): Likewise.
7098 * postreload.c (reload_cse_simplify): Likewise.
7099 (struct reg_use): Add source expr.
7100 (reload_combine): Check for clobber high.
7101 (reload_combine_note_use): Likewise.
7102 (reload_cse_move2add): Likewise.
7103 (move2add_note_store): Likewise.
7104 * print-rtl.c (print_pattern): Likewise.
7105 * recog.c (decode_asm_operands): Likewise.
7106 (store_data_bypass_p): Likewise.
7107 (if_test_bypass_p): Likewise.
7108 * regcprop.c (kill_clobbered_value): Likewise.
7109 (kill_set_value): Likewise.
7110 * reginfo.c (reg_scan_mark_refs): Likewise.
7111 * reload1.c (maybe_fix_stack_asms): Likewise.
7112 (eliminate_regs_1): Likewise.
7113 (elimination_effects): Likewise.
7114 (mark_not_eliminable): Likewise.
7115 (scan_paradoxical_subregs): Likewise.
7116 (forget_old_reloads_1): Likewise.
7117 * reorg.c (find_end_label): Likewise.
7118 (try_merge_delay_insns): Likewise.
7119 (redundant_insn): Likewise.
7120 (own_thread_p): Likewise.
7121 (fill_simple_delay_slots): Likewise.
7122 (fill_slots_from_thread): Likewise.
7123 (dbr_schedule): Likewise.
7124 * resource.c (update_live_status): Likewise.
7125 (mark_referenced_resources): Likewise.
7126 (mark_set_resources): Likewise.
7127 * rtl.c (copy_rtx): Likewise.
7128 * rtlanal.c (reg_referenced_p): Likewise.
7129 (single_set_2): Likewise.
7130 (noop_move_p): Likewise.
7131 (note_stores): Likewise.
7132 * sched-deps.c (sched_analyze_reg): Likewise.
7133 (sched_analyze_insn): Likewise.
7135 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7137 * cse.c (invalidate_reg): New function extracted from...
7138 (invalidate): ...here.
7139 (canonicalize_insn): Check for clobber high.
7140 (invalidate_from_clobbers): invalidate clobber highs.
7141 (invalidate_from_sets_and_clobbers): Likewise.
7142 (count_reg_usage): Check for clobber high.
7143 (insn_live_p): Likewise.
7144 * cselib.c (cselib_expand_value_rtx_1):Likewise.
7145 (cselib_invalidate_regno): Check for clobber in setter.
7146 (cselib_invalidate_rtx): Pass through setter.
7147 (cselib_invalidate_rtx_note_stores):
7148 (cselib_process_insn): Check for clobber high.
7149 * cselib.h (cselib_invalidate_rtx): Add operand.
7151 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7153 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
7154 (mark_not_eliminable): Likewise.
7155 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
7156 * lra-lives.c (process_bb_lives): Check for clobber high.
7157 * lra.c (new_insn_reg): Remember clobber highs.
7158 (collect_non_operand_hard_regs): Check for clobber high.
7159 (lra_set_insn_recog_data): Likewise.
7160 (add_regs_to_insn_regno_info): Likewise.
7161 (lra_update_insn_regno_info): Likewise.
7163 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7165 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
7166 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
7168 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7170 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
7171 (copy_insn_1): Likewise.
7172 (gen_hard_reg_clobber_high): New gen function.
7173 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
7174 * genemit.c (gen_exp): Likewise.
7175 (gen_emit_seq): Pass through info.
7176 (gen_insn): Check for CLOBBER_HIGH.
7177 (gen_expand): Pass through info.
7178 (gen_split): Likewise.
7179 (output_add_clobbers): Likewise.
7180 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
7181 (remove_clobbers): Likewise.
7182 * rtl.h (gen_hard_reg_clobber_high): New declaration.
7184 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7186 * doc/rtl.texi (clobber_high): Add.
7187 (parallel): Add in clobber high
7188 * rtl.c (rtl_check_failed_code3): Add function.
7189 * rtl.def (CLOBBER_HIGH): Add expression.
7190 * rtl.h (RTL_CHECKC3): Add macro.
7191 (rtl_check_failed_code3): Add declaration.
7192 (XC3EXP): Add macro.
7194 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
7197 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
7198 cfun->machine->max_used_stack_alignment if needed.
7200 2018-08-04 Martin Sebor <msebor@redhat.com>
7202 PR tree-optimization/86571
7203 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
7206 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
7209 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7212 2018-08-03 Jeff Law <law@redhat.com>
7215 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7216 Define to speculation_safe_value_not_needed.
7218 2018-08-03 David Malcolm <dmalcolm@redhat.com>
7220 * doc/gcov.texi (-x): Remove duplicate "to".
7221 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
7222 (-Wif-not-aligned): Remove duplicate "is".
7223 (-flto): Remove duplicate "the".
7224 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
7225 duplicate "v5.00.b".
7226 (MSP430 Options): Remove duplicate "and" from the description
7227 of "-mgprel-sec=regexp".
7228 (x86 Options): Remove duplicate copies of "vmldLog102" and
7229 vmlsLog104 from description of "-mveclibabi=type".
7231 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
7233 * internal-fn.h (first_commutative_argument): Declare.
7234 * internal-fn.c (first_commutative_argument): New function.
7235 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
7236 restrictions for pattern statements. Use first_commutative_argument
7237 to look for commutative operands in calls to internal functions.
7239 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
7241 * Makefile.in (wide-int-range.o): New.
7242 * tree-vrp.c: Move all the wide_int_* functions to...
7243 * wide-int-range.cc: ...here.
7244 * tree-vrp.h: Move all the wide_int_* prototypes to...
7245 * wide-int-range.h: ...here.
7247 2018-08-03 Tom de Vries <tdevries@suse.de>
7249 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
7251 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
7252 * except.c (output_function_exception_table): Do early exit if
7253 targetm_common.except_unwind_info (&global_options) == UI_NONE.
7255 2018-08-03 Martin Liska <mliska@suse.cz>
7257 * predict.c (dump_prediction): Change to 2 digits
7260 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
7262 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
7264 (falkor_am_1_gtov_gtov): ... a new insn reservation.
7266 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
7268 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
7269 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
7270 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
7271 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
7272 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
7274 2018-08-02 David Malcolm <dmalcolm@redhat.com>
7276 * diagnostic-show-locus.c (diagnostic_show_locus): Use
7277 pp_take_prefix when saving the existing prefix.
7278 * diagnostic.c (diagnostic_append_note): Likewise.
7279 * langhooks.c (lhd_print_error_function): Likewise.
7280 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
7281 param's type. Free the existing prefix.
7282 (pp_take_prefix): New function.
7283 (pretty_printer::pretty_printer): Drop the prefix parameter.
7284 Rename the length parameter to match the comment.
7285 (pretty_printer::~pretty_printer): Free the prefix.
7286 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
7288 (struct pretty_printer): Drop the "const" from "prefix" field's
7289 type and clarify memory management.
7290 (pp_set_prefix): Drop the "const" from the 2nd param.
7291 (pp_take_prefix): New decl.
7293 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
7295 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
7296 for word_mode_ok here instead of passing as argument.
7297 (expand_block_compare): Change select_block_compare_mode() call.
7298 (expand_strncmp_gpr_sequence): New function.
7299 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
7301 2018-08-02 Jeff Law <law@redhat.com>
7304 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7305 Define to speculation_safe_value_not_needed.
7308 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7309 Define to speculation_safe_value_not_needed.
7311 2018-08-02 Tom de Vries <tdevries@suse.de>
7314 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
7315 function. Return UI_TARGET unconditionally.
7316 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
7317 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
7319 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7321 * genemit.c (print_overload_test): Fix typo.
7323 2018-08-02 Richard Biener <rguenther@suse.de>
7325 PR tree-optimization/86816
7326 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
7327 which checks for value availability before querying it.
7328 (gvn_uses_equal): Use it.
7329 (same_succ_hash): Likewise.
7330 (gimple_equal_p): Likewise.
7332 2018-08-02 Nick Clifton <nickc@redhat.com>
7335 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7336 Define to speculation_safe_value_not_needed.
7339 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7340 Define to speculation_safe_value_not_needed.
7343 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7344 Define to speculation_safe_value_not_needed.
7347 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7348 Define to speculation_safe_value_not_needed.
7351 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7352 Define to speculation_safe_value_not_needed.
7355 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7356 Define to speculation_safe_value_not_needed.
7359 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7360 Define to speculation_safe_value_not_needed.
7363 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7364 Define to speculation_safe_value_not_needed.
7367 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
7368 speculation_safe_value_not_needed.
7371 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
7372 to speculation_safe_value_not_needed.
7374 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7376 * doc/md.texi: Expand the documentation of instruction names
7377 to mention port-local uses. Document '@' in pattern names.
7378 * read-md.h (overloaded_instance, overloaded_name): New structs.
7380 (md_reader::handle_overloaded_name): New member function.
7381 (md_reader::get_overloads): Likewise.
7382 (md_reader::m_first_overload): New member variable.
7383 (md_reader::m_next_overload_ptr): Likewise.
7384 (md_reader::m_overloads_htab): Likewise.
7385 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
7386 m_next_overload_ptr and m_overloads_htab.
7387 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
7388 (get_mode_token, get_code_token, get_int_token): New functions.
7389 (map_attr_string): Add an optional argument that passes back
7390 the associated iterator.
7391 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
7392 (md_reader::handle_overloaded_name, add_overload_instance): New
7394 (apply_iterators): Handle '@' names. Report an error if '@'
7395 is used without iterators.
7396 (initialize_iterators): Initialize the new iterator_group fields.
7397 * genopinit.c (handle_overloaded_code_for)
7398 (handle_overloaded_gen): New functions.
7399 (main): Use them to print declarations of maybe_code_for_* and
7400 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
7401 * genemit.c (print_overload_arguments, print_overload_test)
7402 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
7403 (main): Use it to print definitions of maybe_code_for_* and
7404 maybe_gen_* functions.
7405 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
7406 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
7407 instead of explicit mode checks.
7408 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
7409 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
7410 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
7411 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
7412 (aarch64_expand_compare_and_swap): Likewise
7413 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
7414 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
7415 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
7416 (aarch64_constant_pool_reload_icode): Delete.
7417 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
7418 instead of aarch64_constant_pool_reload_icode. Use
7419 code_for_aarch64_reload_mov instead of explicit mode checks.
7420 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
7421 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
7422 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
7423 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
7424 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
7425 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
7426 (aarch64_atomic_load_op_code): Delete.
7427 (aarch64_emit_atomic_load_op): Likewise.
7428 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
7429 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
7430 instead of aarch64_emit_atomic_load_op.
7431 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
7432 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
7433 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
7434 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
7435 character before the pattern name.
7436 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7437 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
7438 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
7439 (aarch64_frecps<mode>): Likewise.
7440 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
7441 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
7442 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
7443 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
7444 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
7446 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7448 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
7449 Allow HFmode constants if TARGET_FP_F16INST.
7451 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
7454 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
7455 No longer check last store for clobber of address register.
7457 2018-08-02 Martin Liska <mliska@suse.cz>
7459 PR gcov-profile/86817
7460 * gcov.c (process_all_functions): New function.
7462 (process_file): Move functions processing to
7463 process_all_functions.
7465 2018-08-02 David Malcolm <dmalcolm@redhat.com>
7467 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
7468 "const" to the "gimple *" and "rtx_insn *" parameters.
7469 * dumpfile.h (dump_user_location_t::dump_user_location_t):
7471 (dump_location_t::dump_location_t): Likewise.
7473 2018-08-01 Martin Sebor <msebor@redhat.com>
7475 PR tree-optimization/86650
7476 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
7477 rather than a "gcall *". Directly pass the data of interest
7478 to percent_K_format, rather than building a temporary CALL_EXPR
7480 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
7481 (gimple_fold_builtin_strncat): Adjust.
7482 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
7483 gcall* argument with gimple*.
7484 * gimple-ssa-warn-restrict.c (check_call): Same.
7485 (wrestrict_dom_walker::before_dom_children): Same.
7486 (builtin_access::builtin_access): Same.
7487 (check_bounds_or_overlap): Same
7488 (maybe_diag_overlap): Same.
7489 (maybe_diag_offset_bounds): Same.
7490 * tree-diagnostic.c (default_tree_printer): Move usage of
7491 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
7493 * tree-pretty-print.c (percent_K_format): Add argument.
7494 * tree-pretty-print.h: Add argument.
7495 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
7496 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
7497 (maybe_diag_stxncpy_trunc): Same.
7498 (handle_builtin_stxncpy): Same.
7499 (handle_builtin_strcat): Same.
7501 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7503 * match.pd: Optimise pointer range checks.
7505 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7507 PR tree-optimization/86758
7508 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
7509 to remove pattern statements.
7511 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7513 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
7514 result of dfs_enumerate_from when constructing stmt_vec_infos,
7515 instead of additionally calling get_loop_body.
7517 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7519 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
7521 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
7522 When creating an iv, assert that the step is not known to be zero.
7523 (vect_setup_realignment): Update call accordingly.
7524 * tree-vect-stmts.c (vectorizable_store): Likewise.
7525 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
7527 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7529 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
7530 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
7531 (vectorizable_reduction): Likewise.
7532 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
7533 (vect_detect_hybrid_slp_stmts): Likewise.
7534 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
7536 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
7538 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
7539 (wide_int_set_zero_nonzero_bits): ...this.
7540 (zero_nonzero_bits_from_vr): Rename to...
7541 (vrp_set_zero_nonzero_bits): ...this.
7542 (extract_range_from_multiplicative_op_1): Abstract wide int
7544 (wide_int_range_multiplicative_op): ...here.
7545 (extract_range_from_binary_expr_1): Extract wide int binary
7546 operations into their own functions.
7547 (wide_int_range_lshift): New.
7548 (wide_int_range_can_optimize_bit_op): New.
7549 (wide_int_range_shift_undefined_p): New.
7550 (wide_int_range_bit_xor): New.
7551 (wide_int_range_bit_ior): New.
7552 (wide_int_range_bit_and): New.
7553 (wide_int_range_trunc_mod): New.
7554 (extract_range_into_wide_ints): New.
7555 (vrp_shift_undefined_p): New.
7556 (extract_range_from_multiplicative_op): New.
7557 (vrp_can_optimize_bit_op): New.
7558 * tree-vrp.h (value_range::dump): New.
7559 (wide_int_range_multiplicative_op): New.
7560 (wide_int_range_lshift):New.
7561 (wide_int_range_shift_undefined_p): New.
7562 (wide_int_range_bit_xor): New.
7563 (wide_int_range_bit_ior): New.
7564 (wide_int_range_bit_and): New.
7565 (wide_int_range_trunc_mod): New.
7566 (zero_nonzero_bits_from_bounds): Rename to...
7567 (wide_int_set_zero_nonzero_bits): ...this.
7568 (zero_nonzero_bits_from_vr): Rename to...
7569 (vrp_set_zero_nonzero_bits): ...this.
7570 (range_easy_mask_min_max): Rename to...
7571 (wide_int_range_can_optimize_bit_op): this.
7572 * vr-values.c (simplify_bit_ops_using_ranges): Rename
7573 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
7575 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7577 * tree-vectorizer.h (vect_orig_stmt): New function.
7578 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
7579 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
7580 (vect_create_epilog_for_reduction): Likewise.
7581 (vectorizable_live_operation): Likewise.
7582 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
7583 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
7584 * tree-vect-stmts.c (vectorizable_call): Likewise.
7585 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
7587 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7589 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
7591 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
7592 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
7593 (vect_transform_loop): Likewise.
7594 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7596 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7598 * tree-vectorizer.h (vect_schedule_slp): Return void.
7599 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7600 (vect_schedule_slp): Likewise.
7602 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7604 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
7606 (vect_transform_loop): Update calls accordingly. Schedule SLP
7607 instances before the main loop, if any exist.
7609 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7611 PR tree-optimization/86749
7612 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
7613 If the lhs is used in a COND_EXPR, check that it is being used
7614 as the "then" or "else" value.
7616 2018-08-01 Tom de Vries <tdevries@suse.de>
7619 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
7620 speculation_safe_value_not_needed.
7622 2018-08-01 Richard Biener <rguenther@suse.de>
7624 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
7627 2018-08-01 Martin Liska <mliska@suse.cz>
7629 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
7630 format how successful transformation is dumped.
7631 (gimple_mod_pow2_value_transform): Likewise.
7632 (gimple_mod_subtract_transform): Likewise.
7633 (gimple_stringops_transform): Likewise.
7635 2018-08-01 Martin Liska <mliska@suse.cz>
7638 * value-prof.c (interesting_stringop_to_profile_p):
7639 Simplify the code and add BUILT_IN_MEMMOVE.
7640 (gimple_stringops_transform): Likewise.
7642 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
7644 * config/aarch64/aarch64-simd.md
7645 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
7646 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
7647 use GPI iterator instead of SI mode.
7649 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
7651 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
7652 rs6000_speculation_barrier.
7653 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
7654 new barrier pattern name.
7656 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
7658 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
7659 (speculation_barrier): New insn.
7661 2018-08-01 Richard Biener <rguenther@suse.de>
7664 * graphite.h: Include isl/id.h and isl/space.h to allow build
7667 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
7670 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
7671 mode for COFF targets.
7672 * defaults.h (TARGET_COFF): Define.
7673 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
7674 TARGET_COFF): Define.
7675 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
7676 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
7677 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
7679 2018-07-31 Alexandre Oliva <oliva@adacore.com>
7681 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
7682 * gimple-streamer-out.c (output_bb): Save it.
7683 * lto-streamer-in.c (input_struct_function_base): Restore
7684 instance discriminator if available. Create map on demand.
7685 * lto-streamer-out.c (output_struct_function_base): Save it if
7687 * final.c (decl_to_instance_map): Document LTO strategy.
7689 2018-07-31 Alexandre Oliva <oliva@adacore.com>
7690 Olivier Hainque <hainque@adacore.com>
7692 * debug.h (decl_to_instance_map_t): New type.
7693 (decl_to_instance_map): Declare.
7694 (maybe_create_decl_to_instance_map): New inline function.
7695 * final.c (bb_discriminator, last_bb_discriminator): New statics,
7696 to track basic block discriminators.
7697 (final_start_function_1): Initialize them.
7698 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
7700 (decl_to_instance_map): New variable.
7701 (map_decl_to_instance, maybe_set_discriminator): New functions.
7702 (notice_source_line): Set discriminator.
7704 2018-07-31 Ian Lance Taylor <iant@golang.org>
7706 * targhooks.c (default_have_speculation_safe_value): Add
7709 2018-07-31 David Malcolm <dmalcolm@redhat.com>
7711 * dump-context.h: Include "pretty-print.h".
7712 (dump_context::refresh_dumps_are_enabled): New decl.
7713 (dump_context::emit_item): New decl.
7714 (class dump_context): Add fields "m_test_pp" and
7716 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
7717 (temp_dump_context::get_dumped_text): New decl.
7718 (class temp_dump_context): Add field "m_pp".
7719 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
7720 (dump_context::refresh_dumps_are_enabled): ...and add a test for
7722 (set_dump_file): Update for above change.
7723 (set_alt_dump_file): Likewise.
7724 (dump_loc): New overload, taking a pretty_printer *.
7725 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
7726 to any test pretty-printer.
7727 (make_item_for_dump_gimple_stmt): New function, adapted from
7728 optinfo::add_gimple_stmt.
7729 (dump_context::dump_gimple_stmt): Call it, and use the result,
7730 eliminating the direct usage of dump_file and alt_dump_file in
7731 favor of indirectly using them via emit_item.
7732 (make_item_for_dump_gimple_expr): New function, adapted from
7733 optinfo::add_gimple_expr.
7734 (dump_context::dump_gimple_expr): Call it, and use the result,
7735 eliminating the direct usage of dump_file and alt_dump_file in
7736 favor of indirectly using them via emit_item.
7737 (make_item_for_dump_generic_expr): New function, adapted from
7739 (dump_context::dump_generic_expr): Call it, and use the result,
7740 eliminating the direct usage of dump_file and alt_dump_file in
7741 favor of indirectly using them via emit_item.
7742 (make_item_for_dump_printf_va): New function, adapted from
7743 optinfo::add_printf_va.
7744 (make_item_for_dump_printf): New function.
7745 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
7746 and use the result, eliminating the direct usage of dump_file and
7747 alt_dump_file in favor of indirectly using them via emit_item.
7748 (make_item_for_dump_dec): New function.
7749 (dump_context::dump_dec): Call it, and use the result,
7750 eliminating the direct usage of dump_file and alt_dump_file in
7751 favor of indirectly using them via emit_item.
7752 (make_item_for_dump_symtab_node): New function, adapted from
7753 optinfo::add_symtab_node.
7754 (dump_context::dump_symtab_node): Call it, and use the result,
7755 eliminating the direct usage of dump_file and alt_dump_file in
7756 favor of indirectly using them via emit_item.
7757 (dump_context::begin_scope): Reimplement, avoiding direct usage
7758 of dump_file and alt_dump_file in favor of indirectly using them
7760 (dump_context::emit_item): New member function.
7761 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
7762 Set up test pretty-printer on the underlying context. Call
7763 refresh_dumps_are_enabled.
7764 (temp_dump_context::~temp_dump_context): Call
7765 refresh_dumps_are_enabled.
7766 (temp_dump_context::get_dumped_text): New member function.
7767 (selftest::verify_dumped_text): New function.
7768 (ASSERT_DUMPED_TEXT_EQ): New macro.
7769 (selftest::test_capture_of_dump_calls): Run all tests twice, with
7770 and then without optinfo enabled. Add uses of
7771 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
7772 * dumpfile.h: Update comment for the dump_* API.
7773 * optinfo-emit-json.cc
7774 (selftest::test_building_json_from_dump_calls): Update for new
7775 param for temp_dump_context ctor.
7776 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
7777 and "m_owned" field.
7778 (optinfo_item::~optinfo_item): Likewise.
7779 (optinfo::add_item): New member function.
7780 (optinfo::emit): Update comment.
7781 (optinfo::add_string): Delete.
7782 (optinfo::add_printf): Delete.
7783 (optinfo::add_printf_va): Delete.
7784 (optinfo::add_gimple_stmt): Delete.
7785 (optinfo::add_gimple_expr): Delete.
7786 (optinfo::add_tree): Delete.
7787 (optinfo::add_symtab_node): Delete.
7788 (optinfo::add_dec): Delete.
7789 * optinfo.h (class dump_context): New forward decl.
7790 (optinfo::add_item): New decl.
7791 (optinfo::add_string): Delete.
7792 (optinfo::add_printf): Delete.
7793 (optinfo::add_printf_va): Delete.
7794 (optinfo::add_gimple_stmt): Delete.
7795 (optinfo::add_gimple_expr): Delete.
7796 (optinfo::add_tree): Delete.
7797 (optinfo::add_symtab_node): Delete.
7798 (optinfo::add_dec): Delete.
7799 (optinfo::add_poly_int): Delete.
7800 (optinfo_item::optinfo_item): Remove "owned" param.
7801 (class optinfo_item): Remove field "m_owned".
7803 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
7806 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
7807 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
7808 requested variable alignment.
7809 (expand_one_ssa_partition): Likewise.
7810 (expand_one_var): Likewise.
7812 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7814 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
7815 to speculation_safe_value_not_needed.
7817 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7819 * targhooks.h (speculation_safe_value_not_needed): New prototype.
7820 * targhooks.c (speculation_safe_value_not_needed): New function.
7821 * target.def (have_speculation_safe_value): Update documentation.
7822 * doc/tm.texi: Regenerated.
7824 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7826 * config/aarch64/iterators.md (ALLI_TI): New iterator.
7827 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
7829 (despeculate_copy<ALLI:mode>_insn): New insn.
7830 (despeculate_copyti_insn): New insn.
7831 (despeculate_simple<ALLI:mode>): New insn
7832 (despeculate_simpleti): New insn.
7833 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
7835 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
7836 aarch64_speculation_safe_value.
7837 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
7839 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7841 * config/aarch64/aarch64-speculation.cc: New file.
7842 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
7843 before pass_reorder_blocks.
7844 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
7846 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
7847 X14 and X15 when tracking speculation.
7848 * config/aarch64/aarch64.md (register name constants): Add
7849 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
7850 (unspec): Add UNSPEC_SPECULATION_TRACKER.
7851 (speculation_barrier): New insn attribute.
7852 (cmp<mode>): Allow SP in comparisons.
7853 (speculation_tracker): New insn.
7854 (speculation_barrier): Add speculation_barrier attribute.
7855 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
7856 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
7857 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
7859 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7861 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
7862 aarch64_track_speculation is true.
7863 (tb<optab><mode>1): Likewise.
7864 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
7865 generate CB[N]Z when tracking speculation.
7866 (aarch64_split_compare_and_swap): Likewise.
7867 (aarch64_split_atomic_op): Likewise.
7869 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7871 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
7873 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7875 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
7876 (speculation_barrier): New insn.
7878 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7880 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
7881 * config/arm/arm.md (speculation_barrier): New expand.
7882 (speculation_barrier_insn): New pattern.
7884 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7886 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
7887 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
7888 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
7889 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
7891 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
7892 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
7893 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
7894 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
7895 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
7896 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
7897 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
7898 * builtins.c (expand_speculation_safe_value): New function.
7899 (expand_builtin): Call it.
7900 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
7901 * doc/extend.texi: Document __builtin_speculation_safe_value.
7902 * doc/md.texi: Document "speculation_barrier" pattern.
7903 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
7904 TARGET_HAVE_SPECULATION_SAFE_VALUE.
7905 * doc/tm.texi: Regenerated.
7906 * target.def (have_speculation_safe_value, speculation_safe_value): New
7908 * targhooks.c (default_have_speculation_safe_value): New function.
7909 (default_speculation_safe_value): New function.
7910 * targhooks.h (default_have_speculation_safe_value): Add prototype.
7911 (default_speculation_safe_value): Add prototype.
7913 2018-07-31 David Malcolm <dmalcolm@redhat.com>
7915 * dump-context.h (dump_context::dump_loc): New decl.
7916 * dumpfile.c (dump_context::dump_loc): New member function.
7917 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
7918 and dump_gimple_stmt.
7919 (dump_context::dump_gimple_expr_loc): Likewise, using
7921 (dump_context::dump_generic_expr_loc): Likewise, using
7923 (dump_context::dump_printf_loc_va): Likewise, using
7925 (dump_context::begin_scope): Explicitly using the global function
7926 "dump_loc", rather than the member function.
7928 2018-07-31 Martin Sebor <msebor@redhat.com>
7930 PR tree-optimization/86741
7931 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
7933 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
7935 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
7936 depend on whether prefetch instructions will be emitted or not.
7937 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
7938 will be emitted or not.
7939 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
7940 (TARGET_SETMEM_PFD): New macros.
7942 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7944 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
7945 (NULL_STMT_VEC_INFO): Delete.
7946 (stmt_vec_info::operator*): Likewise.
7947 (stmt_vec_info::operator gimple *): Likewise.
7948 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
7949 of NULL_STMT_VEC_INFO.
7950 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
7951 (vect_reassociating_reduction_p): Likewise.
7952 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
7953 (vectorizable_store): Likewise.
7954 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
7955 (vec_info::free_stmt_vec_infos): Likewise.
7957 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7959 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
7960 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
7962 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7964 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
7965 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
7966 (vec_info::free_stmt_vec_info): New private member functions.
7967 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
7968 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
7969 * tree-parloops.c (gather_scalar_reductions): Remove calls to
7970 set_stmt_vec_info_vec and free_stmt_vec_infos.
7971 * tree-vect-loop.c (_loop_vec_info): Remove call to
7972 set_stmt_vec_info_vec.
7973 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
7974 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
7975 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
7976 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
7977 (vec_info::free_stmt_vec_info): ...these new functions. Remove
7978 assignments in {vec_info::,}new_stmt_vec_info that are redundant
7979 with the clearing in the xcalloc.
7980 (stmt_vec_info_vec): Delete.
7981 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
7982 (vectorize_loops): Likewise.
7983 (vec_info::~vec_info): Remove argument from call to
7984 free_stmt_vec_infos.
7985 (vec_info::add_stmt): Remove vinfo argument from call to
7988 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7990 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
7991 rather than a gimple stmt.
7992 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
7993 information for pattern statements when passed the original
7994 statement; instead wait to be passed the pattern statement itself.
7995 Don't call set_vinfo_for_stmt here.
7996 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
7997 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
7998 stmt_vec_infos here.
7999 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
8000 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
8001 stmt_vec_infos entry.
8003 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8005 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
8006 * tree-vectorizer.c (vec_info::replace_stmt): New function.
8007 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
8008 * tree-vect-stmts.c (vectorizable_call): Likewise.
8009 (vectorizable_simd_clone_call): Likewise.
8011 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8013 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
8014 * tree-vectorizer.c (vec_info::remove_stmt): New function.
8015 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
8016 * tree-vect-loop.c (vect_transform_loop): Likewise.
8017 * tree-vect-slp.c (vect_schedule_slp): Likewise.
8018 * tree-vect-stmts.c (vect_remove_stores): Likewise.
8020 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8022 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
8023 (vect_dr_stmt): Delete.
8024 * tree-vectorizer.c (vec_info::lookup_dr): New function.
8025 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
8027 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
8028 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
8029 (vect_verify_datarefs_alignment, vect_peeling_supportable)
8030 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
8031 (vect_analyze_data_refs): Likewise.
8032 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
8034 (vect_find_same_alignment_drs): Likewise.
8035 (vect_slp_analyze_node_dependences): Update calls accordingly.
8036 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
8037 instead of DR_VECT_AUX.
8038 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
8039 of a vector data references. Use vec_info::lookup_dr instead of
8041 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
8042 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
8043 instead of DR_VECT_AUX.
8045 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8047 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
8049 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
8051 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8052 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
8053 (vect_gen_prolog_loop_niters): Likewise.
8055 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8057 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
8058 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
8059 (vect_known_alignment_in_bytes, vect_dr_behavior)
8060 (vect_get_scalar_dr_size): Take references as dr_vec_infos
8061 instead of data_references. Update calls to other routines for
8062 which the same change has been made.
8063 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
8064 dr_vec_infos instead of stmt_vec_infos.
8065 (vect_analyze_data_ref_dependence): Update call accordingly.
8066 (vect_slp_analyze_data_ref_dependence)
8067 (vect_record_base_alignments): Use DR_VECT_AUX.
8068 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
8069 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
8070 (vector_alignment_reachable_p, vect_get_data_access_cost)
8071 (vect_peeling_supportable, vect_analyze_group_access_1)
8072 (vect_analyze_group_access, vect_analyze_data_ref_access)
8073 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
8074 (vect_compile_time_alias, vect_small_gap_p)
8075 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
8076 (vect_supportable_dr_alignment): Take references as dr_vec_infos
8077 instead of data_references. Update calls to other routines for
8078 which the same change has been made.
8079 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
8080 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
8081 (vect_slp_analyze_and_verify_node_alignment)
8082 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
8083 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
8084 (vect_setup_realignment): Use dr_vec_infos. Update calls after
8086 (_vect_peel_info::dr): Replace with...
8087 (_vect_peel_info::dr_info): ...this new field.
8088 (vect_peeling_hash_get_most_frequent)
8089 (vect_peeling_hash_choose_best_peeling): Update accordingly.
8090 (vect_peeling_hash_get_lowest_cost):
8091 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
8092 routines for which the same change has been made.
8093 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
8095 * tree-vect-loop-manip.c (get_misalign_in_elems)
8096 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
8098 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8099 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
8100 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
8101 (get_group_load_store_type, get_negative_load_store_type)
8102 (vect_get_data_ptr_increment, vectorizable_store)
8103 (vectorizable_load): Likewise.
8104 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
8105 Update calls to other routines for which the same change has been made.
8107 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8109 * tree-vectorizer.h (vec_info::move_dr): New member function.
8110 (dataref_aux): Rename to...
8111 (dr_vec_info): ...this and add "dr" and "stmt" fields.
8112 (_stmt_vec_info::dr_aux): Update accordingly.
8113 (_stmt_vec_info::data_ref_info): Delete.
8114 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
8115 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
8116 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
8117 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
8119 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
8120 (STMT_VINFO_DR_INFO): New macro.
8121 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
8122 (set_dr_misalignment): Update after rename of dataref_aux.
8123 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
8124 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
8125 initialization of STMT_VINFO_DATA_REF.
8126 * tree-vectorizer.c (vec_info::move_dr): New function.
8127 * tree-vect-patterns.c (vect_recog_bool_pattern)
8128 (vect_recog_mask_conversion_pattern)
8129 (vect_recog_gather_scatter_pattern): Use it.
8130 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
8131 the "dr" and "stmt" fields of dr_vec_info instead of
8132 STMT_VINFO_DATA_REF.
8134 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8136 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
8137 (is_pattern_stmt_p): Use it.
8138 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
8139 on pattern statements.
8141 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8143 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
8144 original stmt as a stmt_vec_info rather than a gimple stmt.
8145 (vect_pattern_recog_1): Take the statement directly as a
8146 stmt_vec_info, rather than via a gimple_stmt_iterator.
8147 Update call to vect_mark_pattern_stmts.
8148 (vect_pattern_recog): Update calls accordingly.
8150 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8152 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
8153 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
8154 a vect_def_type for the first argument.
8155 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
8156 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
8157 operand if it isn't defined by a vectorized statement.
8158 (vect_build_gather_load_calls): Remove the mask_dt argument and
8159 update calls to vect_get_vec_def_for_stmt_copy.
8160 (vectorizable_bswap): Likewise the dt argument.
8161 (vectorizable_call): Update calls to vectorizable_bswap and
8162 vect_get_vec_def_for_stmt_copy.
8163 (vectorizable_simd_clone_call, vectorizable_assignment)
8164 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
8165 (vectorizable_comparison): Update calls to
8166 vect_get_vec_def_for_stmt_copy.
8167 (vectorizable_store): Likewise. Remove now-unnecessary calls to
8169 (vect_get_loop_based_defs): Remove dt argument and update call
8170 to vect_get_vec_def_for_stmt_copy.
8171 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
8172 and vect_get_vec_def_for_stmt_copy.
8173 (vectorizable_load): Update calls to vect_build_gather_load_calls
8174 and vect_get_vec_def_for_stmt_copy.
8175 * tree-vect-loop.c (vect_create_epilog_for_reduction)
8176 (vectorizable_reduction, vectorizable_live_operation): Update calls
8177 to vect_get_vec_def_for_stmt_copy.
8179 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8181 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
8182 and gimple stmt arguments with a stmt_vec_info.
8183 (vect_record_base_alignments): Update calls accordingly.
8184 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
8185 and gimple stmt arguments with a stmt_vec_info.
8186 (vect_build_slp_tree_1): Remove vinfo argument and update call
8187 to vect_record_max_nunits.
8188 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
8189 and vect_record_max_nunits.
8191 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8193 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
8194 file and take a stmt_vec_info instead of a gimple stmt.
8195 (supportable_widening_operation, vect_finish_replace_stmt)
8196 (vect_finish_stmt_generation, vect_get_store_rhs)
8197 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
8198 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
8199 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
8200 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
8201 (vect_create_data_ref_ptr, bump_vector_ptr)
8202 (vect_permute_store_chain, vect_setup_realignment)
8203 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
8204 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
8205 (vectorizable_reduction, vectorizable_induction)
8206 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
8207 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
8208 than gimple stmts as arguments.
8209 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
8210 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
8211 (can_group_stmts_p, vect_check_gather_scatter)
8212 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
8213 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
8214 (vect_permute_load_chain, vect_shift_permute_load_chain)
8215 (vect_transform_grouped_load)
8216 (vect_record_grouped_load_vectors): Likewise.
8217 * tree-vect-loop.c (vect_fixup_reduc_chain)
8218 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
8219 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
8220 (vectorizable_reduction, vectorizable_induction)
8221 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
8222 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
8223 (vect_get_load_store_mask): Likewise.
8224 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
8225 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
8226 * tree-vect-stmts.c (vect_mark_relevant)
8227 (is_simple_and_all_uses_invariant)
8228 (exist_non_indexing_operands_for_use_p, process_use)
8229 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
8230 (vect_get_vec_def_for_operand, vect_get_vec_defs)
8231 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
8232 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
8233 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
8234 (get_negative_load_store_type, get_load_store_type)
8235 (vect_check_load_store_mask, vect_check_store_rhs)
8236 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
8237 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
8238 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
8239 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8240 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
8241 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
8242 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
8243 (supportable_widening_operation): Likewise.
8245 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8247 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
8248 a stmt_vec_info instead of a gcall.
8249 (vect_check_gather_scatter): Update call accordingly.
8250 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
8252 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
8253 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
8254 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
8255 instead of a gimple stmt.
8256 (vect_transform_loop): Update calls accordingly.
8257 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
8258 stmt_vec_infos instead of gimple stmts.
8259 (vect_analyze_slp_instance): Update use accordingly.
8260 * tree-vect-stmts.c (read_vector_array, write_vector_array)
8261 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
8262 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
8263 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
8264 (vect_gen_widened_results_half, vect_get_loop_based_defs)
8265 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
8266 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
8267 down to subroutines.
8269 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8271 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
8272 of the worklist from a vector of gimple stmts to a vector of
8274 * tree-vect-stmts.c (vect_mark_relevant, process_use)
8275 (vect_mark_stmts_to_be_vectorized): Likewise
8277 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8279 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
8280 statement before passing it to vect_analyze_stmt.
8281 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
8282 the chain of phi vector definitions. Track the exit phi via its
8284 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
8285 STMT_VINFO_REDUC_DEF.
8286 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
8287 stmt_vec_infos to handle the statement chains.
8288 (vect_get_slp_defs): Record the first statement in the node
8289 using a stmt_vec_info.
8290 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
8291 statements here and pass their stmt_vec_info down to subroutines.
8292 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
8293 down to vect_finish_stmt_generation.
8294 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
8295 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
8296 stmt_vec_infos to subroutines.
8297 (vect_remove_stores): Use stmt_vec_infos to handle the statement
8300 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8302 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
8303 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
8304 (vect_permute_store_chain, vect_setup_realignment)
8305 (vect_permute_load_chain, vect_shift_permute_load_chain)
8306 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
8307 stmts internally, and when passing values to other vectorizer routines.
8308 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
8309 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
8310 (vect_analyze_loop_operations, get_initial_def_for_reduction)
8311 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
8312 (vectorizable_reduction, vectorizable_induction)
8313 (vectorizable_live_operation, vect_transform_loop_stmt)
8314 (vect_transform_loop): Likewise.
8315 * tree-vect-patterns.c (vect_reassociating_reduction_p)
8316 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
8317 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
8318 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8319 (vect_slp_analyze_node_operations_1): Likewise.
8320 * tree-vect-stmts.c (vect_mark_relevant, process_use)
8321 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
8322 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
8323 (vect_finish_stmt_generation_1, get_group_load_store_type)
8324 (get_load_store_type, vect_build_gather_load_calls)
8325 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
8326 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
8327 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8328 (vectorizable_store, vectorizable_load, vectorizable_condition)
8329 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
8330 (supportable_widening_operation): Likewise.
8331 (vect_get_vector_types_for_stmt): Likewise.
8332 * tree-vectorizer.h (vect_dr_behavior): Likewise.
8334 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8336 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
8337 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
8338 (vect_permute_store_chain, vect_permute_load_chain)
8339 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
8340 repeated stmt_vec_info lookups.
8341 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
8342 (vect_update_ivs_after_vectorizer): Likewise.
8343 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
8344 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
8345 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
8346 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8347 (vect_bb_slp_scalar_cost): Likewise.
8348 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
8350 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8352 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
8353 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
8354 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
8355 to get gassigns and gcalls, rather than operating on generc gimple
8357 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
8358 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
8359 (vectorizable_load, vect_analyze_stmt): Likewise.
8360 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
8362 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8364 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
8365 return stmt_vec_infos rather than gimple stmts. Do not accept
8367 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
8369 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
8370 Update use of get_later_stmt.
8371 (vect_get_constant_vectors): Update call accordingly.
8372 (vect_schedule_slp_instance): Likewise
8373 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
8374 (vect_slp_analyze_instance_dependence): Likewise.
8375 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
8377 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8379 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
8380 (stmt_info_for_cost::stmt_info): ...this new field.
8381 (add_stmt_costs): Update accordingly.
8382 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
8383 (vect_get_known_peeling_cost): Likewise.
8384 (vect_estimate_min_profitable_iters): Likewise.
8385 * tree-vect-stmts.c (record_stmt_cost): Likewise.
8387 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8389 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
8390 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
8391 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
8393 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
8395 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8397 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
8398 a gimple stmt to a stmt_vec_info.
8399 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
8401 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8403 * tree-vectorizer.h (vec_info::grouped_stores): Change from
8404 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
8405 (_loop_vec_info::reduction_chains): Likewise.
8406 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
8408 * tree-vect-slp.c (vect_analyze_slp): Likewise.
8410 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8412 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
8413 a gimple stmt to a stmt_vec_info.
8414 (_stmt_vec_info::next_element): Likewise.
8415 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
8416 (vect_slp_analyze_and_verify_node_alignment)
8417 (vect_analyze_group_access_1, vect_analyze_group_access)
8418 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
8419 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
8420 (vect_supportable_dr_alignment): Update accordingly.
8421 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
8422 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
8423 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
8424 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
8425 * tree-vect-slp.c (vect_build_slp_tree_1)
8426 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
8427 (vect_split_slp_store_group, vect_analyze_slp_instance)
8428 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
8429 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
8430 (get_group_load_store_type, get_load_store_type)
8431 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
8432 (vect_transform_stmt, vect_remove_stores): Likewise.
8434 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8436 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
8438 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
8439 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
8440 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
8441 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
8442 (vector_alignment_reachable_p, vect_get_data_access_cost)
8443 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
8444 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
8445 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
8446 (vect_analyze_group_access_1, vect_analyze_group_access)
8447 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
8448 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
8449 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
8450 result of vect_dr_stmt and use the stmt_vec_info instead of
8451 the associated gimple stmt.
8452 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
8453 (vect_gen_prolog_loop_niters): Likewise.
8454 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8456 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8458 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
8459 to a vec<stmt_vec_info>.
8460 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
8461 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
8463 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
8464 to a vec<stmt_vec_info>.
8465 (bst_traits::value_type, bst_traits::value_type): Likewise.
8466 (bst_traits::hash): Update accordingly.
8467 (vect_get_and_check_slp_defs): Change the stmts parameter from
8468 a vec<gimple *> to a vec<stmt_vec_info>.
8469 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
8470 (vect_build_slp_tree): Likewise.
8471 (vect_build_slp_tree_2): Likewise. Update uses of
8472 SLP_TREE_SCALAR_STMTS.
8473 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
8474 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
8475 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
8476 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
8477 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
8478 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
8479 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
8480 (vect_get_constant_vectors, vect_get_slp_defs)
8481 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
8482 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
8483 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
8484 instead of gimple stmts.
8485 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
8486 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
8487 (vect_slp_analyze_instance_dependence): Update uses of
8488 SLP_TREE_SCALAR_STMTS.
8489 (vect_slp_analyze_and_verify_node_alignment): Likewise.
8490 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
8491 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
8492 (get_initial_defs_for_reduction): Likewise.
8493 (vect_create_epilog_for_reduction): Likewise.
8494 (vectorize_fold_left_reduction): Likewise.
8495 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
8496 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
8497 (can_vectorize_live_stmts): Likewise.
8499 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8501 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
8502 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
8503 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
8505 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
8507 (gather_scalar_reductions): Update after above interface changes.
8508 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
8509 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
8511 (vect_force_simple_reduction): Likewise.
8512 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
8513 LOOP_VINFO_REDUCTIONS.
8514 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8516 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8518 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
8519 a gimple stmt to a stmt_vec_info.
8520 * tree-vect-loop.c (vect_active_double_reduction_p)
8521 (vect_force_simple_reduction, vectorizable_reduction): Update
8524 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8526 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
8527 vec<gimple *> to a vec<stmt_vec_info>.
8528 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
8529 the reduction_phis argument from a vec<gimple *> to a
8531 (vectorizable_reduction): Likewise the phis local variable that
8532 is passed to vect_create_epilog_for_reduction. Update for new type
8533 of SLP_TREE_VEC_STMTS.
8534 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
8535 (vectorizable_live_operation): Likewise.
8536 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
8537 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
8539 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8541 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
8542 a gimple stmt to a stmt_vec_info.
8543 (vectorizable_condition, vectorizable_live_operation)
8544 (vectorizable_reduction, vectorizable_induction): Pass back the
8545 vectorized statement as a stmt_vec_info.
8546 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
8547 use of STMT_VINFO_VEC_STMT.
8548 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
8549 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
8550 as stmt_vec_infos rather than gimple stmts.
8551 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
8553 (vectorizable_live_operation): Likewise.
8554 (vectorizable_reduction, vectorizable_induction): Likewise,
8555 updating use of STMT_VINFO_VEC_STMT.
8556 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
8557 of STMT_VINFO_VEC_STMT.
8558 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
8559 (vectorizable_simd_clone_call, vectorizable_conversion)
8560 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8561 (vectorizable_store, vectorizable_load, vectorizable_condition)
8562 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
8563 from a gimple stmt to a stmt_vec_info.
8564 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
8565 pointer to a stmt_vec_info to the vectorizable_* routines.
8567 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8569 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
8570 a gimple stmt to a stmt_vec_info.
8571 (is_pattern_stmt_p): Update accordingly.
8572 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
8573 (vect_record_grouped_load_vectors): Likewise.
8574 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
8575 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
8576 (vect_model_reduction_cost): Likewise.
8577 (vect_create_epilog_for_reduction): Likewise.
8578 (vectorizable_reduction, vectorizable_induction): Likewise.
8579 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8580 Return the stmt_vec_info for the pattern statement.
8581 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
8582 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
8583 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
8584 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
8585 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
8586 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
8587 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
8588 (free_stmt_vec_info, vect_is_simple_use): Likewise.
8590 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8592 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
8593 (vect_finish_stmt_generation): Likewise.
8594 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
8595 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
8596 (vect_build_gather_load_calls): Use the return value of the above
8597 functions instead of a separate call to vinfo_for_stmt. Use narrow
8598 scopes for the input gimple stmt and wider scopes for the associated
8599 stmt_vec_info. Use vec_info::lookup_def when setting these
8600 stmt_vec_infos from an SSA_NAME definition.
8601 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
8602 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
8603 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8604 (vectorizable_store, vectorizable_load, vectorizable_condition)
8605 (vectorizable_comparison): Likewise.
8606 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
8607 (vectorizable_reduction): Likewise.
8609 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8611 * tree-vectorizer.h (vect_is_simple_use): Add an optional
8612 stmt_vec_info * parameter before the optional gimple **.
8613 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
8614 (process_use, vect_get_vec_def_for_operand_1): Update callers.
8615 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
8616 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8617 (vectorizable_live_operation): Likewise.
8618 * tree-vect-patterns.c (type_conversion_p): Likewise.
8619 (vect_look_through_possible_promotion): Likewise.
8620 (vect_recog_rotate_pattern): Likewise.
8621 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
8623 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8625 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
8626 a typedef to a wrapper class.
8627 (NULL_STMT_VEC_INFO): New macro.
8628 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
8629 (stmt_vec_info::operator*): New function.
8630 (stmt_vec_info::operator gimple *): Likewise.
8631 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
8632 (add_stmt_costs): Likewise.
8633 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
8634 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
8635 (vect_get_known_peeling_cost): Likewise.
8636 (vect_estimate_min_profitable_iters): Likewise.
8637 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8638 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
8639 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
8640 (vectorizable_store, free_stmt_vec_infos): Likewise.
8641 (new_stmt_vec_info): Change return type of xcalloc to
8644 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8646 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
8647 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
8648 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
8649 a single_imm_use-based sequence.
8650 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
8652 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8654 * tree-vectorizer.h (vec_info::lookup_def): Declare.
8655 * tree-vectorizer.c (vec_info::lookup_def): New function.
8656 * tree-vect-patterns.c (vect_get_internal_def): Use it.
8657 (vect_widened_op_tree): Likewise.
8658 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
8659 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
8660 (vectorizable_reduction): Likewise.
8661 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
8663 (vect_is_slp_reduction): Update calls accordingly. Use
8664 vec_info::lookup_def.
8665 (vect_is_simple_reduction): Likewise
8666 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
8668 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8670 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
8671 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
8672 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
8674 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
8675 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
8676 (vect_update_vf_for_slp, vect_analyze_loop_operations)
8677 (vect_is_slp_reduction, vectorizable_induction)
8678 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
8679 * tree-vect-patterns.c (vect_init_pattern_stmt):
8680 (vect_determine_min_output_precision_1, vect_determine_precisions)
8681 (vect_pattern_recog): Likewise.
8682 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
8683 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
8684 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
8685 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
8686 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
8687 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
8688 info field from a loop to a loop_vec_info.
8690 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8692 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
8693 (vec_info::add_stmt): Declare.
8694 * tree-vectorizer.c (vec_info::add_stmt): New function.
8695 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
8696 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
8697 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
8698 (vectorizable_induction): Likewise.
8699 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
8700 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
8701 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
8702 (vectorizable_load): Likewise.
8703 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8704 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
8705 (vect_recog_gather_scatter_pattern): Likewise.
8706 (append_pattern_def_seq): Likewise. Remove a check that is
8707 performed by add_stmt itself.
8709 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8711 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
8712 which make_ssa_name was called with new_stmt before new_stmt
8715 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8717 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
8719 (vect_is_slp_reduction): ...here...
8720 (vect_is_simple_reduction): ...and here. Remove repetition of tests
8721 that are already known to be false.
8723 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8725 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
8726 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
8727 STMT_VINFO_NUM_SLP_USES when it's true.
8728 (vect_free_slp_instance): Add a final_p parameter and pass it to
8730 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
8731 (vect_analyze_slp_instance): Likewise.
8732 (vect_slp_analyze_operations): Likewise.
8733 (vect_slp_analyze_bb_1): Likewise.
8734 * tree-vectorizer.c (vec_info): Likewise.
8735 * tree-vect-loop.c (vect_transform_loop): Likewise.
8737 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8739 * tree-vect-loop.c (vectorizable_reduction): Assert that the
8740 function is not called for second and subsequent members of
8743 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8745 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
8746 cases for nested loops from here to ...
8747 (vect_create_epilog_for_reduction): ...here. Only call
8748 vect_is_simple_use for inner-loop reductions.
8750 2018-07-31 Martin Liska <mliska@suse.cz>
8752 PR gcov-profile/85338
8753 PR gcov-profile/85350
8754 PR gcov-profile/85372
8755 * profile.c (struct location_triplet): New.
8756 (struct location_triplet_hash): Likewise.
8757 (output_location): Do not output a BB that
8758 is already recorded for a line.
8759 (branch_prob): Use streamed_locations.
8761 2018-07-31 Martin Liska <mliska@suse.cz>
8763 PR gcov-profile/85370
8764 * coverage.c (coverage_begin_function): Do not mark target
8765 clones as artificial functions.
8767 2018-07-31 Martin Liska <mliska@suse.cz>
8769 PR gcov-profile/83813
8770 PR gcov-profile/84758
8771 PR gcov-profile/85217
8772 PR gcov-profile/85332
8773 * profile.c (branch_prob): Do not record GOTO expressions
8774 for GIMPLE statements which locations are already streamed.
8776 2018-07-31 Olivier Hainque <hainque@adacore.com>
8778 * gcc.c (handle_spec_function): Accept a soft_matched_part
8779 argument, as do_spec_1. Pass it down to ...
8780 (eval_spec_function): Accept a soft_matched_part argument,
8781 and pass it down to ...
8782 (do_spec_2): Accept a soft_matched_part argument, and pass
8783 it down to do_spec_1.
8784 (do_spec_1): Pass soft_matched_part to handle_spec_function.
8785 (handle_braces): Update call to handle_spec_function.
8786 (driver::set_up_specs): Update calls to do_spec_2.
8787 (compare_debug_dump_opt_spec_function): Likewise.
8788 (compare_debug_self_opt_spec_function): Likewise.
8790 2018-07-31 Olivier Hainque <hainque@adacore.com>
8792 * common.opt (nolibc): New option.
8793 * doc/invoke.texi (Link Options): Document it.
8794 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
8795 * config/alpha/linux.h: Likewise.
8796 * config/arc/elf.h: Likewise.
8797 * config/arm/uclinux-elf.h: Likewise.
8798 * config/arm/unknown-elf.h: Likewise.
8799 * config/avr/avrlibc.h: Likewise.
8800 * config/bfin/bfin.h: Likewise.
8801 * config/bfin/linux.h: Likewise.
8802 * config/bfin/uclinux.h: Likewise.
8803 * config/darwin.h: Likewise.
8804 * config/darwin10.h: Likewise.
8805 * config/darwin12.h: Likewise.
8806 * config/gnu-user.h: Likewise.
8807 * config/lm32/uclinux-elf.h: Likewise.
8808 * config/pa/pa-hpux11.h: Likewise.
8809 * config/pa/pa64-hpux.h: Likewise.
8810 * config/sparc/sparc.h: Likewise.
8812 2018-07-31 Olivier Hainque <hainque@adacore.com>
8814 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
8815 undefined variables.
8817 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
8820 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
8823 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
8825 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
8826 terminated string literal.
8828 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
8830 PR rtl-optimization/85160
8831 * combine.c (is_just_move): New function.
8832 (try_combine): Allow combining two instructions into two if neither of
8833 the original instructions was a move.
8835 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
8838 * doc/extend.texi (Global Register Variables): Discourage use of type
8840 (Local Register Variables): Likewise.
8842 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
8844 PR tree-optimization/86506
8845 * hwint.h (ceil_log2): Resync with hwint.c implementation.
8847 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
8850 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
8851 hard_regno, make sure no insn between `from` and `to` clobbers it.
8853 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
8854 Tom de Vries <tdevries@suse.de>
8856 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
8857 (PTX_DEFAULT_RUNTIME_DIM): ... this.
8858 (nvptx_goacc_validate_dims): Set default worker and gang dims to
8859 PTX_DEFAULT_RUNTIME_DIM.
8860 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
8862 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
8864 * config/pa/pa.c (pa_output_addr_vec): Align address table.
8865 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
8866 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
8868 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
8870 * config/rs6000/constraints.md (wG constraint): Delete, no longer
8872 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
8873 predicate to reflect toc fusion has been deleted.
8874 (toc_fusion_mem_raw): Delete, no longer used.
8875 (toc_fusion_mem_wrapped): Likewise.
8876 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
8878 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
8879 Delete, no longer used.
8880 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
8881 meant to be used for toc fusion.
8882 (rs6000_debug_print_mode): Delete toc fusion debugging.
8883 (rs6000_debug_reg_global): Likewise.
8884 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
8885 fusion and secondary reload support that were never used.
8886 (rs6000_option_override_internal): Delete TOC fusion, that was only
8887 partially defined, and it did not work unless you also used the
8889 (rs6000_legitimate_address_p): Delete TOC fusion support.
8890 (rs6000_opt_masks): Likewise.
8891 (fusion_wrap_memory_address): Delete function, no longer used.
8892 (fusion_split_address); Delete TOC fusion support.
8893 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
8894 longer used with toc fusion being deleted.
8895 (TARGET_TOC_FUSION_FP): Likewise.
8896 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
8898 (toc fusion spliter): Delete TOC fusion support.
8899 (toc_fusionload_<mode>): Likewise.
8900 (toc_fusionload_di): Likewise.
8901 (fusion_gpr_load_<mode>): Delete generator function, this insn no
8902 longer needs to be named. Rename predicate to delete TOC fusion.
8903 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
8904 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
8905 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
8906 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
8907 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
8909 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
8911 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
8912 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
8913 __int128 in built-in function prototypes.
8914 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
8915 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
8917 2018-07-27 Martin Sebor <msebor@redhat.com>
8919 PR tree-optimization/86696
8920 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
8921 types, including enums.
8922 (handle_char_store): Be prepared for the above function to fail.
8924 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
8926 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
8927 when optimization level is lower than 2 or optimize for size.
8929 2018-07-26 Martin Sebor <msebor@redhat.com>
8931 PR tree-optimization/86043
8932 PR tree-optimization/86042
8933 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
8934 (get_string_cst_length): Rename...
8935 (get_min_string_length): ...to this. Add argument.
8936 (handle_char_store): Extend to handle multi-character stores by
8938 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
8939 * tree.h (initializer_zerop): Add argument.
8941 2018-07-26 Jakub Jelinek <jakub@redhat.com>
8944 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
8945 declare target to variables if they have always,{to,from,tofrom} map
8948 2018-07-26 Martin Liska <mliska@suse.cz>
8951 * lto-wrapper.c: Add linker_output as prefix
8952 for ltrans_output_file.
8954 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
8956 PR rtl-optimization/85805
8957 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
8958 value for hard registers if that was written in the same mode.
8960 2018-07-26 Martin Liska <mliska@suse.cz>
8962 PR gcov-profile/86536
8963 * gcov.c (format_gcov): Use printf format %.*f directly
8964 and do not handle special values.
8966 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
8968 * common/config/arc/arc-common.c (arc_option_optimization_table):
8969 Update default optimizations for size.
8971 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
8973 * config/arc/arc.md (movsf_insn): Add short instruction selection.
8974 * config/arc/constraints.md (CfZ): New constraint.
8975 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
8976 (subsf3_fpu): Likewise.
8977 (cmpsf_fpu): Likewise.
8978 (cmpsf_fpu_uneq): Likewise.
8980 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
8982 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
8984 (arc_is_uncached_mem_p): uncached applies to both the variable and
8987 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
8989 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
8992 2018-07-25 David Malcolm <dmalcolm@redhat.com>
8994 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
8995 field "m_scopes" from vec to auto_vec.
8997 2018-07-25 Martin Liska <mliska@suse.cz>
8999 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
9002 2018-07-25 Richard Biener <rguenther@suse.de>
9005 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
9006 special wrt context_die late.
9007 (gen_subprogram_die): Re-use DIEs in local scope.
9009 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
9011 PR tree-optimization/86644
9012 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
9014 2018-07-25 Martin Liska <mliska@suse.cz>
9017 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
9018 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
9020 2018-07-25 Martin Liska <mliska@suse.cz>
9023 * params.def: Explain ASan abbreviation and provide
9024 a documentation link.
9026 2018-07-24 Martin Sebor <msebor@redhat.com>
9028 PR tree-optimization/86622
9029 PR tree-optimization/86532
9030 * builtins.h (string_length): Declare.
9031 * builtins.c (c_strlen): Correct handling of non-constant offsets.
9032 (check_access): Be prepared for non-constant length ranges.
9033 (string_length): Make extern.
9034 * expr.c (string_constant): Only handle the minor non-constant
9035 array index. Use string_constant to compute the length of
9036 a generic string constant.
9038 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
9040 PR tree-optimization/86618
9041 * tree-vect-stmts.c (vectorizable_call): Don't take the address
9042 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
9044 2018-07-24 David Malcolm <dmalcolm@redhat.com>
9046 PR tree-optimization/86636
9047 * json.cc (json::object::set): Fix comment. Add assertions.
9048 (json::array::append): Move here from json.h. Add comment and an
9050 (json::string::string): Likewise.
9051 * json.h (json::array::append): Move to json.cc.
9052 (json::string::string): Likewise.
9053 * optinfo-emit-json.cc
9054 (optrecord_json_writer::impl_location_to_json): Assert that we
9055 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
9056 wrapper around it. Expand the location once, rather than three
9058 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
9059 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
9061 (optrecord_json_writer::optinfo_to_json): Likewise, in four
9062 places. Fix some overlong lines.
9064 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
9066 * config/aarch64/aarch64-simd.md
9067 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
9068 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
9069 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
9070 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
9071 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
9072 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
9073 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
9074 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
9075 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
9077 2018-07-24 Jakub Jelinek <jakub@redhat.com>
9080 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
9081 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
9082 and abs_d == d, do the power of two handling if profitable.
9084 2018-07-24 Richard Biener <rguenther@suse.de>
9086 * match.pd: Add BIT_FIELD_REF canonicalizations.
9088 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9091 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
9093 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9095 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
9096 terminated STRING_CST object.
9098 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9100 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
9102 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
9104 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
9105 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
9106 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
9107 the elements into a register.
9108 (rs6000_split_v4si_init_di_reg): Delete.
9109 (rs6000_split_v4si_init): Delete.
9110 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
9111 (vsx_init_v4si): Rewrite as a define_expand.
9113 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
9115 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
9116 zero_extend argument from memory): New.
9118 2018-07-22 Martin Sebor <msebor@redhat.com>
9121 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
9122 diagnosing calls with unknown arguments unless -Walloca-larger-than
9123 is restricted to less than PTRDIFF_MAX bytes.
9125 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
9127 * doc/gcov.texi (Invoking Gcov): Editorial changes.
9129 2018-07-20 David Malcolm <dmalcolm@redhat.com>
9131 * pretty-print.c (text_info::set_location): Remove redundant
9132 "line_table" parameter from call to rich_location::set_range.
9134 2018-07-20 Martin Sebor <msebor@redhat.com>
9137 * builtins.c (expand_builtin_alloca): Adjust.
9138 * calls.c (alloc_max_size): Simplify.
9139 * cgraphunit.c (cgraph_node::expand): Adjust.
9140 * common.opt (larger_than_size, warn_frame_larger_than): Remove
9142 (frame_larger_than_size): Same.
9143 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
9144 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
9146 * doc/invoke.texi (GCC Command Options): Document option arguments.
9147 Explain byte-size arguments and suffixes.
9148 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
9149 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
9150 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
9151 * doc/options.texi (UInteger): Expand.
9152 (Host_Wide_Int, ByteSize): Document new properties.
9153 * final.c (final_start_function_1): Include sizes in an error message.
9154 * function.c (frame_offset_overflow): Same.
9155 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
9156 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
9157 Diagnose unbounded alloca calls only for limits of less than
9159 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
9160 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
9162 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
9163 only for limits of less than PTRDIFF_MAX.
9164 * langhooks-def.h (lhd_handle_option): Change function argument
9166 * langhooks.c (lhd_handle_option): Same.
9167 * langhooks.h (handle_option): Same.
9168 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
9170 (var_type, var_type_struct): Same.
9171 (var_set): Handle ByteSize flag.
9172 * optc-gen.awk: Add comments to output to ease debugging. Make
9173 use of HOST_WIDE_INT where appropriate.
9174 * opts-gen-save.awk: Use %lx to format unsigned long.
9175 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
9176 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
9177 arguments. Parse bytes-size suffixes.
9178 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
9179 (enum_value_to_arg): Same.
9180 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
9181 (handle_option): Adjust.
9182 (generate_option): Change function argument to HOST_WIDE_INT.
9183 (cmdline_handle_error): Adjust.
9184 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
9185 (set_option): Change function argument to HOST_WIDE_INT.
9186 (option_enabled): Handle cl_host_wide_int.
9187 (get_option_state): Handle CLVC_SIZE.
9188 (control_warning_option): Same.
9189 * opts.c (common_handle_option): Change function argument to
9190 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
9191 OPT_Wvla_larger_than_.
9192 * opts.h (enum cl_var_type): Add an enumerator.
9193 * stor-layout.c (layout_decl): Print a more meaningful warning.
9194 * toplev.c (output_stack_usage): Adjust.
9196 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
9198 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
9199 call to inline_expand_builtin_string_cmp.
9200 (expand_builtin_strcmp): Likewise.
9201 (expand_builtin_strncmp): Likewise.
9202 (inline_string_cmp): Delete the last parameter, change char_type_node
9203 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
9205 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
9206 the inlining expansion on target where the type of the call has same or
9207 narrower precision than unsigned char.
9209 2018-07-20 David Malcolm <dmalcolm@redhat.com>
9211 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
9212 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
9213 * common.opt (fsave-optimization-record): New option.
9214 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
9215 * doc/invoke.texi (-fsave-optimization-record): New option.
9216 * dumpfile.c: Include "optinfo-emit-json.h".
9217 (struct kv_pair): Move to coretypes.h.
9218 (optgroup_options): Make non-static.
9219 (dump_context::end_scope): Call
9220 optimization_records_maybe_pop_dump_scope.
9221 * dumpfile.h (optgroup_options): New decl.
9222 * json.cc: New file.
9224 * optinfo-emit-json.cc: New file.
9225 * optinfo-emit-json.h: New file.
9226 * optinfo.cc: Include "optinfo-emit-json.h".
9227 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
9228 (optinfo_enabled_p): Check optimization_records_enabled_p.
9229 (optinfo_wants_inlining_info_p): Likewise.
9230 * optinfo.h: Update comment.
9231 * profile-count.c (profile_quality_as_string): New function.
9232 * profile-count.h (profile_quality_as_string): New decl.
9233 (profile_count::quality): New accessor.
9234 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
9235 and optinfo_emit_json_cc_tests.
9236 * selftest.h (selftest::json_cc_tests): New decl.
9237 (selftest::optinfo_emit_json_cc_tests): New decl.
9238 * toplev.c: Include "optinfo-emit-json.h".
9239 (compile_file): Call optimization_records_finish.
9240 (do_compile): Call optimization_records_start.
9241 * tree-ssa-live.c: Include optinfo.h.
9242 (remove_unused_scope_block_p): Retain inlining information if
9243 optinfo_wants_inlining_info_p returns true.
9245 2018-07-20 Richard Biener <rguenther@suse.de>
9248 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
9249 to cover -flto-partition=none.
9251 2018-07-20 Martin Liska <mliska@suse.cz>
9253 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
9254 (get_decl_source_range): Remove unused function.
9256 2018-07-20 Richard Biener <rguenther@suse.de>
9258 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
9259 (struct vn_phi_s): Likewise.
9260 (struct vn_reference_s): Likewise.
9261 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
9262 for searching the slot of an entry known to be in the hash itself.
9263 (vn_phi_hasher::equal): Likewise.
9264 (vn_reference_hasher::equal): Likewise.
9265 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
9267 (optimistic_info, current_info): Remove, keeping only valid_info.
9268 (vn_reference_lookup_1): Remove fallback lookup.
9269 (vn_reference_lookup_2): Likewise.
9270 (vn_nary_op_lookup_1): Likewise.
9271 (vn_phi_lookup): Likewise.
9272 (vn_nary_build_or_lookup_1): Make sure to not chain the built
9274 (vn_reference_insert): Adjust, chain the inserted hash element
9275 at last_inserted_ref.
9276 (vn_reference_insert_pieces): Likewise.
9277 (visit_reference_op_call): Likewise.
9278 (vn_nary_op_insert_into): Chain the inserted hash element at
9280 (vn_nary_op_insert_pieces): Adjust.
9281 (vn_nary_op_insert): Likewise.
9282 (vn_nary_op_insert_stmt): Likewise.
9283 (vn_phi_insert): Adjust, chain the inserted hash element at
9285 (process_scc): Remove clearing and copying the optimistic
9286 table. Instead remove elements inserted during an optimistic
9287 iteration from the single table we maintain.
9288 (init_scc_vn): Adjust.
9289 (free_scc_vn): Likewise.
9290 (sccvn_dom_walker::record_cond): Likewise.
9291 (sccvn_dom_walker::after_dom_children): Likewise.
9293 2018-07-19 Martin Sebor <msebor@redhat.com>
9295 PR tree-optimization/84047
9296 PR tree-optimization/83776
9297 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
9298 (check_array_bounds): Call it.
9300 2018-07-19 Martin Sebor <msebor@redhat.com>
9302 * align.h (align_flags): Use member initialization.
9304 2018-07-19 David Malcolm <dmalcolm@redhat.com>
9306 * Makefile.in (OBJS): Add optinfo.o.
9307 * coretypes.h (class symtab_node): New forward decl.
9308 (struct cgraph_node): New forward decl.
9309 (class varpool_node): New forward decl.
9310 * dump-context.h: New file.
9311 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
9313 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
9314 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
9315 (set_alt_dump_file): Likewise.
9316 (dump_context::~dump_context): New dtor.
9317 (dump_gimple_stmt): Move implementation to...
9318 (dump_context::dump_gimple_stmt): ...this new member function.
9319 Add the stmt to any pending optinfo, creating one if need be.
9320 (dump_gimple_stmt_loc): Move implementation to...
9321 (dump_context::dump_gimple_stmt_loc): ...this new member function.
9322 Start a new optinfo and add the stmt to it.
9323 (dump_gimple_expr): Move implementation to...
9324 (dump_context::dump_gimple_expr): ...this new member function.
9325 Add the stmt to any pending optinfo, creating one if need be.
9326 (dump_gimple_expr_loc): Move implementation to...
9327 (dump_context::dump_gimple_expr_loc): ...this new member function.
9328 Start a new optinfo and add the stmt to it.
9329 (dump_generic_expr): Move implementation to...
9330 (dump_context::dump_generic_expr): ...this new member function.
9331 Add the tree to any pending optinfo, creating one if need be.
9332 (dump_generic_expr_loc): Move implementation to...
9333 (dump_context::dump_generic_expr_loc): ...this new member
9334 function. Add the tree to any pending optinfo, creating one if
9336 (dump_printf): Move implementation to...
9337 (dump_context::dump_printf_va): ...this new member function. Add
9338 the text to any pending optinfo, creating one if need be.
9339 (dump_printf_loc): Move implementation to...
9340 (dump_context::dump_printf_loc_va): ...this new member function.
9341 Start a new optinfo and add the stmt to it.
9342 (dump_dec): Move implementation to...
9343 (dump_context::dump_dec): ...this new member function. Add the
9344 value to any pending optinfo, creating one if need be.
9345 (dump_context::dump_symtab_node): New member function.
9346 (dump_context::get_scope_depth): New member function.
9347 (dump_context::begin_scope): New member function.
9348 (dump_context::end_scope): New member function.
9349 (dump_context::ensure_pending_optinfo): New member function.
9350 (dump_context::begin_next_optinfo): New member function.
9351 (dump_context::end_any_optinfo): New member function.
9352 (dump_context::s_current): New global.
9353 (dump_context::s_default): New global.
9354 (dump_scope_depth): Delete global.
9355 (dumpfile_ensure_any_optinfo_are_flushed): New function.
9356 (dump_symtab_node): New function.
9357 (get_dump_scope_depth): Reimplement in terms of dump_context.
9358 (dump_begin_scope): Likewise.
9359 (dump_end_scope): Likewise.
9360 (selftest::temp_dump_context::temp_dump_context): New ctor.
9361 (selftest::temp_dump_context::~temp_dump_context): New dtor.
9362 (selftest::verify_item): New function.
9363 (ASSERT_IS_TEXT): New macro.
9364 (ASSERT_IS_TREE): New macro.
9365 (ASSERT_IS_GIMPLE): New macro.
9366 (selftest::test_capture_of_dump_calls): New test.
9367 (selftest::dumpfile_c_tests): Call it.
9368 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
9369 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
9370 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
9371 descriptive comment.
9372 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
9373 (dump_node, dump_bb): Move these unrelated decls.
9374 (class dump_manager): Add leading comment.
9375 * optinfo.cc: New file.
9376 * optinfo.h: New file.
9378 2018-07-19 Michael Collison <michael.collison@arm.com>
9379 Richard Henderson <rth@redhat.com>
9381 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
9382 (subti): Handle op1 zero.
9383 (subvti4, usub4ti4): New.
9384 (*sub<GPI>3_compare1_imm): New.
9385 (sub<GPI>3_carryinCV): New.
9386 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
9387 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
9389 2018-07-19 Michael Collison <michael.collison@arm.com>
9390 Richard Henderson <rth@redhat.com>
9392 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
9393 (addti3): Create simpler code if low part is already known to be 0.
9394 (addvti4, uaddvti4): New.
9395 (*add<GPI>3_compareC_cconly_imm): New.
9396 (*add<GPI>3_compareC_cconly): New.
9397 (*add<GPI>3_compareC_imm): New.
9398 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
9399 handle constants within this pattern..
9400 (*add<GPI>3_compareV_cconly_imm): New.
9401 (*add<GPI>3_compareV_cconly): New.
9402 (*add<GPI>3_compareV_imm): New.
9403 (add<GPI>3_compareV): New.
9404 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
9405 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
9406 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
9407 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
9408 with 'comparison' operator.
9409 (*add<GPI>3_compareV_cconly_imm): Ditto.
9410 (*add<GPI>3_compareV_cconly): Ditto.
9411 (*add<GPI>3_compareV_imm): Ditto.
9412 (add<GPI>3_compareV): Ditto.
9413 (add<mode>3_carryinC): Ditto.
9414 (*add<mode>3_carryinC_zero): Ditto.
9415 (*add<mode>3_carryinC): Ditto.
9416 (add<mode>3_carryinV): Ditto.
9417 (*add<mode>3_carryinV_zero): Ditto.
9418 (*add<mode>3_carryinV): Ditto.
9420 2018-07-19 Michael Collison <michael.collison@arm.com>
9421 Richard Henderson <rth@redhat.com>
9423 * config/aarch64/aarch64-modes.def (CC_V): New.
9424 * config/aarch64/aarch64-protos.h
9425 (aarch64_addti_scratch_regs): Declare
9426 (aarch64_subvti_scratch_regs): Declare.
9427 (aarch64_expand_subvti): Declare.
9428 (aarch64_gen_unlikely_cbranch): Declare
9429 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
9430 for signed overflow using CC_Vmode.
9431 (aarch64_get_condition_code_1): Handle CC_Vmode.
9432 (aarch64_gen_unlikely_cbranch): New function.
9433 (aarch64_addti_scratch_regs): New function.
9434 (aarch64_subvti_scratch_regs): New function.
9435 (aarch64_expand_subvti): New function.
9437 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9439 * config/aarch64/aarch64-option-extensions.def: New entry for profile
9441 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
9442 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
9445 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9448 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
9449 address check not strict.
9451 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9453 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
9455 (store_pair_lanes<mode>): Likewise.
9456 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
9457 enum value 'ADDR_QUERY_LDP_STP_N'.
9458 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
9459 (aarch64_print_address_internal): Add declaration.
9460 (aarch64_print_ldpstp_address): Remove.
9461 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
9462 (aarch64_print_operand): Change printing of 'y'.
9463 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
9464 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
9465 'true' rather than '1'.
9466 * config/aarch64/constraints.md (Uml): Likewise.
9467 (Uml): Rename to Umn.
9470 2018-07-19 Richard Biener <rguenther@suse.de>
9472 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
9474 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
9475 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
9476 (vn_reference_hasher): Likewise.
9477 (struct vn_tables_s): Remove obstack and alloc-pool members.
9478 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
9479 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
9480 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
9481 (vn_reference_insert_pieces): Likewise.
9482 (alloc_vn_nary_op_noinit): Adjust.
9483 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
9484 (vn_phi_eq): Adjust.
9485 (shared_lookup_phiargs): Remove.
9486 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
9487 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
9488 (visit_reference_op_call): Likewise.
9489 (copy_nary, copy_phi, copy_reference): Remove.
9490 (process_scc): Rewind the obstack when iterating. Do not
9491 copy the elements to valid_info but just move them from one
9492 hashtable to the other.
9493 (allocate_vn_table): Adjust.
9494 (free_vn_table): Likewise.
9495 (init_scc_vn): Likewise.
9496 (free_scc_vn): Likewise.
9498 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
9501 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
9502 indirect_return as function type attribute.
9503 (ix86_attribute_table): Change indirect_return to function
9505 * doc/extend.texi: Update indirect_return attribute.
9507 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
9509 * wide-int.h (widest2_int): New.
9510 * gimple-fold.c (arith_overflowed_p): Use it.
9511 * tree.h (widest2_int_cst): New.
9512 * tree-vrp.c (wide_int_binop_overflow): Rename from
9513 vrp_int_const_binop.
9514 Rewrite to work on trees.
9515 (extract_range_from_multiplicative_op_1): Abstract code to...
9516 (wide_int_range_min_max): ...here.
9517 (wide_int_range_cross_product): ...and here.
9518 (extract_range_from_binary_expr_1): Abstract overflow code to...
9519 (wide_int_range_mult_wrapping): ...here.
9520 * tree-vrp.h (wide_int_range_cross_product): New.
9521 (wide_int_range_mult_wrapping): New.
9523 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
9524 Julia Koval <julia.koval@intel.com>
9526 * config/i386/x86-tune-costs.h (skylake_memcpy,
9527 skylake_memset): Replace rep_prefix with unrolling for size 512.
9529 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
9532 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
9533 comparision with EQ_EXPR in last stmt.
9535 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
9537 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
9538 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
9539 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
9540 previously known as "PowerPC AltiVec Built-in Functions". Move
9541 some material to new subsubsections "PowerPC AltiVec Built-in
9542 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
9544 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
9545 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
9546 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
9547 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
9549 2018-07-18 Richard Biener <rguenther@suse.de>
9551 PR tree-optimization/86557
9552 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
9555 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
9557 * config/s390/s390.c (s390_function_profiler): Generate CFI.
9559 2018-07-17 Jeff Law <law@redhat.com>
9561 * config/arm/arm.c (get_label_padding): Update for recent
9562 changes to label_to_alignment.
9564 PR tree-optimization/86010
9565 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
9567 * config/mips/mips.c (vr4130_align_insns): Update for recent
9568 changes to label_to_alignment.
9570 * config/frv/frv.c (frv_label_align): Update for recent changes
9571 to label_to_alignment.
9573 * config/nios2/nios2.c (nios2_label_align): Update for recent
9574 changes which dropped ALIGN_LABELS_LOG.
9576 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
9578 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
9579 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
9581 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9583 * config/arc/arc.c (arc_label_align): Use align_labels instead of
9584 deprecated align_labels_log.
9586 2018-07-17 Richard Biener <rguenther@suse.de>
9589 * dwarf2out.c (init_sections_and_labels): Always generate
9590 a debug_line_str_section for early LTO debug.
9591 (dwarf2out_finish): Reset debug_line_str_hash output early.
9592 Bump counter for extra dwarf5 .debug_loc labels to not conflict
9594 (dwarf2out_early_finish): Output debug_line_str.
9596 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
9598 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
9599 index register on z196 or later.
9601 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
9603 * config/s390/s390.c (s390_default_align): Set default function
9605 (s390_override_options_after_change): Call s390_default align.
9606 (s390_option_override_internal): Call s390_default_align.
9607 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
9609 2018-07-17 Jakub Jelinek <jakub@redhat.com>
9612 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
9613 to _looptemp_ clauses, other than the first two.
9615 2018-07-17 Martin Liska <mliska@suse.cz>
9617 * opts.c: Do not enable OPT_falign_* for -Os.
9619 2018-07-17 Martin Liska <mliska@suse.cz>
9621 * align.h (MAX_CODE_ALIGN): New.
9622 (MAX_CODE_ALIGN_VALUE): New.
9623 * common/config/i386/i386-common.c (ix86_handle_option):
9624 (MAX_CODE_ALIGN): Moved to align.h.
9625 * final.c (MAX_CODE_ALIGN): Likewise.
9626 * opts.c (parse_and_check_align_values):
9627 (MAX_CODE_ALIGN): Likewise.
9628 (MAX_CODE_ALIGN_VALUE): Likewise.
9630 2018-07-17 Martin Liska <mliska@suse.cz>
9632 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
9633 in order to fulfil coding style.
9634 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
9635 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9636 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
9637 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
9638 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
9639 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
9640 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
9641 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
9642 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9644 2018-07-17 Martin Liska <mliska@suse.cz>
9646 * align.h: New file.
9647 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
9649 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
9650 align_flags of label_to_alignment.
9651 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
9653 * config/m68k/m68k.c: Do not use removed align_labels_value and
9655 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
9656 (LOOP_ALIGN): Likewise.
9657 (LABEL_ALIGN): Likewise.
9658 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
9659 Remove not used macro.
9660 (rs6000_loop_align): Change return type to align_flags.
9661 (rs6000_loop_align_max_skip): Remove.
9662 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
9663 Change return type to align_flags.
9664 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
9665 Remove not used macro.
9666 (rs6000_loop_align): Change return type to align_flags.
9667 (rs6000_loop_align_max_skip): Remove.
9668 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
9669 * config/rx/rx-protos.h (rx_align_for_label): Make it
9671 * config/rx/rx.c (rx_align_for_label): Change return type
9673 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
9675 into align_flags class.
9676 (LABEL_ALIGN): Likewise.
9677 (LOOP_ALIGN): Likewise.
9678 * config/s390/s390.c (s390_label_align): Use align_flags
9680 (s390_asm_output_function_label): Likewise.
9681 * config/sh/sh.c (sh_override_options_after_change):
9682 Use align_flags class directly without macros.
9683 (find_barrier): Likewise.
9684 (barrier_align): Likewise.
9685 (sh_loop_align): Likewise.
9686 * config/spu/spu.c (spu_option_override):
9687 Use align_flags_tuple::get_value instead of removed macros.
9688 (spu_sched_init): Likewise.
9689 * config/spu/spu.h (GTY): Likewise.
9690 * config/visium/visium.c (visium_option_override):
9691 Set "8" as default secondary alignment.
9692 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
9693 in order to guarantee secondary alignment of 8.
9694 * coretypes.h: Include align.h header file.
9695 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
9696 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
9697 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
9698 * doc/tm.texi.in: Likewise.
9699 * final.c (struct label_alignment): Remove not used structure.
9700 (LABEL_ALIGN): Change type to align_flags.
9701 (LOOP_ALIGN): Likewise.
9702 (JUMP_ALIGN): Likewise.
9703 (default_loop_align_max_skip): Remove.
9704 (default_label_align_max_skip): Likewise.
9705 (default_jump_align_max_skip): Likewise.
9706 (default_label_align_after_barrier_max_skip):
9707 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
9708 (LABEL_TO_MAX_SKIP): Remove.
9709 (label_to_alignment): Return align_flags type instead of integer.
9710 (label_to_max_skip): Remove.
9711 (align_fuzz): Use align_flags type.
9712 (compute_alignments): Use align_flags type and use align_flags::max
9713 to combine multiple alignments.
9714 (grow_label_align): Grow vec instead of C array.
9715 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
9716 (shorten_branches): Use align_flags type and use align_flags::max
9717 to combine multiple alignments.
9718 (final_scan_insn_1): Remove usage of secondary alignment that comes
9719 from label alignment, but instead use proper secondary alignment
9720 which is computed in grow_label_align.
9721 * flags.h (struct align_flags_tuple): Move to align.h.
9722 (struct align_flags): Likewise.
9723 (state_align_loops): Rename to align_loops.
9724 (state_align_jumps): Rename to align_jumps.
9725 (state_align_labels): Rename to align_labels.
9726 (state_align_functions): Rename to align_functions.
9727 (align_loops_log): Remove.
9728 (align_jumps_log): Remove.
9729 (align_labels_log): Remove.
9730 (align_functions_log): Remove.
9731 (align_loops_max_skip): Remove.
9732 (align_jumps_max_skip): Remove.
9733 (align_labels_max_skip): Remove.
9734 (align_functions_max_skip): Remove.
9735 (align_loops_value): Remove.
9736 (align_jumps_value): Remove.
9737 (align_labels_value): Remove.
9738 (align_functions_value): Remove.
9739 * output.h (label_to_alignment): Change return type to align_flags.
9740 (label_to_max_skip): Remove.
9741 * target.def: Remove loop_align_max_skip, label_align_max_skip,
9742 jump_align_max_skip macros.
9743 * targhooks.h (default_loop_align_max_skip): Remove.
9744 (default_label_align_max_skip): Likewise.
9745 (default_jump_align_max_skip): Likewise.
9746 (default_label_align_after_barrier_max_skip): Remove.
9747 * toplev.c (read_log_maxskip): Use ::normalize function.
9748 (parse_N_M): Remove not used argument and also call ::normalize.
9749 (parse_alignment_opts): Do not pass unused arguments.
9750 * varasm.c (assemble_start_function): Use directly align_functions
9751 instead of removed macros.
9752 * system.h: Do not poison removed macros.
9754 2018-07-17 Jakub Jelinek <jakub@redhat.com>
9757 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
9758 and cond temporaries don't have reference type if iterator has
9759 pointer type. For init use &for_pre_body instead of pre_p if
9760 for_pre_body is non-empty.
9762 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
9764 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
9765 double-double modes to SFmode directly directly.
9766 (trunc<mode>sf2_fprs): Delete.
9768 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
9770 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
9771 for conversions between IFmode and the decimal floating point modes.
9772 (init_float128_ieee): Use the correct names for conversions between
9773 KFmode and the decimal floating point modes.
9775 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
9777 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
9778 for the conversions between TDmode and IFmode.
9779 (init_float128_ieee): Use more correct names for the conversions
9780 between TDmode and KFmode.
9782 2018-07-16 Jakub Jelinek <jakub@redhat.com>
9784 PR tree-optimization/86526
9785 * builtins.c (expand_builtin_memcmp): Formatting fixes.
9786 (inline_expand_builtin_string_cmp): Likewise.
9787 (inline_string_cmp): Likewise. Use c_readstr instead of
9788 builtin_memcpy_read_str. Add unit_mode temporary.
9790 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
9793 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
9794 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
9796 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
9798 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
9799 Alphabetize prototypes of built-in functions, separating out
9800 built-in functions that are listed in this section but should be
9801 described elsewhere.
9803 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
9806 * expmed.c (emit_store_flag): Do not emit setcc followed by a
9807 conditional move when trapping comparison was split to a
9808 non-trapping one (and vice versa).
9810 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
9812 * config/s390/s390.c (s390_function_profiler): Generate nops
9813 instead of profiler call sequences.
9814 * config/s390/s390.opt: Add the new option.
9816 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
9818 * config/s390/s390.c (s390_function_profiler): Generate
9819 __mcount_loc section.
9820 * config/s390/s390.opt: Add the new option.
9822 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
9824 * common.opt: Add the new warning.
9825 * config/s390/s390.c (s390_function_profiler): Emit "brasl
9826 %r0,__fentry__" when -mfentry is specified.
9827 (s390_option_override_internal): Disallow -mfentry for 31-bit
9829 * config/s390/s390.opt: Add the new option.
9831 2018-07-16 Richard Biener <rguenther@suse.de>
9834 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
9835 for function-local FUNCTION_DECL and RESULT_DECL immediately.
9837 2018-07-16 Martin Liska <mliska@suse.cz>
9840 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
9843 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
9845 * config/arc/arcHS.md: Update ARCHS scheduling rules.
9847 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
9849 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
9851 * config/arc/arc-cpus.def (hs4x): New cpu.
9853 * config/arc/arc-tables.opt: Regenerate.
9854 * config/arc/arc.c (arc_sched_issue_rate): New function.
9855 (TARGET_SCHED_ISSUE_RATE): Define.
9856 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
9857 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
9859 (attr tune): Add ARCHS4x tune values.
9860 (attr tune_dspmpy): Define.
9861 (*tst): Correct instruction type.
9862 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
9863 * config/arc/arcHS4x.md: New file.
9864 * config/arc/fpu.md: Update instruction type attributes.
9865 * config/arc/t-multilib: Regenerate.
9867 2018-07-16 Tom de Vries <tdevries@suse.de>
9870 * var-tracking.c (vt_initialize): Fix pre_dec handling.
9872 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9874 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
9877 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
9879 PR tree-optimization/86514
9880 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
9881 conversion to a boolean type from a type with greater precision.
9883 2018-07-16 Tom de Vries <tdevries@suse.de>
9885 * var-tracking.c (vt_initialize): Print adjusted insn slim if
9886 dump_flags request TDF_SLIM.
9888 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
9890 * fold-const.c (int_const_binop_1): Abstract...
9891 (wide_int_binop): ...wide int code here.
9892 (poly_int_binop): ...poly int code here.
9893 Abstract the rest of int_const_binop_1 into int_const_binop.
9894 * fold-const.h (wide_int_binop): New.
9895 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
9896 Remove useless PLUS/MINUS_EXPR case.
9897 (zero_nonzero_bits_from_vr): Move wide int code...
9898 (zero_nonzero_bits_from_bounds): ...here.
9899 (extract_range_from_binary_expr_1): Move mask optimization code...
9900 (range_easy_mask_min_max): ...here.
9901 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
9902 (range_easy_mask_min_max): New.
9904 2018-07-15 Jeff Law <law@redhat.com>
9907 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
9910 2018-07-14 Jim Wilson <jimw@sifive.com>
9912 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
9914 2018-07-14 Paul Koning <ni1d@arrl.net>
9916 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
9918 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
9920 * lto-streamer-out.c (copy_function_or_variable): Dump info about
9923 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
9924 Steve Munroe <munroesj52@gmail.com>
9926 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
9927 (_mm_andnot_si128): Likewise.
9928 (_mm_or_si128): Likewise.
9929 (_mm_xor_si128): Likewise.
9931 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
9934 * builtins.c (expand_builtin_memcmp): Inline the calls first
9935 when result_eq is false.
9936 (expand_builtin_strcmp): Inline the calls first.
9937 (expand_builtin_strncmp): Likewise.
9938 (inline_string_cmp): New routine. Expand a string compare
9939 call by using a sequence of char comparison.
9940 (inline_expand_builtin_string_cmp): New routine. Inline expansion
9941 a call to str(n)cmp/memcmp.
9942 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
9944 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
9946 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
9948 * config/arm/driver-arm.c: Include arm-native.h.
9949 (host_detect_local_cpu): Use auto-generated data tables.
9950 (vendors, arm_cpu_table): Delete. Move part information to ...
9951 * config/arm/arm-cpus.in: ... here.
9952 * config/arm/parsecpu.awk (gen_native): New function.
9953 (vendor, part): New CPU fields.
9954 (END): Add support for building the native CPU detection tables.
9955 * config/arm/t-arm (arm-native.h): Add build rule.
9956 (driver-arm.o): Add dependency on arm-native.h.
9958 2018-07-13 Richard Biener <rguenther@suse.de>
9961 * match.pd (addr1 - addr2): Allow either of the operand to
9964 2018-07-13 Tom de Vries <tdevries@suse.de>
9966 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
9969 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
9971 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
9972 arrays instead of numbered variables.
9974 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
9976 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
9977 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
9979 2018-07-13 Richard Biener <rguenther@suse.de>
9982 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
9983 instead of get_context_die.
9985 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9986 Richard Biener <rguenther@suse.de>
9989 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
9990 that the loop latch destination where phi is defined.
9992 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
9994 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
9995 (riscv_expand_epilogue): Add assertion to check interrupt mode.
9996 (riscv_set_current_function): Extract getting interrupt type to new
9998 (riscv_get_interrupt_type): New function.
9999 (riscv_merge_decl_attributes): New function, checking interrupt type is
10001 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
10003 2018-07-12 Paul Koning <ni1d@arrl.net>
10005 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
10008 2018-07-12 Paul Koning <ni1d@arrl.net>
10010 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
10011 zero reference, add doloop_end instead.
10012 * doc/md.texi (decrement_and_branch_until_zero): Remove.
10013 (Looping patterns): Remove decrement_and_branch_until_zero. Add
10014 detail for doloop_end.
10016 2018-07-12 Martin Sebor <msebor@redhat.com>
10019 * attribs.c (decl_attributes): Reject conflicting attributes before
10020 calling attribute handlers.
10022 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
10024 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
10026 (gcc::dump_manager::get_dump_file_name): likewise.
10027 (dump_begin): Likewise.
10028 * dumpfile.h (dump_begin): Update prototype.
10029 (gcc::dump_manager::get_dump_file_name,
10030 gcc::dump_manager::get_dump_file_name): Update prototype.
10032 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10034 * internal-fn.h (vectorizable_internal_fn_p): New function.
10035 * tree-vect-slp.c (compatible_calls_p): Likewise.
10036 (vect_build_slp_tree_1): Remove nops argument. Handle calls
10037 to internal functions.
10038 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
10040 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10042 * fold-const.h (inverse_conditions_p): Declare.
10043 * fold-const.c (inverse_conditions_p): New function.
10044 * match.pd: Use inverse_conditions_p. Add folds of view_converts
10045 that test the inverse condition of a conditional internal function.
10046 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
10047 * internal-fn.c (internal_fn_mask_index): Handle conditional
10048 internal functions.
10049 (vectorized_internal_fn_supported_p): New function.
10050 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
10051 (any_pred_load_store): Replace with...
10052 (need_to_predicate): ...this new variable.
10053 (redundant_ssa_names): New variable.
10054 (ifcvt_can_use_mask_load_store): Move initial checks to...
10055 (ifcvt_can_predicate): ...this new function. Handle tree codes
10056 for which a conditional internal function exists.
10057 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
10058 instead of ifcvt_can_use_mask_load_store. Update after variable
10060 (predicate_load_or_store): New function, split out from
10061 predicate_mem_writes.
10062 (check_redundant_cond_expr): New function.
10063 (value_available_p): Likewise.
10064 (predicate_rhs_code): Likewise.
10065 (predicate_mem_writes): Rename to...
10066 (predicate_statements): ...this. Use predicate_load_or_store
10067 and predicate_rhs_code.
10068 (combine_blocks, tree_if_conversion): Update after above name changes.
10069 (ifcvt_local_dce): Handle redundant_ssa_names.
10070 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
10071 general conditional functions.
10072 * tree-vect-stmts.c (vectorizable_call): Likewise.
10074 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10075 Alan Hayward <alan.hayward@arm.com>
10076 David Sherwood <david.sherwood@arm.com>
10078 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
10079 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
10080 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
10081 plus and minus and convert them into IFN_COND_FMA-based sequences.
10082 (convert_mult_to_fma): Handle conditional plus and minus.
10084 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10086 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
10087 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
10088 (cond_fnms_optab): New optabs.
10089 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
10090 internal functions.
10091 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
10092 * internal-fn.h (get_conditional_internal_fn): Declare.
10093 (get_unconditional_internal_fn): Likewise.
10094 * internal-fn.c (cond_ternary_direct): New macro.
10095 (expand_cond_ternary_optab_fn): Likewise.
10096 (direct_cond_ternary_optab_supported_p): Likewise.
10097 (FOR_EACH_COND_FN_PAIR): Likewise.
10098 (get_conditional_internal_fn): New function.
10099 (get_unconditional_internal_fn): Likewise.
10100 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
10101 (gimple_match_op::gimple_match_op): Add a new overload for 5
10103 (gimple_match_op::set_op): Likewise.
10104 (gimple_resimplify5): Declare.
10105 * genmatch.c (decision_tree::gen): Generate simplifications for
10107 * gimple-match-head.c (gimple_simplify): Define an overload for
10108 5 operands. Handle calls with 5 arguments in the top-level overload.
10109 (convert_conditional_op): Handle conversions from unconditional
10110 internal functions to conditional ones.
10111 (gimple_resimplify5): New function.
10112 (build_call_internal): Pass a fifth operand.
10113 (maybe_push_res_to_seq): Likewise.
10114 (try_conditional_simplification): Try converting conditional
10115 internal functions to unconditional internal functions.
10116 Handle 3-operand unconditional forms.
10117 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
10118 Define ternary equivalents of the current rules for binary conditional
10119 internal functions.
10120 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
10121 ternary operations.
10122 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
10123 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
10124 (optab): Handle them.
10125 (SVE_COND_FP_TERNARY): New int iterator.
10126 (sve_fmla_op, sve_fmad_op): New int attributes.
10127 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
10128 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
10129 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
10131 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10133 * target.def (preferred_else_value): New target hook.
10134 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
10135 * doc/tm.texi: Regenerate.
10136 * targhooks.h (default_preferred_else_value): Declare.
10137 * targhooks.c (default_preferred_else_value): New function.
10138 * internal-fn.h (conditional_internal_fn_code): Declare.
10139 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
10140 (get_conditional_internal_fn): Use it.
10141 (conditional_internal_fn_code): New function.
10142 * gimple-match.h (gimple_match_cond): New struct.
10143 (gimple_match_op): Add a cond member function.
10144 (gimple_match_op::gimple_match_op): Update all forms to take a
10146 * genmatch.c (expr::gen_transform): Use the same condition as res_op
10147 for the suboperation, but don't specify a particular else_value.
10148 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
10149 (visit_nary_op, visit_reference_op_load): Pass
10150 gimple_match_cond::UNCOND to the gimple_match_op constructor.
10151 * gimple-match-head.c: Include tree-eh.h
10152 (convert_conditional_op): New function.
10153 (maybe_resimplify_conditional_op): Likewise.
10154 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
10155 (gimple_resimplify2): Likewise.
10156 (gimple_resimplify3): Likewise.
10157 (gimple_resimplify4): Likewise.
10158 (maybe_push_res_to_seq): Return null for conditional operations.
10159 (try_conditional_simplification): New function.
10160 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
10162 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
10164 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
10166 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
10168 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
10170 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
10172 (hash_tree): Do not hash DECL_FCONTEXT
10173 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
10174 Do not stream DECL_FCONTEXT.
10175 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
10176 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
10178 2018-07-12 Richard Biener <rguenther@suse.de>
10181 * dwarf2out.c (gen_block_die): Only output blocks when they have
10182 at least one !DECL_IGNORED_P variable.
10184 2018-07-12 Richard Biener <rguenther@suse.de>
10187 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
10188 Remove -mieee-fp handling.
10190 2018-07-12 Richard Biener <rguenther@suse.de>
10192 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
10193 left-over from last patch.
10195 2018-07-12 Jakub Jelinek <jakub@redhat.com>
10197 PR tree-optimization/86492
10198 * gimple-ssa-store-merging.c
10199 (imm_store_chain_info::coalesce_immediate_stores): Call
10200 check_no_overlap even for the merge_overlapping case. Formatting fix.
10202 2018-07-12 Richard Biener <rguenther@suse.de>
10204 PR middle-end/86479
10205 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
10206 move possibly trapping operations into the conditional.
10208 2018-07-12 Richard Biener <rguenther@suse.de>
10210 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
10211 (vn_lookup_simplify_result): Remove recursion limit applied
10213 (vn_nary_build_or_lookup_1): Adjust.
10214 (try_to_simplify): Likewise.
10215 * gimple-match-head.c (gimple_resimplify1): Instead apply one
10217 (gimple_resimplify2): Likewise.
10218 (gimple_resimplify3): Likewise.
10219 (gimple_resimplify4): Likewise.
10221 2018-07-11 Jakub Jelinek <jakub@redhat.com>
10223 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
10224 Use __mmask64 type instead of __mmask8 for __M argument.
10225 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
10226 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
10228 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
10229 __mmask16 for __M argument.
10230 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
10231 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
10232 to __mmask16 instead of __mmask8.
10233 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
10234 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
10235 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
10236 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
10237 instead of __mmask16 for __U argument.
10238 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
10239 __mmask16 instead of __mmask8 for __U argument.
10240 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
10242 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
10244 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
10246 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
10248 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
10250 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
10252 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
10254 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
10256 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
10258 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
10260 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
10261 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
10262 return type as well as __M argument type and all casts from __mmask8
10264 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
10265 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
10266 return type as well as __M argument type and all casts from __mmask8
10268 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
10269 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
10270 return type as well as __M argument type and all casts from __mmask8
10272 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
10273 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
10274 return type as well as __M argument type and all casts from __mmask8
10276 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
10277 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
10280 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
10282 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
10283 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
10286 2018-07-11 Paul Koning <ni1d@arrl.net>
10288 * doc/md.texi (define_subst): Document how multiple occurrences of
10289 the same argument in the replacement pattern are handled.
10291 2018-07-11 Paul Koning <ni1d@arrl.net>
10293 * doc/extend.texi (Common Variable Attributes): Move "mode" into
10294 alphabetical order.
10295 (Common Type Attributes): Add "mode" attribute.
10297 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
10299 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
10300 stream DECL_ORIGINAL_TYPE.
10301 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
10302 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
10303 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
10304 Do not walk original type.
10305 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
10307 (write_ts_decl_non_common_tree_pointers): Do not stream
10309 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
10310 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
10312 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
10314 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
10315 thread twice from the same starting edge.
10317 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
10319 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
10320 * gimple.c (gimple_call_nonnull_result_p): ...here...
10321 (gimple_call_nonnull_arg): ...and here.
10322 * gimple.h (gimple_call_nonnull_result_p): New.
10323 (gimple_call_nonnull_arg): New.
10325 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
10327 * config/arm/arm-cpus.in: Move information from fpu field of each
10328 cpu definition to the isa field.
10329 * config/arm/parsecpu.awk (fpu): Delete match rule.
10330 (gen_comm_data): Don't add bits from the CPU's FPU entry.
10332 2018-07-11 Richard Biener <rguenther@suse.de>
10335 * dwarf2out.c (init_sections_and_labels): Use
10336 output_asm_line_debug_info consistently.
10337 (dwarf2out_early_finish): Likewise.
10338 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
10341 2018-07-11 Richard Biener <rguenther@suse.de>
10343 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
10344 Rework father_bb setting in a way to avoid propagating constants
10345 multiple times on a loop body.
10347 2018-07-10 Mark Wielaard <mark@klomp.org>
10350 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
10352 2018-07-10 Richard Biener <rguenther@suse.de>
10354 * hash-map.h (hash_map::iterator::operator*): Return
10355 references to key and value.
10357 2018-07-10 Jakub Jelinek <jakub@redhat.com>
10360 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
10361 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
10362 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
10363 (gimplify_omp_for): For composite loops, move outer
10364 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
10365 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
10366 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
10367 TREE_LIST for both the original class iterator and the "last" helper
10368 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
10369 loop, remember has_decl_expr from outer composite loops for the
10370 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
10372 2018-07-09 Martin Sebor <msebor@redhat.com>
10374 PR middle-end/77357
10375 PR middle-end/86428
10376 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
10377 accessing implicitly initialized array elements.
10378 * expr.c (string_constant): Handle string initializers of
10379 character arrays within aggregates.
10380 * gimple-fold.c (fold_array_ctor_reference): Add argument.
10381 Store element offset. As a special case, handle zero size.
10382 (fold_nonarray_ctor_reference): Same.
10383 (fold_ctor_reference): Add argument. Store subobject offset.
10384 * gimple-fold.h (fold_ctor_reference): Add argument.
10386 2018-07-09 Paul Koning <ni1d@arrl.net>
10388 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
10389 (pdp11_insn_cost): New function.
10390 (pdp11_md_asm_adjust): New function.
10391 (TARGET_INVALID_WITHIN_DOLOOP): Define.
10392 (pdp11_rtx_costs): Update to match machine better.
10393 (output_addr_const_pdp11): Correct format mismatch warnings.
10394 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
10395 * config/pdp11/pdp11.md: General change to add base_cost and/or
10396 length attributes for use by new pdp11_insn_cost function.
10397 (MIN_BRANCH): Correct definition.
10399 (doloop_end): Use standard pattern name for looping pattern.
10400 (doloop_end_nocc): New.
10401 (movsf): Add another constraint alternative.
10402 (zero_extendqihi2): Add constraint alternatives for not in place
10404 (zero_extendhisi2): Remove.
10405 (shift patterns): Add CC handling variants.
10409 (define_peephole2): New peephole to recognize mov that sets CC for
10412 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10414 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
10415 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
10418 2018-07-09 Jakub Jelinek <jakub@redhat.com>
10421 * real.c (real_nextafter): Return true if result is denormal.
10423 2018-07-09 Martin Liska <mliska@suse.cz>
10425 * common.opt: Add back wrongly removed attribute.
10427 2018-07-09 Richard Biener <rguenther@suse.de>
10430 * dwarf2out.c (gen_block_die): For an early generated DIE
10431 always output high/low PC attributes.
10433 2018-07-09 Tom de Vries <tdevries@suse.de>
10435 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
10436 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
10437 onto VAR_DECL with abstract origin.
10439 2018-07-07 Jim Wilson <jimw@sifive.com>
10441 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
10443 2018-07-07 Tom de Vries <tdevries@suse.de>
10445 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
10447 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
10449 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
10451 (combine_bound): Use wide-int overflow calculation instead of
10453 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
10455 * fold-const.c (int_const_binop_2): Same.
10456 (extract_muldiv_1): Same.
10457 (fold_div_compare): Same.
10458 (fold_abs_const): Same.
10460 * poly-int.h (add): Same.
10464 * predict.c (predict_iv_comparison): Same.
10465 * profile-count.c (slow_safe_scale_64bit): Same.
10466 * simplify-rtx.c (simplify_const_binary_operation): Same.
10467 * tree-chrec.c (tree_fold_binomial): Same.
10468 * tree-data-ref.c (split_constant_offset_1): Same.
10469 * tree-if-conv.c (idx_within_array_bound): Same.
10470 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
10471 * tree-ssa-phiopt.c (minmax_replacement): Same.
10472 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
10473 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
10474 * vr-values.c (vr_values::adjust_range_with_scev): Same.
10475 * wide-int.cc (wi::add_large): Same.
10476 (wi::mul_internal): Same.
10477 (wi::sub_large): Same.
10478 (wi::divmod_internal): Same.
10479 * wide-int.h: Change overflow type to overflow_type for neg, add,
10480 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
10481 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
10482 mul_internal, divmod_internal.
10483 (overflow_type): New enum.
10484 (accumulate_overflow): New.
10486 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
10488 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
10489 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
10491 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
10493 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
10494 argument is checked for zero before entering loop, avoid checking again.
10496 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
10498 * gimplify.h (generic_expr_could_trap_p): Set as global function.
10499 * gimplify.h (generic_expr_could_trap_p): Likwise.
10500 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
10502 2018-07-06 Jakub Jelinek <jakub@redhat.com>
10504 PR tree-optimization/86401
10505 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
10506 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
10507 (fold_bit_and_mask): ... here. New helper function for match.pd.
10508 * fold-const.h (fold_bit_and_mask): Declare.
10509 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
10511 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
10514 * target.def (translate_mode_attribute): New hook.
10515 * targhooks.h (default_translate_mode_attribute): Declare.
10516 * targhooks.c (default_translate_mode_attribute): New function.
10517 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
10518 * doc/tm.texi: Regenerate.
10519 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
10520 (rs6000_translate_mode_attribute): New function.
10522 2018-07-06 Paul Koning <ni1d@arrl.net>
10524 * doc/md.texi (define_split): Document DONE and FAIL.
10525 (define_peephole2): Ditto.
10527 2018-07-05 Jeff Law <law@redhat.com>
10529 PR tree-optimization/86010
10530 * tree-ssa-dse.c (compute_trims): More aggressively trim at
10531 both the head and tail of mem* and str* calls.
10533 2018-07-05 Jim Wilson <jimw@sifive.com>
10535 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
10537 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
10539 * config/aarch64/aarch64-simd.md: correct flags text for
10540 MIN_EXPR replacement.
10542 2018-07-05 James Clarke <jrtc27@jrtc27.com>
10544 * configure: Regenerated.
10546 2018-07-05 Carl Love <cel@us.ibm.com>
10548 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
10549 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
10550 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
10551 VSX_BUILTIN_DOUBLEL_V4SF.
10553 2018-07-05 Martin Sebor <msebor@redhat.com>
10556 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
10557 than its domain to compute its the upper bound of a char array.
10559 2018-07-05 Nathan Sidwell <nathan@acm.org>
10561 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
10562 * doc/cpp.texi: Update comment.
10563 * doc/tm.texi: Rebuilt.
10564 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
10565 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
10566 * doc/extend.texi (Backwards Compatibility): Clarify it is system
10567 headers affected by extern "C".
10568 * system.h: Poison NO_IMPLICIT_EXTERN_C.
10569 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
10570 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
10571 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
10572 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
10573 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
10574 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
10575 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
10576 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
10577 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
10578 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
10579 config/sparc/sp64-elf.h, config/spu/spu.h,
10580 config/stormy16/stormy16.h, config/v850/v850.h,
10581 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
10582 define NO_IMPLICIT_EXTERN_C.
10583 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
10585 2018-07-05 Tamar Christina <tamar.christina@arm.com>
10588 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
10589 instead of GET_MODE_SIZE when comparing Units.
10591 2018-07-05 Tamar Christina <tamar.christina@arm.com>
10594 * rtlanal.c (set_noop_p): Constrain on mode change,
10595 include hard-reg-set.h
10597 2018-07-05 Tamar Christina <tamar.christina@arm.com>
10599 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
10601 2018-07-05 Jakub Jelinek <jakub@redhat.com>
10604 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
10607 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
10609 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
10611 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
10614 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
10616 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
10618 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
10620 PR middle-end/86380
10621 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
10623 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
10625 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
10626 neg_*_op* variables.
10628 2018-07-04 Martin Liska <mliska@suse.cz>
10630 * tree-switch-conversion.c: Define
10631 max_ratio_for_speed and max_ratio_for_size constants.
10633 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
10634 Martin Liska <mliska@suse.cz>
10636 PR middle-end/66240
10639 * common.opt: Rename align options with 'str_' prefix.
10640 * common/config/i386/i386-common.c (set_malign_value): New
10642 (ix86_handle_option): Use it to set -falign-* options/
10643 * config/aarch64/aarch64-protos.h (struct tune_params): Change
10644 type from int to string.
10645 * config/aarch64/aarch64.c: Update default values from int
10647 * config/alpha/alpha.c (alpha_override_options_after_change):
10649 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
10650 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10651 max skip conditionally.
10652 * config/i386/freebsd.h (SUBALIGN_LOG): New.
10653 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10654 max skip conditionally.
10655 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10656 max skip conditionally.
10657 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
10658 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10659 max skip conditionally.
10660 * config/i386/i386.c (struct ptt): Change type from int to
10662 (ix86_default_align): Set default values.
10663 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
10664 max skip conditionally.
10665 * config/i386/iamcu.h (SUBALIGN_LOG): New.
10666 (ASM_OUTPUT_MAX_SKIP_ALIGN):
10667 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
10668 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10669 max skip conditionally.
10670 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
10671 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
10672 * config/i386/x86-64.h (SUBALIGN_LOG): New.
10673 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10674 max skip conditionally.
10675 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
10676 * config/ia64/ia64.c (ia64_option_override): Set default values
10677 for alignment options.
10678 * config/m68k/m68k.c: Handle new str_align_* options.
10679 * config/mips/mips.c (mips_set_compression_mode): Change
10681 (mips_option_override): Set default values for options.
10682 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
10684 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10686 * config/rx/rx.c (rx_option_override): Likewise.
10687 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
10688 (LABEL_ALIGN): Use align_labels_log.
10689 (LOOP_ALIGN): Use align_loops_align.
10690 * config/s390/s390.c (s390_asm_output_function_label): Use new
10692 * config/sh/sh.c (sh_override_options_after_change):
10693 Change type of constants.
10694 * config/spu/spu.c (spu_sched_init): Likewise.
10695 * config/sparc/sparc.c (sparc_option_override): Set default
10696 values for options.
10697 * config/visium/visium.c (visium_option_override): Likewise.
10698 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
10699 emit p2align format with last argument if it's not needed.
10700 * doc/invoke.texi: Document extended format of -falign-*.
10701 * final.c: Use align_labels alignment.
10702 * flags.h (struct target_flag_state): Change type to use
10704 (struct align_flags_tuple): New.
10705 (struct align_flags): Likewise.
10706 (align_loops_log): Redefine macro to use new types.
10707 (align_loops_max_skip): Redefine macro to use new types.
10708 (align_jumps_log): Redefine macro to use new types.
10709 (align_jumps_max_skip): Redefine macro to use new types.
10710 (align_labels_log): Redefine macro to use new types.
10711 (align_labels_max_skip): Redefine macro to use new types.
10712 (align_functions_log): Redefine macro to use new types.
10713 (align_loops): Redefine macro to use new types.
10714 (align_jumps): Redefine macro to use new types.
10715 (align_labels): Redefine macro to use new types.
10716 (align_functions): Redefine macro to use new types.
10717 (align_functions_max_skip): Redefine macro to use new types.
10718 (align_loops_value): New macro.
10719 (align_jumps_value): New macro.
10720 (align_labels_value): New macro.
10721 (align_functions_value): New macro.
10722 * function.c (invoke_set_current_function_hook): Propagate
10723 alignment values from flags to global variables default in
10725 * ipa-icf.c (sem_function::equals_wpa): Use
10726 cl_optimization_option_eq instead of memcmp.
10727 * lto-streamer.h (cl_optimization_stream_out): Support streaming
10729 (cl_optimization_stream_in): Likewise.
10730 * optc-save-gen.awk: Support strings in cl_optimization.
10731 * opth-gen.awk: Likewise.
10732 * opts.c (finish_options): Remove error checking of invalid
10734 (MAX_CODE_ALIGN): Remove.
10735 (MAX_CODE_ALIGN_VALUE): Likewise.
10736 (parse_and_check_align_values): New function.
10737 (check_alignment_argument): Likewise.
10738 (common_handle_option): Use check_alignment_argument.
10739 * opts.h (parse_and_check_align_values): Declare.
10740 * toplev.c (init_alignments): Remove.
10741 (read_log_maxskip): New.
10742 (parse_N_M): Likewise.
10743 (parse_alignment_opts): Likewise.
10744 (backend_init_target): Remove usage of init_alignments.
10745 * toplev.h (parse_alignment_opts): Declare.
10746 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
10748 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
10749 * tree.c (cl_option_hasher::equal): New.
10750 * varasm.c: Use new global macros.
10752 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
10754 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
10755 Use a simpler align directive also if MAXSKIP = ALIGN-1.
10756 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10757 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10758 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10759 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
10760 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
10761 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
10762 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
10763 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10764 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10765 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10766 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10768 2018-07-04 Martin Liska <mliska@suse.cz>
10769 Jonathan Wakely <jwakely@redhat.com>
10771 * coverage.c: Use correct type.
10772 * doc/invoke.texi: Language correction.
10774 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
10777 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
10778 ENDBRANCH for non-tail call which may return via indirect branch.
10779 * doc/extend.texi: Document indirect_return attribute.
10781 2018-07-03 Martin Sebor <msebor@redhat.com>
10783 PR tree-optimization/86274
10784 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
10786 (format_floating): Correct handling of infinities and NaNs.
10788 2018-07-03 Martin Sebor <msebor@redhat.com>
10790 * print-tree.c (print_real_cst): New function.
10791 (print_node_brief): Call it.
10792 (print_node): Ditto.
10794 2018-07-03 Jeff Law <law@redhat.com>
10796 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
10797 into a single pattern.
10799 * config/h8300/h8300.md (ors code_iterator): New.
10800 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
10801 a single pattern and single splitter.
10802 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
10803 (iorqi3_1, xorqi3_1): Likewise.
10804 (iorqi3, xorqi3 expanders): Similarly.
10806 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
10807 (movmd_internal) into a single pattern using the P mode iterator.
10808 (movmd splitters): Similarly.
10809 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
10810 (movsd splitters): Similarly.
10812 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
10813 ADDB, ADDW and ADDL into a single ADD attribute which selects the
10814 right table based on the size of the operand.
10815 * config/h8300/h8300.md (length_table): Corresponding changes. All
10816 references to "addb", "addw" and "addl" changed to "add".
10817 (btst patterns): Merge two variants into a single pattern.
10818 (tstqi, tsthi): Likewise.
10819 (addhi3_incdec, addsi3_incdec): Likewise.
10820 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
10821 (mulhi3, mulsi3): Likewise.
10822 (udivhi3, udivsi3): Likewise.
10823 (divhi3, divsi3): Likewise.
10824 (andorqi3, andorhi3, andorsi3): Likewise.
10826 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
10829 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
10830 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
10832 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10834 PR tree-optimization/85694
10835 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
10836 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
10838 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
10839 (<u>avg<mode>3_ceil): New patterns.
10841 2018-07-03 David Malcolm <dmalcolm@redhat.com>
10843 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
10844 scan-tree-dump directive.
10845 * gcc.dg/vect/slp-perm-2.c: Likewise.
10846 * gcc.dg/vect/slp-perm-3.c: Likewise.
10847 * gcc.dg/vect/slp-perm-5.c: Likewise.
10848 * gcc.dg/vect/slp-perm-6.c: Likewise.
10849 * gcc.dg/vect/slp-perm-7.c: Likewise.
10850 * gcc.dg/vect/slp-perm-8.c: Likewise.
10852 2018-07-03 Marek Polacek <polacek@redhat.com>
10854 PR middle-end/86202
10855 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
10857 2018-07-03 Richard Biener <rguenther@suse.de>
10860 * tree-ssa-structalias.c (find_func_clobbers): Properly
10861 handle indirect calls.
10863 2018-07-03 Jeff Law <law@redhat.com>
10865 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
10866 (shifts): New code iterator.
10867 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
10868 expander. Fix HImode handling on H8/SX.
10869 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
10870 (subqi3, subhi3, subsi3 expanders): Likewise.
10871 (andqi3, andhi3, andsi3 expanders): Likewise.
10872 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
10873 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
10874 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
10875 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
10876 (zero_extendqihi2, zero_extendqisi2): Likewise.
10877 (extendqihi2, extendqisi2): Likewise.
10878 (rotlqi3, rotlhi3, rotlsi3): Likewise.
10879 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
10880 (rotlqi3_1, rotlhi3_1): Likewise.
10881 (logicalhi3_sn, logicalsi3_sn): Likewise.
10882 (logicalhi3, logicalsi3): Likewise.
10884 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10886 * tree-vect-patterns.c (vect_recog_rotate_pattern)
10887 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
10888 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
10889 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
10890 type to append_pattern_def_seq instead of creating a stmt_vec_info
10892 (build_mask_conversion): Likewise. Remove vinfo argument.
10893 (vect_add_conversion_to_patterm): Likewise, renaming to...
10894 (vect_add_conversion_to_pattern): ...this.
10895 (vect_recog_mask_conversion_pattern): Update call to
10896 build_mask_conversion. Pass the vector type to
10897 append_pattern_def_seq here too.
10898 (vect_recog_gather_scatter_pattern): Update call to
10899 vect_add_conversion_to_pattern.
10901 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10903 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
10904 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
10905 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
10906 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
10907 STMT_VINFO_PATTERN_DEF_SEQ to null here.
10908 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
10909 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
10910 append_pattern_def_seq instead of new_pattern_def_seq.
10911 (vect_recog_divmod_pattern): Do both of the above.
10912 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
10915 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10917 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
10918 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
10919 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
10920 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
10921 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
10922 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
10923 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
10924 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
10925 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
10926 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
10927 parameter with a single stmt_vec_info.
10928 (vect_recog_func_ptr): Likewise.
10929 (vect_recog_gather_scatter_pattern): Likewise, folding in...
10930 (vect_try_gather_scatter_pattern): ...this.
10931 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
10932 the stmt_vec_info of the statement to be matched. Don't clear
10933 STMT_VINFO_RELATED_STMT.
10934 (vect_pattern_recog): Update call accordingly.
10936 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10938 PR tree-optimization/85694
10939 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
10940 (uavgM3_ceil): Document new optabs.
10941 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
10942 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
10944 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
10945 (savg_ceil_optab): New optabs.
10946 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
10947 (vect_vect_recog_func_ptrs): Add it.
10948 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
10949 constant directly from the associated lhs.
10951 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10953 * tree-vect-patterns.c (vect_split_statement): New function.
10954 (vect_convert_input): Use it to try to split an existing cast.
10956 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10958 * poly-int.h (print_hex): New function.
10959 * dumpfile.h (dump_dec, dump_hex): Declare.
10960 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
10961 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
10962 min_input_precision, operation_precision and operation_sign.
10963 * tree-vect-patterns.c (vect_get_range_info): New function.
10964 (vect_same_loop_or_bb_p, vect_single_imm_use)
10965 (vect_operation_fits_smaller_type): Delete.
10966 (vect_look_through_possible_promotion): Add an optional
10967 single_use_p parameter.
10968 (vect_recog_over_widening_pattern): Rewrite to use new
10969 stmt_vec_info infomration. Handle one operation at a time.
10970 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
10971 (vect_truncatable_operation_p, vect_set_operation_type)
10972 (vect_set_min_input_precision): New functions.
10973 (vect_determine_min_output_precision_1): Likewise.
10974 (vect_determine_min_output_precision): Likewise.
10975 (vect_determine_precisions_from_range): Likewise.
10976 (vect_determine_precisions_from_users): Likewise.
10977 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
10978 (vect_vect_recog_func_ptrs): Put over_widening first.
10980 (vect_pattern_recog): Call vect_determine_precisions.
10982 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10984 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
10985 statements that have been replaced by further pattern statements.
10986 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
10988 2018-07-03 Richard Biener <rguenther@suse.de>
10990 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
10991 always set *dt. Dump vectype in vectype overload.
10992 * dumpfile.h (dump_gimple_expr): New function.
10993 (dump_gimple_expr_loc): Likewise.
10994 * dumpfile.c (dump_gimple_expr): New function.
10995 (dump_gimple_expr_loc): Likewise.
10997 2018-07-02 Jeff Law <law@redhat.com>
10999 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
11000 the H8/300, H8/300H and H8/S variants into a single pattern.
11001 (movhi_h8300, movqi_h8300hs): Similarly.
11002 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
11003 (QHI mode iterator): New.
11005 * config/h8300/h8300.md: Remove trailing whitespace.
11007 2018-07-02 Jim Wilson <jimw@sifive.com>
11009 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
11010 instead of emit_insn for interrupt returns.
11011 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
11012 (riscv_sret, riscv_uret): Likewise.
11014 2018-07-02 David Malcolm <dmalcolm@redhat.com>
11016 * pretty-print.c (selftest::test_pp_format): Move save and restore
11017 of quotes to class auto_fix_quotes, and add an instance.
11018 * selftest.c: Include "intl.h".
11019 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
11020 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
11021 * selftest.h (selftest::auto_fix_quotes): New class.
11023 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
11025 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
11026 (aarch64_sve_prepare_conditional_op): Remove.
11027 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
11028 Allow aarch64_simd_reg_or_zero as select operand; remove
11029 the aarch64_sve_prepare_conditional_op call.
11030 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
11031 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
11032 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
11033 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
11034 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
11035 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
11036 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
11037 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
11038 and a splitters to match all of the *_any patterns.
11039 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
11041 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
11042 (SVE_COND_FP_BINARY_REV): Remove.
11043 (sve_int_op_rev, sve_fp_op_rev): New.
11044 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
11045 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
11046 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
11047 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
11048 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
11049 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
11050 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
11051 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
11052 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
11054 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
11055 Remove match_dup 1 from the inner unspec.
11056 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
11058 * config/aarch64/aarch64.md (movprfx): New attr.
11059 (length): Default movprfx to 8.
11060 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
11061 (*madd<SVE_I>, *msub<SVE_I): Likewise.
11062 (*<su>mul<SVE_I>3_highpart): Likewise.
11063 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
11064 (*v<ASHIFT><SVE_I>3): Likewise.
11065 (*<su><MAXMIN><SVE_I>3): Likewise.
11066 (*<su><MAXMIN><SVE_F>3): Likewise.
11067 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
11068 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
11069 (*div<SVE_F>4): Likewise.
11071 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
11073 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
11076 2018-07-02 Richard Biener <rguenther@suse.de>
11078 PR tree-optimization/86363
11079 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
11080 memset argument refers to a non-variable address.
11082 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
11084 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
11085 {PLUS,MINUS}_EXPR code to...
11086 (adjust_symbolic_bound): ...here,
11087 (combine_bound): ...here,
11088 (set_value_range_with_overflow): ...and here.
11090 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
11092 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
11094 (extract_range_from_abs_expr): ...here.
11096 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
11098 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
11099 -fno-omit-frame-pointer when not optimizing.
11101 2018-07-02 Martin Liska <mliska@suse.cz>
11104 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
11105 (propagate_nothrow): Likewise.
11107 2018-07-02 Martin Liska <mliska@suse.cz>
11110 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
11112 2018-07-02 David Malcolm <dmalcolm@redhat.com>
11114 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
11115 function in r262149, changing "loc" param from source_location to
11116 const dump_location_t &.
11117 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
11118 declaration, as above.
11120 2018-07-01 Paul Koning <ni1d@arrl.net>
11122 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
11123 -munit-asm, -mgnu-asm, -mdec-asm.
11124 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
11125 (pdp11_output_labelref): New.
11126 (pdp11_output_def): New.
11127 (pdp11_output_addr_vec_elt): New.
11128 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
11129 %# and %@ format codes.
11130 (pdp11_option_override): New.
11131 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
11132 (pdp11_output_ident): New.
11133 (pdp11_asm_named_section): New.
11134 (pdp11_asm_init_sections): New.
11135 (pdp11_file_start): New.
11136 (pdp11_file_end): New.
11137 (output_ascii): Use .ascii/.asciz for -mdec-asm.
11138 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
11139 %o, like %c but octal.
11140 (pdp11_option_override): New.
11141 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
11143 (DATA_SECTION_ASM_OP): Ditto.
11144 (READONLY_DATA_SECTION_ASM_OP): New.
11145 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
11146 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
11147 (ASM_OUTPUT_LABELREF): Ditto.
11148 (ASM_OUTPUT_DEF): Ditto.
11149 (ASM_OUTPUT_EXTERNAL): New.
11150 (ASM_OUTPUT_SOURCE_FILENAME): New.
11151 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
11152 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
11153 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
11154 %# and %@ format codes.
11155 * config/pdp11/pdp11.opt (mgnu-asm): New.
11156 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
11157 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
11158 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
11160 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
11162 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
11163 dereferencing path[] beyond its length.
11165 (debug_all_paths): New.
11166 (rewire_first_differing_edge): New.
11167 (adjust_paths_after_duplication): New.
11168 (duplicate_thread_path): Call adjust_paths_after_duplication.
11170 (thread_through_all_blocks): Add new argument to
11171 duplicate_thread_path.
11173 2018-06-30 Jim Wilson <jimw@sifive.com>
11175 * config/riscv/predicates.md (p2m1_shift_operand): New.
11176 (high_mask_shift_operand): New.
11177 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
11178 pattern using p2m1_shift_operand.
11179 (lshsi3_zero_extend_3+2): New combiner pattern using
11180 high_mask_shift_operand.
11182 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11184 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
11186 (vect_recog_rotate_pattern): ...here.
11187 (vect_convert_input): Try to insert casts of invariants in the
11189 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
11190 preheader to be empty.
11192 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11194 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
11195 vector type. If given, install it in the new statement's
11196 STMT_VINFO_VECTYPE.
11197 (vect_element_precision): New function.
11198 (vect_unpromoted_value): New struct.
11199 (vect_unpromoted_value::vect_unpromoted_value): New function.
11200 (vect_unpromoted_value::set_op): Likewise.
11201 (vect_look_through_possible_promotion): Likewise.
11202 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
11203 (vect_widened_op_tree, vect_convert_input): Likewise.
11204 (vect_convert_inputs, vect_convert_output): Likewise.
11205 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
11206 to handle the optional cast of the multiplication result and
11207 vect_widened_op_tree to detect the widened multiplication itself.
11208 Do not require the input and output of promotion casts to have
11209 the same sign, but base the signedness of the operation on the
11210 input rather than the result. If the pattern includes two
11211 promotions, check that those promotions have the same sign.
11212 Do not restrict the MULT_EXPR handling to a double-width result;
11213 handle quadruple-width results and wider. Use vect_convert_inputs
11214 to convert the inputs to the common type.
11215 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
11216 to handle the optional cast of the ABS result. Also allow a sign
11217 change or a sign extension between the ABS and MINUS.
11218 Use vect_widened_op_tree to detect the widened subtraction and use
11219 vect_convert_inputs to convert the inputs to the common type.
11220 (vect_handle_widen_op_by_const): Delete.
11221 (vect_recog_widen_op_pattern): New function.
11222 (vect_recog_widen_mult_pattern): Use it.
11223 (vect_recog_widen_shift_pattern): Likewise.
11224 (vect_recog_widen_sum_pattern): Use
11225 vect_look_through_possible_promotion to handle the promoted
11228 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11230 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
11231 the containing gimple_seq *.
11232 * gimple-iterator.h (gsi_for_stmt): Declare it.
11233 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
11234 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
11235 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
11236 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
11237 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
11239 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
11241 (vect_mark_pattern_stmts): ...here. Handle cases in which the
11242 statement being replaced is part of an existing pattern
11243 definition sequence, inserting the new pattern statements before
11245 (vect_pattern_recog_1): Don't return a bool. If the statement
11246 is already part of a pattern, instead apply pattern matching
11247 to the pattern definition statements. Don't clear the
11248 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
11249 (vect_pattern_recog): Don't break after the first match;
11250 continue processing the pattern definition statements instead.
11251 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
11253 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11255 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
11256 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
11257 (vect_recog_widen_sum_pattern): Use it.
11259 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11261 * tree-vect-loop.c (vectorizable_reduction): Assert that the
11262 phi is not a pattern statement and has not been replaced by
11263 a pattern statement.
11264 * tree-vect-patterns.c (type_conversion_p): Don't check
11265 STMT_VINFO_IN_PATTERN_P.
11266 (vect_recog_vector_vector_shift_pattern): Likewise.
11267 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
11268 the pattern statement rather than the original statement; check
11269 directly for a WIDEN_MULT_EXPR here.
11270 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
11271 vect_is_simple_use to return the pattern statement rather
11272 than the original statement; use is_pattern_stmt_p to check
11273 for such a pattern statement.
11274 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
11275 to return the pattern statement rather than the original statement;
11276 don't do the same transformation here.
11277 (vect_is_simple_use): If the defining statement has been replaced
11278 by a pattern statement, return the pattern statement instead.
11279 Remove the corresponding (local) transformation from the vectype
11282 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11284 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
11285 end and default to null.
11286 * tree-vect-loop.c (vect_create_epilog_for_reduction)
11287 (vectorizable_reduction): Update calls accordingly, dropping the
11288 gimple ** argument if the passed-back statement isn't needed.
11289 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
11290 (vect_recog_rotate_pattern): Likewise.
11291 (vect_recog_mask_conversion_pattern): Likewise.
11292 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
11293 (vect_mask_constant_operand_p): Likewise.
11294 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
11295 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
11296 (get_group_load_store_type, get_load_store_type): Likewise.
11297 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
11298 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
11299 (vectorizable_conversion, vectorizable_assignment): Likewise.
11300 (vectorizable_shift, vectorizable_operation): Likewise.
11301 (vectorizable_store, vect_is_simple_cond): Likewise.
11302 (vectorizable_condition, vectorizable_comparison): Likewise.
11303 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
11304 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
11305 and move it to the end. Cope with null def_stmt_outs.
11307 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
11309 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
11311 2018-06-29 Jeff Law <law@redhat.com>
11313 * config/v850/v850.c (v850_legitimate_address_p): Handle large
11314 displacements for TARGET_V850E2V3 and newer.
11315 (TARGET_LRA_P): Remove. Defaults to LRA now.
11316 * config/v850/v850.md (sign23byte_load): Remove.
11317 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
11318 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
11320 2018-06-29 Martin Liska <mliska@suse.cz>
11323 * coverage.c (coverage_init): Mangle full path name.
11324 * doc/invoke.texi: Document the change.
11325 * gcov-io.c (mangle_path): New.
11326 * gcov-io.h (mangle_path): Likewise.
11327 * gcov.c (mangle_name): Use mangle_path for path mangling.
11329 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11331 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
11332 if starting source register is not even.
11334 2018-06-29 Martin Liska <mliska@suse.cz>
11336 PR tree-optimization/86263
11337 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
11338 Make edge redirection.
11340 2018-06-29 David Malcolm <dmalcolm@redhat.com>
11342 * dumpfile.c (dump_loc): Add indentation based on scope depth.
11343 (dump_scope_depth): New variable.
11344 (get_dump_scope_depth): New function.
11345 (dump_begin_scope): New function.
11346 (dump_end_scope): New function.
11347 * dumpfile.h (get_dump_scope_depth): New declaration.
11348 (dump_begin_scope): New declaration.
11349 (dump_end_scope): New declaration.
11350 (class auto_dump_scope): New class.
11351 (AUTO_DUMP_SCOPE): New macro.
11352 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
11355 2018-06-29 Richard Biener <rguenther@suse.de>
11357 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
11358 compute_all_dependences succeeds.
11359 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
11360 exceed --param loop-max-datarefs-for-datadeps.
11362 2018-06-29 Jakub Jelinek <jakub@redhat.com>
11364 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
11366 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
11369 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
11370 alternative 0 in preferred_for_speed attribute.
11372 2018-06-28 Paul Koning <ni1d@arrl.net>
11374 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
11375 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
11376 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
11377 * config/pdp11/pdp11.md: Correct "length" attribute calculation
11378 for shift insn patterns.
11380 2018-06-28 David Malcolm <dmalcolm@redhat.com>
11382 * cgraph.c (cgraph_node::get_body): Replace assignments to
11383 "dump_file" with calls to set_dump_file.
11384 * dumpfile.c (alt_dump_file): Make static, and group with...
11385 (alt_flags): ...this definition.
11386 (dumps_are_enabled): New variable.
11387 (refresh_dumps_are_enabled): New function.
11388 (set_dump_file): New function.
11389 (set_alt_dump_file): New function.
11390 (gcc::dump_manager::dump_start): Replace assignments to
11391 "dump_file" and "alt_dump_file" with calls to set_dump_file and
11393 (gcc::dump_manager::dump_finish): Likewise.
11394 * dumpfile.h (alt_dump_file): Delete decl.
11395 (dumps_are_enabled): New variable decl.
11396 (set_dump_file): New function decl.
11397 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
11399 * tree-nested.c (lower_nested_functions): Replace assignments to
11400 "dump_file" with calls to set_dump_file.
11402 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
11404 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
11405 goto_locus of each outgoing edge of each basic block.
11407 2018-06-28 Richard Biener <rguenther@suse.de>
11409 * dwarf2out.c (decl_scope_table): Remove.
11410 (push_decl_scope): Likewise.
11411 (pop_decl_scope): Likewise.
11412 (gen_type_die_for_member): Do not call push/pop_decl_scope.
11413 (gen_struct_or_union_type_die): Likewise.
11414 (gen_tagged_type_die): Likewise.
11415 (dwarf2out_init): Do not initialize decl_scope_table.
11416 (dwarf2out_c_finalize): Do not free it.
11418 2018-06-28 Richard Biener <rguenther@suse.de>
11420 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
11421 deciding whether to not re-use a DIE.
11423 2018-06-28 Richard Biener <rguenther@suse.de>
11425 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
11426 DW_AT_abstract_origin attribute.
11428 2018-06-28 Martin Liska <mliska@suse.cz>
11430 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
11431 Use newly introduced constants.
11432 * tree-switch-conversion.h (struct jump_table_cluster):
11433 Define max_ratio_for_size and max_ratio_for_speed.
11435 2018-06-28 Martin Liska <mliska@suse.cz>
11437 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
11438 Add new checking assert to catch invalid state.
11439 (jump_table_cluster::can_be_handled): Handle single case
11441 (jump_table_cluster::is_beneficial): Bail out for such case.
11442 (bit_test_cluster::find_bit_tests):
11443 Add new checking assert to catch invalid state.
11444 (bit_test_cluster::can_be_handled): Handle single case
11446 (bit_test_cluster::is_beneficial): Bail out for such case.
11447 (switch_decision_tree::analyze_switch_statement):
11450 2018-06-28 Martin Liska <mliska@suse.cz>
11452 * common.opt: Introduce -completion option.
11453 * gcc.c (driver_handle_option): Handle it.
11454 (driver::main): Print completions if completion
11456 * opt-suggestions.c (option_proposer::get_completions):
11458 (option_proposer::suggest_completion): Likewise.
11459 (option_proposer::find_param_completions): Likewise.
11460 (verify_autocompletions): Likewise.
11461 (test_completion_valid_options): Likewise.
11462 (test_completion_valid_params): Likewise.
11463 (in_completion_p): Likewise.
11464 (empty_completion_p): Likewise.
11465 (test_completion_partial_match): Likewise.
11466 (test_completion_garbage): Likewise.
11467 (opt_proposer_c_tests): Likewise.
11468 * opt-suggestions.h: Declare new functions.
11469 * opts.c (common_handle_option): Handle OPT__completion_.
11470 * selftest-run-tests.c (selftest::run_tests): Add
11471 opt_proposer_c_tests.
11472 * selftest.c (assert_str_startswith): New.
11473 * selftest.h (assert_str_startswith): Likewise.
11474 (opt_proposer_c_tests): New.
11475 (ASSERT_STR_STARTSWITH): Likewise.
11477 2018-06-28 Martin Liska <mliska@suse.cz>
11479 * Makefile.in: Add opt-suggestions.o.
11480 * gcc-main.c: Include opt-suggestions.h.
11481 * gcc.c (driver::driver): Likewise.
11482 (driver::~driver): Remove m_option_suggestions.
11483 (driver::build_option_suggestions): Moved to option_proposer.
11484 (driver::suggest_option): Likewise.
11485 (driver::handle_unrecognized_options): Use option_proposer.
11486 * gcc.h (class driver): Add new memver m_option_proposer.
11487 * opt-suggestions.c: New file.
11488 * opt-suggestions.h: New file.
11490 2018-06-28 Martin Liska <mliska@suse.cz>
11492 * vec.h (class auto_string_vec): New (moved from auto_argvec).
11493 (auto_string_vec::~auto_string_vec): Likewise.
11495 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
11497 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
11498 prevent_decl_creation_for_types fields up and add reset_location field.
11499 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
11500 statement if id->reset_location is true.
11501 (copy_edges_for_bb): Do not set goto_locus on the new edges if
11502 id->reset_location is true.
11503 (copy_phis_for_bb): Force input_location on the arguments if
11504 id->reset_location is true.
11505 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
11506 is set on the function to be inlined.
11508 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
11510 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
11512 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
11514 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
11515 registers for Pmode.
11516 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
11517 hard registers for the clobbered pseudo.
11519 2018-06-27 Paul Koning <ni1d@arrl.net>
11521 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
11522 mutually exclusive options.
11523 * config/pdp11/constraints.md (h): New constraint.
11524 (O): Update definition to match shift code generation.
11525 (D): New constraint.
11526 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
11528 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
11530 (output_jump): Change arguments.
11531 (pdp11_fixed_cc_regs): New function.
11532 (pdp11_cc_mode): Ditto.
11533 (pdp11_expand_shift): Ditto.
11534 (pdp11_assemble_shift): Ditto.
11535 (pdp11_small_shift): Ditto.
11536 (pdp11_branch_cost): Remove.
11537 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
11539 (pdp11_register_move_cost): Update for CC registers.
11540 (pdp11_rtx_costs): Add case for LSHIFTRT.
11541 (pdp11_output_jump): Add CCNZ mode conditional branches.
11542 (notice_update_cc_on_set): Remove.
11543 (pdp11_cc_mode): New function.
11544 (simple_memory_operand): Correct pre/post decrement case.
11545 (no_side_effect_operand): New function.
11546 (pdp11_regno_reg_class): Add CC_REGS class.
11547 (pdp11_fixed_cc_regs): New function.
11548 (pdp11_small_shift): New function.
11549 (pdp11_expand_shift): New function to expand shift insns.
11550 (pdp11_assemble_shift): New function to output shifts.
11551 (pdp11_branch_cost): Remove.
11552 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
11553 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
11554 (WCHAR_TYPE): Ditto.
11555 (PTRDIFF_TYPE): Ditto.
11556 (ADJUST_INSN_LENGTH): New macro.
11557 (FIXED_REGISTERS): Add CC registers.
11558 (CALL_USED_REGISTERS): Ditto.
11559 (reg_class): Ditto.
11560 (REG_CLASS_NAMES): Ditto.
11561 (REG_CLASS_CONTENTS): Ditto.
11562 (SELECT_CC_MODE): Use new function.
11563 (TARGET_FLAGS_REGNUM): New macro.
11564 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
11565 (cc0_reg_rtx): Remove.
11566 (CC_STATUS_MDEP): Remove.
11567 (CC_STATUS_MDEFP_INIT): Remove.
11568 (CC_IN_FPU): Remove.
11569 (NOTICE_UPDATE_CC): Remove.
11570 (REGISTER_NAMES): Add CC registers.
11571 (BRANCH_COST): Change to constant 1.
11572 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
11574 * config/pdp11/pdp11.opt (mbcopy): Remove.
11575 (mbcopy-builtin): Remove.
11576 (mbranch-cheap): Remove.
11577 (mbranch-expensive): Remove.
11578 * config/pdp11/predicates.md (expand_shift_operand): Update to
11579 match shift code generation.
11580 (ccnz_operator): New predicate.
11581 * doc/invoke.texi (PDP-11 Options): Remove deleted options
11582 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
11583 Remove non-existent option -mabshi, -mno-abshi. Document mutually
11585 * doc/md.texi (PDP-11): Document new D and h constraints. Update
11586 description of O constraint.
11588 2018-06-27 Jeff Law <law@redhat.com>
11589 Austin Law <austinklaw@gmail.com>
11591 * config/v850/v850.md (addsi3_set_flags): New pattern.
11592 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
11593 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
11594 (zero_extendhisi2_v850_set_flags): Likewise.
11595 (zero_extendqisi2_v850_set_flags): Likewise.
11596 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
11597 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
11598 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
11600 * config/v850/v850-protos.h (notice_update_cc): Remove.
11601 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
11602 (v850_print_operand): Handle 'D' and "d".
11603 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
11604 Add handling of arithmetic/logical operations compared against zero.
11605 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
11606 Do not look at v850_compare_op, instead get mode from last argument.
11607 (v850_gen_compare): Remove
11608 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
11609 after reload for prologue insns.
11610 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
11612 (construct_save_jarl): Likewise.
11613 (TARGET_FLAGS_REGNUM): Define.
11614 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
11615 (NOTICE_UPDATE_CC): Remove.
11616 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
11617 than cc0. Conditionalize on reload_completed.
11618 (cmpsi_insn, setfcc_insn): Likewise.
11619 (tst1 splitter): Turn into define_and_split which sets the flags
11621 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
11622 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
11623 (cstoresf4, cstoredf4): Clobber the flags.
11624 (cmpsi, cmpsf, cmpdf): Remove expanders.
11625 (setf_insn): Remove pattern.
11626 (addsi3): Turn into define_and_split which clobbers the flags after
11627 reload and a suitable pattern (addsi3_clobber_flags) for use after
11629 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
11630 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
11631 (ashrsi3, ashrsi3_v850e2): Likewise.
11632 (bins): Clobber the flags.
11633 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
11634 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
11635 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
11636 (call_value_internal_short, call_value_internal_long): Likewise.
11637 (callt_save_interrupt, callt_return_interrupt): Likewise.
11638 (save_interrupt, return_interrupt): Likewise.
11639 (callt_save_all_interrupt, save_all_interrupt): Likewise.
11640 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
11641 (restore_all_interrupt, _restore_all_interrupt): Likewise.
11642 (All FP comparisons): Only allow after reload has completed.
11644 (divh, divhu): Tweak output template.
11645 (branch_z_normal, branch_z_invert): Remove
11646 (branch_nz_normal, branch_nz_invert): Likewise.
11647 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
11649 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
11650 * config/v850/v850.c (notice_update_cc): Remove.
11651 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
11652 (CC_NO_CARRY): Likewise.
11653 (NOTICE_UPDATE_CC): Define to nothing.
11654 * config/v850/v850.md: Remove block comment on cc0 handling
11655 Remove "cc" attribute from all patterns. Remove cc_status handling
11656 from all patterns. Minor formatting fixes.
11658 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11660 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
11661 (cortex-a76.cortex-a55): Likewise.
11662 * config/aarch64/aarch64-tune.md: Regenerate.
11663 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
11664 cortex-a76.cortex-a55.
11666 2018-06-27 Jeff Law <law@redhat.com>
11668 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
11669 (MULTILIB_DIRNAMES): Similarly.
11671 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
11673 * gimple.h (gimple_return_retbnd): Delete.
11674 (gimple_return_set_retbnd): Likewise.
11675 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
11676 gimple_return_set_retbnd.
11677 * gimple-pretty-print.c (dump_gimple_return): Remove call to
11678 gimple_return_retbnd and adjust.
11679 * tree-inline.h (struct copy_body_data): Remove retbnd field.
11680 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
11681 Explicitly return NULL in a couple more cases. Move assertion
11682 on debug statements and remove unreachable code.
11683 (reset_debug_binding): Do not test id->retbnd.
11684 (expand_call_inline): Do not set it.
11686 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
11688 * configure.ac: Add --disable-gcov option.
11689 * configure: Regenerate.
11690 * Makefile.in: Honour @enable_gcov@.
11691 * doc/install.texi: Document --disable-gcov.
11693 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11695 * config/arm/arm-cpus.in (cortex-a76): New entry.
11696 (cortex-a76.cortex-a55): Likewise.
11697 * config/arm/arm-tables.opt: Regenerate.
11698 * config/arm/arm-tune.md: Likewise.
11699 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
11700 * doc/invoke.texi (ARM Options): Document cortex-a76 and
11701 cortex-a76.cortex-a55.
11703 2018-06-27 Tamar Christina <tamar.christina@arm.com>
11706 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
11708 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
11710 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
11712 (EPILOGUE_USES): Likewise.
11714 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
11716 * tree-inline.c (remap_location): New function extracted from...
11717 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
11718 (copy_phis_for_bb): ...here. Call remap_location.
11719 (copy_cfg_body): Adjust call to copy_edges_for_bb.
11721 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
11723 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
11724 unaligned vsx for 16B memset.
11726 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
11729 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
11730 ieee128_float_type_node to long_double_type_node unless
11731 TARGET_LONG_DOUBLE_128 is set.
11733 2018-06-26 David Malcolm <dmalcolm@redhat.com>
11735 * cfgloop.c (get_loop_location): Convert return type from
11736 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
11737 by implicit construction from rtx_insn *, and using
11738 dump_user_location_t::from_function_decl for the fallback case.
11739 * cfgloop.h (get_loop_location): Convert return type from
11740 location_t to dump_user_location_t.
11741 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
11742 dump_printf_loc to pass in a dump_location_t rather than a
11743 location_t, via the gimple stmt.
11744 * coverage.c (get_coverage_counts): Update calls to
11745 dump_printf_loc to pass in dump_location_t rather than a
11747 * doc/optinfo.texi (Dump types): Convert example of
11748 dump_printf_loc from taking "locus" to taking "insn". Update
11749 description of the "_loc" calls to cover dump_location_t.
11750 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
11752 (dump_user_location_t::dump_user_location_t): New constructors,
11753 from gimple *stmt and rtx_insn *.
11754 (dump_user_location_t::from_function_decl): New function.
11755 (dump_loc): Make static.
11756 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
11757 const dump_location_t &.
11758 (dump_generic_expr_loc): Delete.
11759 (dump_printf_loc): Convert param "loc" from location_t to
11760 const dump_location_t &.
11761 (selftest::test_impl_location): New function.
11762 (selftest::dumpfile_c_tests): New function.
11763 * dumpfile.h: Include "profile-count.h".
11764 (class dump_user_location_t): New class.
11765 (struct dump_impl_location_t): New struct.
11766 (class dump_location_t): New class.
11767 (dump_printf_loc): Convert 2nd param from source_location to
11768 const dump_location_t &.
11769 (dump_generic_expr_loc): Delete.
11770 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
11771 const dump_location_t &.
11772 * gimple-fold.c (fold_gimple_assign): Update call to
11773 dump_printf_loc to pass in a dump_location_t rather than a
11774 location_t, via the gimple stmt.
11775 (gimple_fold_call): Likewise.
11776 * gimple-loop-interchange.cc
11777 (loop_cand::analyze_iloop_reduction_var): Update for change to
11778 check_reduction_path.
11779 (tree_loop_interchange::interchange): Update for change to
11780 find_loop_location.
11781 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
11782 change in return-type of find_loop_location.
11783 (graphite_regenerate_ast_isl): Likewise.
11784 * graphite-optimize-isl.c (optimize_isl): Likewise.
11785 * graphite.c (graphite_transform_loops): Likewise.
11786 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
11787 pass in a dump_location_t rather than a location_t, via the
11789 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
11790 * ipa.c (walk_polymorphic_call_targets): Likewise.
11791 * loop-unroll.c (report_unroll): Convert "locus" param from
11792 location_t to dump_location_t.
11793 (decide_unrolling): Update for change to get_loop_location's
11795 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
11796 location_t to dump_user_location_t.
11797 (grid_find_single_omp_among_assignments_1): Updates calls to
11798 dump_printf_loc to pass in a dump_location_t rather than a
11799 location_t, via the gimple stmt.
11800 (grid_parallel_clauses_gridifiable): Convert "tloc" from
11801 location_t to dump_location_t. Updates calls to dump_printf_loc
11802 to pass in a dump_location_t rather than a location_t, via the
11804 (grid_inner_loop_gridifiable_p): Likewise.
11805 (grid_dist_follows_simple_pattern): Likewise.
11806 (grid_gfor_follows_tiling_pattern): Likewise.
11807 (grid_target_follows_gridifiable_pattern): Likewise.
11808 (grid_attempt_target_gridification): Convert initialization
11809 of local "grid" from memset to zero-initialization; FIXME: does
11810 this require C++11? Update call to dump_printf_loc to pass in a
11811 optinfo_location rather than a location_t, via the gimple stmt.
11812 * profile.c (read_profile_edge_counts): Updates call to
11813 dump_printf_loc to pass in a dump_location_t rather than a
11815 (compute_branch_probabilities): Likewise.
11816 * selftest-run-tests.c (selftest::run_tests): Call
11818 * selftest.h (dumpfile_c_tests): New decl.
11819 * tree-loop-distribution.c (pass_loop_distribution::execute):
11820 Update for change in return type of find_loop_location.
11821 * tree-parloops.c (parallelize_loops): Likewise.
11822 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
11823 "locus" from location_t to dump_user_location_t.
11824 (canonicalize_loop_induction_variables): Likewise.
11825 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
11826 for change in return type of find_loop_location.
11827 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
11828 to dump_printf_loc to pass in a dump_location_t rather than a
11829 location_t, via the stmt.
11830 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
11832 * tree-vect-loop-manip.c (find_loop_location): Convert return
11833 type from source_location to dump_user_location_t.
11834 (vect_do_peeling): Update for above change.
11835 (vect_loop_versioning): Update for change in type of
11837 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
11838 from location_t to dump_user_location_t.
11839 (vect_estimate_min_profitable_iters): Update for change in type
11841 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
11842 location_t to dump_location_t.
11843 (vect_slp_bb): Update for change in type of vect_location.
11844 * tree-vectorizer.c (vect_location): Convert from source_location
11845 to dump_user_location_t.
11846 (try_vectorize_loop_1): Update for change in vect_location's type.
11847 (vectorize_loops): Likewise.
11848 (increase_alignment): Likewise.
11849 * tree-vectorizer.h (vect_location): Convert from source_location
11850 to dump_user_location_t.
11851 (find_loop_location): Convert return type from source_location to
11852 dump_user_location_t.
11853 (check_reduction_path): Convert 1st param from location_t to
11854 dump_user_location_t.
11855 * value-prof.c (check_counter): Update call to dump_printf_loc to
11856 pass in a dump_user_location_t rather than a location_t; update
11857 call to error_at for change in type of "locus".
11858 (check_ic_target): Update call to dump_printf_loc to
11859 pass in a dump_user_location_t rather than a location_t, via the
11862 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
11864 * config/s390/s390.h (enum processor_flags): Do not use
11865 default tune parameter when -march was specified.
11867 2018-06-26 Jakub Jelinek <jakub@redhat.com>
11870 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
11871 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
11874 2018-06-26 Richard Biener <rguenther@suse.de>
11876 PR tree-optimization/86287
11878 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
11879 (vect_analyze_loop): Initialize n_stmts.
11881 2018-06-26 Richard Biener <rguenther@suse.de>
11883 PR middle-end/86271
11884 * fold-const.c (fold_convertible_p): Pointer extension
11887 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
11890 * dwarf2out.c (loc_list_has_views): Adjust comments.
11891 (dw_loc_list): Split single cross-partition range with
11894 2018-06-25 Jeff Law <law@redhat.com>
11896 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
11897 on -mbig-switch by default.
11899 * config/v850/predicates.md (const_float_1_operand): Fix match_code
11901 (const_float_0_operand): Remove unused predicate.
11902 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
11903 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
11904 (recipsf2): New expander. Original pattern now called
11906 (recipdf2, recipdf2_insn): Similarly.
11907 (rsqrtsf2, rsqrtsf2_insn): Similarly
11908 (rsqrtdf2, rsqrtdf2_insn): Similarly
11910 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
11912 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
11913 Simplify logic for FreeBSD (twice).
11915 2018-06-25 Martin Sebor <msebor@redhat.com>
11917 PR tree-optimization/86204
11918 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
11919 a strnlen result if it's less than the length of the string.
11921 2018-06-25 Martin Sebor <msebor@redhat.com>
11923 PR tree-optimization/85700
11924 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
11925 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
11926 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
11928 2018-06-25 Martin Sebor <msebor@redhat.com>
11930 * doc/extend.texi (Zero-length arrays): Update and clarify.
11932 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
11934 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
11935 added IEEE/IBM long double multilib support on PowerPC little
11936 endian Linux systems.
11937 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
11938 (MULTILIB_DEFAULTS): Likewise.
11939 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11941 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
11942 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
11943 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
11945 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
11947 PR middle-end/86311
11948 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
11949 (REORDER_45): Likewise.
11951 2018-06-25 Jeff Law <law@redhat.com>
11953 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
11954 dividend to 32 bits. Adjust length.
11955 (udivmodhi4): Cleanup output template. Fix length.
11957 2018-06-25 Carl Love <cel@us.ibm.com>
11959 * config/rs6000/vsx.md: Change word selector to prefered location.
11961 2018-06-25 Richard Biener <rguenther@suse.de>
11963 PR tree-optimization/86304
11964 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
11965 epilogue-if-converted loops as well.
11967 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
11969 * lto-section-out.c (lto_begin_section): Do not print section
11970 name for noaddr and unnumbered dumps.
11972 2018-06-25 Richard Biener <rguenther@suse.de>
11974 * tree-vectorizer.h (struct vec_info_shared): New structure
11975 with parts split out from struct vec_info and loop_nest from
11976 struct _loop_vec_info.
11977 (struct vec_info): Adjust accordingly.
11978 (struct _loop_vec_info): Likewise.
11979 (LOOP_VINFO_LOOP_NEST): Adjust.
11980 (LOOP_VINFO_DATAREFS): Likewise.
11981 (LOOP_VINFO_DDRS): Likewise.
11982 (struct _bb_vec_info): Likewise.
11983 (BB_VINFO_DATAREFS): Likewise.
11984 (BB_VINFO_DDRS): Likewise.
11985 (struct _stmt_vec_info): Add dr_aux member.
11986 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
11987 (DR_MISALIGNMENT_UNINITIALIZED): New.
11988 (set_dr_misalignment): Adjust.
11989 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
11990 (vect_analyze_loop): Adjust prototype.
11991 (vect_analyze_loop_form): Likewise.
11992 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
11993 Compute dependences lazily.
11994 (vect_record_base_alignments): Use shared datarefs/ddrs.
11995 (vect_verify_datarefs_alignment): Likewise.
11996 (vect_analyze_data_refs_alignment): Likewise.
11997 (vect_analyze_data_ref_accesses): Likewise.
11998 (vect_analyze_data_refs): Likewise.
11999 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
12000 constructor parameter for shared part.
12001 (vect_analyze_loop_form): Pass in shared part and adjust.
12002 (vect_analyze_loop_2): Pass in storage for the number of
12003 stmts. Move loop nest finding to the caller. Compute
12005 (vect_analyze_loop): Pass in shared part.
12006 (vect_transform_loop): Verify shared datarefs are unchanged.
12007 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
12008 constructor parameter for shared part.
12009 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
12010 (vect_slp_bb): Verify shared datarefs are unchanged before
12012 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
12014 (new_stmt_vec_info): Initialize DR_AUX misalignment to
12015 DR_MISALIGNMENT_UNINITIALIZED.
12016 * tree-vectorizer.c (vec_info::vec_info): Add constructor
12017 parameter for shared part.
12018 (vec_info::~vec_info): Adjust.
12019 (vec_info_shared::vec_info_shared): New.
12020 (vec_info_shared::~vec_info_shared): Likewise.
12021 (vec_info_shared::save_datarefs): Likewise.
12022 (vec_info_shared::check_datarefs): Likewise.
12023 (try_vectorize_loop_1): Construct shared part live for analyses
12024 of a single loop for multiple vector sizes.
12025 * tree-parloops.c (gather_scalar_reductions): Adjust.
12027 2018-06-25 Richard Biener <rguenther@suse.de>
12029 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
12030 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
12031 (vect_analyze_data_refs): Remove similar code from here and
12032 simplify accordingly.
12034 2018-06-25 Richard Biener <rguenther@suse.de>
12036 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
12037 for reverse storage order accesses rather than asserting
12038 they cannot happen here.
12040 2018-06-25 Tom de Vries <tdevries@suse.de>
12043 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
12044 Use data16 instead of .byte for insn prefix.
12046 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
12049 * parser.c (make_char_string_pack): Pass this literal chars
12050 through cpp_interpret_string.
12051 (cp_parser_userdef_numeric_literal): Check the result of
12052 make_char_string_pack.
12054 2018-06-24 Maya Rashish <coypu@sdf.org>
12056 * ginclude/stddef.h: Simplify conditions around avoiding
12057 re-definition of __size_t.
12059 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12061 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
12062 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
12064 2018-06-22 Maya Rashish <coypu@sdf.org>
12066 * doc/invoke.texi (mno-fancy-math-387): Update for changes
12067 made to OpenBSD and NetBSD through the years.
12069 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
12071 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
12072 behavior of vec_pack (vector double, vector double) to match
12073 behavior of vec_float2 (vector double, vector double).
12075 2018-06-22 Olivier Hainque <hainque@adacore.com>
12077 * gimplify.c (gimplify_function_tree): Prevent creation
12078 of a trampoline for the address of the current function
12079 passed to entry/exit instrumentation hooks.
12081 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
12084 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
12087 2018-06-22 Martin Liska <mliska@suse.cz>
12089 PR tree-optimization/86263
12090 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
12091 Bail out if is_enabled is false.
12092 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
12094 (jump_table_cluster::is_enabled): New function.
12096 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12098 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
12099 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
12100 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
12101 (lto_input_ts_binfo_tree_pointers): Likewise.
12102 * tree-streamer-out.c (streamer_write_tree_bitfields,
12103 write_ts_binfo_tree_pointers): Likewise.
12104 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
12106 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12108 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
12110 2018-06-22 Martin Liska <mliska@suse.cz>
12112 * symbol-summary.h (get): Make it pure and inline move
12113 functionality from ::get function.
12114 (get): Remove and inline into ::get and ::get_create.
12115 (get_create): Move code from ::get function.
12117 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12120 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
12121 -x assembler-with-cpp.
12123 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12125 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
12126 _FILE_OFFSET_BITS=64 for C++.
12128 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
12130 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
12131 conversion insn that shows up when pr85657-3.c is compiled using
12132 IEEE 128-bit long double.
12133 (neg<mode>2_internal): Use the correct mode to check whether the
12134 mode is IBM extended.
12135 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
12136 multiply and divide external functions from being created more
12139 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12141 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
12143 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
12144 the edge can be forwarded.
12145 (cfg_layout_merge_blocks): Likewise.
12147 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12149 * except.c (finish_eh_generation): Commit edge insertions only after
12150 the EH edges have been redirected from post-landing to landing pads.
12152 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12154 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
12155 create_tmp_var_for to create the FRAME decl.
12156 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
12158 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12160 * tree-inline.c (copy_edges_for_bb): Minor tweak.
12161 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
12162 debug statement when resetting its value.
12163 (expand_call_inline): Copy the locus of the call onto the assignment
12164 of the return value, if any. Use local variable in more cases.
12166 2018-06-21 Martin Liska <mliska@suse.cz>
12168 * ipa-pure-const.c (propagate_nothrow): Use
12169 funct_state_summaries->get.
12170 (dump_malloc_lattice): Likewise.
12171 (propagate_malloc): Likewise.
12173 2018-06-21 Richard Biener <rguenther@suse.de>
12175 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
12176 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
12177 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
12179 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
12180 BLOCK_ABSTRACT_ORIGIN unconditionally.
12182 2018-06-21 David Malcolm <dmalcolm@redhat.com>
12184 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
12186 * ipa-reference.c (ipa_reference_c_finalize): Delete
12187 ipa_ref_opt_sum_summaries and set it to NULL.
12189 2018-06-21 Tom de Vries <tdevries@suse.de>
12191 PR tree-optimization/85859
12192 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
12193 test with comment from bb_no_side_effects_p.
12195 2018-06-21 Richard Biener <rguenther@suse.de>
12197 PR tree-optimization/86232
12198 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
12199 max for constant niter.
12201 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
12203 * config/aarch64/aarch64-simd.md
12204 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
12206 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
12208 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
12209 Make opernads of the unspec commutative.
12211 2018-06-21 Richard Biener <rguenther@suse.de>
12213 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
12214 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
12215 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
12216 (vect_analyze_data_ref_dependence): Re-order checks to deal with
12218 (vect_record_base_alignments): Do not record base alignment
12219 for gathers or scatters.
12220 (vect_compute_data_ref_alignment): Drop return value that is always
12221 true. Bail out early for gathers or scatters.
12222 (vect_enhance_data_refs_alignment): Bail out early for gathers
12224 (vect_find_same_alignment_drs): Likewise.
12225 (vect_analyze_data_refs_alignment): Remove dead code.
12226 (vect_slp_analyze_and_verify_node_alignment): Likewise.
12227 (vect_analyze_data_refs): For possible gathers or scatters do
12228 not create an alternate DR, just check their possible validity
12229 and mark them. Adjust DECL_NONALIASED handling to not rely
12230 on DR_BASE_ADDRESS.
12231 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
12232 update inits of gathers or scatters.
12233 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
12234 Also copy gather/scatter flag to pattern vinfo.
12236 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
12238 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
12239 behavior of vec_packsu (vector unsigned long long, vector unsigned
12240 long long) to match behavior of vec_packs with same signature.
12242 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
12243 Thomas Schwinge <thomas@codesourcery.com>
12244 Cesar Philippidis <cesar@codesourcery.com>
12246 * gimplify.c (gimplify_scan_omp_clauses): Add support for
12247 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
12248 (gimplify_adjust_omp_clauses): Likewise.
12249 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
12250 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
12251 (gimplify_omp_target_update): Update handling of acc update and
12253 * omp-low.c (install_var_field): Remove unused parameter
12254 base_pointers_restrict.
12255 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
12256 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
12258 (omp_target_base_pointers_restrict_p): Delete.
12259 (scan_omp_target): Update call to scan_sharing_clauses.
12260 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
12262 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
12263 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
12264 (convert_local_omp_clauses): Likewise.
12265 * tree-pretty-print.c (dump_omp_clause): Likewise.
12266 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
12268 (omp_clause_code_name): Likewise.
12270 2018-06-20 Jakub Jelinek <jakub@redhat.com>
12273 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
12276 PR tree-optimization/86231
12277 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
12278 anti-range don't overwrite *vr0min before using it to compute *vr0max.
12280 2018-06-20 Tom de Vries <tdevries@suse.de>
12282 PR tree-optimization/86097
12283 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
12284 iv type if signedness of iv type is not the same as that of *nit.
12286 2018-06-20 Jakub Jelinek <jakub@redhat.com>
12288 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
12289 EDGE_EH edges, verify they are all EDGE_EH.
12291 2018-06-20 Maya Rashish <coypu@sdf.org>
12293 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
12295 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12297 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
12298 * config/aarch64/aarch64.c (xgene1_tunings): Add
12299 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
12300 (aarch64_mode_valid_for_sched_fusion_p):
12301 Allow 16-byte modes.
12302 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
12303 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
12305 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
12307 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
12308 * config/aarch64/iterators.md (VQ2): New mode iterator.
12310 2018-06-20 Martin Liska <mliska@suse.cz>
12312 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
12313 Change default ratio from 10 to 8.
12315 2018-06-20 Martin Liska <mliska@suse.cz>
12317 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
12319 (bit_test_cluster::find_bit_tests): Likewise.
12320 (switch_decision_tree::analyze_switch_statement): Find clusters.
12321 * tree-switch-conversion.h (struct jump_table_cluster): Document
12324 2018-06-20 Martin Liska <mliska@suse.cz>
12326 * tree-switch-conversion.c (switch_conversion::collect):
12327 Record m_uniq property.
12328 (switch_conversion::expand): Bail out for special conditions.
12329 (group_cluster::~group_cluster): New.
12330 (group_cluster::group_cluster): Likewise.
12331 (group_cluster::dump): Likewise.
12332 (jump_table_cluster::emit): New.
12333 (switch_decision_tree::fix_phi_operands_for_edges): New.
12334 (struct case_node): Remove struct.
12335 (jump_table_cluster::can_be_handled): New.
12336 (case_values_threshold): Moved to header.
12337 (reset_out_edges_aux): Likewise.
12338 (jump_table_cluster::is_beneficial): New.
12339 (bit_test_cluster::can_be_handled): Likewise.
12340 (add_case_node): Remove.
12341 (bit_test_cluster::is_beneficial): New.
12342 (case_bit_test::cmp): New.
12343 (bit_test_cluster::emit): New.
12344 (expand_switch_as_decision_tree_p): Remove.
12345 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
12346 (fix_phi_operands_for_edge): Likewise.
12347 (switch_decision_tree::analyze_switch_statement): New.
12348 (compute_cases_per_edge): Move ...
12349 (switch_decision_tree::compute_cases_per_edge): ... here.
12350 (try_switch_expansion): Likewise.
12351 (switch_decision_tree::try_switch_expansion): Likewise.
12352 (record_phi_operand_mapping): Likewise.
12353 (switch_decision_tree::record_phi_operand_mapping): Likewise.
12354 (emit_case_decision_tree): Likewise.
12355 (switch_decision_tree::emit): Likewise.
12356 (balance_case_nodes): Likewise.
12357 (switch_decision_tree::balance_case_nodes): Likewise.
12358 (dump_case_nodes): Likewise.
12359 (switch_decision_tree::dump_case_nodes): Likewise.
12360 (emit_jump): Likewise.
12361 (switch_decision_tree::emit_jump): Likewise.
12362 (emit_cmp_and_jump_insns): Likewise.
12363 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
12364 (emit_case_nodes): Likewise.
12365 (switch_decision_tree::emit_case_nodes): Likewise.
12366 (conditional_probability): Remove.
12367 * tree-switch-conversion.h (enum cluster_type): New.
12369 (struct cluster): Likewise.
12370 (cluster::cluster): Likewise.
12371 (struct simple_cluster): Likewise.
12372 (simple_cluster::simple_cluster): Likewise.
12373 (struct group_cluster): Likewise.
12374 (struct jump_table_cluster): Likewise.
12375 (struct bit_test_cluster): Likewise.
12376 (struct min_cluster_item): Likewise.
12377 (struct case_tree_node): Likewise.
12378 (case_tree_node::case_tree_node): Likewise.
12379 (jump_table_cluster::case_values_threshold): Likewise.
12380 (struct case_bit_test): Likewise.
12381 (struct switch_decision_tree): Likewise.
12382 (struct switch_conversion): Likewise.
12383 (switch_decision_tree::reset_out_edges_aux): Likewise.
12385 2018-06-20 Martin Liska <mliska@suse.cz>
12387 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
12388 (hoist_edge_and_branch_if_true): Likewise.
12389 (expand_switch_using_bit_tests_p): Likewise.
12390 (struct case_bit_test): Likewise.
12391 (case_bit_test_cmp): Likewise.
12392 (emit_case_bit_tests): Likewise.
12393 (switch_conversion::switch_conversion): New class.
12394 (struct switch_conv_info): Remove old struct.
12395 (collect_switch_conv_info): More to ...
12396 (switch_conversion::collect): ... this.
12397 (check_range): Likewise.
12398 (switch_conversion::check_range): Likewise.
12399 (check_all_empty_except_final): Likewise.
12400 (switch_conversion::check_all_empty_except_final): Likewise.
12401 (check_final_bb): Likewise.
12402 (switch_conversion::check_final_bb): Likewise.
12403 (create_temp_arrays): Likewise.
12404 (switch_conversion::create_temp_arrays): Likewise.
12405 (free_temp_arrays): Likewise.
12406 (gather_default_values): Likewise.
12407 (switch_conversion::gather_default_values): Likewise.
12408 (build_constructors): Likewise.
12409 (switch_conversion::build_constructors): Likewise.
12410 (constructor_contains_same_values_p): Likewise.
12411 (switch_conversion::contains_same_values_p): Likewise.
12412 (array_value_type): Likewise.
12413 (switch_conversion::array_value_type): Likewise.
12414 (build_one_array): Likewise.
12415 (switch_conversion::build_one_array): Likewise.
12416 (build_arrays): Likewise.
12417 (switch_conversion::build_arrays): Likewise.
12418 (gen_def_assigns): Likewise.
12419 (switch_conversion::gen_def_assigns): Likewise.
12420 (prune_bbs): Likewise.
12421 (switch_conversion::prune_bbs): Likewise.
12422 (fix_phi_nodes): Likewise.
12423 (switch_conversion::fix_phi_nodes): Likewise.
12424 (gen_inbound_check): Likewise.
12425 (switch_conversion::gen_inbound_check): Likewise.
12426 (process_switch): Use the newly created class.
12427 (switch_conversion::expand): New.
12428 (switch_conversion::~switch_conversion): New.
12429 * tree-switch-conversion.h: New file.
12431 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12433 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
12434 tree-vect-patterns.c.
12435 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
12436 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
12437 (vect_recog_sad_pattern): Likewise.
12438 (vect_recog_widen_sum_pattern): Likewise.
12439 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
12440 (vect_recog_widen_shift_pattern): Remove the type_in argument.
12441 (vect_recog_rotate_pattern): Likewise.
12442 (vect_recog_mult_pattern): Likewise.
12443 (vect_recog_vector_vector_shift_pattern): Likewise.
12444 (vect_recog_divmod_pattern): Likewise.
12445 (vect_recog_mixed_size_cond_pattern): Likewise.
12446 (vect_recog_bool_pattern): Likewise.
12447 (vect_recog_mask_conversion_pattern): Likewise.
12448 (vect_try_gather_scatter_pattern): Likewise.
12449 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
12450 (vect_recog_over_widening_pattern): Likewise.
12451 (vect_recog_gather_scatter_pattern): Likewise.
12452 (vect_recog_func_ptr): Move from tree-vectorizer.h
12453 (vect_vect_recog_func_ptrs): Move further down the file.
12454 (vect_recog_func): Likewise. Remove the third argument.
12455 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
12456 (vect_pattern_recog_1): Expect the pattern function to do any
12457 necessary target tests. Also expect it to provide a vector type.
12458 Remove the type_in handling.
12460 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12462 * tree-vect-patterns.c (vect_pattern_detected): New function.
12463 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
12464 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
12465 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
12466 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
12467 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
12468 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
12469 (vect_recog_mask_conversion_pattern)
12470 (vect_try_gather_scatter_pattern): Likewise.
12472 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12474 * tree-vect-patterns.c (vect_get_internal_def): New function.
12475 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
12476 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
12477 (search_type_for_mask_1): Use it.
12479 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12481 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
12482 redundant WIDEN_SUM_EXPR handling.
12483 (vect_recog_sad_pattern): Likewise.
12485 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12487 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
12488 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
12489 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
12490 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
12491 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
12493 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12495 * tree-vect-stmts.c (vectorizable_call): Make sure that we
12496 use the stmt_vec_info of the original bb statement for the
12497 new zero assignment, even if the call is part of a pattern.
12499 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12501 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
12502 that the sequence is attached to the original statement rather
12503 than the pattern statement.
12504 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
12505 PATTERN_DEF_SEQ from the original statement rather than
12506 the main pattern statement.
12507 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
12508 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
12509 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
12511 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12513 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
12514 definition statements before the early exit for statements that aren't
12516 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
12518 (vect_transform_loop): ...here. Process pattern definition
12519 statements without first checking whether the main pattern
12520 statement is live or relevant.
12522 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
12524 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
12525 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
12527 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
12529 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
12530 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
12531 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
12532 (expand_block_compare): Change select_block_compare_mode call.
12533 (expand_strncmp_align_check): Use new functions, fix comment.
12534 (emit_final_str_compare_gpr): New function.
12535 (expand_strn_compare): Refactor and clean up code.
12536 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
12538 2018-06-19 Tony Reix <tony.reix@atos.com>
12539 Damien Bergamini <damien.bergamini@atos.com>
12540 David Edelsohn <dje.gcc@gmail.com>
12542 * collect2.c (static_obj): New variable.
12543 (static_libs): New variable.
12544 (is_in_list): Uncomment declaration.
12545 (main): Track AIX libraries linked statically.
12546 (is_in_list): Uncomment definition.
12547 (scan_prog_file): Don't add AIX shared libraries initializer
12548 to constructor list if linking statically.
12550 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
12552 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
12554 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
12556 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
12558 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
12561 2018-06-19 Martin Liska <mliska@suse.cz>
12563 * config/i386/i386.c (ix86_can_inline_p): Do not use
12564 ipa_fn_summaries::get_create.
12565 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
12567 (devirtualization_time_bonus): Likewise.
12568 (ipcp_propagate_stage): Likewise.
12569 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
12570 (edge_set_predicate): Likewise.
12571 (evaluate_conditions_for_known_args): Likewise.
12572 (evaluate_properties_for_edge): Likewise.
12573 (ipa_call_summary::reset): Tranform to ...
12574 (ipa_call_summary::~ipa_call_summary): ... this.
12575 (ipa_fn_summary::reset): Transform to ...
12576 (ipa_fn_summary::~ipa_fn_summary): ... this.
12577 (ipa_fn_summary_t::remove): Rename to ...
12578 (ipa_fn_summary_t::remove_callees): ... this.
12579 (ipa_fn_summary_t::duplicate): Use placement new
12580 instead of memory copy.
12581 (ipa_call_summary_t::duplicate): Likewise.
12582 (ipa_call_summary_t::remove): Remove.
12583 (dump_ipa_call_summary): Change get_create to get.
12584 (ipa_dump_fn_summary): Dump only when summary exists.
12585 (analyze_function_body): Use symbol_summary::get instead
12587 (compute_fn_summary): Likewise.
12588 (estimate_edge_devirt_benefit): Likewise.
12589 (estimate_edge_size_and_time): Likewise.
12590 (inline_update_callee_summaries): Likewise.
12591 (remap_edge_change_prob): Likewise.
12592 (remap_edge_summaries): Likewise.
12593 (ipa_merge_fn_summary_after_inlining): Likewise.
12594 (write_ipa_call_summary): Likewise.
12595 (ipa_fn_summary_write): Likewise.
12596 (ipa_free_fn_summary): Likewise.
12597 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
12598 (struct ipa_call_summary): Likewise.
12599 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
12601 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
12602 (estimate_size_after_inlining): Likewise.
12603 (estimate_growth): Likewise.
12604 (growth_likely_positive): Likewise.
12605 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
12606 (inline_call): Likewise.
12607 * ipa-inline.c (caller_growth_limits): Likewise.
12608 (can_inline_edge_p): Likewise.
12609 (can_inline_edge_by_limits_p): Likewise.
12610 (compute_uninlined_call_time): Likewise.
12611 (compute_inlined_call_time): Likewise.
12612 (want_inline_small_function_p): Likewise.
12613 (edge_badness): Likewise.
12614 (update_caller_keys): Likewise.
12615 (update_callee_keys): Likewise.
12616 (inline_small_functions): Likewise.
12617 (inline_to_all_callers_1): Likewise.
12618 (dump_overall_stats): Likewise.
12619 (early_inline_small_functions): Likewise.
12620 (early_inliner): Likewise.
12621 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12622 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
12623 * ipa-pure-const.c (malloc_candidate_p): Likewise.
12624 * ipa-split.c (execute_split_functions): Likewise.
12625 * symbol-summary.h: Likewise.
12626 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
12628 2018-06-19 Richard Biener <rguenther@suse.de>
12630 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
12631 (vectorize_loops): ... here. Fix dbgcnt handling.
12632 (try_vectorize_loop): Wrap try_vectorize_loop_1.
12634 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
12637 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
12638 ieee128 argument takes up only one (vector) register, not two (floating
12641 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
12643 * gimplify.c (gimplify_init_constructor): Really never clear for an
12644 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
12646 2018-06-19 Richard Biener <rguenther@suse.de>
12648 PR tree-optimization/86179
12649 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
12650 after failed recognition.
12652 2018-06-18 Martin Sebor <msebor@redhat.com>
12654 PR middle-end/85602
12655 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
12656 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
12657 Handle integer subtraction.
12658 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
12659 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
12661 2018-06-18 David Malcolm <dmalcolm@redhat.com>
12663 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
12664 param from rtx to rtx_insn *.
12665 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
12667 (frv_ifcvt_modify_insn): Likwise.
12668 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
12669 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
12670 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
12671 as_a <rtx_insn *> cast to local "unprotected_region" once
12672 it's been established that it's not NULL or pc_rtx.
12673 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
12674 param "sethi" from rtx to rtx_insn *.
12675 (nds32_group_float_insns): Likewise for param "insn".
12676 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
12678 (vax_output_int_subtract): Likewise.
12679 * config/vax/vax.c (vax_output_int_add): Likewise for param
12681 (vax_output_int_subtract): Likewise.
12682 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
12683 (emit_pattern_after): Likewise for param "after".
12684 (emit_insn_after): Likewise.
12685 (emit_jump_insn_after): Likewise.
12686 (emit_call_insn_after): Likewise.
12687 (emit_debug_insn_after): Likewise.
12688 (emit_pattern_before): Likewise for param "before".
12689 (emit_insn_before): Likewise.
12690 (emit_jump_insn_before): Likewise.
12691 * final.c (get_insn_template): Likewise for param "insn", removing
12693 * output.h (get_insn_template): Likewise for 2nd param.
12694 * rtl.h (emit_insn_before): Likewise.
12695 (emit_jump_insn_before): Likewise.
12696 (emit_debug_insn_before_noloc): Likewise.
12697 (emit_insn_after): Likewise.
12698 (emit_jump_insn_after): Likewise.
12699 (emit_call_insn_after): Likewise.
12700 (emit_debug_insn_after): Likewise.
12701 (set_insn_deleted): Likewise for param.
12703 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
12706 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
12707 floating point modes, so that IFmode is numerically greater than
12708 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
12709 to declare the ordering. This prevents IFmode from being
12710 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
12711 machine. Include rs6000-modes.h to share the fractional values
12712 between genmodes* and the rest of the compiler.
12713 (IFmode): Likewise.
12714 (KFmode): Likewise.
12715 (TFmode): Likewise.
12716 * config/rs6000/rs6000-modes.h: New file.
12717 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
12718 meaning of rs6000_long_double_size so that 126..128 selects an
12719 appropriate 128-bit floating point type.
12720 (rs6000_option_override_internal): Likewise.
12721 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
12722 (TARGET_LONG_DOUBLE_128): Change the meaning of
12723 rs6000_long_double_size so that 126..128 selects an appropriate
12724 128-bit floating point type.
12725 (LONG_DOUBLE_TYPE_SIZE): Update comment.
12726 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
12727 source and destination to match the standard usage.
12728 (truncifkf2): Likewise.
12729 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
12730 ISA 2.07 to use an explicit clobber, instead of passing in a
12732 (copysign<mode>3_soft): Likewise.
12734 2018-06-18 David Malcolm <dmalcolm@redhat.com>
12736 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
12737 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
12738 (vect_slp_analyze_instance_dependence): Likewise.
12739 (vect_enhance_data_refs_alignment): Likewise.
12740 (vect_analyze_data_refs_alignment): Likewise.
12741 (vect_slp_analyze_and_verify_instance_alignment
12742 (vect_analyze_data_ref_accesses): Likewise.
12743 (vect_prune_runtime_alias_test_list): Likewise.
12744 (vect_analyze_data_refs): Likewise.
12745 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
12746 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
12747 (vect_analyze_scalar_cycles_1): Likewise.
12748 (vect_get_loop_niters): Likewise.
12749 (vect_analyze_loop_form_1): Likewise.
12750 (vect_update_vf_for_slp): Likewise.
12751 (vect_analyze_loop_operations): Likewise.
12752 (vect_analyze_loop): Likewise.
12753 (vectorizable_induction): Likewise.
12754 (vect_transform_loop): Likewise.
12755 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
12756 * tree-vect-slp.c (vect_analyze_slp): Likewise.
12757 (vect_make_slp_decision): Likewise.
12758 (vect_detect_hybrid_slp): Likewise.
12759 (vect_slp_analyze_operations): Likewise.
12760 (vect_slp_bb): Likewise.
12761 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
12762 (vectorizable_bswap): Likewise.
12763 (vectorizable_call): Likewise.
12764 (vectorizable_simd_clone_call): Likewise.
12765 (vectorizable_conversion): Likewise.
12766 (vectorizable_assignment): Likewise.
12767 (vectorizable_shift): Likewise.
12768 (vectorizable_operation): Likewise.
12769 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
12771 2018-06-18 Martin Sebor <msebor@redhat.com>
12773 PR tree-optimization/81384
12774 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
12775 * builtins.c (expand_builtin_strnlen): New function.
12776 (expand_builtin): Call it.
12777 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
12778 * builtins.def (BUILT_IN_STRNLEN): New.
12779 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
12780 Warn for bounds in excess of maximum object size.
12781 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
12782 single-value ranges. Handle strnlen.
12783 (handle_builtin_strlen): Handle strnlen.
12784 (strlen_check_and_optimize_stmt): Same.
12785 * doc/extend.texi (Other Builtins): Document strnlen.
12787 2018-06-18 Maya Rashish <coypu@sdf.org>
12789 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
12790 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
12791 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
12793 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
12795 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
12797 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12799 * tree.c (escaped_string::escape): Replace cast to char * by
12800 const_cast<char *> (unescaped).
12802 2018-06-18 Nick Clifton <nickc@redhat.com>
12805 * tree.c (escaped_string): New class. Converts an unescaped
12806 string into its escaped equivalent.
12807 (warn_deprecated_use): Use the new class to convert the
12808 deprecation message, if present.
12809 (test_escaped_strings): New self test.
12810 (test_c_tests): Add test_escaped_strings.
12811 * doc/extend.texi (deprecated): Add a note that the
12812 deprecation message is affected by the -fmessage-length
12813 option, and that control characters will be escaped.
12814 (#pragma GCC error): Document this pragma.
12815 (#pragma GCC warning): Likewise.
12816 * doc/invoke.texi (-fmessage-length): Document this option's
12817 effect on the #warning and #error preprocessor directives and
12818 the deprecated attribute.
12820 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
12822 * tree.c (decl_value_expr_lookup): Revert latest change.
12823 (decl_value_expr_insert): Likewise.
12825 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
12827 * gimplify.c (nonlocal_vlas): Delete.
12828 (nonlocal_vla_vars): Likewise.
12829 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
12831 (gimplify_body): Do not create and destroy nonlocal_vlas.
12832 * tree-nested.c: Include diagnostic.h.
12833 (use_pointer_in_frame): Tweak.
12834 (lookup_field_for_decl): Add assertion and declare the transformation.
12835 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
12836 internal error when the reference is in a wrong context. Do not
12837 create a debug decl by default.
12838 (note_nonlocal_block_vlas): Delete.
12839 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
12840 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
12841 create a debug decl by default.
12842 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
12843 call is in a wrong context.
12844 (fixup_vla_decls): New function.
12845 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
12846 debug variables were created.
12847 * tree.c (decl_value_expr_lookup): Add checking assertion.
12848 (decl_value_expr_insert): Likewise.
12850 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
12852 PR middle-end/82479
12853 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
12854 * tree-scalar-evolution.c (interpret_expr): Likewise.
12855 (expression_expensive_p): Likewise.
12856 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
12857 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
12858 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
12859 (ssa_defined_by_minus_one_stmt_p): New.
12861 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
12863 PR middle-end/64946
12864 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
12865 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
12866 * dojump.c (do_jump): Likewise.
12867 * expr.c (expand_expr_real_2): Check operand type's sign.
12868 * fold-const.c (const_unop): Handle ABSU_EXPR.
12869 (fold_abs_const): Likewise.
12870 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
12871 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
12872 (strip_sign_op_1): Likesise.
12873 * match.pd: Add new pattern to generate ABSU_EXPR.
12874 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
12875 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
12876 * tree-eh.c (operation_could_trap_helper_p): Likewise.
12877 * tree-inline.c (estimate_operator_cost): Likewise.
12878 * tree-pretty-print.c (dump_generic_node): Likewise.
12879 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
12880 * tree.def (ABSU_EXPR): New.
12882 2018-06-16 Jakub Jelinek <jakub@redhat.com>
12884 PR middle-end/86095
12885 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
12886 documented as preserved for backward compatibility only.
12887 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
12889 PR rtl-optimization/86108
12890 * bb-reorder.c (create_forwarder_block): Renamed to ...
12891 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
12892 jump from new landing pad to the second part.
12893 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
12896 2018-06-15 Jakub Jelinek <jakub@redhat.com>
12898 PR middle-end/85878
12899 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
12900 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
12901 Only call store_expr for halves if the mode is the same.
12903 PR middle-end/86123
12904 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
12905 Fix up comment formatting.
12907 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
12909 * typed-splay-tree.h (typed_splay_tree::remove): New function.
12910 (typed_splay_tree::closure,
12911 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
12912 (typed_splay_tree::typed_splay_tree,
12913 typed_splay_tree::operator =): Declared private.
12914 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
12915 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
12916 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
12917 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
12918 typed_splay_tree::splay_tree_splay,
12919 typed_splay_tree::splay_tree_foreach_helper,
12920 typed_splay_tree::splay_tree_insert,
12921 typed_splay_tree::splay_tree_remove,
12922 typed_splay_tree::splay_tree_lookup,
12923 typed_splay_tree::splay_tree_predecessor,
12924 typed_splay_tree::splay_tree_successor,
12925 typed_splay_tree::splay_tree_min,
12926 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
12927 (typed_splay_tree::root, typed_splay_tree::comp,
12928 typed_splay_tree::delete_key,
12929 typed_splay_tree::delete_value): New data members.
12930 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
12931 typed_splay_tree::remove.
12933 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
12935 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
12936 -mginv and -mno-ginv to the assembler.
12937 * config/mips/mips.opt (-mcrc): New option.
12938 (-mginv): Likewise.
12939 * doc/invoke.text (-mcrc): Document.
12940 (-mginv): Likewise.
12942 2018-06-15 Nick Clifton <nickc@redhat.com>
12945 * tree.c (escaped_string): New class. Converts an unescaped
12946 string into its escaped equivalent.
12947 (warn_deprecated_use): Use the new class to convert the
12948 deprecation message, if present.
12949 (test_escaped_strings): New self test.
12950 (test_c_tests): Add test_escaped_strings.
12951 * doc/extend.texi (deprecated): Add a note that the
12952 deprecation message is affected by the -fmessage-length
12953 option, and that control characters will be escaped.
12954 (#pragma GCC error): Document this pragma.
12955 (#pragma GCC warning): Likewise.
12956 * doc/invoke.texi (-fmessage-length): Document this option's
12957 effect on the #warning and #error preprocessor directives and
12958 the deprecated attribute.
12960 2018-06-15 Richard Biener <rguenther@suse.de>
12962 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
12963 here, also noting vector size used.
12964 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
12965 size used in MSG_OPTIMIZED_LOCATIONS dump.
12966 (pass_slp_vectorize::execute): Adjust.
12968 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
12971 * config/arc/arc.c (arc_return_address_register): Fix
12974 2018-06-15 Richard Biener <rguenther@suse.de>
12976 PR middle-end/86159
12977 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
12978 leave useless conversion stripping to force_gimple_operand_gsi.
12979 (gimplify_build2): Likewise.
12980 (gimplify_build1): Likewise.
12982 2018-06-15 Richard Biener <rguenther@suse.de>
12984 PR middle-end/86076
12985 * tree-cfg.c (move_stmt_op): unshare invariant addresses
12986 before adjusting their block.
12988 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
12990 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
12991 multilibs for *-*-rtems*.
12992 * config/riscv/t-rtems: New file.
12994 2018-06-14 Jakub Jelinek <jakub@redhat.com>
12996 PR middle-end/86122
12997 * match.pd ((A +- CST1) +- CST2): Punt if last resort
12998 unsigned_type_for returns NULL.
13001 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
13002 subregs of multi-word pseudos unless the float mode has word size.
13004 2018-06-14 Richard Biener <rguenther@suse.de>
13006 PR middle-end/86139
13007 * tree-vect-generic.c (build_word_mode_vector_type): Remove
13008 duplicate and harmful type_hash_canon.
13009 * tree.c (type_hash_canon): Assert we didn't find ourselves.
13011 2018-06-14 Richard Biener <rguenther@suse.de>
13014 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
13017 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
13019 * config/rtems.h (STDINT_LONG32): Define.
13021 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
13022 Prachi Godbole <prachi.godbole@imgtec.com>
13024 * config/mips/mips-cpus.def: Define P6600.
13025 * config/mips/mips-tables.opt: Regenerate.
13026 * config/mips/mips.c (mips_ucbranch_type): New enum.
13027 (mips_rtx_cost_data): Add support for P6600.
13028 (mips_issue_rate): Likewise.
13029 (mips_multipass_dfa_lookahead): Likewise.
13030 (mips_avoid_hazard): Likewise.
13031 (mips_reorg_process_insns): Likewise.
13032 (mips_classify_branch_p6600): New function.
13033 * config/mips/mips.h (TUNE_P6600): New define.
13034 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
13035 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
13036 * config/mips/mips.md: Include p6600.md.
13037 (processor): Add p6600.
13038 * config/mips/p6600.md: New file.
13039 * doc/invoke.texi: Add p6600 to supported architectures.
13041 2018-06-13 Martin Sebor <msebor@redhat.com>
13043 PR tree-optimization/86114
13044 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
13046 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
13048 2018-06-13 Richard Biener <rguenther@suse.de>
13050 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
13051 Properly set vector type of the intermediate stmt.
13052 * tree-vect-stmts.c (vectorizable_operation): The destination
13053 var always has vectype_out type.
13055 2018-06-13 Jeff Law <law@redhat.com>
13057 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
13058 integer 0 for argument to print_rtl_with_bb.
13059 (rl78_reorg): Likewise.
13061 2018-06-13 David Malcolm <dmalcolm@redhat.com>
13063 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
13064 from rtx to rtx_insn *.
13065 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
13067 (add_sched_insns_for_speculation): Likewise for local "target",
13068 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
13069 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
13070 from rtx_insn ** to rtx_code_label **.
13071 (reorg_emit_nops): Likewise.
13072 (c6x_reorg): Likewise for local "call_labels".
13073 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
13075 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
13076 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
13077 the loops over LABEL_REFS.
13078 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
13080 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
13081 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
13082 (split_branches): Strengthen local "olabel" from rtx to
13083 rtx_insn *, adding a safe_as_a cast.
13084 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
13086 (add_insn_after): Likewise for first two params.
13087 (add_insn_before): Likewise.
13088 (remove_insn): Likewise for param.
13089 (emit_pattern_before_noloc): Likewise for second and third params.
13090 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
13091 (emit_call_insn_before_noloc): Likewise.
13092 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
13094 (emit_barrier_before): Likewise.
13095 (emit_label_before): Strengthen "label" param from "rtx" to
13096 "rtx_code_label *". Strengthen "before" param from "rtx" to
13098 (emit_insn_after_1): Strengthen "after" param from "rtx" to
13100 (emit_pattern_after_noloc): Likewise.
13101 (emit_insn_after_noloc): Likewise.
13102 (emit_jump_insn_after_noloc): Likewise.
13103 (emit_call_insn_after_noloc): Likewise.
13104 (emit_debug_insn_after_noloc): Likewise.
13105 (emit_barrier_after): Likewise.
13106 (emit_label_after): Likewise for both params.
13107 (emit_pattern_after_setloc): Likewise for "after" param. Convert
13108 "loc" param from "int" to "location_t".
13109 (emit_insn_after_setloc): Likewise.
13110 (emit_jump_insn_after_setloc): Likewise.
13111 (emit_call_insn_after_setloc): Likewise.
13112 (emit_debug_insn_after_setloc): Likewise.
13113 (emit_pattern_before_setloc): Likewise for "before" param. Convert
13114 "loc" param from "int" to "location_t".
13115 (emit_pattern_before): Convert NULL_RTX to NULL.
13116 (emit_insn_before_setloc): Convert "loc" param from "int" to
13118 (emit_jump_insn_before_setloc): Likewise.
13119 (emit_call_insn_before_setloc): Likewise.
13120 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
13121 rtx_insn *. Convert "loc" param from "int" to "location_t".
13122 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
13123 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
13124 Convert 3rd param from "int" to "location_t".
13125 (emit_barrier_before, emit_barrier_after, next_real_insn):
13126 Strengthen param from rtx to rtx_insn *.
13127 (emit_label_before): Strengthen 1st param from "rtx" to
13128 "rtx_code_label *". Strengthen 2nd param from "rtx" to
13130 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
13131 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
13132 Strengthen 2nd param from "rtx" to "rtx_insn *".
13133 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
13134 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
13135 Likewise. Convert 3rd param from "int" to "location_t".
13136 (emit_label_after): Strengthen 1st param from "rtx" to
13137 "rtx_code_label *".
13138 (next_real_insn, remove_insn): Strengthen param from "rtx" to
13140 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
13141 from "rtx" to "rtx_insn *".
13143 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
13145 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
13146 bodies streamed in with -Q.
13147 * dumpfile.c (dump_files): Add lto-stream-out dump file.
13148 * dumpfile.h (tree_dump_index): Add lto_stream_out.
13149 * gimple-streamer-out.c: Include gimple-pretty-print.h
13150 (output_bb): Dump stmts streamed.
13151 * lto-section-out.c: Include print-tree.h
13152 (lto_begin_section): Dump sections created.
13153 (lto_output_decl_index): Dump decl encoded.
13154 * lto-streamer-out.c: Include print-tree.h
13155 (create_output_block): Dump output block created.
13156 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
13157 (output_function): Dump function output.
13158 (output_constructor): Dump constructor streamed.
13159 (write_global_stream): Output indexes encoded.
13160 (produce_asm_for_decls): Dump streams encoded.
13161 * lto-streamer.c (streamer_dump_file): New global var.
13162 * lto-streamer.h (streamer_dump_file): Declare.
13163 * passes.c (ipa_write_summaries): Initialize streamer dump.
13164 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
13167 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
13170 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
13171 offsets for register save directives. Emit a second batch of save
13172 directives, if need be, when the function accesses prior frames.
13174 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13176 * config/arc/fpu.md (fmasf4): Force operand to register.
13177 (fnmasf4): Likewise.
13179 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13181 * config/arc/arc-protos.h (arc_pad_return): Remove.
13182 * config/arc/arc.c (machine_function): Remove force_short_suffix
13184 (arc_print_operand): Adjust printing of '&'.
13185 (arc_verify_short): Remove conditional printing of short suffix.
13186 (arc_final_prescan_insn): Remove reference to size_reason.
13187 (pad_return): New function.
13188 (arc_reorg): Call pad_return.
13189 (arc_pad_return): Remove.
13190 (arc_init_machine_status): Remove reference to force_short_suffix.
13191 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
13192 (attr length): When attribute iscompact is true force to 2
13193 regardless; in the case of maybe check if we want to force the
13194 instruction to have 4 bytes length.
13195 (nopv): Change it to generate 4 byte long nop as well.
13196 (blockage): New pattern.
13197 (simple_return): Remove call to arc_pad_return.
13198 (p_return_i): Likewise.
13200 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13202 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
13204 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13206 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
13209 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13211 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
13212 for ARC700 and ARCv2.
13214 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
13217 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
13218 operands[2] instead of operands[1].
13221 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13223 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
13224 case, check whether the outer register overlaps an unallocatable
13225 register, not just whether it fits the required class.
13227 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13229 * poly-int.h (can_div_trunc_p): Add new overload in which all values
13231 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
13232 (memrefs_conflict_p): Likewise.
13233 (init_alias_analysis): Likewise.
13234 * cfgexpand.c (expand_debug_expr): Likewise.
13235 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
13236 * cse.c (fold_rtx): Likewise.
13237 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
13238 * expr.c (emit_block_move_hints): Likewise.
13239 (clear_storage_hints, push_block, emit_push_insn): Likewise.
13240 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
13241 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
13242 (emit_group_store): Likewise.
13243 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
13244 to read the PRE/POST_MODIFY increment.
13245 * calls.c (store_one_arg): Use strip_offset.
13246 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
13248 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
13250 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
13251 (simplify_binary_operation_1): Extend CONST_INT handling to
13253 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
13254 than a HOST_WIDE_INT.
13255 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
13257 (adjust_mems, add_stores): Update accodingly.
13258 (vt_canonicalize_addr): Track polynomial offsets.
13259 (emit_note_insn_var_location): Likewise.
13260 (vt_add_function_parameter): Likewise.
13261 (vt_initialize): Likewise.
13263 2018-06-12 Jeff Law <law@redhat.com>
13265 * config.gcc (alpha*-*-freebsd*): Remove.
13266 * config/alpha/freebsd.h: Remove.
13268 2018-06-12 David Malcolm <dmalcolm@redhat.com>
13271 * spellcheck-tree.c (levenshtein_distance): Rename to...
13272 (get_edit_distance): ...this, and update for underlying renaming.
13273 * spellcheck-tree.h (levenshtein_distance): Rename to...
13274 (get_edit_distance): ...this.
13275 * spellcheck.c (levenshtein_distance): Rename to...
13276 (get_edit_distance): ...this. Convert from Levenshtein distance
13277 to Damerau-Levenshtein distance by supporting transpositions of
13278 adjacent characters. Rename "v1" to "v_next" and "v0" to
13280 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
13281 (selftest::test_edit_distance_unit_test_oneway): ...this, and
13282 update for underlying renaming.
13283 (selftest::levenshtein_distance_unit_test): Rename to...
13284 (selftest::test_get_edit_distance_unit): ...this, and update for
13285 underlying renaming.
13286 (selftest::test_find_closest_string): Add example from PR 69968
13287 where transposition helps
13288 (selftest::test_metric_conditions): Update for renaming.
13289 (selftest::test_metric_conditions): Likewise.
13290 (selftest::spellcheck_c_tests): Likewise.
13291 * spellcheck.h (levenshtein_distance): Rename both overloads to...
13292 (get_edit_distance): ...this.
13293 (best_match::consider): Update for renaming.
13295 2018-06-12 Martin Sebor <msebor@redhat.com>
13297 PR tree-optimization/85259
13298 * builtins.c (compute_objsize): Handle constant offsets.
13299 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
13300 true iff a warning has been issued.
13301 * gimple.h (gimple_nonartificial_location): New function.
13302 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
13303 gimple_nonartificial_location and handle -Wno-system-headers.
13304 (handle_builtin_stxncpy): Same.
13306 2018-06-12 Martin Sebor <msebor@redhat.com>
13309 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
13311 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13313 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13314 BUILTIN_VEC_XST entries for pointer to double and long long.
13316 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
13319 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
13321 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
13324 2018-06-12 Martin Liska <mliska@suse.cz>
13326 * doc/options.texi: Document IntegerRange.
13328 2018-06-12 Martin Liska <mliska@suse.cz>
13330 * config/i386/i386.opt: Make MPX-related options as Deprecated.
13331 * opt-functions.awk: Handle Deprecated flag.
13332 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
13334 (read_cmdline_option): Report warning for a deprecated option.
13335 * opts.h (struct cl_option): Add new field cl_deprecated.
13336 (CL_ERR_DEPRECATED): New.
13338 2018-06-12 Martin Liska <mliska@suse.cz>
13340 * doc/options.texi: Document Deprecated option flag.
13342 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13344 * config/arc/arc-arch.h (arc_extras): New enum.
13345 (arc_cpu_t):Add field extra.
13346 (arc_cpu_types): Consider the extras.
13347 * config/arc/arc-cpus.def: Add extras info.
13348 * config/arc/arc-opts.h (processor_type): Consider extra field.
13349 * config/arc/arc.c (arc_override_options): Handle extra field.
13351 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13353 * config/arc/arc-arch.h: Update ARC_OPTX macro.
13354 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
13356 * config/arc/arc.c (arc_init): Update pic warning.
13357 (irq_range): Update irq range parsing warnings.
13358 (arc_override_options): Update various warning messages.
13359 (arc_handle_aux_attribute): Likewise.
13361 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
13363 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
13365 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
13367 * doc/sourcebuild.texi: Document usage of line number 0 in verify
13368 compiler messages directives.
13370 2018-06-12 Matthew Fortune <mfortune@gmail.com>
13372 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
13373 * config/mips/mips-tables.opt: Regenerate.
13374 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
13376 * doc/invoke.texi: Document -march=i6500.
13378 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
13380 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
13381 (i6400_gpmul): Add cpu_unit.
13382 (i6400_gpdiv): Likewise.
13383 (i6400_msa_add_d): Update reservations.
13384 (i6400_msa_int_add) Likewise.
13385 (i6400_msa_short_logic3) Likewise.
13386 (i6400_msa_short_logic2) Likewise.
13387 (i6400_msa_short_logic) Likewise.
13388 (i6400_msa_move) Likewise.
13389 (i6400_msa_cmp) Likewise.
13390 (i6400_msa_short_float2) Likewise.
13391 (i6400_msa_div_d) Likewise.
13392 (i6400_msa_long_logic1) Likewise.
13393 (i6400_msa_long_logic2) Likewise.
13394 (i6400_msa_mult) Likewise.
13395 (i6400_msa_long_float2) Likewise.
13396 (i6400_msa_long_float4) Likewise.
13397 (i6400_msa_long_float5) Likewise.
13398 (i6400_msa_long_float8) Likewise.
13399 (i6400_fpu_fadd): Include frint type.
13400 (i6400_fpu_store): New define_insn_reservation.
13401 (i6400_fpu_load): Likewise.
13402 (i6400_fpu_move): Likewise.
13403 (i6400_fpu_fcmp): Likewise.
13404 (i6400_fpu_fmadd): Likewise.
13405 (i6400_int_mult): Include imul3nc type and update reservation.
13406 (i6400_int_div): Include idiv3 type and update reservation.
13407 (i6400_int_load): Update to check type not move_type.
13408 (i6400_int_store): Likewise.
13409 (i6400_int_prefetch): Set zero latency.
13411 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
13413 * gcc.c: Document new %@{...} sequence.
13414 (LINK_COMMAND_SPEC): Use it for the -L switches.
13415 (cpp_unique_options): Use it for the -I switches.
13416 (at_file_argbuf): New global variable.
13417 (in_at_file): Likewise.
13418 (alloc_args): Create at_file_argbuf.
13419 (clear_args): Truncate at_file_argbuf.
13420 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
13421 (open_at_file): New function.
13422 (close_at_file): Likewise.
13423 (create_at_file): Delete.
13424 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
13427 (validate_switches_from_spec): Deal with %@{...} sequence.
13428 (validate_switches): Likewise.
13429 (driver::finalize): Call clear_args.
13431 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
13433 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
13435 2018-06-11 Martin Sebor <msebor@redhat.com>
13437 * doc/invoke.texi (-Wall): List -Wc++17-compat.
13438 (Wno-class-memaccess): Add @opindex.
13439 (Wno-templates, Wno-multiple-inheritance): Same.
13440 (Wno-virtual-inheritance, Wno-namespaces): Same.
13441 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
13442 (Wno-format-overflow, Wno-format-truncation): Same.
13443 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
13444 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
13445 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
13446 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
13447 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
13448 (Wno-misspelled-isr): Same.
13450 2018-06-11 Martin Sebor <msebor@redhat.com>
13452 * PR tree-optimization/86083
13453 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
13455 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
13457 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
13459 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
13462 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
13463 on the correct operand.
13464 (*movdi_internal64): Ditto.
13466 2018-06-11 Martin Liska <mliska@suse.cz>
13468 PR tree-optimization/86089
13469 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
13471 2018-06-11 Julia Koval <julia.koval@intel.com>
13473 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
13474 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
13475 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
13477 2018-06-11 Olivier Hainque <hainque@adacore.com>
13479 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
13480 for Ada with strict dwarf2.
13482 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
13485 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
13488 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
13490 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
13492 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
13494 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
13497 2018-06-08 Martin Liska <mliska@suse.cz>
13499 * tree-cfg.h (debug_function): Fix argument type to match
13502 2018-06-08 Martin Liska <mliska@suse.cz>
13504 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
13505 Remove usage of MPX-related (and removed) fields.
13506 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
13508 2018-06-08 David Malcolm <dmalcolm@redhat.com>
13510 * cfg.c (debug): Use TDF_NONE rather than 0.
13511 * cfghooks.c (debug): Likewise.
13512 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
13513 (struct dump_option_value_info): Convert to...
13514 (struct kv_pair): ...this template type.
13515 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
13517 (optinfo_verbosity_options): Likewise.
13518 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
13520 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
13521 than int for "optgroup_flags" param.
13522 (dump_generic_expr_loc): Use dump_flags_t rather than int for
13524 (dump_dec): Likewise.
13525 (dump_finish): Use TDF_NONE rather than 0.
13526 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
13527 rather than int for "optgroup_flags" param. Use TDF_NONE rather
13528 than 0. Update for change to option_ptr.
13529 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
13530 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
13531 0. Update for changes to optinfo_verbosity_options and
13533 (opt_info_switch_p): Convert optgroup_flags from int to
13535 (dump_basic_block): Use dump_flags_t rather than int
13536 for "dump_kind" param.
13537 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
13538 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
13539 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
13540 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
13541 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
13542 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
13543 TDF_NONE): Convert from macros to...
13544 (enum dump_flag): ...this new enum.
13545 (dump_flags_t): Update to use enum.
13546 (operator|, operator&, operator~, operator|=, operator&=):
13547 Implement for dump_flags_t.
13548 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
13549 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
13550 Convert from macros to...
13551 (enum optgroup_flag): ...this new enum.
13552 (optgroup_flags_t): New typedef.
13553 (operator|, operator|=): Implement for optgroup_flags_t.
13554 (struct dump_file_info): Convert field "alt_flags" to
13555 dump_flags_t. Convert field "optgroup_flags" to
13557 (dump_basic_block): Use dump_flags_t rather than int for param.
13558 (dump_generic_expr_loc): Likewise.
13559 (dump_dec): Likewise.
13560 (dump_register): Convert param "optgroup_flags" to
13562 (opt_info_enable_passes): Likewise.
13563 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
13565 * gimple-pretty-print.c (debug): Likewise.
13566 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
13567 (merged_store_group::apply_stores): Likewise.
13568 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
13569 * gimple.c (verify_gimple_pp): Likewise.
13570 * graphite-poly.c (print_pbb_body): Likewise.
13571 * passes.c (pass_manager::register_one_dump_file): Convert
13572 local "optgroup_flags" to optgroup_flags_t.
13573 * print-tree.c (print_node): Use TDF_NONE rather than 0.
13575 (debug_body): Likewise.
13576 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
13577 to optgroup_flags_t.
13578 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
13580 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
13581 (convert_mult_to_fma): Likewise.
13582 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
13583 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
13584 * tree-vect-data-refs.c (dump_lower_bound): Convert param
13585 "dump_kind" to dump_flags_t.
13587 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
13589 * config/rs6000/rs6000.c (min, max): Delete.
13591 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
13593 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
13594 -mabi=spe and -mabi=no-spe.
13596 2018-06-08 Martin Liska <mliska@suse.cz>
13598 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
13599 where we expect an existing summary.
13601 2018-06-08 Martin Liska <mliska@suse.cz>
13603 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
13604 * ipa-inline.h (estimate_edge_growth): Likewise.
13606 2018-06-08 Martin Liska <mliska@suse.cz>
13608 * cgraph.c (function_version_hasher::hash): Use
13609 cgraph_node::get_uid ().
13610 (function_version_hasher::equal):
13611 * cgraph.h (cgraph_node::get_uid): New method.
13612 * ipa-inline.c (update_caller_keys): Use
13613 cgraph_node::get_uid ().
13614 (update_callee_keys): Likewise.
13615 * ipa-utils.c (searchc): Likewise.
13616 (ipa_reduced_postorder): Likewise.
13617 * lto-cgraph.c (input_node): Likewise.
13618 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
13619 * symbol-summary.h (symtab_insertion): Likewise.
13620 (symtab_removal): Likewise.
13621 (symtab_duplication): Likewise.
13622 * tree-pretty-print.c (dump_function_header): Likewise.
13623 * tree-sra.c (convert_callers_for_node): Likewise.
13625 2018-06-08 Martin Liska <mliska@suse.cz>
13627 * cgraph.c (symbol_table::create_edge): Always assign a new
13629 (symbol_table::free_edge): Do not recycle numbers.
13630 * cgraph.h (cgraph_edge::get): New method.
13631 * symbol-summary.h (symtab_removal): Use it.
13632 (symtab_duplication): Likewise.
13633 (call_summary::hashable_uid): Remove.
13635 2018-06-08 Martin Liska <mliska@suse.cz>
13637 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
13638 (initialize_growth_caches): Remove.
13639 (free_growth_caches): Likewise.
13640 (do_estimate_edge_time): Use edge_growth_cache.
13641 (do_estimate_edge_size): Likewise.
13642 (do_estimate_edge_hints): Likewise.
13643 * ipa-inline.c (reset_edge_caches): Likewise.
13644 (recursive_inlining): Likewise.
13645 (inline_small_functions): Likewise.
13646 * ipa-inline.h (initialize_growth_caches): Remove.
13647 (estimate_edge_size): Likewise.
13648 (estimate_edge_time): Likewise.
13649 (estimate_edge_hints): Likewise.
13650 (reset_edge_growth_cache): Likewise.
13651 * symbol-summary.h (call_summary::remove): New method.
13653 2018-06-08 Martin Liska <mliska@suse.cz>
13655 * ipa-cp.c (class edge_clone_summary): New summary.
13656 (grow_edge_clone_vectors): Remove.
13657 (ipcp_edge_duplication_hook): Remove.
13658 (class edge_clone_summary_t): New call_summary class.
13659 (ipcp_edge_removal_hook): Remove.
13660 (edge_clone_summary_t::duplicate): New function.
13661 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
13662 (create_specialized_node): Likewise.
13663 (ipcp_driver): Initialize edge_clone_summaries and do not
13666 2018-06-08 Martin Liska <mliska@suse.cz>
13668 * symbol-summary.h (get): New function.
13669 (call_summary::m_initialize_when_cloning): New class member.
13671 2018-06-08 Martin Liska <mliska@suse.cz>
13673 * cgraph.c (cgraph_node::remove): Do not recycle uid.
13674 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
13675 (symbol_table::allocate_cgraph_symbol): Do not set uid.
13676 * passes.c (uid_hash_t): Record removed_nodes by their uids.
13677 (remove_cgraph_node_from_order): Use the removed_nodes set.
13678 (do_per_function_toporder): Likwise.
13679 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
13680 instead of summary_uid.
13681 (symtab_removal): Likewise.
13682 (symtab_duplication): Likewise.
13684 2018-06-08 Martin Liska <mliska@suse.cz>
13686 * ipa-cp.c (ipcp_store_bits_results): Use
13687 ipcp_transformation_sum.
13688 (ipcp_store_vr_results): Likewise.
13689 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
13691 (ipcp_transformation_initialize): ... this.
13692 (ipa_set_node_agg_value_chain):
13693 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
13694 (write_ipcp_transformation_info): Likewise.
13695 (read_ipcp_transformation_info): Likewise.
13696 (ipcp_update_bits): Likewise.
13697 (ipcp_update_vr): Likewise.
13698 (ipcp_transform_function): Likewise.
13699 * ipa-prop.h: Rename ipcp_transformation_summary to
13700 ipcp_transformation.
13701 (class ipcp_transformation_t): New function summary.
13702 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
13703 (ipa_get_agg_replacements_for_node): Likewise.
13705 2018-06-08 Martin Liska <mliska@suse.cz>
13707 * ipa-pure-const.c (struct funct_state_d): Do it class instead
13709 (class funct_state_summary_t): New function_summary class.
13710 (has_function_state): Remove.
13711 (get_function_state): Likewise.
13712 (set_function_state): Likewise.
13713 (add_new_function): Likewise.
13714 (funct_state_summary_t::insert): New function.
13715 (duplicate_node_data): Remove.
13716 (remove_node_data): Remove.
13717 (funct_state_summary_t::duplicate): New function.
13718 (register_hooks): Create new funct_state_summaries.
13719 (pure_const_generate_summary): Use it.
13720 (pure_const_write_summary): Likewise.
13721 (pure_const_read_summary): Likewise.
13722 (propagate_pure_const): Likewise.
13723 (propagate_nothrow): Likewise.
13724 (dump_malloc_lattice): Likewise.
13725 (propagate_malloc): Likewise.
13726 (execute): Do not register hooks, just remove summary
13728 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
13731 2018-06-08 Martin Liska <mliska@suse.cz>
13733 * ipa-reference.c (remove_node_data): Remove.
13734 (duplicate_node_data): Likewise.
13735 (class ipa_ref_var_info_summary_t): New class.
13736 (class ipa_ref_opt_summary_t): Likewise.
13737 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
13738 (get_reference_optimization_summary): Use
13739 ipa_ref_opt_sum_summaries.
13740 (set_reference_vars_info): Remove.
13741 (set_reference_optimization_summary): Likewise.
13742 (ipa_init): Create summaries.
13743 (init_function_info): Use function summary.
13744 (ipa_ref_opt_summary_t::duplicate): New function.
13745 (ipa_ref_opt_summary_t::remove): New function.
13746 (get_read_write_all_from_node): Fix GNU coding style.
13747 (propagate): Use function summary.
13748 (write_node_summary_p): Fix GNU coding style.
13749 (stream_out_bitmap): Likewise.
13750 (ipa_reference_read_optimization_summary): Use function summary.
13751 (ipa_reference_c_finalize): Do not release hooks.
13753 2018-06-08 Martin Liska <mliska@suse.cz>
13755 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
13756 (analyze_function_body): Extract multiple calls of get_create.
13757 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
13758 * ipa-inline.c (recursive_inlining): Use ::get method.
13759 * ipa-inline.h (estimate_edge_growth): Likewise.
13761 2018-06-08 Martin Liska <mliska@suse.cz>
13763 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
13765 (hsa_function_summary::hsa_function_summary): Use the new enum
13767 (hsa_gpu_implementation_p): Use hsa_summaries::get.
13768 * hsa-gen.c (hsa_get_host_function): Likewise.
13769 (get_brig_function_name): Likewise.
13770 * ipa-hsa.c (process_hsa_functions): Likewise.
13771 (ipa_hsa_write_summary): Likewise.
13772 * symbol-summary.h (symtab_duplication): Use ::get function/
13773 (get): New function.
13775 2018-06-08 Martin Liska <mliska@suse.cz>
13777 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
13779 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
13780 (hsa_register_kernel): Likewise.
13781 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
13782 * hsa-gen.c (hsa_get_host_function): Likewise.
13783 (get_brig_function_name): Likewise.
13784 (generate_hsa): Likewise.
13785 (pass_gen_hsail::execute): Likewise.
13786 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
13787 (devirtualization_time_bonus): Likewise.
13788 (ipcp_propagate_stage): Likewise.
13789 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
13790 (edge_set_predicate): Likewise.
13791 (evaluate_conditions_for_known_args): Likewise.
13792 (evaluate_properties_for_edge): Likewise.
13793 (ipa_fn_summary::reset): Likewise.
13794 (ipa_fn_summary_t::duplicate): Likewise.
13795 (dump_ipa_call_summary): Likewise.
13796 (ipa_dump_fn_summary): Likewise.
13797 (analyze_function_body): Likewise.
13798 (compute_fn_summary): Likewise.
13799 (estimate_edge_devirt_benefit): Likewise.
13800 (estimate_edge_size_and_time): Likewise.
13801 (estimate_calls_size_and_time): Likewise.
13802 (estimate_node_size_and_time): Likewise.
13803 (inline_update_callee_summaries): Likewise.
13804 (remap_edge_change_prob): Likewise.
13805 (remap_edge_summaries): Likewise.
13806 (ipa_merge_fn_summary_after_inlining): Likewise.
13807 (ipa_update_overall_fn_summary): Likewise.
13808 (read_ipa_call_summary): Likewise.
13809 (inline_read_section): Likewise.
13810 (write_ipa_call_summary): Likewise.
13811 (ipa_fn_summary_write): Likewise.
13812 (ipa_free_fn_summary): Likewise.
13813 * ipa-hsa.c (process_hsa_functions): Likewise.
13814 (ipa_hsa_write_summary): Likewise.
13815 (ipa_hsa_read_section): Likewise.
13816 * ipa-icf.c (sem_function::merge): Likewise.
13817 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
13818 (do_estimate_edge_time): Likewise.
13819 (estimate_size_after_inlining): Likewise.
13820 (estimate_growth): Likewise.
13821 (growth_likely_positive): Likewise.
13822 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
13823 (inline_call): Likewise.
13824 * ipa-inline.c (caller_growth_limits): Likewise.
13825 (can_inline_edge_p): Likewise.
13826 (can_inline_edge_by_limits_p): Likewise.
13827 (compute_uninlined_call_time): Likewise.
13828 (compute_inlined_call_time): Likewise.
13829 (want_inline_small_function_p): Likewise.
13830 (edge_badness): Likewise.
13831 (update_caller_keys): Likewise.
13832 (update_callee_keys): Likewise.
13833 (recursive_inlining): Likewise.
13834 (inline_small_functions): Likewise.
13835 (inline_to_all_callers_1): Likewise.
13836 (dump_overall_stats): Likewise.
13837 (early_inline_small_functions): Likewise.
13838 (early_inliner): Likewise.
13839 * ipa-inline.h (estimate_edge_growth): Likewise.
13840 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13841 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
13842 * ipa-prop.h (IPA_NODE_REF): Likewise.
13843 (IPA_EDGE_REF): Likewise.
13844 * ipa-pure-const.c (malloc_candidate_p): Likewise.
13845 (propagate_malloc): Likewise.
13846 * ipa-split.c (execute_split_functions): Likewise.
13847 * symbol-summary.h: Rename get to get_create.
13849 (get_create): Likewise.
13850 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
13852 2018-06-08 Martin Liska <mliska@suse.cz>
13854 * symbol-summary.h (release): Move definition out of class
13856 (symtab_removal): Likewise.
13857 (symtab_duplication): Likewise.
13859 2018-06-08 Martin Liska <mliska@suse.cz>
13861 * symbol-summary.h (function_summary): Move constructor
13862 implementation out of class declaration.
13863 (release): Likewise.
13864 (symtab_insertion): Likewise.
13865 (symtab_removal): Likewise.
13866 (symtab_duplication): Likewise.
13869 2018-06-08 Martin Liska <mliska@suse.cz>
13871 * Makefile.in: Remove support for MPX (macros, related functions,
13872 fields in cgraph_node, ...).
13873 * builtin-types.def (BT_BND): Likewise.
13874 (BT_FN_BND_CONST_PTR): Likewise.
13875 (BT_FN_CONST_PTR_BND): Likewise.
13876 (BT_FN_VOID_PTR_BND): Likewise.
13877 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
13878 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
13879 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
13880 (expand_builtin_mempcpy_with_bounds): Likewise.
13881 (expand_builtin_memset_with_bounds): Likewise.
13882 (expand_builtin_memset_args): Likewise.
13883 (std_expand_builtin_va_start): Likewise.
13884 (expand_builtin): Likewise.
13885 (expand_builtin_with_bounds): Likewise.
13886 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
13887 (DEF_LIB_BUILTIN_CHKP): Likewise.
13888 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
13889 (DEF_CHKP_BUILTIN): Likewise.
13890 (BUILT_IN_MEMCPY): Likewise.
13891 (BUILT_IN_MEMMOVE): Likewise.
13892 (BUILT_IN_MEMPCPY): Likewise.
13893 (BUILT_IN_MEMSET): Likewise.
13894 (BUILT_IN_STPCPY): Likewise.
13895 (BUILT_IN_STRCAT): Likewise.
13896 (BUILT_IN_STRCHR): Likewise.
13897 (BUILT_IN_STRCPY): Likewise.
13898 (BUILT_IN_STRLEN): Likewise.
13899 (BUILT_IN_MEMCPY_CHK): Likewise.
13900 (BUILT_IN_MEMMOVE_CHK): Likewise.
13901 (BUILT_IN_MEMPCPY_CHK): Likewise.
13902 (BUILT_IN_MEMSET_CHK): Likewise.
13903 (BUILT_IN_STPCPY_CHK): Likewise.
13904 (BUILT_IN_STRCAT_CHK): Likewise.
13905 (BUILT_IN_STRCPY_CHK): Likewise.
13906 * calls.c (store_bounds): Likewise.
13907 (emit_call_1): Likewise.
13908 (special_function_p): Likewise.
13909 (maybe_warn_nonstring_arg): Likewise.
13910 (initialize_argument_information): Likewise.
13911 (finalize_must_preallocate): Likewise.
13912 (compute_argument_addresses): Likewise.
13913 (expand_call): Likewise.
13914 * cfgexpand.c (expand_call_stmt): Likewise.
13915 (expand_return): Likewise.
13916 (expand_gimple_stmt_1): Likewise.
13917 (pass_expand::execute): Likewise.
13918 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
13919 (cgraph_node::remove): Likewise.
13920 (cgraph_node::dump): Likewise.
13921 (cgraph_node::verify_node): Likewise.
13922 * cgraph.h (chkp_function_instrumented_p): Likewise.
13923 (symtab_node::get_alias_target): Likewise.
13924 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
13925 (cgraph_local_p): Likewise.
13926 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
13927 (cgraph_edge::rebuild_references): Likewise.
13928 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
13929 (walk_polymorphic_call_targets): Likewise.
13930 (cgraph_node::expand_thunk): Likewise.
13931 (symbol_table::output_weakrefs): Likewise.
13932 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
13933 (ix86_handle_option): Likewise.
13934 * config/i386/constraints.md: Likewise.
13935 * config/i386/i386-builtin-types.def (BND): Likewise.
13939 * config/i386/i386-builtin.def (BDESC_END): Likewise.
13940 (BDESC_FIRST): Likewise.
13942 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
13943 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
13944 * config/i386/i386.c (enum reg_class): Likewise.
13945 (ix86_target_string): Likewise.
13946 (ix86_option_override_internal): Likewise.
13947 (ix86_conditional_register_usage): Likewise.
13948 (ix86_valid_target_attribute_inner_p): Likewise.
13949 (ix86_set_indirect_branch_type): Likewise.
13950 (ix86_set_current_function): Likewise.
13951 (ix86_function_arg_regno_p): Likewise.
13952 (init_cumulative_args): Likewise.
13953 (ix86_function_arg_advance): Likewise.
13954 (ix86_function_arg): Likewise.
13955 (ix86_pass_by_reference): Likewise.
13956 (ix86_function_value_regno_p): Likewise.
13957 (ix86_function_value_1): Likewise.
13958 (ix86_function_value_bounds): Likewise.
13959 (ix86_return_in_memory): Likewise.
13960 (ix86_setup_incoming_vararg_bounds): Likewise.
13961 (ix86_va_start): Likewise.
13962 (indirect_thunk_need_prefix): Likewise.
13963 (print_reg): Likewise.
13964 (ix86_print_operand): Likewise.
13965 (ix86_expand_call): Likewise.
13966 (ix86_output_function_return): Likewise.
13967 (reg_encoded_number): Likewise.
13968 (BDESC_VERIFYS): Likewise.
13969 (ix86_init_mpx_builtins): Likewise.
13970 (ix86_init_builtins): Likewise.
13971 (ix86_emit_cmove): Likewise.
13972 (ix86_emit_move_max): Likewise.
13973 (ix86_expand_builtin): Likewise.
13974 (ix86_builtin_mpx_function): Likewise.
13975 (ix86_get_arg_address_for_bt): Likewise.
13976 (ix86_load_bounds): Likewise.
13977 (ix86_store_bounds): Likewise.
13978 (ix86_load_returned_bounds): Likewise.
13979 (ix86_store_returned_bounds): Likewise.
13980 (ix86_class_likely_spilled_p): Likewise.
13981 (ix86_hard_regno_mode_ok): Likewise.
13982 (x86_order_regs_for_local_alloc): Likewise.
13983 (ix86_mitigate_rop): Likewise.
13984 (ix86_bnd_prefixed_insn_p): Likewise.
13985 (ix86_mpx_bound_mode): Likewise.
13986 (ix86_make_bounds_constant): Likewise.
13987 (ix86_initialize_bounds): Likewise.
13988 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
13989 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
13990 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
13991 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
13992 (TARGET_CHKP_BOUND_MODE): Likewise.
13993 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
13994 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
13995 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
13996 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
13997 * config/i386/i386.h (TARGET_MPX): Likewise.
13998 (TARGET_MPX_P): Likewise.
13999 (VALID_BND_REG_MODE): Likewise.
14000 (FIRST_BND_REG): Likewise.
14001 (LAST_BND_REG): Likewise.
14002 (enum reg_class): Likewise.
14003 (BND_REG_P): Likewise.
14004 (BND_REGNO_P): Likewise.
14005 (BNDmode): Likewise.
14006 (ADJUST_INSN_LENGTH): Likewise.
14007 * config/i386/i386.md: Likewise.
14008 * config/i386/i386.opt: Likewise.
14009 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
14010 (defined): Likewise.
14011 (LINK_MPX): Likewise.
14012 (MPX_SPEC): Likewise.
14013 (LIBMPX_SPEC): Likewise.
14014 (LIBMPXWRAPPERS_SPEC): Likewise.
14015 (CHKP_SPEC): Likewise.
14016 * config/i386/predicates.md: Likewise.
14017 * dbxout.c (dbxout_type): Likewise.
14018 * doc/extend.texi: Likewise.
14019 * doc/invoke.texi: Likewise.
14020 * doc/md.texi: Likewise.
14021 * doc/tm.texi: Likewise.
14022 * doc/tm.texi.in: Likewise.
14023 * dwarf2out.c (is_base_type): Likewise.
14024 (gen_formal_types_die): Likewise.
14025 (gen_subprogram_die): Likewise.
14026 (gen_type_die_with_usage): Likewise.
14027 (gen_decl_die): Likewise.
14028 (dwarf2out_late_global_decl): Likewise.
14029 * expr.c (expand_assignment): Likewise.
14030 (emit_storent_insn): Likewise.
14031 (store_expr_with_bounds): Likewise.
14032 (store_expr): Likewise.
14033 (expand_expr_real_1): Likewise.
14034 * expr.h (store_expr_with_bounds): Likewise.
14035 * function.c (use_register_for_decl): Likewise.
14036 (struct bounds_parm_data): Likewise.
14037 (assign_parms_augmented_arg_list): Likewise.
14038 (assign_parm_find_entry_rtl): Likewise.
14039 (assign_parm_is_stack_parm): Likewise.
14040 (assign_parm_load_bounds): Likewise.
14041 (assign_bounds): Likewise.
14042 (assign_parms): Likewise.
14043 (expand_function_start): Likewise.
14044 * gcc.c (CHKP_SPEC): Likewise.
14045 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
14046 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
14047 (wrestrict_dom_walker::check_call): Likewise.
14048 * gimple.c (gimple_build_call_from_tree): Likewise.
14049 * gimple.h (enum gf_mask): Likewise.
14050 (gimple_call_with_bounds_p): Likewise.
14051 (gimple_call_set_with_bounds): Likewise.
14052 * gimplify.c (gimplify_init_constructor): Likewise.
14053 * ipa-cp.c (initialize_node_lattices): Likewise.
14054 (propagate_constants_across_call): Likewise.
14055 (find_more_scalar_values_for_callers_subset): Likewise.
14056 * ipa-hsa.c (process_hsa_functions): Likewise.
14057 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
14058 * ipa-icf.c (sem_function::merge): Likewise.
14059 * ipa-inline.c (early_inliner): Likewise.
14060 * ipa-pure-const.c (warn_function_noreturn): Likewise.
14061 (warn_function_cold): Likewise.
14062 (propagate_pure_const): Likewise.
14063 * ipa-ref.h (enum GTY): Likewise.
14064 * ipa-split.c (find_retbnd): Likewise.
14065 (consider_split): Likewise.
14066 (split_function): Likewise.
14067 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
14068 * ipa.c (walk_polymorphic_call_targets): Likewise.
14069 (symbol_table::remove_unreachable_nodes): Likewise.
14070 (process_references): Likewise.
14071 (cgraph_build_static_cdtor_1): Likewise.
14072 * lto-cgraph.c (lto_output_node): Likewise.
14073 (output_refs): Likewise.
14074 (compute_ltrans_boundary): Likewise.
14075 (input_overwrite_node): Likewise.
14076 (input_node): Likewise.
14077 (input_cgraph_1): Likewise.
14078 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
14079 * passes.c (pass_manager::execute_early_local_passes): Likewise.
14080 (class pass_chkp_instrumentation_passes): Likewise.
14081 (make_pass_chkp_instrumentation_passes): Likewise.
14082 * passes.def: Likewise.
14083 * rtl.h (struct GTY): Likewise.
14084 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
14085 * stor-layout.c (layout_type): Likewise.
14086 * symtab.c: Likewise.
14087 * target.def: Likewise.
14088 * targhooks.c (default_chkp_bound_type): Likewise.
14089 (default_chkp_bound_mode): Likewise.
14090 (default_builtin_chkp_function): Likewise.
14091 (default_chkp_function_value_bounds): Likewise.
14092 (default_chkp_make_bounds_constant): Likewise.
14093 (default_chkp_initialize_bounds): Likewise.
14094 * targhooks.h (default_chkp_bound_type): Likewise.
14095 (default_chkp_bound_mode): Likewise.
14096 (default_builtin_chkp_function): Likewise.
14097 (default_chkp_function_value_bounds): Likewise.
14098 (default_chkp_make_bounds_constant): Likewise.
14099 (default_chkp_initialize_bounds): Likewise.
14100 * toplev.c (compile_file): Likewise.
14101 (process_options): Likewise.
14102 * tree-core.h (DEF_BUILTIN): Likewise.
14103 (DEF_BUILTIN_CHKP): Likewise.
14104 * tree-inline.c (declare_return_variable): Likewise.
14105 (remap_gimple_stmt): Likewise.
14106 (copy_bb): Likewise.
14107 (initialize_inlined_parameters): Likewise.
14108 (expand_call_inline): Likewise.
14109 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
14110 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
14111 (make_pass_ipa_chkp_produce_thunks): Likewise.
14112 (make_pass_chkp): Likewise.
14113 (make_pass_chkp_opt): Likewise.
14114 (make_pass_chkp_instrumentation_passes): Likewise.
14115 * tree-pretty-print.c (dump_generic_node): Likewise.
14116 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
14117 * tree-ssa-dce.c (propagate_necessity): Likewise.
14118 (eliminate_unnecessary_stmts): Likewise.
14119 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
14120 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
14121 * tree-ssa-sccvn.h: Likewise.
14122 * tree-ssa-strlen.c (get_string_length): Likewise.
14123 (valid_builtin_call): Likewise.
14124 (adjust_last_stmt): Likewise.
14125 (handle_builtin_strchr): Likewise.
14126 (handle_builtin_strcpy): Likewise.
14127 (handle_builtin_stxncpy): Likewise.
14128 (handle_builtin_memcpy): Likewise.
14129 (handle_builtin_strcat): Likewise.
14130 (strlen_check_and_optimize_stmt): Likewise.
14131 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
14132 * tree-streamer-in.c: Likewise.
14133 * tree-streamer.c (record_common_node): Likewise.
14134 * tree.c (tree_code_size): Likewise.
14135 (wide_int_to_tree_1): Likewise.
14136 (type_contains_placeholder_1): Likewise.
14137 (build_common_tree_nodes): Likewise.
14138 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
14139 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
14140 (POINTER_BOUNDS_P): Likewise.
14141 (BOUNDED_TYPE_P): Likewise.
14142 (BOUNDED_P): Likewise.
14143 (CALL_WITH_BOUNDS_P): Likewise.
14144 (pointer_bounds_type_node): Likewise.
14145 * value-prof.c (gimple_ic): Likewise.
14146 * var-tracking.c (vt_add_function_parameters): Likewise.
14147 * varasm.c (make_decl_rtl): Likewise.
14148 (assemble_start_function): Likewise.
14149 (output_constant): Likewise.
14150 (maybe_assemble_visibility): Likewise.
14151 * varpool.c (ctor_for_folding): Likewise.
14152 * chkp-builtins.def: Remove.
14153 * ipa-chkp.c: Remove.
14154 * ipa-chkp.h: Remove.
14155 * rtl-chkp.c: Remove.
14156 * rtl-chkp.h: Remove.
14157 * tree-chkp-opt.c: Remove.
14158 * tree-chkp.c: Remove.
14159 * tree-chkp.h: Remove.
14161 2018-06-07 Carl Love <cel@us.ibm.com>
14163 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14164 vextract_fp_from_shortl): Add BE support.
14166 2018-06-07 Paul Koning <ni1d@arrl.net>
14168 * compare-elim.c (try_merge_compare): Don't merge compare if
14169 address contains a side effect.
14170 (try_eliminate_compare): Likewise.
14172 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
14174 * config.gcc: Support "tremont".
14175 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
14176 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14178 * config/i386/i386.c (m_TREMONT): Define.
14179 (processor_target_table): Add "tremont".
14180 (PTA_TREMONT): Define.
14181 (ix86_lea_outperforms): Add TARGET_TREMONT.
14182 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
14183 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
14184 and M_INTEL_GOLDMONT_PLUS.
14185 (fold_builtin_cpu): Add "tremont".
14186 (ix86_add_stmt_cost): Add TARGET_TREMONT.
14187 (ix86_option_override_internal): Add "tremont".
14188 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
14189 (processor_type): Add PROCESSOR_TREMONT.
14190 * config/i386/x86-tune.def: Add m_TREMONT.
14191 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
14193 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14195 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
14196 symbol defined for msp430i* devices to be lower case.
14198 2018-06-07 Richard Biener <rguenther@suse.de>
14200 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
14201 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
14202 Properly wrap signed arithmetic if overflow wraps.
14204 2018-06-07 Jakub Jelinek <jakub@redhat.com>
14206 PR tree-optimization/69615
14207 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
14208 of a cast from a same precision integral SSA_NAME in a bb dominated
14209 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
14210 cast to utype if rhs2 has already a compatible type.
14212 2018-06-07 Richard Biener <rguenther@suse.de>
14214 PR tree-optimization/85935
14215 * graphite-scop-detection.c (find_params_in_bb): Analyze
14216 condition operands with respect to the correct loop. Assert
14217 the analysis doesn't fail.
14219 2018-06-04 Carl Love <cel@us.ibm.com>
14221 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
14222 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
14223 as it is slightly cheaper.
14224 (first_match_or_eos_index_<mode>):
14225 Calculate index using natural element order.
14226 (first_match_index_<mode>):
14227 Calculate index using natural element order.
14228 (first_match_or_eos_index_<mode>):
14229 Calculate index using natural order.
14230 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
14231 for BE and LE modes.
14232 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
14233 P9V_BUILTIN_VCLZLSBB_V16QI.
14234 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
14237 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
14239 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
14240 indentation and line wrap for many prototypes. Add missing
14241 @smallexample directives around block of prototypes for vec_xl and
14244 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
14246 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
14247 track if we pass or return IEEE 128-bit floating point.
14248 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
14249 C++ mangling that is compatible with GCC 8.1.
14250 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
14251 (init_cumulative_args): Note if we pass or return IEEE 128-bit
14252 floating point types.
14253 (rs6000_function_arg_advance_1): Likewise.
14254 (rs6000_mangle_type): Optionally generate mangled names that match
14255 what GCC 8.1 generated for IEEE 128-bit floating point types.
14256 (rs6000_globalize_decl_name): If we have an external function that
14257 passes or returns IEEE 128-bit types, generate a weak reference
14258 from the mangled name used in GCC 8.1 to the current mangled
14260 (rs6000_init_builtins): Make __ibm128 use the long double type if
14261 long double is IBM extended double. Make __float128 use the long
14262 double type if long double is IEEE 128-bit.
14265 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
14266 macro for __ibm128 built-in functions.
14267 (PACK_IF): Add __ibm128 pack/unpack functions.
14268 (UNPACK_IF): Likewise.
14269 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
14270 enable long double built-in functions if long double is IEEE
14271 128-bit floating point.
14272 (rs6000_invalid_builtin): Update long double built-in function
14274 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
14275 functions, adjust the built-in function to use the long double
14276 built-in function if __ibm128 and long double are the same type.
14277 * doc/extend.texi (PowerPC builtins): Update documention for
14278 __builtin_{,un}pack_longdouble. Add documentation for
14279 __builtin_{,un}pack_ibm128.
14281 2018-06-06 Jim Wilson <jimw@sifive.com>
14283 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
14284 (struct machine_function): New field interrupt_mode.
14285 (riscv_handle_type_attribute): New function. Add forward declaration.
14286 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
14287 (riscv_expand_epilogue): Check interrupt_mode field.
14288 (riscv_set_current_function): Check interrupt attribute args and
14289 set interrupt_mode field.
14290 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
14291 (riscv_sret, riscv_uret): New.
14292 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
14293 new arguments to interrupt attribute.
14295 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
14298 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
14299 Don't handle -mcpu=power8 if -mpower9-vector is also used.
14301 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
14303 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
14304 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
14305 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
14306 several redundant entries.
14308 2018-06-06 David Malcolm <dmalcolm@redhat.com>
14310 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
14311 type from "rtx" to "rtx_insn *".
14312 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
14313 for local "call_insn", removing cast.
14314 (ix86_expand_call): Likewise, introducing a "call_insn" local.
14316 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
14318 PR tree-optimization/86066
14319 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
14320 for BIT_INSERT_EXPR stores.
14322 2018-06-06 Richard Biener <rguenther@suse.de>
14324 PR tree-optimization/86062
14325 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
14326 component refs ontop
14327 of to be offsetted base.
14329 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14331 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
14332 to be static and remove check on interrupt attribute name.
14334 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
14336 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
14337 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
14339 2018-06-05 Steve Ellcey <sellcey@cavium.com>
14342 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
14344 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
14345 Remove second argument, change how error is called.
14346 (aarch64_layout_arg): Remove second argument from
14347 aarch64_err_no_fpadvsimd call.
14348 (aarch64_init_cumulative_args): Ditto.
14349 (aarch64_gimplify_va_arg_expr): Ditto.
14350 * config/aarch64/aarch64.md (mov<mode>): Ditto.
14352 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
14354 * config/i386/i386.md (simple_return_indirect_internal): New expander.
14355 (*simple_return_indirect_internal<mode>): Rename from
14356 simple_return_indirect_internal. Use W mode iterator.
14357 (rstorssp): New expander.
14358 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
14359 (clrssbsy): New expander.
14360 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
14362 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
14364 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
14366 (cmse_check_pointed_object): Likewise.
14368 2018-06-05 Martin Liska <mliska@suse.cz>
14370 PR gcov-profile/47618
14371 * doc/invoke.texi: Document how -fprofile-dir format
14374 2018-06-05 Richard Biener <rguenther@suse.de>
14376 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
14377 removal pretend DOM info isn't available so we do not update
14378 it and only remove edges, not dominated blocks. Actually free
14379 DOM info in case we removed something. Remove unreachable blocks.
14380 (mfb_keep_latches): Work with either DOM info or marked backedges.
14381 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
14382 first. Mark backedges if DOM info isn't available.
14383 (Re-)compute DOM info after cleanup_control_flow_pre.
14385 2018-06-05 Richard Biener <rguenther@suse.de>
14387 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
14388 (locus_discrim_hasher::hash): Adjust.
14389 (locus_discrim_hasher::equal): Likewise.
14390 (next_discriminator_for_locus): Work on line directly.
14391 (same_line_p): Pass in expanded locus1 as well.
14392 (assign_discriminators): Avoid redundant location expansions.
14394 2018-06-05 Richard Biener <rguenther@suse.de>
14396 PR tree-optimization/86046
14397 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
14398 if required after clearing TREE_ADDRESSABLE.
14400 2018-06-05 Richard Biener <rguenther@suse.de>
14402 PR tree-optimization/86047
14403 * tree-ssa-loop.c (for_each_index): Glob handling of all
14404 decls and constants and really handle all of them.
14406 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14409 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
14410 qualifier_void_pointer and qualifier_const_void_pointer.
14411 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
14412 (arm_init_builtins): Handle the above.
14413 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
14414 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
14415 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
14418 2018-06-05 Martin Liska <mliska@suse.cz>
14420 * auto-profile.c (read_autofdo_file): Do not use
14421 gcov_ctr_summary struct.
14422 (afdo_callsite_hot_enough_for_early_inline): Likewise.
14423 * coverage.c (struct counts_entry): Likewise.
14424 (read_counts_file): Read just single summary entry.
14425 (get_coverage_counts): Use gcov_summary struct.
14426 * coverage.h (get_coverage_counts): Likewise.
14427 * gcov-dump.c (dump_working_sets): Likewise.
14428 (tag_summary): Dump just single summary.
14429 * gcov-io.c (gcov_write_summary): Write just histogram
14431 (gcov_read_summary): Read just single summary.
14432 (compute_working_sets): Use gcov_summary struct.
14433 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
14434 of GCOV_COUNTERS_SUMMABLE.
14435 (GCOV_COUNTERS_SUMMABLE): Remove.
14436 (GCOV_FIRST_VALUE_COUNTER): Replace with
14437 GCOV_COUNTER_V_INTERVAL.
14438 (struct gcov_ctr_summary): Remove.
14439 (struct gcov_summary): Directly use fields of former
14441 (compute_working_sets): Use gcov_summary struct.
14442 * gcov.c (read_count_file): Do not use ctrs fields.
14443 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
14445 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
14447 * profile.c: Likewise.
14448 * profile.h: Likewise.
14450 2018-06-05 Martin Liska <mliska@suse.cz>
14452 PR gcov-profile/84846
14453 * gcov.c (output_lines): Print working directory only
14454 in intermediate format.
14456 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
14458 * config/s390/s390-builtin-types.def: Add void function type.
14459 * config/s390/s390-builtins.def: Use the function type for the
14462 2018-06-04 Jim Wilson <jimw@sifive.com>
14464 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
14466 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
14467 and maybe_eh_return. Change regno to unsigned int. Use new args to
14468 handle EH_RETURN_DATA_REGNO registers properly.
14469 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
14470 (riscv_expand_epilogue): Update comment. Change argument name and
14471 type. Update code to use new name and type. Pass new args to
14472 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
14474 * config/riscv/riscv.md (NORMAL_RETURN): New.
14475 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
14476 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
14477 (eh_return): Call gen_eh_return_internal and emit barrier.
14478 (eh_return_internal): Call riscv_expand_epilogue.
14480 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
14482 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
14483 bit_insertion field and declare can_be_merged_into method.
14484 (merged_store_group::can_be_merged_into): New method.
14485 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
14486 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
14487 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
14489 2018-06-04 Richard Biener <rguenther@suse.de>
14491 PR tree-optimization/85955
14492 * builtins.c (fold_builtin_sincos): Convert pointers to
14493 destination to appropriate type before dereferencing.
14495 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
14497 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
14499 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
14501 * expr.c (expand_expr_real_1): Force the operand into memory if
14502 its TYPE_MODE is BLKmode and if there is no integer mode for
14503 the number of bits being extracted.
14505 2018-06-04 Jakub Jelinek <jakub@redhat.com>
14509 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
14510 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
14512 2018-06-04 Richard Biener <rguenther@suse.de>
14514 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
14515 (cleanup_tree_cfg_noloop): ... single caller. Do
14516 start_recording_case_labels later.
14518 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
14520 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
14521 to _IMMINTRIN_H_INCLUDED.
14522 * config/i386/pconfigintrin.h: Ditto.
14523 * config/i386/waitpkgintrin.h: Ditto.
14524 * config/i386/immintrin.h: Add includes for sgxintrin.h,
14525 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
14526 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
14527 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
14528 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
14529 waitpkgintrin.h and cldemoteintrin.h.
14531 2018-06-04 Richard Biener <rguenther@suse.de>
14533 PR tree-optimization/86038
14534 * tracer.c (find_best_successor): Check probability for
14535 being initialized, bail out if not.
14537 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
14540 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
14541 of bits to ignore when comparing architectures.
14543 2018-06-04 Jakub Jelinek <jakub@redhat.com>
14545 PR tree-optimization/69615
14546 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
14547 maximum or minimum of the type, try to merge it also as if
14548 range1 is + [-, x - 1] or + [x + 1, -].
14551 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
14553 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
14555 PR tree-optimization/86034
14556 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
14557 the unsigned bitfield type in a bit insertion sequence if it does not
14558 have a larger precision than the bitfield size.
14559 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
14561 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
14563 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
14565 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
14567 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
14568 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
14569 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
14570 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
14572 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
14574 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
14575 Disable -fdelete-null-pointer-checks for ELF toolchain.
14577 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
14578 Kito Cheng <kito.cheng@gmail.com>
14580 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
14581 (nds32le-*-*, nds32be-*-*): Integrate checking process.
14582 (nds32*-*-*): Add glibc and uclibc conditions.
14583 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
14584 (TARGET_EXCEPT_UNWIND_INFO): Define.
14585 * config/nds32/elf.h: New file.
14586 * config/nds32/linux.h: New file.
14587 * config/nds32/nds32-elf.opt: New file.
14588 * config/nds32/nds32-linux.opt: New file.
14589 * config/nds32/nds32-fp-as-gp.c
14590 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
14591 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
14593 (nds32_asm_file_end): Ditto.
14594 (nds32_print_operand): Ditto.
14595 (nds32_insert_attributes): Ditto.
14596 (nds32_init_libfuncs): New function.
14597 (TARGET_HAVE_TLS): Define.
14598 (TARGET_INIT_LIBFUNCS): Define.
14599 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
14601 (TARGET_ELF): Apply different mcmodel setting.
14602 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
14603 been migrated into elf.h and linux.h files.
14604 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
14605 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
14606 (mcmodel): The content has been migrated into nds32-elf.opt and
14607 nds32-linux.opt files.
14608 * config/nds32/t-elf: New file.
14609 * config/nds32/t-linux: New file.
14611 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
14612 Shiva Chen <shiva0217@gmail.com>
14614 * config/nds32/constants.md (unspec_volatile_element): Add
14615 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
14616 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
14618 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
14619 (make_pass_nds32_fp_as_gp): Declare.
14620 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
14622 (nds32_asm_function_end_prologue): Remove unused asm output.
14623 (nds32_asm_function_begin_epilogue): Remove unused asm output.
14624 (nds32_asm_file_start): Output necessary fp_as_gp information.
14625 (nds32_option_override): Adjust register usage.
14626 (nds32_expand_prologue): Consider fp_as_gp situation.
14627 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
14628 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
14631 (simple_return): Ditto.
14632 (omit_fp_begin): Output special directive for fp_as_gp.
14633 (omit_fp_end): Output special directive for fp_as_gp.
14634 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
14635 mforbid-fp-as-gp): New options.
14637 2018-06-01 Mark Wielaard <mark@klomp.org>
14639 * dwarf2out.c (dwarf2out_finish): Remove generation of
14640 DW_AT_loclists_base.
14642 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
14644 * gimple-ssa-store-merging.c: Include gimple-fold.h.
14645 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
14646 (struct merged_store_group): Add bit_insertion field.
14647 (dump_char_array): Use standard hexadecimal format.
14648 (merged_store_group::merged_store_group): Set bit_insertion to false.
14649 (merged_store_group::apply_stores): Use optimal buffer size. Deal
14650 with BIT_INSERT_EXPR stores. Move up code updating the mask and
14651 also print the mask in the dump file.
14652 (pass_store_merging::gate): Minor tweak.
14653 (imm_store_chain_info::coalesce_immediate): Fix wrong association
14654 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
14655 stores with INTEGER_CST stores.
14656 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
14657 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
14658 and use it throughout. Generate bit insertion sequences if need be.
14659 (pass_store_merging::process_store): Remove redundant condition.
14660 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
14662 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
14664 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
14665 the 128-bit floating point types. Fix function comment.
14667 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14669 * config/aarch64/aarch64-simd.md
14670 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
14672 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
14675 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
14677 PR tree-optimization/85989
14678 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
14680 (backprop::intersect_uses): Check it when deciding whether this
14681 is a backedge reference.
14682 (backprop::process_block): Add each phi to m_visited_phis
14683 after visiting it, then clear it at the end.
14685 2018-06-01 Richard Biener <rguenther@suse.de>
14687 * tree-vectorizer.h (vect_dr_stmt): New function.
14688 (vect_get_load_cost): Adjust.
14689 (vect_get_store_cost): Likewise.
14690 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
14691 Use vect_dr_stmt instead of DR_SMTT.
14692 (vect_record_base_alignments): Likewise.
14693 (vect_calculate_target_alignment): Likewise.
14694 (vect_compute_data_ref_alignment): Likewise and make static.
14695 (vect_update_misalignment_for_peel): Likewise.
14696 (vect_verify_datarefs_alignment): Likewise.
14697 (vector_alignment_reachable_p): Likewise.
14698 (vect_get_data_access_cost): Likewise. Pass down
14699 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
14700 (vect_get_peeling_costs_all_drs): Likewise.
14701 (vect_peeling_hash_get_lowest_cost): Likewise.
14702 (vect_enhance_data_refs_alignment): Likewise.
14703 (vect_find_same_alignment_drs): Likewise.
14704 (vect_analyze_data_refs_alignment): Likewise.
14705 (vect_analyze_group_access_1): Likewise.
14706 (vect_analyze_group_access): Likewise.
14707 (vect_analyze_data_ref_access): Likewise.
14708 (vect_analyze_data_ref_accesses): Likewise.
14709 (vect_vfa_segment_size): Likewise.
14710 (vect_small_gap_p): Likewise.
14711 (vectorizable_with_step_bound_p): Likewise.
14712 (vect_prune_runtime_alias_test_list): Likewise.
14713 (vect_analyze_data_refs): Likewise.
14714 (vect_supportable_dr_alignment): Likewise.
14715 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
14716 (vect_gen_prolog_loop_niters): Likewise.
14717 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
14718 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
14720 (vect_recog_mask_conversion_pattern): Likewise.
14721 (vect_try_gather_scatter_pattern): Likewise.
14722 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
14723 to vect_get_store_cost.
14724 (vect_get_store_cost): Get stmt_info instead of DR.
14725 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
14726 (vect_get_load_cost): Get stmt_info instead of DR.
14728 2018-06-01 Richard Biener <rguenther@suse.de>
14730 PR middle-end/86017
14731 * gimple-fold.c (var_decl_component_p): Also allow offsetted
14732 vars wrapped in MEM_REFs.
14734 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
14736 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
14737 Fix subreg tests so that we only return a choice between
14738 GENERAL_REGS and FP_REGS if the original classes included both.
14740 2018-06-01 Richard Biener <rguenther@suse.de>
14743 * tree-ssa-structalias.c (get_function_part_constraint):
14744 Handle NULL fi->decl.
14745 (find_func_aliases_for_call): Properly handle indirect
14746 fi from direct call.
14747 (find_func_clobbers): Likewise.
14748 (ipa_pta_execute): Likewise.
14749 (create_variable_info_for): For functions that are ifunc_resolver
14750 resolve to a varinfo that contains the result of the resolver call.
14751 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
14754 2018-05-31 Michael Collison <michael.collison@arm.com>
14756 * config/aarch64/aarch64.md:
14757 (*fix_to_zero_extenddfdi2): New pattern.
14758 * gcc.target/aarch64/fix_extend1.c: New testcase.
14760 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
14762 PR middle-end/78809
14763 PR middle-end/83026
14764 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
14765 and BUILT_IN_STRNCMP_EQ.
14766 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
14767 BUILT_IN_STRNCMP_EQ.
14768 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
14769 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
14770 (gimple_fold_builtin): Likewise.
14771 * tree-ssa-strlen.c (compute_string_length): New function.
14772 (determine_min_obsize): New function.
14773 (handle_builtin_string_cmp): New function to handle calls to
14774 string compare functions.
14775 (strlen_optimize_stmt): Add handling to builtin string compare
14777 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
14778 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
14779 * tree.c (build_common_builtin_nodes): Add new defines of
14780 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
14782 2018-05-31 Jakub Jelinek <jakub@redhat.com>
14785 * bb-reorder.c (pass_partition_blocks::gate): Return false for
14786 functions with naked attribute.
14788 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
14790 * config/i386/sse.md (avx_vec_concat<mode>):
14791 Substitute concat_tg_mode mode attribute with xtg_mode.
14792 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
14793 (concat_tg_mode): Remove mode attribute.
14795 2018-05-31 Martin Sebor <msebor@redhat.com>
14798 * calls.c (alloc_max_size): Correct a logic error/typo.
14799 Treat excessive arguments as infinite. Warn for invalid arguments.
14800 * doc/invoke.texi (-Walloc-size-larger-than): Update.
14802 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
14805 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
14806 and movx for Haswell.
14808 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
14809 Cesar Philippidis <cesar@codesourcery.com>
14811 PR middle-end/85879
14812 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
14813 when emitting error on private/firstprivate reductions.
14814 * omp-low.c (lower_omp_target): Avoid reference-type processing
14815 on pointers for firstprivate clause.
14817 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
14819 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
14822 * config/aarch64/aarch64-simd.md
14823 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
14824 (aarch64_ld1_x3_<mode>): Likewise
14825 (aarch64_st1x2<VALLDIF:mode>): Likewise
14826 (aarch64_st1_x2_<mode>): Likewise
14827 (aarch64_st1x3<VALLDIF:mode>): Likewise
14828 (aarch64_st1_x3_<mode>): Likewise
14829 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
14830 (vld1_s8_x3): Likewise.
14831 (vld1_u16_x3): Likewise.
14832 (vld1_s16_x3): Likewise.
14833 (vld1_u32_x3): Likewise.
14834 (vld1_s32_x3): Likewise.
14835 (vld1_u64_x3): Likewise.
14836 (vld1_s64_x3): Likewise.
14837 (vld1_f16_x3): Likewise.
14838 (vld1_f32_x3): Likewise.
14839 (vld1_f64_x3): Likewise.
14840 (vld1_p8_x3): Likewise.
14841 (vld1_p16_x3): Likewise.
14842 (vld1_p64_x3): Likewise.
14843 (vld1q_u8_x3): Likewise.
14844 (vld1q_s8_x3): Likewise.
14845 (vld1q_u16_x3): Likewise.
14846 (vld1q_s16_x3): Likewise.
14847 (vld1q_u32_x3): Likewise.
14848 (vld1q_s32_x3): Likewise.
14849 (vld1q_u64_x3): Likewise.
14850 (vld1q_s64_x3): Likewise.
14851 (vld1q_f16_x3): Likewise.
14852 (vld1q_f32_x3): Likewise.
14853 (vld1q_f64_x3): Likewise.
14854 (vld1q_p8_x3): Likewise.
14855 (vld1q_p16_x3): Likewise.
14856 (vld1q_p64_x3): Likewise.
14857 (vst1_s64_x2): Likewise.
14858 (vst1_u64_x2): Likewise.
14859 (vst1_f64_x2): Likewise.
14860 (vst1_s8_x2): Likewise.
14861 (vst1_p8_x2): Likewise.
14862 (vst1_s16_x2): Likewise.
14863 (vst1_p16_x2): Likewise.
14864 (vst1_s32_x2): Likewise.
14865 (vst1_u8_x2): Likewise.
14866 (vst1_u16_x2): Likewise.
14867 (vst1_u32_x2): Likewise.
14868 (vst1_f16_x2): Likewise.
14869 (vst1_f32_x2): Likewise.
14870 (vst1_p64_x2): Likewise.
14871 (vst1q_s8_x2): Likewise.
14872 (vst1q_p8_x2): Likewise.
14873 (vst1q_s16_x2): Likewise.
14874 (vst1q_p16_x2): Likewise.
14875 (vst1q_s32_x2): Likewise.
14876 (vst1q_s64_x2): Likewise.
14877 (vst1q_u8_x2): Likewise.
14878 (vst1q_u16_x2): Likewise.
14879 (vst1q_u32_x2): Likewise.
14880 (vst1q_u64_x2): Likewise.
14881 (vst1q_f16_x2): Likewise.
14882 (vst1q_f32_x2): Likewise.
14883 (vst1q_f64_x2): Likewise.
14884 (vst1q_p64_x2): Likewise.
14885 (vst1_s64_x3): Likewise.
14886 (vst1_u64_x3): Likewise.
14887 (vst1_f64_x3): Likewise.
14888 (vst1_s8_x3): Likewise.
14889 (vst1_p8_x3): Likewise.
14890 (vst1_s16_x3): Likewise.
14891 (vst1_p16_x3): Likewise.
14892 (vst1_s32_x3): Likewise.
14893 (vst1_u8_x3): Likewise.
14894 (vst1_u16_x3): Likewise.
14895 (vst1_u32_x3): Likewise.
14896 (vst1_f16_x3): Likewise.
14897 (vst1_f32_x3): Likewise.
14898 (vst1_p64_x3): Likewise.
14899 (vst1q_s8_x3): Likewise.
14900 (vst1q_p8_x3): Likewise.
14901 (vst1q_s16_x3): Likewise.
14902 (vst1q_p16_x3): Likewise.
14903 (vst1q_s32_x3): Likewise.
14904 (vst1q_s64_x3): Likewise.
14905 (vst1q_u8_x3): Likewise.
14906 (vst1q_u16_x3): Likewise.
14907 (vst1q_u32_x3): Likewise.
14908 (vst1q_u64_x3): Likewise.
14909 (vst1q_f16_x3): Likewise.
14910 (vst1q_f32_x3): Likewise.
14911 (vst1q_f64_x3): Likewise.
14912 (vst1q_p64_x3): Likewise.
14914 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14916 * config/msp430/msp430.c (msp430_output_labelref): Prepend
14917 user_label_prefix to name.
14919 * tree-core.h: Update comment about the format of NAME string
14920 passed to handler in attribute_spec.
14922 * config/msp430/msp430.md: Remove erroneous subreg expression from
14923 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
14924 zero_extend{q,h}isi2.
14926 2018-05-30 Borislav Petkov <bp@suse.de>
14928 * doc/extend.texi: Document some architecture specific
14929 constraints and sort entries.
14931 2018-05-30 Martin Sebor <msebor@redhat.com>
14933 PR middle-end/85369
14934 * builtins.c (expand_builtin_stpcpy_1): New function.
14935 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
14936 only if the former succeeds.
14938 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
14940 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
14943 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
14945 * doc/invoke.texi (-flinker-output): Document
14947 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
14949 * passes.c (ipa_write_summaries): Only modify statements if body
14951 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
14952 incrementally linking.
14953 (ipa_passes): Likewise.
14954 * lto-cgraph.c (lto_output_node): When incrementally linking do not
14955 pass down resolution info.
14956 * common.opt (flag_incremental_link): Update info.
14957 * gcc.c (plugin specs): Turn flinker-output=* to
14958 -plugin-opt=-linker-output-known
14959 * toplev.c (compile_file): Also cut compilation when doing incremental
14961 * flag-types. (enum lto_partition_model): Add
14962 LTO_LINKER_OUTPUT_NOLTOREL.
14963 (invoke.texi): Add -flinker-output docs.
14964 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
14965 link same way as WPA; do not stream in dead initializers.
14967 * dwarf2out.c (dwarf2out_die_ref_for_decl,
14968 darf2out_register_external_decl): Support incremental link.
14970 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
14972 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
14974 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
14976 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
14977 it down to simple_object_copy_lto_debug_sections.
14978 (run_gcc): Determine incremental LTO link time and configure
14979 lto1 into non-wpa mode, disable renaming of debug sections.
14981 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
14983 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
14984 descriptions of various incorrectly documented functions.
14986 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
14989 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
14990 address check not strict.
14992 2018-05-30 Richard Biener <rguenther@suse.de>
14994 PR tree-optimization/85964
14995 * tracer.c (better_p): Drop initialized count check, we only
14996 call the function with initialized counts now.
14997 (find_best_successor): Do find a best edge if one
14998 has uninitialized count.
14999 (find_best_predecessor): Likewise. Do BB frequency check only
15000 if count is initialized.
15002 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
15004 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
15005 (aarch64_ldrstr_offset_compare): New.
15006 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
15007 load/store orderings.
15008 (aarch64_gen_adjusted_ldpstp): Likewise.
15010 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
15012 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
15013 Check for subset of GENERAL_REGS and FP_REGS.
15014 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
15017 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
15019 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
15020 and wi::to_poly_offset. Add the current offset and then check
15021 whether the sum fits, rather than using an unchecked addition of
15022 a checked term. Check for a shwi rather than a uhwi.
15023 * expr.c (get_bit_range): Use tree_to_poly_uint64.
15024 (store_constructor): Use poly_int_tree_p.
15025 (expand_expr_real_1): Likewise.
15026 * function.c (assign_temp): Likewise.
15027 * fold-const.c (const_binop): Use poly_int_tree_p and
15028 wi::to_poly_offset.
15029 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
15031 * ipa-icf-gimple.c (func_checker::compare_operand): Use
15032 to_poly_offset for MEM offsets.
15033 * ipa-icf.c (sem_variable::equals): Likewise.
15034 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
15035 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
15036 wi::to_poly_offset for BIT_FIELD_REF offsets.
15037 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
15038 wi::to_poly_offset.
15039 * var-tracking.c (emit_note_insn_var_location): Use
15040 tree_to_poly_uint64.
15042 2018-05-29 Jim Wilson <jimw@sifive.com>
15044 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
15046 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
15049 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
15050 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
15052 (sse4_1_round<mode>2): Use nonimmediate_operand
15053 for operand 1 predicate.
15055 2018-05-29 Martin Sebor <msebor@redhat.com>
15056 Richard Biener <rguenther@suse.de>
15059 * calls.c (get_size_range): Call determine_value_range instead
15060 of get_value_range..
15061 * tree-vrp.h (determine_value_range): Declared new function.
15062 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
15064 2018-05-29 Richard Biener <rguenther@suse.de>
15066 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
15067 sure to use non-pattern stmts for get_earlier_stmt arguments.
15068 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
15069 called on pattern stmts.
15070 (get_later_stmt): Likewise.
15072 2018-05-29 Martin Liska <mliska@suse.cz>
15074 PR gcov-profile/85759
15075 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
15078 2018-05-29 Jakub Jelinek <jakub@redhat.com>
15080 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
15082 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
15086 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
15087 VEC_PACK_FLOAT_EXPR): New tree codes.
15088 * tree-pretty-print.c (op_code_prio): Handle
15089 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
15090 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
15091 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
15092 * tree-inline.c (estimate_operator_cost): Likewise.
15093 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
15094 * fold-const.c (const_binop): Likewise.
15095 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
15096 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
15097 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
15098 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
15099 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
15100 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
15101 * expr.c (expand_expr_real_2): Likewise.
15102 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
15103 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
15104 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
15106 * optabs.c (expand_widen_pattern_expr): For
15107 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
15108 sign from result type rather than operand's type.
15109 (expand_binop_directly): For vec_packu_float_optab and
15110 vec_packs_float_optab allow result type to be different from operand's
15112 * optabs-tree.c (optab_for_tree_code): Handle
15113 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
15114 VEC_PACK_FLOAT_EXPR. Formatting fixes.
15115 * tree-vect-generic.c (expand_vector_operations_1): Handle
15116 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
15117 VEC_PACK_FLOAT_EXPR.
15118 * tree-vect-stmts.c (supportable_widening_operation): Handle
15120 (supportable_narrowing_operation): Handle FLOAT_EXPR.
15121 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
15122 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
15123 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
15124 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
15126 (vec_pack<floatprefix>_float_<mode>): New expander.
15127 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
15129 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
15130 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
15131 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
15132 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
15133 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
15135 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
15136 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
15137 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
15138 VEC_PACK_FLOAT_EXPR): Document.
15140 2018-05-29 Richard Biener <rguenther@suse.de>
15142 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
15144 (stmt_vec_info_vec): Make pointer.
15145 (init_stmt_vec_info_vec): Remove.
15146 (free_stmt_vec_info_vec): Likewise.
15147 (set_stmt_vec_info_vec): New function.
15148 (free_stmt_vec_infos): Likewise.
15149 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
15150 (set_vinfo_for_stmt): Likewise.
15151 (get_earlier_stmt): Likewise.
15152 (get_later_stmt): Likewise.
15153 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
15154 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
15155 (vec_info::~vec_info): Free stmt_vec_infos.
15156 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
15157 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
15158 (pass_slp_vectorize::execute): Likewise.
15159 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
15160 (free_stmt_vec_info_vec): Likewise.
15161 (set_stmt_vec_info_vec): New function.
15162 (free_stmt_vec_infos): Likewise.
15163 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
15164 the global stmt_vec_info_vec.
15165 * tree-parloops.c (gather_scalar_reductions): Use
15166 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
15169 2018-05-29 Richard Biener <rguenther@suse.de>
15171 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
15173 2018-05-29 Martin Liska <mliska@suse.cz>
15174 David Malcolm <dmalcolm@redhat.com>
15176 * vec.c (test_reverse): New.
15177 (vec_c_tests): Add new test.
15178 * vec.h (vl_ptr>::reverse): New function.
15180 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
15182 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
15184 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
15187 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
15189 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
15191 2018-05-28 Richard Biener <rguenther@suse.de>
15193 PR tree-optimization/85933
15194 * tree-vect-data-refs.c (vect_record_base_alignments): Only
15195 look at stmts marked as vectorizable.
15197 2018-05-28 Richard Biener <rguenther@suse.de>
15199 PR tree-optimization/85934
15200 * tree-vect-generic.c (expand_vector_operations_1): Hoist
15201 vector boolean check before scalar optimization.
15203 2018-05-28 Jakub Jelinek <jakub@redhat.com>
15205 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
15208 2018-05-28 Mark Wielaard <mark@klomp.org>
15210 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
15211 if it is an expression containing a minus sign.
15213 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
15215 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
15217 2018-05-27 Paul Koning <ni1d@arrl.net>
15219 * config/pdp11/pdp11.md (truncsihi2): Remove.
15221 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
15222 Chung-Ju Wu <jasonwucj@gmail.com>
15224 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
15226 (unaligned_store_dw): Ditto.
15227 * config/nds32/nds32-memory-manipulation.c
15228 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
15229 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
15230 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
15231 (emit_setmem_word_loop): Rename to ...
15232 (emit_setmem_doubleword_loop): ... this.
15233 (nds32_gen_dup_4_byte_to_word_value): New function.
15234 (nds32_gen_dup_8_byte_to_double_word_value): New function.
15235 (nds32_expand_setmem_loop): Refine implementation.
15236 (nds32_expand_setmem_loop_v3m): Ditto.
15237 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
15240 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
15242 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
15244 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
15246 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
15247 (nds32_init_machine_status): Initialize machine->attr_naked_p and
15248 machine->attr_no_prologue_p.
15249 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
15250 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
15251 (nds32_expand_epilogue): Consider attr_naked_p.
15252 (nds32_expand_epilogue_v3pop): Likewise.
15253 (nds32_can_use_return_insn): Likewise.
15254 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
15255 attr_no_prologue_p fields.
15256 * config/nds32/nds32.opt (mret-in-naked-func): New option.
15258 2018-05-27 Jakub Jelinek <jakub@redhat.com>
15261 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
15263 * config/i386/sse.md
15264 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
15266 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
15268 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
15270 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
15272 (*<floatsuffix>floatv2div2sf2): Rename to ...
15273 (*float<floatunssuffix>v2div2sf2): ... this.
15274 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
15275 (float<floatunssuffix>v2div2sf2_mask): ... this.
15276 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
15277 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
15278 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
15280 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
15282 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
15284 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
15286 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
15288 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
15290 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
15291 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
15292 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
15293 gen_ufix_truncv8dfv8si2.
15294 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
15295 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
15296 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
15297 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
15298 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
15299 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
15300 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
15301 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
15303 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
15307 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
15309 2018-05-25 Jim Wilson <jimw@sifive.com>
15311 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
15312 * config/riscv/riscv.c (struct machine_function): Add
15313 interrupt_handler_p and attribute_checked_p fields.
15314 (riscv_attribute_table): Add interrupt.
15315 (riscv_interrupt_type_p): New.
15316 (riscv_save_reg_p): Save extra regs for interrupt handler.
15317 (riscv_use_save_libcall): Return false for interrupt handler.
15318 (riscv_first_stack_step): Add forward declaration.
15319 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
15320 for interrupt handler with large frame. Use it for saved reg list.
15321 (riscv_expand_prologue): Move flag_stack_usage_info support to
15322 eliminate duplication.
15323 (riscv_expand_epilogue): Generate mret for interrupt handler.
15324 (riscv_epilogue_uses): New.
15325 (riscv_can_use_return_insn): Return false for interrupt handler.
15326 (riscv_function_ok_for_sibcall): Likewise.
15327 (riscv_set_current_function): Add interrupt handler support.
15328 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
15329 * config/riscv/riscv.md (UNSPECV_MRET): New.
15331 (riscv_frflags, riscv_fsflags): Use tab after opcode.
15333 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
15335 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
15337 PR tree-optimization/85712
15338 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
15339 this candidate has already been replaced in-situ by a copy.
15341 2018-05-25 Jason Merrill <jason@redhat.com>
15343 PR c++/80485 - inline function non-zero address.
15344 * symtab.c (nonzero_address): Check DECL_COMDAT.
15346 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
15349 * config/alpha/alpha.md (ashlsi3): New insn pattern.
15350 (*ashlsi_se): Rename from *ashldi_se. Define as sign
15351 extension of SImode operation. Use const123_operand predicate.
15352 (*saddsi_1): Remove.
15353 (*saddl_se_1): Ditto.
15354 (*ssubsi_1): Ditto.
15355 (*ssubl_se_1): Ditto.
15356 * config/alpha/predicates.md (const123_operand): New predicate.
15357 * config/alpha/constraints.md (P): Use IN_RANGE.
15359 2018-05-25 Richard Biener <rguenther@suse.de>
15361 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
15363 (ref_maybe_used_by_stmt_p): Likewise.
15364 (stmt_may_clobber_ref_p): Likewise.
15365 (stmt_may_clobber_ref_p_1): Likewise.
15366 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
15368 (ref_maybe_used_by_stmt_p): Likewise.
15369 (stmt_may_clobber_ref_p): Likewise.
15370 (stmt_may_clobber_ref_p_1): Likewise.
15371 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
15372 the alias oracle to disambiguate DRs with stmts DR analysis
15374 (vect_analyze_data_refs): Do not give up on not analyzable
15375 DRs for BB vectorization. Remove code truncating the dataref
15378 2018-05-25 Jakub Jelinek <jakub@redhat.com>
15381 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
15382 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
15383 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
15385 2018-05-25 Richard Biener <rguenther@suse.de>
15387 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
15388 function, combining stmt data ref gathering and fatal analysis
15390 (vect_analyze_data_refs): Remove now redudnant code and simplify.
15391 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
15392 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
15393 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
15394 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
15396 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15398 PR tree-optimization/85720
15399 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
15400 SCC if all partitions are builtins.
15401 (version_loop_by_alias_check): New parameter. Generate cancelable
15402 runtime alias check if all partitions are builtins.
15403 (distribute_loop): Update call to above function.
15405 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15407 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
15408 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
15409 (parm_default_def_partition_arg): Ditto.
15410 (set_parm_default_def_partition): Ditto.
15411 (get_parm_default_def_partitions): Ditto and make it static.
15412 (get_undefined_value_partitions): Ditto and make it static.
15413 (remove_ssa_form): Refactor call to init_var_map here.
15414 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
15415 computation for loop region.
15416 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
15417 (register_default_def): Delete.
15418 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
15419 (parm_default_def_partition_arg): Ditto.
15420 (set_parm_default_def_partition): Ditto.
15421 (get_parm_default_def_partitions): Ditto and make it static.
15422 (get_undefined_value_partitions): Ditto and make it static.
15423 (coalesce_with_default, coalesce_with_default): Update comment.
15424 (create_coalesce_list_for_region): New func factored out from
15425 create_outofssa_var_map.
15426 (populate_coalesce_list_for_outofssa): New func factored out from
15427 create_outofssa_var_map and coalesce_ssa_name.
15428 (create_outofssa_var_map): Delete.
15429 (coalesce_ssa_name): Refactor to support live range computation.
15430 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
15431 (get_parm_default_def_partitions): Delete.
15432 (get_undefined_value_partitions): Ditto.
15433 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
15434 computation for loop region.
15435 (new_tree_live_info, loe_visit_block): Ditto.
15436 (live_worklist, set_var_live_on_entry): Ditto.
15437 (calculate_live_on_exit, verify_live_on_entry): Ditto.
15438 * tree-ssa-live.h (struct _var_map): New fields.
15439 (init_var_map): Change decl.
15440 (region_contains_p): New.
15442 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15444 * tree-ssa-live.h (live_merge_and_clear): Delete.
15446 2018-05-25 Richard Biener <rguenther@suse.de>
15449 * tree-dump.c (dequeue_and_dump): Remove access to removed
15450 operand 2 of a SWITCH_EXPR.
15452 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15454 * doc/sourcebuild.texi (vect_double_cond_arith): Include
15455 multiplication and division.
15456 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
15457 (cond_udiv@var{m}, cond_umod@var{m}): Document.
15458 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
15459 (cond_udiv_optab, cond_umod_optab): New optabs.
15460 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
15461 (IFN_COND_RDIV): New internal functions.
15462 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
15463 TRUNC_MOD_EXPR and RDIV_EXPR.
15464 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
15465 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
15467 (SVE_INT_BINARY): Include mult.
15468 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
15469 (optab, sve_int_op): Handle mult.
15470 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
15472 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
15473 for SVE_INT_BINARY_SD.
15475 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15477 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
15478 (optab, sve_int_op): Handle div and udiv.
15479 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
15480 for SVE_INT_BINARY_SD.
15481 (*<optab><mode>3): New insn for the same.
15483 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15485 * tree-vect-patterns.c: Include predict.h.
15486 (vect_recog_divmod_pattern): Restrict check for division support
15487 to when optimizing for size.
15489 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15491 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
15492 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
15493 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
15494 (gimple_match_op::set_op): Likewise.
15495 (gimple_resimplify4): Declare.
15496 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
15497 (expr::gen_transform): Likewise.
15498 (decision_tree::gen): Generate a simplification routine for 4 operands.
15499 * gimple-match-head.c (gimple_simplify): Add an overload for
15500 4 operands. In the top-level function, handle up to 4 call
15501 arguments and call gimple_resimplify4.
15502 (gimple_resimplify4): New function.
15503 (build_call_internal): Pass a fourth operand.
15504 (maybe_push_to_seq): Likewise.
15505 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
15506 Fold VEC_COND_EXPRs of an operation and a default value into
15507 an IFN_COND_* function if possible.
15508 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
15510 (SVE_COND_FP_BINARY): Include them.
15511 (optab, sve_fp_op): Handle them.
15512 (SVE_INT_BINARY_REV): New code iterator.
15513 (SVE_COND_FP_BINARY_REV): New int iterator.
15514 (commutative): New int attribute.
15515 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
15517 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
15519 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
15520 (*cond_<optab><mode>): New patterns for reversed operands.
15522 2018-05-25 Richard Biener <rguenther@suse.de>
15524 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
15525 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
15526 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
15527 (STMT_VINFO_GROUPED_ACCESS): Adjust.
15528 * tree-vect-data-refs.c (everywhere): Adjust users.
15529 * tree-vect-loop.c (everywhere): Likewise.
15530 * tree-vect-slp.c (everywhere): Likewise.
15531 * tree-vect-stmts.c (everywhere): Likewise.
15532 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
15534 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15536 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
15538 (gcc_cv_as_section_exclude): ... this.
15539 Try Solaris as #exclude syntax.
15540 * configure: Regenerate.
15541 * config.in: Regenerate.
15542 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
15544 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
15545 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
15547 * varasm.c (default_elf_asm_named_section): Don't check if
15548 HAVE_GAS_SECTION_EXCLUDE is defined.
15550 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15552 * doc/md.texi: Update the documentation of the cond_* optabs
15553 to mention the new final operand. Fix GET_MODE_NUNITS call.
15554 Describe the scalar case too.
15555 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
15556 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
15558 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
15559 (get_conditional_internal_fn): Update comment.
15560 * tree-vect-loop.c (vectorizable_reduction): Pass the original
15561 accumulator value as a final argument to conditional functions.
15562 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
15563 a define_expand and add an "else" operand. Assert for now that
15564 the else operand is equal to operand 2. Use SVE_INT_BINARY and
15565 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
15566 (*cond_<optab><mode>): New patterns.
15567 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
15568 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
15569 (UNSPEC_COND_EOR): Delete.
15570 (optab): Remove associated mappings.
15571 (SVE_INT_BINARY): New code iterator.
15572 (sve_int_op): Remove int attribute and add "minus" to the code
15574 (SVE_COND_INT_OP): Delete.
15575 (SVE_COND_FP_OP): Rename to...
15576 (SVE_COND_FP_BINARY): ...this.
15578 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15580 * optabs.c (can_reuse_operands_p): New function.
15581 (maybe_legitimize_operands): Try to reuse the results for
15584 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
15586 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
15587 Add {q} suffix to insn mnemonic.
15589 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
15591 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
15592 (msp430_warn_func_return): New.
15594 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
15596 * fold-const.c (tree_nonzero_bits): New function.
15597 * fold-const.h (tree_nonzero_bits): Likewise.
15598 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
15599 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
15601 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
15605 * varasm.c (assemble_alias): Check ifunc_resolver only on
15608 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
15611 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
15612 when memory input operand is handled.
15614 2018-05-24 Luis Machado <luis.machado@linaro.org>
15616 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
15618 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
15620 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
15622 * match.pd: Delay FMA folds until after vectorization.
15624 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
15627 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
15628 address check not strict.
15630 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
15632 * gimple-match.h (gimple_match_op): New class.
15633 (mprts_hook): Replace parameters with a gimple_match_op *.
15634 (maybe_build_generic_op): Likewise.
15635 (gimple_simplified_result_is_gimple_val): Replace parameters with
15636 a const gimple_match_op *.
15637 (gimple_simplify): Replace code_helper * and tree * parameters with
15638 a gimple_match_op * parameter.
15639 (gimple_resimplify1): Replace code_helper *, tree and tree *
15640 parameters with a gimple_match_op * parameter.
15641 (gimple_resimplify2): Likewise.
15642 (gimple_resimplify3): Likewise.
15643 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
15644 parameters with a gimple_match_op * parameter.
15645 * gimple-match-head.c (gimple_simplify): Change prototypes of
15646 auto-generated functions to take a gimple_match_op * instead of
15647 separate code_helper * and tree * parameters. Make the same
15648 change in the top-level overload and update calls to the
15649 gimple_resimplify routines. Update calls to the auto-generated
15650 functions and to maybe_push_res_to_seq in the publicly-facing
15651 operation-specific gimple_simplify overloads.
15652 (gimple_match_op::MAX_NUM_OPS): Define.
15653 (gimple_resimplify1): Replace rcode and ops with a single res_op
15654 parameter. Update call to gimple_simplify.
15655 (gimple_resimplify2): Likewise.
15656 (gimple_resimplify3): Likewise.
15657 (mprts_hook): Replace parameters with a gimple_match_op *.
15658 (maybe_build_generic_op): Likewise.
15659 (build_call_internal): Replace type, nargs and ops with
15660 a gimple_match_op *.
15661 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
15662 with a single gimple_match_op *. Update calls to mprts_hook,
15663 build_call_internal and gimple_simplified_result_is_gimple_val.
15664 Factor out code that is common to the tree_code and combined_fn cases.
15665 * genmatch.c (expr::gen_transform): Replace tem_code and
15666 tem_ops with a gimple_match_op called tem_op. Update calls
15667 to the gimple_resimplify functions and maybe_push_res_to_seq.
15668 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
15669 res_ops. Update call to the gimple_resimplify functions.
15670 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
15671 (decision_tree::gen): Make the functions take a gimple_match_op *
15672 called res_op instead of separate res_code and res_ops parameters.
15673 Update call accordingly.
15674 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
15675 and ops with a single res_op parameter. Update calls to
15676 maybe_build_generic_op and maybe_push_res_to_seq.
15677 (fold_stmt_1): Update calls to gimple_simplify and
15678 replace_stmt_with_simplification.
15679 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
15680 and gimple_simplified_result_is_gimple_val.
15681 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
15683 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
15684 with a gimple_match_op *.
15685 (vn_nary_build_or_lookup): Likewise. Update call to
15686 vn_nary_build_or_lookup_1.
15687 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
15688 gimple_match_op *. Update calls to the gimple_resimplify routines
15689 and to gimple_simplified_result_is_gimple_val.
15690 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
15691 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
15692 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
15693 (visit_nary_op): Likewise.
15694 (visit_reference_op_load): Likewise.
15696 2018-05-23 Luis Machado <luis.machado@linaro.org>
15698 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
15699 modifier for printing the step amount.
15701 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15704 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
15707 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
15709 * doc/sourcebuild.texi (Endianness): New subsubsection.
15711 2018-05-23 Luis Machado <luis.machado@linaro.org>
15713 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
15714 <prefetch_dynamic_strides>: New const bool field.
15715 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
15716 prefetch_dynamic_strides.
15717 (exynosm1_prefetch_tune): Likewise.
15718 (thunderxt88_prefetch_tune): Likewise.
15719 (thunderx_prefetch_tune): Likewise.
15720 (thunderx2t99_prefetch_tune): Likewise.
15721 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
15723 (aarch64_override_options_internal): Update to set
15724 PARAM_PREFETCH_DYNAMIC_STRIDES.
15725 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
15726 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
15727 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
15728 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
15729 prefetch-dynamic-strides setting.
15731 2018-05-23 Luis Machado <luis.machado@linaro.org>
15733 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
15734 <minimum_stride>: New const int field.
15735 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
15736 minimum_stride field defaulting to -1.
15737 (exynosm1_prefetch_tune): Likewise.
15738 (thunderxt88_prefetch_tune): Likewise.
15739 (thunderx_prefetch_tune): Likewise.
15740 (thunderx2t99_prefetch_tune): Likewise.
15741 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
15742 <default_opt_level>: Set to 3.
15743 (aarch64_override_options_internal): Update to set
15744 PARAM_PREFETCH_MINIMUM_STRIDE.
15745 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
15746 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
15747 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
15748 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
15749 stride is constant and is below the minimum stride threshold.
15751 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15753 * config/arm/arm-cpus.in (mode26): Delete.
15754 (armv4): Delete mode26 reference.
15755 * config/arm/arm.c (arm_configure_build_target): Delete use of
15758 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
15760 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
15762 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
15763 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
15765 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
15766 (floatunsdidf2): Ditto.
15768 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
15770 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
15771 (fixuns_trunc<mode>si2_avx512f): Ditto.
15772 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
15773 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
15774 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
15776 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
15778 PR rtl-optimization/79985
15779 * df-scan.c (df_insn_refs_collect): Remove special case for
15780 global registers and asm statements.
15782 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
15784 * extend.texi (Global Register Variables): Rewrite the bullet list.
15785 Note that the register is available for allocation. Note that access
15786 via inline asm must use constraints. Add note about async-signal
15787 handlers. Remove paragraph about automagic register selection.
15789 2018-05-23 Richard Biener <rguenther@suse.de>
15791 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
15792 of fixed offset from memset VN.
15794 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
15796 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
15797 first_interp field.
15798 (alloc_cand_and_find_basis): Initialize first_interp field.
15799 (slsr_process_mul): Modify first_interp field.
15800 (slsr_process_add): Likewise.
15801 (slsr_process_cast): Modify first_interp field for each new
15803 (slsr_process_copy): Likewise.
15804 (dump_candidate): Dump first_interp field.
15805 (replace_mult_candidate): Process all interpretations, not just
15807 (replace_rhs_if_not_dup): Likewise.
15808 (replace_one_candidate): Likewise.
15810 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
15812 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
15814 (aarch64_needs_frame_chain): New function.
15815 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
15817 2018-05-23 Sudakshina Das <sudi.das@arm.com>
15820 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
15821 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
15822 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
15823 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
15824 as true for strict-align.
15825 (aarch64_can_inline_p): Perform checks even when callee has no
15826 attributes to check for strict alignment.
15827 * doc/extend.texi (AArch64 Function Attributes): Document
15829 * doc/invoke.texi: (AArch64 Options): Likewise.
15831 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
15833 PR tree-optimization/85853
15834 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
15835 the handling of the root of the node to...
15836 (vect_slp_analyze_node_operations_1): ...this new function,
15837 and run the whole thing with the child nodes' def types
15838 set according to their SLP node's def type.
15840 2018-05-23 Richard Biener <rguenther@suse.de>
15842 PR middle-end/85874
15843 * tree-data-ref.c (create_runtime_alias_checks): Defer
15844 and ignore overflow warnings.
15846 2018-05-23 Yury Gribov <tetra2005@gmail.com>
15848 PR tree-optimization/85822
15849 * tree-vrp.c (is_masked_range_test): Fix handling of negative
15852 2018-05-23 Richard Biener <rguenther@suse.de>
15854 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
15855 memset constants via native_interpret_expr.
15857 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
15860 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
15862 (cgraph_node::create_alias): Likewise.
15863 (cgraph_node::get_availability): Check ifunc_resolver instead
15864 of looking up ifunc attribute.
15865 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
15866 * varasm.c (do_assemble_alias): Likewise.
15867 (assemble_alias): Likewise.
15868 (default_binds_local_p_3): Likewise.
15869 * cgraph.h (cgraph_node): Add ifunc_resolver.
15870 (cgraph_node::only_called_directly_or_aliased_p): Return false
15871 for IFUNC resolver.
15872 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
15874 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
15875 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
15876 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
15877 instead of looking up ifunc attribute.
15879 2018-05-22 Luis Machado <luis.machado@linaro.org>
15881 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
15883 2018-05-22 Martin Sebor <msebor@redhat.com>
15885 PR middle-end/85359
15886 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
15887 only when expasion succeeds.
15888 (expand_builtin_strcmp): Same.
15889 (expand_builtin_strncmp): Same.
15891 2018-05-22 Martin Sebor <msebor@redhat.com>
15893 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
15895 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
15896 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15898 * config/aarch64/aarch64-ldpstp.md: Replace uses of
15899 aarch64_mem_pair_operand with memory_operand and delete operand swapping
15901 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
15902 Add check for legitimate_address.
15903 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
15904 (aarch64_swap_ldrstr_operands): New.
15905 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
15908 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
15909 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15911 * config/aarch64/aarch64.md: New patterns to generate stp
15913 (store_pair_sw, store_pair_dw): New patterns to generate stp for
15914 single words and double words.
15915 (load_pair_sw, load_pair_dw): Likewise.
15916 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
15918 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
15920 * config/aarch64/aarch64-ldpstp.md: Modify peephole
15921 for different mode ldpstp and add peephole for merged zero stores.
15922 Likewise for loads.
15923 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
15925 (aarch64_gen_store_pair): Rename calls to match new patterns.
15926 (aarch64_gen_load_pair): Rename calls to match new patterns.
15927 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
15928 (load_pair<DREG:mode><DREG2:mode>): ... This.
15929 (store_pair<mode>): Rename to...
15930 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
15931 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
15932 New mode iterators.
15933 (V_INT_EQUIV): Handle SImode.
15934 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
15937 2018-05-22 Martin Sebor <msebor@redhat.com>
15940 * calls.c (maybe_warn_nonstring_arg): Use string length to set
15941 or ajust the presumed bound on an operation to avoid unnecessary
15944 2018-05-22 Martin Sebor <msebor@redhat.com>
15946 PR tree-optimization/85826
15947 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
15948 assuming that a DECL necesarily has a constant size.
15950 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
15952 PR middle-end/85862
15953 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
15955 2018-05-22 Richard Biener <rguenther@suse.de>
15957 PR tree-optimization/85834
15958 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
15959 non-constant and non-zero memset arguments.
15961 2018-05-22 Martin Liska <mliska@suse.cz>
15964 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
15966 2018-05-22 Richard Biener <rguenther@suse.de>
15968 PR tree-optimization/85863
15969 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
15970 comparisons when vectype is specified.
15971 (vectorizable_condition): Do not specify vectype for
15972 vect_is_simple_cond when SLP vectorizing.
15974 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
15977 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
15978 define __ibm128 as long double.
15979 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
15980 as a distinct type when IEEE 128-bit support is enabled.
15981 (init_float128_ieee): Fix up conversions between IFmode and IEEE
15982 128-bit types to use the correct functions.
15983 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
15984 convert between 128-bit floating point types that have different
15985 modes but the same representation, instead of using gen_lowpart to
15987 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
15989 (IFKF_reg): New attributes to give the register constraints for
15991 (extend<mode>tf2_internal): New insns to mark an explicit
15992 conversion between 128-bit floating point types that have a
15993 different mode but share the same representation.
15995 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
15997 PR tree-optimization/85814
15998 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
15999 a null return from get_strinfo when unsharing the next
16000 strinfo in the chain.
16002 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
16005 * varasm.c (weak_finish): Clean up weak_decls.
16007 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16009 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
16010 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
16011 UNSPEC_UADALP values.
16012 * config/aarch64/iterators.md (ABAL): New int iterator.
16015 (sur): Add mappings for the above.
16016 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
16018 (aarch64_<sur>abal<mode>_4): Likewise.
16019 (aarch64_<sur>adalp<mode>_3): Likewise.
16020 (<sur>sadv16qi): New define_expand.
16022 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
16024 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
16025 (*movdf_internal): Ditto.
16026 (*rcpsf2_sse): Ditto.
16027 (*rsqrtsf2_sse): Ditto.
16028 (*sqrt<mode>2_sse): Ditto.
16030 2018-05-21 Tamar Christina <tamar.christina@arm.com>
16032 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
16034 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
16035 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
16036 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
16037 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
16039 * config/aarch64/arm_neon.h: Likewise.
16040 * config/aarch64/iterators.md (VQ_I): New.
16042 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
16044 * config.gcc: Add arc/t-multilib-linux to tmake_file for
16046 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
16049 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
16051 * config/nds32/constraints.md (S): New constraint.
16052 * config/nds32/nds32.md (call_internal): Use constraint S.
16053 (call_value_internal): Likewise.
16054 (sibcall_internal): Likewise.
16055 (sibcall_value_internal): Likewise.
16057 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
16058 Chung-Ju Wu <jasonwucj@gmail.com>
16060 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
16061 into consideration.
16063 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
16064 Chung-Ju Wu <jasonwucj@gmail.com>
16066 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
16067 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
16068 (nds32_rtx_costs_impl): Simplify.
16069 (nds32_address_cost_impl): Simplify.
16070 (nds32_init_rtx_costs): New function.
16071 (nds32_rtx_costs_speed_prefer): Likewise.
16072 (nds32_rtx_costs_size_prefer): Likewise.
16073 (nds32_address_cost_speed_prefer): Likewise.
16074 (nds32_address_cost_speed_fwprop): Likewise.
16075 (nds32_address_cost_size_prefer): Likewise.
16076 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
16077 * config/nds32/nds32.c (nds32_option_override): Use
16078 nds32_init_rtx_costs function.
16080 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
16082 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
16083 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
16084 (TARGET_PIPELINE_N8): Likewise.
16085 (TARGET_PIPELINE_N10): Likewise.
16086 (TARGET_PIPELINE_N13): Likewise.
16087 (TARGET_PIPELINE_GRAYWOLF): Likewise.
16089 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
16091 * config/nds32/nds32-fpu.md: Update copyright year.
16093 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16095 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
16097 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16099 * config/nds32/nds32.c
16100 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
16101 * config/nds32/nds32.opt (minline-asm-r15): New option.
16103 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16105 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
16107 * config/nds32/nds32.md (abssi2): New pattern.
16109 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
16111 * config/i386/i386.md (rex64namesuffix): New mode attribute.
16112 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
16113 Merge insn pattern from sse_cvtsi2ss<round_name> and
16114 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
16115 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
16116 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
16117 using SWI48 mode iterator.
16118 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
16119 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
16120 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
16121 pattern from sse_cvttss2si<round_saeonly_name>
16122 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
16123 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
16124 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
16125 using SWI48 mode iterator.
16126 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
16127 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
16128 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
16129 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
16130 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
16131 using SWI48 mode iterator.
16132 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
16133 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
16134 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
16135 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
16136 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
16137 SWI48 mode iterator.
16138 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
16139 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
16140 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
16141 pattern from sse_cvttsd2si<round_saeonly_name>
16142 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
16144 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16146 * config/nds32/nds32-md-auxiliary.c
16147 (nds32_valid_smw_lwm_base_p): Refine.
16148 (nds32_output_smw_single_word): Refine.
16149 (nds32_output_smw_double_word): New.
16150 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
16152 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16154 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
16155 (nds32_output_stack_pop): Refine.
16156 (nds32_expand_unaligned_load): Refine.
16157 (nds32_expand_unaligned_store): Refine.
16159 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16160 Chung-Ju Wu <jasonwucj@gmail.com>
16162 * config/nds32/constants.md: Add TP_REGNUM constant.
16163 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
16164 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
16166 * config/nds32/nds32-doubleword.md: Consider flag_pic.
16167 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
16168 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
16169 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
16170 and PIC code generation.
16171 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
16173 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
16175 * config/nds32/nds32.md: Support TLS and PIC.
16176 * config/nds32/nds32.c: Support TLS and PIC.
16177 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
16178 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
16181 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16183 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
16184 mode with E_ prefix.
16186 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16187 Chung-Ju Wu <jasonwucj@gmail.com>
16189 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
16190 * config/nds32/nds32-md-auxiliary.c
16191 (symbolic_reference_mentioned_p): New.
16192 (nds32_legitimize_ict_address): New.
16193 (nds32_expand_ict_move): New.
16194 (nds32_indirect_call_referenced_p): New.
16195 (nds32_symbol_binds_local_p): Delete.
16196 (nds32_long_call_p): Modify.
16197 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
16198 * config/nds32/nds32-protos.h
16199 (symbolic_reference_mentioned_p): Declare.
16200 (nds32_legitimize_ict_address): Declare.
16201 (nds32_expand_ict_move): Declare.
16202 (nds32_indirect_call_referenced_p): Declare.
16203 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
16204 (nds32_relax_group): Use nds32_ict_const_p as condition.
16205 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
16206 (nds32_asm_file_start): Output ict_model directive in asm code.
16207 (nds32_legitimate_address_p): Consider indirect call.
16208 (nds32_print_operand): Consider indirect call.
16209 (nds32_print_operand_address): Consider indirect call.
16210 (nds32_insert_attributes): Handle "indirect_call" attribute.
16211 (TARGET_LEGITIMATE_ADDRESS_P): Define.
16212 (TARGET_LEGITIMATE_CONSTANT_P): Define.
16213 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
16214 (TARGET_DELEGITIMIZE_ADDRESS): Define.
16215 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
16216 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
16217 (TARGET_ICT_MODEL_SMALL): Define.
16218 (TARGET_ICT_MODEL_LARGE): Define.
16219 * config/nds32/nds32.md (movsi): Consider ict model.
16220 (call, call_value): Consider ict model.
16221 (sibcall, sibcall_value): Consider ict model.
16222 * config/nds32/nds32.opt (mict-model): New option.
16223 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
16226 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
16227 Monk Chiang <sh.chiang04@gmail.com>
16228 Jim Wilson <jimw@sifive.com>
16230 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
16231 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
16232 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
16233 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
16234 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
16235 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
16236 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
16237 compute save_libcall_adjustment properly.
16238 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
16239 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
16240 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
16241 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
16242 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
16243 (ABI_SPEC): Handle mabi=ilp32e.
16244 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
16245 (RVE): Add RVE mask.
16246 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
16247 <-march>: Add rv32e as an example.
16249 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
16252 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
16253 (intra_create_variable_infos): Handle C++ constructors.
16255 2018-05-18 Martin Liska <mliska@suse.cz>
16257 * passes.def: Remove a redundant pass.
16259 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
16262 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
16264 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16266 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
16268 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
16270 (armv2, armv2a, armv3, armv3m): Delete architectures.
16271 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
16272 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
16273 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
16275 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
16276 (*mulsidi3adddi): Likewise.
16277 (mulsidi3): Likewise.
16278 (*mulsidi3_nov6): Likewise.
16279 (umulsidi3): Likewise.
16280 (umulsidi3_nov6): Likewise.
16281 (umaddsidi4): Likewise.
16282 (*umulsidi3adddi): Likewise.
16283 (smulsi3_highpart): Likewise.
16284 (*smulsi3_highpart_nov6): Likewise.
16285 (umulsi3_highpart): Likewise.
16286 (*umulsi3_highpart_nov6): Likewise.
16287 * config/arm/arm.h (arm_arch3m): Delete.
16288 * config/arm/arm.c (arm_arch3m): Delete.
16289 (arm_option_override_internal): Update armv3-related comment.
16290 (arm_configure_build_target): Delete use of isa_bit_mode32.
16291 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
16292 (arm_rtx_costs_internal): Delete check of arm_arch3m.
16293 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
16294 (mulsa3): Likewise.
16295 (mulusa3): Likewise.
16296 * config/arm/arm-protos.h (arm_arch3m): Delete.
16297 * config/arm/arm-tables.opt: Regenerate.
16298 * config/arm/arm-tune.md: Likewise.
16299 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
16300 deleted architectures.
16302 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16304 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
16305 (armv5t, armv5te): New features.
16306 (ARMv5, ARMv5e): Delete fgroups.
16307 (ARMv5t, ARMv5te): Adjust for above changes.
16308 (ARMv6m): Likewise.
16309 (armv5, armv5e): Delete arches.
16310 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
16312 (*call_reg_arm): Likewise.
16313 (*call_value_reg_armv5): Likewise.
16314 (*call_value_reg_arm): Likewise.
16315 (*call_symbol): Likewise.
16316 (*call_value_symbol): Likewise.
16317 (*sibcall_insn): Likewise.
16318 (*sibcall_value_insn): Likewise.
16319 (clzsi2): Likewise.
16320 (prefetch): Likewise.
16321 (define_split and define_peephole2 dependent on arm_arch5):
16323 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
16325 (TARGET_ARM_QBIT): Likewise.
16326 (TARGET_DSP_MULTIPLY): Likewise.
16327 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
16328 (arm_arch5, arm_arch5e): Delete.
16329 (arm_arch5t, arm_arch5te): Declare.
16330 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
16331 (arm_arch5t): Declare.
16332 (arm_option_reconfigure_globals): Update for the above.
16333 (arm_options_perform_arch_sanity_checks): Update comment, replace
16334 use of arm_arch5 with arm_arch5t.
16335 (use_return_insn): Likewise.
16336 (arm_emit_call_insn): Likewise.
16337 (output_return_instruction): Likewise.
16338 (arm_final_prescan_insn): Likewise.
16339 (arm_coproc_builtin_available): Likewise.
16340 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
16341 arm_arch5e with arm_arch5t and arm_arch5te.
16342 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
16343 (arm_arch5t, arm_arch5te): Declare.
16344 * config/arm/arm-tables.opt: Regenerate.
16345 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
16346 * config/arm/t-multilib: Likewise.
16347 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
16348 instead of arm_arch5.
16349 (*call_reg_thumb1): Likewise.
16350 (*call_value_reg_thumb1_v5): Likewise.
16351 (*call_value_reg_thumb1): Likewise.
16352 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
16354 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
16356 2018-05-18 Martin Liska <mliska@suse.cz>
16358 PR gcov-profile/84846
16359 * doc/gcov.texi: Document -t option of gcov tool.
16361 2018-05-18 Martin Liska <mliska@suse.cz>
16363 PR gcov-profile/84846
16364 * gcov.c (print_usage): Add new -t option.
16365 (process_args): Handle the option.
16366 (generate_results): Use stdout as output when requested by
16369 2018-05-18 Martin Liska <mliska@suse.cz>
16371 PR gcov-profile/84846
16372 * coverage.c (coverage_init): Write PWD to .gcno file.
16373 * doc/gcov.texi: Document how working directory is printed.
16374 * gcov-dump.c (dump_gcov_file): Print PWD.
16375 * gcov.c (output_intermediate_file): Likewise.
16376 (read_graph_file): Read PWD string.
16377 (output_lines): Print PWD.
16379 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16381 PR middle-end/85817
16382 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
16383 for retval and return false if all args to phi are zero.
16385 2018-05-18 Richard Biener <rguenther@suse.de>
16387 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
16389 (evrp_dom_walker::before_dom_children): Call it.
16391 2018-05-18 Richard Biener <rguenther@suse.de>
16393 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
16394 results when processing array refs with variable index.
16396 2018-05-18 Toon Moene <toon@moene.org>
16398 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
16399 directly after that of -floop-interchange. Indicate that both
16400 options are enabled by default when specifying -O3.
16402 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16404 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
16405 iterator. Delete separate integer-mode vec_set<mode> expander.
16406 (aarch64_simd_vec_setv2di): Delete.
16407 (vec_setv2di): Delete.
16408 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
16409 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
16410 the "w, r" alternative.
16412 2018-05-18 Martin Liska <mliska@suse.cz>
16414 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
16415 * tree-pass.h (make_pass_lower_switch_O0): New function.
16416 * tree-switch-conversion.c (node_has_low_bound): Remove.
16417 (node_has_high_bound): Likewise.
16418 (node_is_bounded): Likewise.
16419 (class pass_lower_switch): Make it a template type and create
16421 (pass_lower_switch::execute): Add template argument.
16422 (make_pass_lower_switch): New function.
16423 (make_pass_lower_switch_O0): New function.
16424 (do_jump_if_equal): Remove.
16425 (emit_case_nodes): Simplify to just handle all 3 cases and leave
16426 all the hard work to tree optimization passes.
16428 2018-05-18 Martin Liska <mliska@suse.cz>
16430 * dbgcnt.c (limit_low): Renamed from limit.
16431 (limit_high): New variable.
16432 (dbg_cnt_is_enabled): Check for upper limit.
16433 (dbg_cnt): Adjust dumping.
16434 (dbg_cnt_set_limit_by_index): Add new argument for high
16436 (dbg_cnt_set_limit_by_name): Likewise.
16437 (dbg_cnt_process_single_pair): Parse new format.
16438 (dbg_cnt_process_opt): Use strtok.
16439 (dbg_cnt_list_all_counters): Remove 'value' and add
16441 * doc/invoke.texi: Document changes.
16443 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
16445 * doc/sourcebuild.texi (scalar_all_fma): Document.
16446 * tree.def (FMA_EXPR): Delete.
16447 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
16448 * internal-fn.c (ternary_direct): New macro.
16449 (expand_ternary_optab_fn): Likewise.
16450 (direct_ternary_optab_supported_p): Likewise.
16451 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
16452 * builtins.c (fold_builtin_fma): Delete.
16453 (fold_builtin_3): Don't call it.
16454 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
16455 * expr.c (expand_expr_real_2): Likewise.
16456 * fold-const.c (operand_equal_p): Likewise.
16457 (fold_ternary_loc): Likewise.
16458 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
16459 * gimple.c (DEFTREECODE): Likewise.
16460 * gimplify.c (gimplify_expr): Likewise.
16461 * optabs-tree.c (optab_for_tree_code): Likewise.
16462 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16463 * tree-eh.c (operation_could_trap_p): Likewise.
16464 (stmt_could_throw_1_p): Likewise.
16465 * tree-inline.c (estimate_operator_cost): Likewise.
16466 * tree-pretty-print.c (dump_generic_node): Likewise.
16467 (op_code_prio): Likewise.
16468 * tree-ssa-loop-im.c (stmt_cost): Likewise.
16469 * tree-ssa-operands.c (get_expr_operands): Likewise.
16470 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
16471 * fold-const-call.h (fold_fma): Delete.
16472 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
16473 CFN_FNMA and CFN_FNMS.
16474 (fold_fma): Delete.
16475 * genmatch.c (combined_fn): New enum.
16476 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
16477 (commutative_op): New function.
16478 (commutate): Use it. Handle more than 2 operands.
16479 (dt_operand::gen_gimple_expr): Use commutative_op.
16480 (parser::parse_expr): Allow :c to be used with non-binary
16481 operators if the commutative operand is known.
16482 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
16483 CFN_FMS, CFN_FNMA and CFN_FNMS.
16484 (backprop::process_assign_use): Remove FMA_EXPR handling.
16485 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
16486 (gen_hsa_fma): New function.
16487 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
16488 IFN_FNMA and IFN_FNMS.
16489 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
16490 * gimple-fold.h (follow_all_ssa_edges): Declare.
16491 * gimple-fold.c (follow_all_ssa_edges): New function.
16492 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
16493 gimple_build interface and use follow_all_ssa_edges to fold the result.
16494 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
16495 instead of checking for optabs directly.
16496 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
16497 rather than FMA_EXPRs.
16498 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
16499 call to IFN_FMA instead of an FMA_EXPR.
16501 2018-05-17 Jim Wilson <jimw@sifive.com>
16503 * expr.c (do_tablejump): When converting index to Pmode, if we have a
16504 sign extended promoted subreg, and the range does not have the sign bit
16505 set, then do a sign extend.
16507 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
16508 test, check for sign extended subreg and/or constant operands, and
16509 do a sign extend in that case.
16511 2018-05-17 Steve Ellcey <sellcey@cavium.com>
16513 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
16514 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
16516 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
16517 Change logics_shift_reg to logics_shift_imm.
16518 (thunderx2t99_fp_loadpair_basic): Delete.
16519 (thunderx2t99_fp_storepair_basic): Delete.
16520 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
16521 (thunderx2t99_asimd_polynomial): Delete.
16522 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
16523 and neon_fp_mul_d_scalar_q.
16524 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
16525 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
16526 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
16527 (thunderx2t99_asimd_lut): Add missing tbl types.
16528 (thunderx2t99_asimd_ext): Delete.
16529 (thunderx2t99_asimd_load1_1_mult): Delete.
16530 (thunderx2t99_asimd_load1_2_mult): Delete.
16531 (thunderx2t99_asimd_load1_ldp): New.
16532 (thunderx2t99_asimd_load1): New.
16533 (thunderx2t99_asimd_load2): Add missing *load2* types.
16534 (thunderx2t99_asimd_load3): New.
16535 (thunderx2t99_asimd_load4): New.
16536 (thunderx2t99_asimd_store1_1_mult): Delete.
16537 (thunderx2t99_asimd_store1_2_mult): Delete.
16538 (thunderx2t99_asimd_store2_mult): Delete.
16539 (thunderx2t99_asimd_store2_onelane): Delete.
16540 (thunderx2t99_asimd_store_stp): New.
16541 (thunderx2t99_asimd_store1): New.
16542 (thunderx2t99_asimd_store2): New.
16543 (thunderx2t99_asimd_store3): New.
16544 (thunderx2t99_asimd_store4): New.
16546 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
16548 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
16549 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
16551 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
16552 Segher Boessenkool <segher@kernel.crashing.org>
16555 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
16558 2018-05-17 Richard Biener <rguenther@suse.de>
16560 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
16561 for pruning loop and prune defs feeding only already visited PHIs.
16563 2018-05-17 Richard Biener <rguenther@suse.de>
16565 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
16567 2018-05-17 Bin Cheng <bin.cheng@arm.com>
16568 Richard Biener <rguenther@suse.de>
16570 PR tree-optimization/85793
16571 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
16572 for VMAT_ELEMENTWISE.
16574 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
16576 * internal-fn.h (lookup_internal_fn): Declare
16577 * internal-fn.c (lookup_internal_fn): New function.
16578 * gimple.c (gimple_build_call_from_tree): Handle calls to
16579 internal functions.
16580 * gimple-pretty-print.c (dump_gimple_call): Print "." before
16581 internal function names.
16582 * tree-pretty-print.c (dump_generic_node): Likewise.
16583 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
16585 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
16587 * gimple-fold.h (gimple_build): Make the function forms take
16588 combined_fn rather than built_in_function.
16589 (gimple_simplify): Likewise.
16590 * gimple-match-head.c (gimple_simplify): Likewise.
16591 * gimple-fold.c (gimple_build): Likewise.
16592 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
16593 rather than gimple_build_call_internal.
16594 (get_initial_defs_for_reduction): Likewise.
16595 (vect_create_epilog_for_reduction): Likewise.
16596 (vectorizable_live_operation): Likewise.
16598 2018-05-17 Martin Liska <mliska@suse.cz>
16600 * gimple-ssa-sprintf.c (format_directive): Do not use
16601 space in between 'G_' and '('.
16603 2018-05-17 Jakub Jelinek <jakub@redhat.com>
16606 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
16607 even if the mask is not all ones.
16610 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
16612 (ix86_gimple_fold_builtin): Likewise.
16615 * config/i386/i386.c: Include tree-vector-builder.h.
16616 (ix86_vector_shift_count): New function.
16617 (ix86_fold_builtin): Fold shift builtins by scalar count.
16618 (ix86_gimple_fold_builtin): Likewise.
16620 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
16621 _mm512_setzero): New intrinsics.
16623 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
16624 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16626 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
16627 code generation for cases where splatting a value is not useful.
16628 * simplify-rtx.c (simplify_ternary_operation): Simplify
16629 vec_merge across a vec_duplicate and a paradoxical subreg forming
16630 a vector mode to a vec_concat.
16632 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
16634 * config.gcc: Support "goldmont-plus".
16635 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
16637 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
16638 PROCESSOR_GOLDMONT_PLUS.
16639 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
16640 (processor_target_table): Add "goldmont-plus".
16641 (PTA_GOLDMONT_PLUS): Define.
16642 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
16643 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
16644 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
16645 (fold_builtin_cpu): Add "goldmont-plus".
16646 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
16647 (ix86_option_override_internal): Add "goldmont-plus".
16648 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
16649 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
16650 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
16651 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
16653 2018-05-17 Richard Biener <rguenther@suse.de>
16655 PR tree-optimization/85757
16656 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
16657 remove defs that only feed that PHI from further processing.
16659 2018-05-16 Jim Wilson <jimw@sifive.com>
16661 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
16663 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
16665 2018-05-16 Mark Wielaard <mark@klomp.org>
16667 * dwarf2out.c (count_index_strings): New function.
16668 (output_indirect_strings): Call count_index_strings and generate
16669 header for dwarf_version >= 5.
16671 2018-05-16 Mark Wielaard <mark@klomp.org>
16673 * dwarf2out.c (dwarf_FORM): New function.
16674 (set_indirect_string): Use dwarf_FORM.
16675 (reset_indirect_string): Likewise.
16676 (size_of_die): Likewise.
16677 (value_format): Likewise.
16678 (output_die): Likewise.
16679 (add_skeleton_AT_string): Likewise.
16680 (output_macinfo_op): Likewise.
16681 (index_string): Likewise.
16682 (output_index_string_offset): Likewise.
16683 (output_index_string): Likewise.
16684 (count_index_strings): Likewise.
16686 2018-05-16 Carl Love <cel@us.ibm.com>
16688 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
16689 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
16691 2018-05-16 Martin Jambor <mjambor@suse.cz>
16693 * ipa-prop.c (ipa_free_all_edge_args): Remove.
16694 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
16696 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
16698 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
16699 (fnma<mode>4): Likewise.
16700 (fms<mode>4): Likewise.
16701 (fnms<mode>4): Likewise.
16702 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
16703 (aarch64_fnma<mode>4): Likewise.
16704 (aarch64_fms<mode>4): Likewise.
16705 (aarch64_fnms<mode>4): Likewise.
16706 (aarch64_fnmadd<mode>4): Likewise.
16708 2018-05-16 Jason Merrill <jason@redhat.com>
16710 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
16712 2018-05-16 Richard Biener <rguenther@suse.de>
16714 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
16715 (dump_stmt_cost): Declare.
16716 (add_stmt_cost): Dump cost we add.
16717 (add_stmt_costs): New function.
16718 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
16719 No longer exported.
16720 (vect_analyze_stmt): Adjust prototype.
16721 (vectorizable_condition): Likewise.
16722 (vectorizable_live_operation): Likewise.
16723 (vectorizable_reduction): Likewise.
16724 (vectorizable_induction): Likewise.
16725 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
16726 cost vector to pass to vectorizable_ and record afterwards.
16727 (vect_model_reduction_cost): Take cost vector argument and adjust.
16728 (vect_model_induction_cost): Likewise.
16729 (vectorizable_reduction): Likewise.
16730 (vectorizable_induction): Likewise.
16731 (vectorizable_live_operation): Likewise.
16732 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
16733 SLP_TREE_NUMBER_OF_VEC_STMTS.
16734 (vect_analyze_slp_cost_1): Remove.
16735 (vect_analyze_slp_cost): Likewise.
16736 (vect_slp_analyze_node_operations): Take visited args and
16737 a target cost vector. Avoid processing already visited stmt sets.
16738 (vect_slp_analyze_operations): Use a local cost vector to gather
16739 costs and register those of non-discarded instances.
16740 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
16741 (vect_schedule_slp_instance): Remove copying of
16742 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
16744 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
16745 adding cost. Record cost entry location.
16746 (vect_prologue_cost_for_slp_op): Function to compute cost of
16747 a constant or invariant generated for SLP vect in the prologue,
16748 split out from vect_analyze_slp_cost_1.
16749 (vect_model_simple_cost): Make static. Adjust for SLP costing.
16750 (vect_model_promotion_demotion_cost): Likewise.
16751 (vect_model_store_cost): Likewise, make static.
16752 (vect_model_load_cost): Likewise.
16753 (vectorizable_bswap): Add cost vector arg and adjust.
16754 (vectorizable_call): Likewise.
16755 (vectorizable_simd_clone_call): Likewise.
16756 (vectorizable_conversion): Likewise.
16757 (vectorizable_assignment): Likewise.
16758 (vectorizable_shift): Likewise.
16759 (vectorizable_operation): Likewise.
16760 (vectorizable_store): Likewise.
16761 (vectorizable_load): Likewise.
16762 (vectorizable_condition): Likewise.
16763 (vectorizable_comparison): Likewise.
16764 (can_vectorize_live_stmts): Likewise.
16765 (vect_analyze_stmt): Likewise.
16766 (vect_transform_stmt): Adjust calls to vectorizable_*.
16767 * tree-vectorizer.c: Include gimple-pretty-print.h.
16768 (dump_stmt_cost): New function.
16770 2018-05-16 Richard Biener <rguenther@suse.de>
16772 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
16773 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
16774 * tree-ssa-dse.c: Include tree-ssa-loop.h.
16775 (check_name): New callback.
16776 (dse_classify_store): Track cycles via a visited bitmap of PHI
16777 defs and simplify handling of in-loop and across loop dead stores
16778 and properly fail for loop-variant refs. Handle byte-tracking with
16779 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
16782 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
16784 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
16785 (vect_get_mask_type_for_stmt): Likewise.
16786 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
16788 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
16789 to determine the statement's vector type and the vector type that
16790 should be used for calculating nunits. Deal with cases in which
16791 the type has to be deferred.
16792 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
16793 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
16794 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
16795 (vect_determine_vf_for_stmt): New functions, split out from...
16796 (vect_determine_vectorization_factor): ...here.
16797 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
16798 (vect_get_mask_type_for_stmt): New functions, split out from
16799 vect_determine_vectorization_factor.
16801 2018-05-16 Richard Biener <rguenther@suse.de>
16803 * tree-cfg.c (verify_gimple_assign_ternary): Properly
16804 verify the [VEC_]COND_EXPR embedded comparison.
16806 2018-05-15 Martin Sebor <msebor@redhat.com>
16808 PR tree-optimization/85753
16809 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
16810 RECORD_TYPE in addition to ARRAY_TYPE.
16812 2018-05-15 Martin Sebor <msebor@redhat.com>
16814 PR middle-end/85643
16815 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
16817 2018-05-15 Richard Biener <rguenther@suse.de>
16819 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
16820 add by_clobber_p one. Change algorithm to collect all defs
16821 representing uses we need to walk and try reducing them to
16822 a single one before failing.
16823 (dse_dom_walker::dse_optimize_stmt): Adjust.
16825 2018-05-13 Mark Wielaard <mark@klomp.org>
16827 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
16828 (size_of_loc_descr): Likewise.
16829 (output_loc_operands): Likewise.
16830 (output_loc_operands_raw): Likewise.
16831 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
16832 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
16833 (hash_loc_operands): Likewise.
16834 (compare_loc_operands): Likewise.
16836 2018-05-14 Mark Wielaard <mark@klomp.org>
16838 * dwarf2out.c (count_index_addrs): New function.
16839 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
16841 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16843 PR tree-optimization/83648
16844 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
16845 return value as malloc candidate.
16847 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16850 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
16851 param as true in call to suggest_attribute.
16853 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
16855 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
16856 -mreadonly-in-sdata.
16858 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16860 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
16862 (aarch64_crypto_aesd_fused): Likewise.
16864 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
16866 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
16867 (movsi_aarch64): Likewise.
16868 (load_pairsi): Likewise.
16869 (load_pairdi): Likewise.
16870 (store_pairsi): Likewise.
16871 (store_pairdi): Likewise.
16872 (load_pairsf): Likewise.
16873 (load_pairdf): Likewise.
16874 (store_pairsf): Likewise.
16875 (store_pairdf): Likewise.
16876 (zero_extend): Likewise.
16877 (trunc): Swap alternatives.
16878 (fcvt_target): Add '?' to prefer w over r.
16880 2018-05-14 Jakub Jelinek <jakub@redhat.com>
16883 * config/i386/i386.md: Disallow non-commutative arithmetics in
16884 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
16885 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
16886 in the peephole2 before it.
16888 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
16890 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
16891 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
16892 (ix86_handle_option): Handle -mcldemote.
16893 * config.gcc: New header.
16894 * config/i386/cldemoteintrin.h: New file.
16895 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
16896 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
16898 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
16899 OPTION_MASK_ISA_CLDEMOTE.
16900 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
16901 (ix86_valid_target_attribute_inner_p): Ditto.
16902 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
16903 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
16904 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
16905 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
16906 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
16908 * config/i386/i386.opt: Add -mcldemote.
16909 * config/i386/x86intrin.h: New header.
16910 * doc/invoke.texi: Add -mcldemote.
16912 2018-05-14 Richard Biener <rguenther@suse.de>
16914 * doc/match-and-simplify.texi: Adjust :s documentation.
16916 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
16918 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
16919 intended memcpy size.
16920 (REORDER_45): Likewise.
16922 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
16924 * sort.cc: New file.
16925 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
16926 * vec.c (qsort_chk): Use gcc_qsort.
16927 * Makefile.in (OBJS-libcommon): Add sort.o.
16928 (build/sort.o): New target. Use it...
16929 (BUILD_RTL): ... here, and...
16930 (build/gencfn-macros): ... here, and...
16931 (build/genmatch): ... here.
16933 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
16934 Chung-Ju Wu <jasonwucj@gmail.com>
16936 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
16937 * config/nds32/nds32-graywolf.md: New file.
16938 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
16939 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
16941 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
16942 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
16943 * config/nds32/nds32.md (pipeline_model): Add graywolf.
16944 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
16945 * config/nds32/pipelines.md: Include n15 settings.
16947 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
16948 Chung-Ju Wu <jasonwucj@gmail.com>
16950 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
16951 * config/nds32/nds32-n13.md: New file.
16952 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
16953 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
16955 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
16956 * config/nds32/nds32.md (pipeline_model): Add n13.
16957 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
16958 * config/nds32/pipelines.md: Include n13 settings.
16960 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
16961 Chung-Ju Wu <jasonwucj@gmail.com>
16963 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
16964 * config/nds32/nds32-n10.md: New file.
16965 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
16966 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
16968 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
16969 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
16970 * config/nds32/nds32.md (pipeline_model): Add n10.
16971 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
16972 * config/nds32/pipelines.md: Include n10 settings.
16974 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
16975 Kito Cheng <kito.cheng@gmail.com>
16976 Chung-Ju Wu <jasonwucj@gmail.com>
16978 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
16979 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
16980 Add enum values for DSP extension instructions.
16981 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
16983 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
16984 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
16985 New code iterators.
16986 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
16987 * config/nds32/nds32-dspext.md: New file for DSP implementation.
16988 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
16989 * config/nds32/nds32-intrinsic.md: Likewise.
16990 * config/nds32/nds32_intrinsic.h: Likewise.
16991 * config/nds32/nds32-md-auxiliary.c: Likewise.
16992 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
16993 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
16994 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
16995 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
16996 * config/nds32/nds32-protos.h: New declarations for DSP extension.
16997 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
16998 TYPE_DMAC in switch statement.
16999 * config/nds32/nds32.c: New checking and implementation for DSP
17000 extension instructions.
17001 * config/nds32/nds32.h: Likewise.
17002 * config/nds32/nds32.md: Likewise.
17003 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
17004 * config/nds32/predicates.md: Implement new predicates for DSP
17007 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17009 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
17010 Reformat alternatives and attributes so it is easier to identify
17011 which constraints/attributes go with which instruction.
17012 (mov<mode>_hardfloat32, FMOVE64): Likewise.
17013 (mov<mode>_softfloat32, FMOVE64): Likewise.
17014 (mov<mode>_hardfloat64, FMOVE64): Likewise.
17015 (mov<mode>_softfloat64, FMOVE64): Likewise.
17017 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
17019 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17021 (Basic PowerPC Built-in Functions): The new name of the
17022 subsection previously known as "PowerPC Built-in Functions".
17023 (Basic PowerPC Built-in Functions Available on all Configurations):
17025 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
17026 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
17027 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
17028 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
17030 2018-05-11 Martin Jambor <mjambor@suse.cz>
17033 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
17036 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
17039 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
17041 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
17043 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
17044 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
17045 (ix86_handle_option): Handle -mwaitpkg.
17046 * config.gcc: New header.
17047 * config/i386/cpuid.h (bit_WAITPKG): New bit.
17048 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
17049 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
17051 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17052 OPTION_MASK_ISA_WAITPKG.
17053 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
17054 (ix86_option_override_internal): Add PTA_WAITPKG.
17055 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
17056 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
17057 IX86_BUILTIN_TPAUSE.
17058 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
17059 __builtin_ia32_umwait and __builtin_ia32_tpause.
17060 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
17061 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
17062 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
17063 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
17064 UNSPECV_TPAUSE): New.
17065 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
17066 * config/i386/i386.opt: Add -mwaitpkg.
17067 * config/i386/waitpkgintrin.h: New file.
17068 * config/i386/x86intrin.h: New header.
17069 * doc/invoke.texi: Add -mwaitpkg.
17071 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
17074 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
17076 (cortex-m0): Use armv6s-m isa.
17077 (cortex-m0plus): Likewise.
17078 (cortex-m1): Likewise.
17079 (cortex-m0.small-multiply): Likewise.
17080 (cortex-m0plus.small-multiply): Likewise.
17081 (cortex-m1.small-multiply): Likewise.
17083 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
17084 Jakub Jelinek <jakub@redhat.com>
17086 PR tree-optimization/85692
17087 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
17088 source permute as well.
17090 2018-05-11 Martin Liska <mliska@suse.cz>
17093 * doc/extend.texi: Document LLVM style format for no_sanitize
17096 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
17098 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
17099 mode_supports_vsx_dform_quad to mode_supports_dq_form.
17100 (mode_supports_vsx_dform_quad): Likewise.
17101 (mode_supports_vmx_dform): Move these functions to be next to the
17102 other mode_supports functions.
17103 (mode_supports_dq_form): Likewise.
17104 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
17105 mode_supports_dq_form.
17106 (reg_offset_addressing_ok_p): Likewise.
17107 (offsettable_ok_by_alignment): Likewise.
17108 (rs6000_legitimate_offset_address_p): Likewise.
17109 (legitimate_lo_sum_address_p): Likewise.
17110 (rs6000_legitimize_address): Likewise.
17111 (rs6000_legitimize_reload_address): Likewise.
17112 (rs6000_secondary_reload_inner): Likewise.
17113 (rs6000_preferred_reload_class): Likewise.
17114 (rs6000_output_move_128bit): Likewise.
17116 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
17118 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
17119 Generate SImode target register for null target.
17120 <case IX86_BUILTIN_XGETBV>: Ditto.
17121 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
17122 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
17124 2018-05-10 Carl Love <cel@us.ibm.com>
17126 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
17127 dcbtt and dcbtstt if operands[2] is 0.
17129 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
17132 * config/i386/sse.md (usadv64qi): New expander.
17134 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
17136 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
17137 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
17138 -maltivec=be support.
17139 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
17140 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
17141 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17142 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
17143 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
17144 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
17145 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
17146 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
17147 altivec_vsumsws): Adjust.
17148 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
17149 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
17150 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
17152 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
17153 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
17154 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
17155 (altivec_lve<VI_char>x): Delete expand.
17156 (*altivec_lve<VI_char>x_internal): Rename to...
17157 (altivec_lve<VI_char>x): ... this.
17158 (altivec_lvxl_<mode>): Delete expand.
17159 (*altivec_lvxl_<mode>_internal): Rename to ...
17160 (altivec_lvxl_<mode>): ... this.
17161 (altivec_stvxl_<mode>): Delete expand.
17162 (*altivec_stvxl_<mode>_internal): Rename to ...
17163 (altivec_stvxl_<mode>): ... this.
17164 (altivec_stve<VI_char>x): Delete expand.
17165 (*altivec_stve<VI_char>x_internal): Rename to ...
17166 (altivec_stve<VI_char>x): ... this.
17167 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
17168 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
17169 reduc_plus_scal_<mode>): Adjust.
17170 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
17172 (rs6000_cpu_cpp_builtins): Adjust.
17173 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
17174 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
17175 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
17176 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
17177 -maltivec=be support.
17178 (rs6000_split_vec_extract_var): Adjust.
17179 (rs6000_split_v4si_init): Adjust.
17180 (swap_selector_for_mode): Delete.
17181 (altivec_expand_lvx_be, altivec_expand_stvx_be,
17182 altivec_expand_stvex_be): Delete.
17183 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
17184 -maltivec=be support.
17185 (rs6000_gimple_fold_builtin): Ditto.
17186 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
17188 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
17189 (TARGET_DIRECT_MOVE_64BIT): Adjust.
17190 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
17191 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
17192 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
17193 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
17194 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
17195 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
17196 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
17197 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
17198 anonymous split): Adjust.
17199 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
17200 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
17202 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
17204 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
17205 when --with-gxx-include-dir is also specified.
17206 * configure: Regenerate.
17208 2018-05-09 Jim Wilson <jimw@sifive.com>
17211 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
17212 * config/riscv/t-withmultilib: New.
17213 * config/riscv/withmultilib.h: New.
17214 * doc/install.texi: Document RISC-V --with-multilib-list support.
17216 2018-05-09 Richard Biener <rguenther@suse.de>
17218 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
17220 (vect_bb_vectorization_profitable_p): Adjust. Compute
17221 actual scalar cost using the cost vector and the add_stmt_cost
17224 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17226 PR rtl-optimization/85645
17227 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
17228 in the REG_CFA_REGISTER note for LR, don't leave it empty.
17230 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17232 PR rtl-optimization/85645
17233 * shrink-wrap.c (spread_components): Return a boolean saying if
17234 anything was changed.
17235 (try_shrink_wrapping_separate): Iterate spread_components until
17236 nothing changes anymore.
17238 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17240 PR rtl-optimization/85645
17241 * regrename.c (build_def_use): Also kill the chains that include the
17242 destination of a REG_CFA_REGISTER note.
17244 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17246 PR rtl-optimization/85645
17247 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
17248 insn that has a REG_CFA_REGISTER note.
17250 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
17252 * cfgexpand.c (expand_clobber): New function.
17253 (expand_gimple_stmt_1): Use it.
17254 * tree-vect-stmts.c (vect_clobber_variable): New function,
17256 (vectorizable_simd_clone_call): ...here.
17257 (vectorizable_store): Emit a clobber either side of an
17258 IFN_STORE_LANES sequence.
17259 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
17261 2018-05-09 Tom de Vries <tom@codesourcery.com>
17264 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
17265 (define_insn "trap_if_false"): Add exit after trap.
17267 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
17269 PR rtl-optimization/85638
17270 * bb-reorder.c: Include common/common-target.h.
17271 (create_forwarder_block): New function extracted from...
17272 (fix_up_crossing_landing_pad): ...here. Rename into...
17273 (dw2_fix_up_crossing_landing_pad): ...this.
17274 (sjlj_fix_up_crossing_landing_pad): New function.
17275 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
17276 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
17277 from both partitions and exit the loop after one iteration.
17279 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17282 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17284 (Basic PowerPC Built-in Functions): The new name of the
17285 subsection previously known as "PowerPC Built-in Functions".
17286 (Basic PowerPC Built-in Functions Available on all Configurations):
17288 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
17290 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
17292 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
17294 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
17297 2018-05-08 Jim Wilson <jimw@sifive.com>
17299 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
17300 (LD_EMUL_SUFFIX): New.
17301 (LINK_SPEC): Use it.
17303 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17305 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17307 (Basic PowerPC Built-in Functions): The new name of the
17308 subsection previously known as "PowerPC Built-in Functions".
17309 (Basic PowerPC Built-in Functions Available on all Configurations):
17311 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
17313 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
17315 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
17317 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
17320 2018-05-08 Jakub Jelinek <jakub@redhat.com>
17323 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
17324 after cmpelim optimization.
17326 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
17328 * config.gcc: Support "goldmont".
17329 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
17330 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17331 PROCESSOR_GOLDMONT.
17332 * config/i386/i386.c (m_GOLDMONT): Define.
17333 (processor_target_table): Add "goldmont".
17334 (PTA_GOLDMONT): Define.
17335 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
17336 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
17337 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
17338 (fold_builtin_cpu): Add "goldmont".
17339 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
17340 (ix86_option_override_internal): Add "goldmont".
17341 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
17342 (processor_type): Add PROCESSOR_GOLDMONT.
17343 * config/i386/i386.md: Add CPU "glm".
17344 * config/i386/glm.md: New file.
17345 * config/i386/x86-tune.def: Add m_GOLDMONT.
17346 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
17348 2018-05-08 Jakub Jelinek <jakub@redhat.com>
17351 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
17353 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
17354 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
17355 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
17358 * config/i386/i386.c (ix86_fold_builtin): Handle
17359 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
17362 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
17363 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
17365 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
17368 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
17369 (check_arch): Likewise.
17370 (check_fpu): Return the result rather than printing it.
17371 (end arch): Fix operator precedence.
17372 (end cpu): Likewise.
17373 (END): Print the result from check_fpu.
17375 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
17376 Alan Hayward <alan.hayward@arm.com>
17377 David Sherwood <david.sherwood@arm.com>
17379 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
17380 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
17381 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
17382 (*fcmuo<mode>_and): New patterns.
17384 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
17386 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
17387 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
17388 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
17389 (cmp_op, sve_imm_con): New code attributes.
17390 (SVE_COND_INT_CMP, imm_con): Delete.
17391 (cmp_op): Remove above unspecs from int attribute.
17392 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
17394 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
17395 comparison-specific unspecs.
17396 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
17397 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
17398 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
17399 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
17400 (*vec_fcm<cmp_op><mode>): Rename to...
17401 (*fcm<cmp_op><mode>): ...this and adjust likewise.
17402 (*vec_fcmuo<mode>): Rename to...
17403 (*fcmuo<mode>): ...this and adjust likewise.
17404 (*pred_fcm<cmp_op><mode>): New pattern.
17405 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
17406 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
17408 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
17410 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
17411 (aarch64_emit_sve_predicated_cond): New function.
17412 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
17413 (aarch64_emit_unspec_cond_or): Replace with...
17414 (aarch64_emit_sve_or_conds): ...this new function. Use
17415 aarch64_emit_sve_ptrue_op for the individual comparisons and
17416 aarch64_emit_binop to OR them together.
17417 (aarch64_emit_inverted_unspec_cond): Replace with...
17418 (aarch64_emit_sve_inverted_cond): ...this new function. Use
17419 aarch64_emit_sve_ptrue_op for the comparison and
17420 aarch64_emit_unop to invert the result.
17421 (aarch64_expand_sve_vec_cmp_float): Update after the above
17422 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
17424 2018-05-07 Nathan Sidwell <nathan@acm.org>
17426 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
17427 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
17428 (Backwards Compatibility): Likewise.
17430 2018-05-07 Luis Machado <luis.machado@linaro.org>
17434 2018-05-07 Luis Machado <luis.machado@linaro.org>
17436 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17437 <prefetch_dynamic_strides>: New const bool field.
17438 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17439 prefetch_dynamic_strides.
17440 (exynosm1_prefetch_tune): Likewise.
17441 (thunderxt88_prefetch_tune): Likewise.
17442 (thunderx_prefetch_tune): Likewise.
17443 (thunderx2t99_prefetch_tune): Likewise.
17444 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
17446 (aarch64_override_options_internal): Update to set
17447 PARAM_PREFETCH_DYNAMIC_STRIDES.
17448 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
17449 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
17450 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
17451 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
17452 prefetch-dynamic-strides setting.
17454 2018-05-07 Luis Machado <luis.machado@linaro.org>
17456 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17457 <minimum_stride>: New const int field.
17458 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17459 minimum_stride field.
17460 (exynosm1_prefetch_tune): Likewise.
17461 (thunderxt88_prefetch_tune): Likewise.
17462 (thunderx_prefetch_tune): Likewise.
17463 (thunderx2t99_prefetch_tune): Likewise.
17464 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
17465 (aarch64_override_options_internal): Update to set
17466 PARAM_PREFETCH_MINIMUM_STRIDE.
17467 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
17468 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
17469 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
17470 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
17471 stride is constant and is below the minimum stride threshold.
17473 2018-05-07 Luis Machado <luis.machado@linaro.org>
17475 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
17478 2018-05-07 Luis Machado <luis.machado@linaro.org>
17480 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17481 <prefetch_dynamic_strides>: New const bool field.
17482 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17483 prefetch_dynamic_strides.
17484 (exynosm1_prefetch_tune): Likewise.
17485 (thunderxt88_prefetch_tune): Likewise.
17486 (thunderx_prefetch_tune): Likewise.
17487 (thunderx2t99_prefetch_tune): Likewise.
17488 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
17490 (aarch64_override_options_internal): Update to set
17491 PARAM_PREFETCH_DYNAMIC_STRIDES.
17492 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
17493 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
17494 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
17495 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
17496 prefetch-dynamic-strides setting.
17498 2018-05-07 Luis Machado <luis.machado@linaro.org>
17500 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17501 <minimum_stride>: New const int field.
17502 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17503 minimum_stride field.
17504 (exynosm1_prefetch_tune): Likewise.
17505 (thunderxt88_prefetch_tune): Likewise.
17506 (thunderx_prefetch_tune): Likewise.
17507 (thunderx2t99_prefetch_tune): Likewise.
17508 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
17509 (aarch64_override_options_internal): Update to set
17510 PARAM_PREFETCH_MINIMUM_STRIDE.
17511 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
17512 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
17513 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
17514 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
17515 stride is constant and is below the minimum stride threshold.
17517 2018-05-06 Jakub Jelinek <jakub@redhat.com>
17520 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
17521 the type is addressable. Don't force op into register if it has
17524 2018-05-05 Roland McGrath <mcgrathr@google.com>
17527 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
17528 any section for which we don't know a specific type it should have,
17529 regardless of name. Previously this was done only for the exact
17530 names ".init_array", ".fini_array", and ".preinit_array".
17531 (default_elf_asm_named_section): Add comment about
17532 relationship with default_section_type_flags and SECTION_NOTYPE.
17533 (get_section): Don't consider it a type conflict if one side has
17534 SECTION_NOTYPE and the other doesn't, as long as neither has the
17535 SECTION_BSS et al used in the default_section_type_flags logic.
17537 2018-05-05 Tom de Vries <tom@codesourcery.com>
17540 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
17541 (workaround_barsyncs): New function.
17542 (nvptx_reorg): Use workaround_barsyncs.
17543 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
17544 (define_expand "nvptx_membar_cta"): New define_expand.
17545 (define_insn "*nvptx_membar_cta"): New insn.
17547 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17549 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
17550 To improve optimization opportunities.
17551 * builtin-types.def: The new needed builtin types for the above.
17553 2018-05-04 Richard Biener <rguenther@suse.de>
17555 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
17556 * gimple-ssa-store-merging.c
17557 (imm_store_chain_info::output_merged_store): Remove redundant create,
17558 release split_store vector contents on failure.
17559 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
17560 scalar stmt vector on cache hit.
17562 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
17564 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
17566 * config.gcc (powerpc-xilinx-eabi*): Remove.
17567 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
17569 (fusion_addis_mem_combo_load): Ditto.
17570 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
17572 (rs6000_cpu_cpp_builtins): Ditto.
17573 * config/rs6000/rs6000-linux.c
17574 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
17575 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
17576 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
17578 (rs6000_setup_reg_addr_masks): Ditto.
17579 (rs6000_init_hard_regno_mode_ok): Ditto.
17580 (rs6000_option_override_internal): Ditto.
17581 (legitimate_lo_sum_address_p): Ditto.
17582 (rs6000_legitimize_address): Ditto.
17583 (rs6000_legitimize_reload_address): Ditto.
17584 (rs6000_legitimate_address_p): Ditto.
17585 (abi_v4_pass_in_fpr): Ditto.
17586 (setup_incoming_varargs): Ditto.
17587 (rs6000_gimplify_va_arg): Ditto.
17588 (rs6000_split_multireg_move): Ditto.
17589 (rs6000_savres_strategy): Ditto.
17590 (rs6000_emit_prologue_components): Ditto.
17591 (rs6000_emit_epilogue_components): Ditto.
17592 (rs6000_emit_prologue): Ditto.
17593 (rs6000_emit_epilogue): Ditto.
17594 (rs6000_elf_file_end): Ditto.
17595 (rs6000_function_value): Ditto.
17596 (rs6000_libcall_value): Ditto.
17597 * config/rs6000/rs6000.h: Ditto.
17598 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
17599 (TARGET_MINMAX): ... this. New.
17600 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
17601 * config/rs6000/rs6000.md: Remove Xilinx FP support.
17602 (*movsi_internal1_single): Delete.
17603 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
17604 mfpu=, mxilinx-fpu): Delete.
17605 * config/rs6000/singlefp.h: Delete.
17606 * config/rs6000/sysv4.h: Remove Xilinx FP support.
17607 * config/rs6000/t-rs6000: Ditto.
17608 * config/rs6000/t-xilinx: Delete.
17609 * config/rs6000/titan.md: Adjust for fp_type removal.
17610 * config/rs6000/vsx.md: Remove Xilinx FP support.
17611 (VStype_simple): Delete.
17612 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
17613 * config/rs6000/xfpu.h: Delete.
17614 * config/rs6000/xfpu.md: Delete.
17615 * config/rs6000/xilinx.h: Delete.
17616 * config/rs6000/xilinx.opt: Delete.
17617 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
17618 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
17620 2018-05-04 Tom de Vries <tom@codesourcery.com>
17623 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
17626 2018-05-04 Richard Biener <rguenther@suse.de>
17628 PR middle-end/85627
17629 * tree-complex.c (update_complex_assignment): We are always in SSA form.
17630 (expand_complex_div_wide): Likewise.
17631 (expand_complex_operations_1): Likewise.
17632 (expand_complex_libcall): Preserve EH info of the original stmt.
17633 (tree_lower_complex): Handle removed blocks.
17634 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
17635 on complex multiplication and division libcall builtins.
17637 2018-05-04 Richard Biener <rguenther@suse.de>
17639 PR middle-end/85574
17640 * fold-const.c (negate_expr_p): Restrict negation of operand
17641 zero of a division to when we know that can happen without
17643 (fold_negate_expr_1): Likewise.
17645 2018-05-04 Jakub Jelinek <jakub@redhat.com>
17648 * real.h (real_nextafter): Declare.
17649 * real.c (real_nextafter): New function.
17650 * fold-const-call.c (fold_const_nextafter): New function.
17651 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
17652 CASE_CFN_NEXTTOWARD.
17653 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
17654 even when arg1_mode is different from arg0_mode.
17656 2018-05-03 Nathan Sidwell <nathan@acm.org>
17658 * doc/extend.texi (Deprecated Features): Remove
17659 -ffriend-injection.
17660 (Backwards Compatibility): Likewise.
17661 * doc/invoke.texi (C++ Language Options): Likewise.
17662 (C++ Dialect Options): Likewise.
17664 2018-05-03 Jakub Jelinek <jakub@redhat.com>
17667 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
17668 _mm512_mask_mullox_epi64): New intrinsics.
17670 2018-05-03 Tom de Vries <tom@codesourcery.com>
17673 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
17674 dump files): Add offload-tree.
17676 2018-05-03 Richard Biener <rguenther@suse.de>
17678 PR tree-optimization/85615
17679 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
17680 to loops not nested in BBs loop father to avoid creating multi-entry
17683 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17685 PR tree-optimization/70291
17686 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
17687 arguments. Change return type to tree. Emit libcall as a new
17688 statement rather than replacing existing one when inplace_p is true.
17689 (expand_complex_multiplication_components): New function.
17690 (expand_complex_multiplication): Expand floating-point complex
17691 multiplication using the above.
17692 (expand_complex_division): Rename inner_type parameter to type.
17693 Update expand_complex_libcall call-site.
17694 (expand_complex_operations_1): Update expand_complex_multiplication
17695 and expand_complex_division call-sites.
17697 2018-05-02 Jakub Jelinek <jakub@redhat.com>
17700 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
17701 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
17702 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
17703 the highest significant bit of the shift count mask is clear. In
17704 check whether and[sq]i3 is needed verify that all significant bits
17705 of the shift count other than the highest are set.
17707 2018-05-02 Tom de Vries <tom@codesourcery.com>
17710 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
17711 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
17712 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
17713 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
17714 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
17715 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
17716 __builtin_goacc_parlevel_size.
17718 2018-05-02 Richard Biener <rguenther@suse.de>
17720 PR tree-optimization/85597
17721 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
17722 do not use split vect_get_vec_defs call but call vect_get_slp_defs
17725 2018-05-02 Tom de Vries <tom@codesourcery.com>
17728 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
17729 dump files): Add ltrans-tree.
17731 2018-05-02 Tom de Vries <tom@codesourcery.com>
17734 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
17735 dump files): Add wpa-ipa.
17737 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
17739 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
17740 powerpc*-*-linux*paired* target.
17741 * config/rs6000/750cl.h: Delete.
17742 * config/rs6000/paired.h: Delete.
17743 * config/rs6000/paired.md: Delete.
17744 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
17746 * config/rs6000/rs6000-builtin.def: Remove paired float support.
17747 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
17748 comment. Remove paired float support.
17749 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
17750 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
17752 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
17753 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
17755 * config/rs6000/rs6000.c: Remove paired float support.
17756 (paired_expand_vector_init, paired_expand_vector_move,
17757 paired_emit_vector_compare, paired_emit_vector_cond_expr,
17758 (paired_expand_lv_builtin, paired_expand_stv_builtin,
17759 paired_expand_builtin, paired_expand_predicate_builtin,
17760 paired_init_builtins): Delete.
17761 * config/rs6000/rs6000.h: Remove paired float support.
17762 * config/rs6000/rs6000.md: Remove paired float support.
17763 (move_from_CR_ov_bit): Delete.
17764 * config/rs6000/rs6000.opt (mpaired): Delete.
17765 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
17766 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
17768 2018-05-02 Richard Biener <rguenther@suse.de>
17770 PR middle-end/85567
17771 * gimplify.c (gimplify_save_expr): When in SSA form allow
17772 SAVE_EXPRs to compute to SSA vars.
17774 2018-05-02 Jakub Jelinek <jakub@redhat.com>
17777 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
17778 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
17779 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
17780 clobber operands[2], instead use a new pseudo. Formatting fixes.
17782 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
17784 PR tree-optimization/85586
17785 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
17786 exit early for statements in the same group if the accesses are
17789 2018-05-02 Tom de Vries <tom@codesourcery.com>
17792 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
17795 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
17797 PR tree-optimization/85143
17798 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
17800 2018-05-01 Tom de Vries <tom@codesourcery.com>
17803 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
17804 not found" error message.
17806 2018-05-01 Tom de Vries <tom@codesourcery.com>
17809 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
17810 * vec.c (test_ordered_remove_if): New function.
17811 (vec_c_tests): Call test_ordered_remove_if.
17812 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
17813 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
17814 * tree-vect-patterns.c (vect_pattern_recog_1): Use
17815 VEC_ORDERED_REMOVE_IF.
17817 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17819 PR tree-optimization/82665
17820 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
17821 pointer subtraction where arguments come from a memchr call.
17823 2018-05-01 Jakub Jelinek <jakub@redhat.com>
17825 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
17826 --push-state --as-needed and --pop-state instead of --as-needed and
17827 --no-as-needed if ld supports it.
17828 * configure: Regenerated.
17831 * doc/install.texi2html: Replace _002d with - and _002a with * in
17832 generated html files using sed.
17834 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17837 * gcc-rich-location.c (blank_line_before_p): New function.
17838 (use_new_line): New function.
17839 (gcc_rich_location::add_fixit_insert_formatted): New function.
17840 * gcc-rich-location.h
17841 (gcc_rich_location::add_fixit_insert_formatted): New function.
17843 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17845 * selftest.c (assert_streq): Rename "expected" and "actual" to
17846 "val1" and "val2". Extend NULL-handling to cover both inputs
17847 symmetrically, while still requiring both to be non-NULL for a pass.
17848 * selftest.h (assert_streq): Rename "expected" and "actual" to
17850 (ASSERT_EQ): Likewise.
17851 (ASSERT_EQ_AT): Likewise.
17852 (ASSERT_KNOWN_EQ): Likewise.
17853 (ASSERT_KNOWN_EQ_AT): Likewise.
17854 (ASSERT_NE): Likewise.
17855 (ASSERT_MAYBE_NE): Likewise.
17856 (ASSERT_MAYBE_NE_AT): Likewise.
17857 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
17858 the assertion to pass.
17859 (ASSERT_STREQ_AT): Likewise.
17861 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
17863 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
17864 interaction with -pie.
17866 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17868 * selftest.h: Fix alphabetization of per-source-file selftest
17871 2018-04-30 Jason Merrill <jason@redhat.com>
17873 PR c++/61982 - dead stores to destroyed objects.
17874 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
17877 2018-04-30 Jason Merrill <jason@redhat.com>
17879 * tree.c (build_clobber): New.
17880 * tree.h: Declare it.
17881 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
17883 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17885 * diagnostic-show-locus.c (layout::layout): Update for
17886 location_get_source_line returning a char_span.
17887 (struct char_span): Move to input.h.
17888 (struct correction): Update for fields in char_span becoming
17890 (struct source_line): Update for location_get_source_line
17891 returning a char_span.
17892 (layout::print_line): Likewise.
17893 * edit-context.c (edited_file::print_content): Likewise.
17894 (edited_file::print_diff_hunk): Likewise.
17895 (edited_file::print_run_of_changed_lines): Likewise.
17896 (edited_file::get_num_lines): Likewise.
17897 (edited_line::edited_line): Likewise.
17898 * final.c (asm_show_source): Likewise.
17899 * input.c (location_get_source_line): Convert return type
17900 from const char * to char_span, losing the final "line_len"
17902 (dump_location_info): Update for the above.
17903 (get_substring_ranges_for_loc): Likewise. Use a char_span
17904 when handling the literal within the line.
17905 (test_reading_source_line): Update for location_get_source_line
17906 returning a char_span.
17907 * input.h (class char_span): Move here from
17908 diagnostic-show-locus.c, converting from a struct to a class.
17909 Make data members private.
17910 (char_span::operator bool): New.
17911 (char_span::length): New.
17912 (char_span::get_buffer): New.
17913 (char_span::operator[]): New.
17914 (char_span::subspan): Make const.
17915 (char_span::xstrdup): New.
17916 (location_get_source_line): Convert return type from const char *
17917 to char_span, losing the final "line_size" param.
17919 2018-04-30 Jan Hubicka <jh@suse.cz>
17921 * lto-wrapper.c (ltrans_priorities): New static var.
17922 (cmp_priority): New.
17923 (run_gcc): Read priorities and if doing parallel build order
17924 the Makefile by them.
17926 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17928 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
17930 2018-04-30 Richard Biener <rguenther@suse.de>
17932 * tree-cfg.c (verify_address): Remove base argument, add
17933 flag whether to check TREE_ADDRESSABLE and do that.
17934 (verify_expr): Remove.
17935 (verify_types_in_gimple_reference): Add pieces from verify_expr.
17936 (verify_gimple_assign_single): Likewise.
17937 (verify_gimple_switch): Likewise.
17938 (verify_expr_location_1): Dereference tp once. Add (disabled)
17939 piece from verify_expr.
17940 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
17942 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
17944 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
17946 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
17948 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
17949 (small_data_pattern): Likewise.
17950 (arc_rewrite_small_data): Likewise.
17951 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
17952 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
17953 (get_symbol_alignment): New function.
17954 (legitimate_small_data_address_p): Likewise.
17955 (legitimate_scaled_address): Update, call
17956 legitimate_small_data_address_p.
17957 (output_sdata): New static variable.
17958 (arc_print_operand): Update how we handle small data operands.
17959 (arc_print_operand_address): Likewise.
17960 (arc_legitimate_address_p): Update, use
17961 legitimate_small_data_address_p.
17962 (arc_rewrite_small_data_p): Remove.
17963 (arc_rewrite_small_data_1): Likewise.
17964 (arc_rewrite_small_data): Likewise.
17965 (small_data_pattern): Likewise.
17966 (compact_sda_memory_operand): Update to use
17967 legitimate_small_data_address_p and get_symbol_alignment.
17968 (prepare_move_operands): Don't rewite sdata pattern.
17969 (prepare_extend_operands): Remove.
17970 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
17972 (zero_extendqisi2): Likewise.
17973 (zero_extendhisi2): Likewise.
17974 (extendqihi2): Likewise.
17975 (extendqisi2): Likewise.
17976 (extendhisi2): Likewise.
17977 (addsi3): Likewise.
17978 (subsi3): Likewise.
17979 (andsi3): Likewise.
17980 * config/arc/constraints.md (Usd): Change it to memory constraint.
17982 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
17984 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
17985 as source of std instructions.
17986 * config/arc/arc.md (movsi_insn): Update pattern predicate to
17987 allow 6-bit constants as source for store instructions.
17988 (movdi_insn): Update instruction pattern to allow 6-bit constants
17989 as source for store instructions.
17991 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
17993 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
17995 2018-04-30 Nathan Sidwell <nathan@acm.org>
17996 Sandra Loosemore <sandra@codesourcery.com>
17998 * dumpfile.c (dump_open): Allow '-' for stdout.
17999 * doc/invoke.texi (Developer Options): Document dump filename
18000 determination early. Document stdin/stdout selection.
18002 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
18004 Microblaze Target: PIC data text relative
18006 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
18007 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
18009 * config/microblaze/microblaze.h (microblaze_constant_address_p):
18010 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
18011 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
18012 New addressing mode for data-text relative position indepenedent code.
18013 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
18014 'ADDRESS_SYMBOLIC_TXT_REL'.
18015 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
18016 (microblaze_legitimate_pic_operand): Exclude function calls from
18017 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
18018 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
18020 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
18021 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
18022 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
18023 for 'address + offset'.
18024 (microblaze_expand_prologue): Add new function prologue call for
18026 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
18027 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
18028 table in case of TARGET_PIC_DATA_TEXT_REL.
18029 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
18030 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
18031 Add new macros 'UNSPEC_TEXT',
18032 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
18033 + exclude function calls from 'UNSPEC_PLT' in case of data text
18035 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
18036 new target hook for generating address diff vector tables in case of
18038 * doc/tm.texi : Regenerate.
18039 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
18040 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
18041 of addr diff vector generation.
18042 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
18043 target hook definition.
18044 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
18045 Add default function for generate_pic_addr_diff_vec -> flag_pic.
18046 * doc/invoke.texi (Add new pic option): Add new microblaze pic
18047 option for data text relative.
18049 2018-04-30 Richard Biener <rguenther@suse.de>
18051 * tree-chrec.h (evolution_function_is_constant_p): Remove
18053 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
18055 2018-04-30 Richard Biener <rguenther@suse.de>
18058 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
18060 2018-04-30 Richard Biener <rguenther@suse.de>
18062 PR tree-optimization/28364
18063 PR tree-optimization/85275
18064 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
18065 copying first exit test.
18067 2018-04-28 Mark Wielaard <mark@klomp.org>
18069 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
18070 dwarf_version >= 5.
18071 (dwarf_AT): Handle DW_AT_addr_base.
18072 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
18074 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
18077 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
18078 (*ashl<dwi>3_doubleword_mask_1): Ditto.
18079 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
18080 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
18082 2018-04-28 Richard Biener <rguenther@suse.de>
18084 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
18085 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
18086 to reflect use. Only add interesting stmts.
18088 2018-04-27 Martin Jambor <mjambor@suse.cz>
18091 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
18092 the jump function allows for passing through aggregate values.
18094 2018-04-27 David Malcolm <dmalcolm@redhat.com>
18096 * input.h (in_system_header_at): Convert from macro to inline
18098 (from_macro_expansion_at): Likewise.
18099 (from_macro_definition_at): Likewise.
18101 2018-04-27 Jeff Law <law@redhat.com>
18103 * config.gcc: Mark tile* targets as deprecated/obsolete.
18105 2018-04-27 Richard Biener <rguenther@suse.de>
18107 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
18110 2018-04-27 Richard Biener <rguenther@suse.de>
18112 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
18114 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
18116 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
18117 with Yd constraint. Set "preferred_for_speed" attribute from
18118 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
18119 with Yd constraint.
18120 (*movdi_internal): Ditto.
18121 (movti_interunit splitters): Remove
18122 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
18123 (movdi_interunit splitters): Ditto.
18124 * config/i386/constraints.md (Ye): Remove.
18125 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
18127 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18130 * config/aarch64/constraints.md (Usg): Limit to 31.
18131 (Usj): Limit to 63.
18133 2018-04-27 Jakub Jelinek <jakub@redhat.com>
18135 PR tree-optimization/85529
18136 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
18137 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
18138 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
18139 zero extension or masking of the MSB bit.
18140 (optimize_range_tests): Add FIRST_BB argument, pass it through
18141 to optimize_range_tests_var_bound.
18142 (maybe_optimize_range_tests, reassociate_bb): Adjust
18143 optimize_range_tests callers.
18145 2018-04-26 Richard Biener <rguenther@suse.de>
18146 Jakub Jelinek <jakub@redhat.com>
18148 * cgraph.h (symbol_table): Just declare debug method here.
18149 * symtab.c (symbol_table::debug): Define.
18151 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
18153 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
18155 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
18157 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
18158 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
18159 (*movdi_internal): Substitute Yi and Yj constraint with x
18160 and Ym and Yn constraint with y constraint. Update "isa"
18161 attribute and set "preferred_for_speed" attribute from
18162 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
18163 (*movsi_internal): Ditto.
18164 (*movdf_internal): Ditto.
18165 (*movsf_internal): Ditto.
18166 (*zero_extendsidi2): Ditto.
18167 * config/i386/sse.md (vec_set<mode>_0): Ditto.
18168 (sse2_loadld): Ditto.
18169 (*vec_extract<ssevecmodelower>_0): Ditto.
18170 (*vec_extractv4si_0_zext_sse4): Ditto.
18171 (vec_concatv2di): Ditto.
18172 (*vec_dup<mode>): Ditto.
18173 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
18174 * config/i386/constraints.md (Yi): Remove.
18179 2018-04-26 Nathan Sidwell <nathan@acm.org>
18181 * dumpfile.c (dump_open): New.
18182 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
18183 (dump_finish): Detect stdio/stderr by value not name.
18185 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
18187 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
18189 2018-04-26 Tom de Vries <tom@codesourcery.com>
18192 * config/nvptx/nvptx.c (verify_neutering_jumps)
18193 (verify_neutering_labels): New function
18194 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
18196 2018-04-26 Tom de Vries <tom@codesourcery.com>
18199 * config/nvptx/nvptx.c (needs_neutering_p): New function.
18200 (nvptx_single): Use needs_neutering_p to skip over insns that do not
18203 2018-04-26 Richard Biener <rguenther@suse.de>
18204 Tom de Vries <tom@codesourcery.com>
18207 * lto-streamer-out.c (output_function): Fixup loops if required to match
18208 discovery done in the reader.
18210 2018-04-26 Richard Biener <rguenther@suse.de>
18212 PR tree-optimization/85116
18213 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
18214 have a loop exit from the single latch predecessor. Remove
18215 case of header with just condition.
18216 (ch_base::copy_headers): Exclude infinite loops from any
18218 (pass_ch::execute): Record exits.
18220 2018-04-26 Richard Biener <rguenther@suse.de>
18222 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
18223 prologue cost vector and pass it to vect_get_load_cost.
18224 (vect_get_peeling_costs_all_drs): Likewise.
18225 (vect_peeling_hash_get_lowest_cost): Likewise.
18226 (vect_enhance_data_refs_alignment): Likewise.
18228 2018-04-26 Richard Biener <rguenther@suse.de>
18230 PR middle-end/85450
18231 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
18232 checking of integer<->pointer conversions.
18233 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
18234 sign-/zero-extending pointer types.
18235 (expand_omp_for_static_chunk): Likewise.
18237 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
18238 Jean Lee <xiaoyur347@gmail.com>
18240 * config/mips/mips.c (mips_asan_shadow_offset): New function.
18241 (TARGET_ASAN_SHADOW_OFFSET): Define.
18242 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
18243 true for -fsanitize=address.
18245 2018-04-25 Mark Wielaard <mark@klomp.org>
18247 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
18250 2018-04-25 Jakub Jelinek <jakub@redhat.com>
18252 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
18253 than "alu", remove explicit "memory" and "imm_disp" attributes.
18254 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
18256 PR middle-end/85414
18257 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
18258 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
18259 gen_lowpart_no_emit.
18261 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
18264 * config/i386/i386.c (ix86_expand_builtin): Change memory
18265 operand to XI, extend p0 to Pmode.
18266 * config/i386/i386.md: Change unspec volatile and operand
18267 1 mode to XI, change operand 0 mode to P.
18269 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18271 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
18272 GET_MODE_MASK before any checking.
18273 (nds32_can_use_bset_p): Likewise.
18274 (nds32_can_use_btgl_p): Likewise.
18276 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18278 * config/nds32/nds32-doubleword.md: New define_split pattern for
18279 illegal register number.
18281 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18283 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
18285 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18287 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
18289 2018-04-25 Richard Biener <rguenther@suse.de>
18291 * lto-streamer.h (LTO_major_version): Bump to 8.
18293 2018-04-25 Jakub Jelinek <jakub@redhat.com>
18295 * BASE-VER: Set to 9.0.0.
18297 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
18299 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
18300 in __abskf2 and __powikf2.
18302 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18305 * config/aarch64/constraints.md (Usg, Usj): New constraints.
18306 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
18307 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
18308 Use the above on operand 2. Reindent.
18309 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
18311 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
18314 * common/config/i386/i386-common.c (ix86_handle_option): Don't
18316 * config/i386/i386.opt (mcet): Removed.
18317 * doc/install.texi: Remove -mcet documentation.
18318 * doc/invoke.texi: Likewise.
18320 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
18323 * doc/install.texi: Remove -mcet from bootstrap-cet.
18325 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18328 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
18329 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
18333 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
18334 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
18335 containing a CONST_VECTOR.
18337 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
18339 * doc/install.texi: Update newlib dependency for nvptx.
18341 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18344 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
18345 instead of INTVAL when shifting x left.
18347 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
18349 PR tree-optimization/85478
18350 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
18351 vect_grouped_store_supported for single element vectors.
18353 2018-04-24 Richard Biener <rguenther@suse.de>
18356 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
18357 load cost increase to the case of non-constant step.
18359 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18362 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
18363 destination if any_malformed_asm.
18365 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
18367 PR middle-end/85496
18368 * expr.c (store_field): In the bitfield case, if the value comes from
18369 a function call and is returned in registers by means of a PARALLEL,
18370 do not change the mode of the temporary unless BLKmode and VOIDmode.
18372 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
18374 PR rtl-optimization/85423
18375 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
18376 dependencies to debug insns when the previous insn is non-debug.
18378 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
18380 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
18381 enums into a single definition.
18382 (fls): Fix predicates and printing.
18385 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
18387 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
18388 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
18389 and short u6 immediate.
18390 (check_if_valid_sleep_operand): Remove.
18391 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
18393 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18395 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
18396 flag_always_save_lp condition.
18397 * config/nds32/nds32.opt (malways-save-lp): New option.
18399 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18401 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
18402 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
18403 * config/nds32/nds32.h
18404 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
18405 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
18407 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18409 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
18410 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
18412 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18413 Chung-Ju Wu <jasonwucj@gmail.com>
18415 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
18417 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
18418 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
18420 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18422 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
18424 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18426 * config/nds32/nds32-protos.h (nds32_data_alignment,
18427 nds32_local_alignment): Declare.
18428 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
18429 nds32_local_alignment): New functions.
18430 (TARGET_CONSTANT_ALIGNMENT): Define.
18431 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
18433 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18435 * config/nds32/nds32.c
18436 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
18437 (TARGET_MODES_TIEABLE_P): Likewise.
18439 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18441 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
18442 level Ofast and Og.
18444 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
18445 Chung-Ju Wu <jasonwucj@gmail.com>
18447 * config/nds32/constants.md (unspec_volatile_element): Add enum values
18448 for unaligned access.
18449 * config/nds32/nds32-intrinsic.c: Implementation of expanding
18451 * config/nds32/nds32-intrinsic.md: Likewise.
18452 * config/nds32/nds32_intrinsic.h: Likewise.
18453 * config/nds32/nds32.h (nds32_builtins): Likewise.
18454 * config/nds32/nds32.opt (munaligned-access): New option.
18455 * config/nds32/nds32.c (nds32_asm_file_start): Display
18456 flag_unaligned_access status.
18458 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
18460 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
18461 -mno-relax is present.
18462 * config/riscv/linux.h (LINK_SPEC): Ditto.
18464 2018-04-20 Martin Sebor <msebor@redhat.com>
18467 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
18469 (gimple_fold_builtin_stxcpy_chk): Same.
18470 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
18472 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
18475 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
18476 __powikf2 when long double is IEEE 128-bit.
18478 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
18480 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
18481 step to make sure stack always aligned.
18483 2018-04-20 Carl Love <cel@us.ibm.com>
18486 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
18487 size check for arg0.
18489 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
18490 Tom de Vries <tom@codesourcery.com>
18493 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
18494 Emit insns for calls too.
18495 (nvptx_find_par): Always look for worker-level predecessor insn.
18496 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
18498 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
18499 (nvptx_process_pars): Propagate frames for calls.
18501 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
18504 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
18506 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
18507 (ix86_handle_option): Don't handle OPT_mibt.
18508 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
18510 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
18512 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
18513 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
18514 (ix86_target_macros): Define __CET__ with flag_cf_protection
18515 for -fcf-protection.
18516 * config/i386/i386.c (isa2_opts): Remove -mibt.
18517 * config/i386/i386.h (TARGET_IBT): Removed.
18518 (TARGET_IBT_P): Likewise.
18519 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
18520 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
18521 * config/i386/i386.opt (mcet): Update help message.
18522 (mshstk): Likewise.
18524 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
18525 -mcet as an alias for -mshstk.
18527 2018-04-20 Richard Biener <rguenther@suse.de>
18529 PR middle-end/85475
18530 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
18531 complexity by forcing a single use of the multiply operand.
18533 2018-04-20 Martin Jambor <mjambor@suse.cz>
18536 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
18537 recursion dependency to only apply to non-clones.
18539 2018-04-20 Martin Jambor <mjambor@suse.cz>
18542 * ipa-cp.c (create_specialized_node): Check that clones of
18543 self-recursive edges exist during IPA-CP.
18545 2018-04-19 Toon Moene <toon@moene.org>
18547 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
18550 2018-04-19 Jakub Jelinek <jakub@redhat.com>
18552 PR tree-optimization/85467
18553 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
18554 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
18555 VECTOR_CST element to type.
18557 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
18560 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
18561 * config/i386/i386.md (builtin_setjmp_setup): Removed.
18562 (builtin_longjmp): Likewise.
18563 (save_stack_nonlocal): New pattern.
18564 (restore_stack_nonlocal): Likewise.
18566 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
18569 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
18570 Replace ASM_OUTPUT_LABEL with fprintf.
18572 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
18575 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
18576 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
18577 * config/i386/i386-c.c (ix86_target_macros_internal): Also
18578 define __IBT__ and __SHSTK__ for -fcf-protection.
18579 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
18581 (ix86_trampoline_init): Likewise.
18582 (x86_output_mi_thunk): Likewise.
18583 (ix86_notrack_prefixed_insn_p): Likewise.
18584 (ix86_option_override_internal): Don't disallow -fcf-protection.
18585 * config/i386/i386.md (rdssp<mode>): Also enable for
18587 (incssp<mode>): Likewise.
18588 (nop_endbr): Likewise.
18589 * config/i386/i386.opt (mcet): Change help message to built-in
18592 (mshstk): Likewise.
18593 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
18594 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
18595 enable CET built-in functions.
18597 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
18599 * common/config/i386/i386-common.c
18600 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
18601 OPTION_MASK_ISA_MOVDIRI_UNSET,
18602 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
18603 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
18604 * config.gcc (movdirintrin.h): New header.
18605 * config/i386/cpuid.h (bit_MOVDIRI,
18606 bit_MOVDIR64B): New bits.
18607 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
18609 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
18610 (VOID, PVOID, PCVOID)): New function types.
18611 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
18612 __builtin_ia32_directstoreu_u64,
18613 __builtin_ia32_movdir64b): New builtins.
18614 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
18615 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
18617 (ix86_valid_target_attribute_inner_p): Ditto.
18618 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
18619 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
18620 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
18621 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
18622 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
18623 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
18624 (movdiri<mode>, movdir64b_<mode>): New.
18625 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
18626 * config/i386/immintrin.h: Include movdirintrin.h.
18627 * config/i386/movdirintrin.h: New file.
18628 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
18630 2018-04-19 Richard Biener <rguenther@suse.de>
18632 PR middle-end/85455
18633 * cfg.c (clear_bb_flags): When loop state says we have
18634 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
18636 2018-04-19 Richard Biener <rguenther@suse.de>
18638 PR tree-optimization/84737
18639 * tree-vect-data-refs.c (vect_copy_ref_info): New function
18640 copying restrict info.
18641 (vect_setup_realignment): Use it.
18642 * tree-vectorizer.h (vect_copy_ref_info): Declare.
18643 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
18644 the first DR to all generated stores.
18645 (vectorizable_load): Likewise for loads.
18647 2018-04-19 Jakub Jelinek <jakub@redhat.com>
18649 PR tree-optimization/85446
18650 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
18651 the integral and pointer types to have the same precision.
18653 * doc/install.texi: Document --disable-cet being the default and
18656 2018-04-18 Martin Liska <mliska@suse.cz>
18658 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
18661 2018-04-18 Martin Liska <mliska@suse.cz>
18664 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
18667 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
18668 arguments if they are comparable.
18670 2018-04-18 Martin Liska <mliska@suse.cz>
18673 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
18676 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
18679 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
18682 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
18683 ENDBR after calling __morestack.
18685 2018-04-18 David Malcolm <dmalcolm@redhat.com>
18688 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
18689 by using gcc_base_ver to generate a gcc_driver_version, and use
18690 it when generating GCC_DRIVER_NAME.
18691 * configure: Regenerate.
18693 2018-04-18 Jakub Jelinek <jakub@redhat.com>
18696 * config.gcc: Obsolete powerpc*-*-*spe*.
18698 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18701 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
18702 (stabstr_D): Change type of unum from unsigned int to
18703 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
18706 2018-04-17 Jim Wilson <jimw@sifive.com>
18709 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
18710 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
18711 Set arg_pointer_offset after using pretend_args_size.
18713 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18715 PR rtl-optimization/85431
18716 * dse.c (record_store): Ignore zero width stores.
18719 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
18720 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
18721 __builtin_stack_restore rather than after it.
18722 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
18723 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
18724 argument instead of virtual_dynamic_stack_rtx.
18726 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
18728 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
18730 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18731 Add note to error message to explain internal mapping of overloaded
18732 built-in function name to non-overloaded built-in function name.
18733 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
18736 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
18739 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
18740 where the inputs overlap with the output.
18742 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18744 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
18745 (=v, v) alternative and explicit "memory" attribute.
18746 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
18747 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
18749 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
18750 "sselog1" type instead of "sselog".
18751 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
18752 "sselog". Remove explicit "memory" attribute.
18753 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
18754 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
18756 (vec_extract_hi_v32hi): Merge all alternatives into one, use
18757 "sselog1" type instead of "sselog". Remove explicit "memory"
18759 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
18760 use "sselog1" type instead of "sselog". Remove explicit "memory"
18762 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
18763 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
18765 (vec_extract_hi_v64qi): Merge all alternatives into one, use
18766 "sselog1" type instead of "sselog". Remove explicit "memory"
18768 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
18769 use "sselog1" type instead of "sselog". Remove explicit "memory"
18773 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
18775 PR middle-end/85414
18776 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
18779 2018-04-17 Martin Jambor <mjambor@suse.cz>
18782 * ipa-cp.c (create_specialized_node): Call
18783 expand_all_artificial_thunks if necessary.
18785 2018-04-17 Martin Liska <mliska@suse.cz>
18788 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
18789 in message, remote space in between '_G' and '('.
18791 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18794 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
18795 avx512f_vmcmp<mode>3<round_saeonly_name>,
18796 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
18797 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
18798 avx512f_rndscale<mode><round_saeonly_name>,
18799 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
18800 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
18801 Use %<iptr>2 instead of %2 for -masm=intel.
18802 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
18803 avx512f_vcvttss2usi<round_saeonly_name>,
18804 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
18806 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
18807 avx512f_vcvttsd2usi<round_saeonly_name>,
18808 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
18809 Use %q1 instead of %1 for -masm=intel.
18810 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
18811 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
18812 of %3 for -masm=intel.
18813 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
18815 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
18817 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
18819 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
18820 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
18822 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
18824 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
18825 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
18826 %g1 and one with %0 and %1.
18827 (avx512er_vmrcp28<mode><round_saeonly_name>,
18828 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
18829 %1 for -masm=intel.
18830 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
18831 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
18832 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
18833 of %0 and %{%4%} for -masm=intel.
18834 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
18835 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
18836 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
18837 order of %0 and %{%5%}%{z%} for -masm=intel.
18839 2018-04-17 Jan Hubicka <jh@suse.cz>
18842 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
18844 2018-04-17 Martin Liska <mliska@suse.cz>
18847 * multiple_target.c (create_dispatcher_calls): Set apostrophes
18848 for target_clone error message. Make default implementation
18849 clone to be a local declaration.
18850 (separate_attrs): Add new argument and check for an empty
18852 (expand_target_clones): Handle it.
18853 (ipa_target_clone): Make redirection just for target_clones
18856 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
18857 Tom de Vries <tom@codesourcery.com>
18859 PR middle-end/84955
18860 * omp-expand.c (expand_oacc_for): Add dummy false branch for
18861 tiled basic blocks without omp continue statements.
18863 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
18866 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
18867 vec_extract expression as having side effects to make sure it gets
18870 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
18873 * config/i386/i386.c (get_builtin_code_for_version): Check
18876 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
18879 * config.gcc: Support "skylake".
18880 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
18882 * config/i386/i386.c (m_SKYLAKE): Define.
18883 (processor_target_table): Add "skylake".
18884 (ix86_option_override_internal): Add "skylake".
18885 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
18886 PROCESSOR_CANNONLAKE.
18887 (get_builtin_code_for_version): Fix priority for
18888 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
18889 PROCESSOR_SKYLAKE-AVX512.
18890 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
18891 (processor_type): Add PROCESSOR_SKYLAKE.
18893 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
18894 Jason Merrill <jason@redhat.com>
18897 * convert.c (convert_to_integer_1): Use direct recursion for
18898 enumeral types and types with a precision less than the number
18899 of bits in their mode.
18901 2018-04-16 Julia Koval <julia.koval@intel.com>
18904 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
18905 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
18907 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
18910 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
18911 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
18912 and -mno-direct-move.
18914 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
18917 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
18918 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
18919 Ensure negative shifts result in {0}.
18921 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
18923 PR rtl-optimization/79916
18924 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
18925 regs (if any) to define how to gnerate SD moves when LRA is in
18928 2018-04-13 Jakub Jelinek <jakub@redhat.com>
18930 PR rtl-optimization/85393
18931 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
18932 * except.c (expand_dw2_landing_pad_for_region): Make static.
18933 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
18934 a label and unconditional jump to old_bb, rather than
18935 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
18938 PR rtl-optimization/85376
18939 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
18940 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
18941 instead of a specific value.
18943 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
18944 Bin Cheng <bin.cheng@arm.com>
18946 PR tree-optimization/82965
18947 PR tree-optimization/83991
18948 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
18949 by_profile_only parameter.
18950 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
18951 information if the loop was predicted to iterate too many times.
18952 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
18954 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
18957 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
18960 2018-04-13 Martin Liska <mliska@suse.cz>
18961 Jakub Jelinek <jakub@redhat.com>
18963 PR middle-end/81657
18964 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
18965 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
18966 * builtins.c (expand_builtin_memory_copy_args): Use
18967 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
18968 handle dest_addr == pc_rtx.
18970 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
18973 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
18974 asked to not generate direct moves.
18975 (fix_trunc<mode>si2_stfiwx): Similar.
18976 (fix_trunc<mode>si2_internal): Similar.
18978 2018-04-12 Jakub Jelinek <jakub@redhat.com>
18981 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
18982 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
18983 lookup if dest in some wider mode is known to be const0_rtx and
18984 if so, record permanent equivalence for it to be ZERO_EXTEND of
18985 the narrower mode destination.
18987 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
18989 * lto-streamer-out.c (output_function): Revert 259346.
18990 * omp-expand.c (expand_oacc_for): Likewise.
18992 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
18994 PR rtl-optimization/85354
18995 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
18996 * sel-sched.c (sel_global_init): ... here.
18998 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
19001 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
19002 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
19003 mode for PE-COFF targets.
19004 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
19005 (i386_pe_asm_lto_end): Likewise.
19006 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
19007 (TARGET_ASM_LTO_END): Likewise.
19008 * config/i386/winnt.c (saved_debug_info_level): New static variable.
19009 (i386_pe_asm_lto_start): New function.
19010 (i386_pe_asm_lto_end): Likewise.
19012 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
19013 Richard Biener <rguenther@suse.de>
19015 PR middle-end/84955
19016 * lto-streamer-out.c (output_function): Fix CFG loop state before
19018 * omp-expand.c (expand_oacc_for): Handle calls to internal
19019 functions like regular functions.
19021 2018-04-12 Richard Biener <rguenther@suse.de>
19024 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
19025 for the early LTO debug to properly generate references to it
19026 during DIE emission. Do not re-use that for the skeleton for
19028 (dwarf2out_early_finish): Likewise.
19030 2018-04-12 Jakub Jelinek <jakub@redhat.com>
19033 * config/i386/sse.md
19034 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
19035 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
19036 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
19037 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
19038 and output is a reg, avoid creating invalid lowpart subreg, but
19039 instead split into a 512-bit move. Don't split if not AVX512VL,
19040 input is xmm16+ reg and output is a mem.
19041 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
19042 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
19043 xmm16+ reg and output is a mem.
19045 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19047 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
19048 also for flag_dwarf2_cfi_asm.
19050 2018-04-12 Jakub Jelinek <jakub@redhat.com>
19052 PR rtl-optimization/85342
19053 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
19054 a bool scalar var inside of the loop instead. Don't try to update
19055 recog_data.operand after failed apply_change_group.
19057 2018-04-12 Tom de Vries <tom@codesourcery.com>
19060 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
19061 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
19062 array with flexible array member as array without given dimension.
19063 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
19064 argument for undefined param to true.
19066 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
19069 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
19070 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
19071 from PowerPC section.
19072 * config/rs6000/sysv4.opt (mcall-): Improve help text.
19073 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
19074 help text that is too long.
19075 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
19076 help text that is too long.
19077 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
19078 help text that is too long.
19080 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
19082 * config/alpha/alpha.md (stack_probe_internal): Rename
19083 from "probe_stack". Update all callers.
19085 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19087 PR rtl-optimization/84566
19088 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
19089 sched_macro_fuse_insns.
19091 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19094 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
19095 (compute_block_dependences): ... from here.
19097 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19099 PR tree-optimization/85331
19100 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
19101 from int to HOST_WIDE_INT.
19103 2018-04-11 Martin Jambor <mjambor@suse.cz>
19106 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
19107 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
19108 not the same as the source val.
19109 (cgraph_edge_brings_value_p): New parameter.
19110 (gather_edges_for_value): Pass destination value to
19111 cgraph_edge_brings_value_p.
19112 (perhaps_add_new_callers): Likewise.
19113 (get_info_about_necessary_edges): Likewise and exclude values brought
19114 only by self-recursive edges.
19115 (create_specialized_node): Redirect only clones of self-calling edges.
19116 (+self_recursive_pass_through_p): New function.
19117 (find_more_scalar_values_for_callers_subset): Use it.
19118 (find_aggregate_values_for_callers_subset): Likewise.
19119 (known_aggs_to_agg_replacement_list): Removed.
19120 (decide_whether_version_node): Re-calculate known constants for all
19121 remaining context clones.
19123 2018-04-11 Richard Biener <rguenther@suse.de>
19126 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
19127 from early DWARF output.
19128 (dwarf2out_early_finish): Output line info unconditionally into
19129 early DWARF and add reference to it.
19131 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19134 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
19135 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
19136 other than V2DFmode using iptr mode attribute.
19137 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
19139 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19141 PR rtl-optimization/84659
19142 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
19144 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19147 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
19149 (output_loc_list): Pass address of a dummy size variable even in the
19150 locview handling loop.
19151 (index_location_lists): Add comment on why skip_loc_list_entry can't
19154 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
19157 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
19160 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
19163 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
19164 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
19165 and -mstring-compare-inline-limit.
19167 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19170 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
19171 for stack clash protection in a register whenever we need it to be in
19174 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19176 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
19177 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
19179 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19182 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
19184 (mlong-double-): Ditto.
19185 * config/rs6000/sysv4.opt (msdata=): Ditto.
19186 (mtls-size=): Ditto.
19188 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19190 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
19191 erroneous entries for
19192 "vector int vec_ldl (int, long int *)", and
19193 "vector unsigned int vec_ldl (int, unsigned long int *)".
19194 Add comments and entries for
19195 "vector bool char vec_ldl (int, bool char *)",
19196 "vector bool short vec_ldl (int, bool short *)",
19197 "vector bool int vec_ldl (int, bool int *)",
19198 "vector bool long long vec_ldl (int, bool long long *)",
19199 "vector pixel vec_ldl (int, pixel *)",
19200 "vector long long vec_ldl (int, long long *)",
19201 "vector unsigned long long vec_ldl (int, unsigned long long *)".
19202 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
19203 type tree bool_long_long_type_node and correct definition of
19204 bool_V2DI_type_node to make reference to this new type tree.
19205 (rs6000_mangle_type): Replace erroneous reference to
19206 bool_long_type_node with bool_long_long_type_node.
19207 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
19208 comments to emphasize sign distinctions for char and int types and
19209 replace RS6000_BTI_bool_long constant with
19210 RS6000_BTI_bool_long_long constant. Also add comment to restrict
19211 use of RS6000_BTI_pixel.
19212 (bool_long_type_node): Remove this macro definition.
19213 (bool_long_long_type_node): New macro definition
19215 2018-04-10 Jakub Jelinek <jakub@redhat.com>
19217 PR rtl-optimization/85300
19218 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
19219 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
19220 simplify_unary_operation fails.
19222 2018-04-10 Martin Liska <mliska@suse.cz>
19224 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
19225 cgraph_edge and ipa_ref.
19227 2018-04-10 Jakub Jelinek <jakub@redhat.com>
19231 * config/i386/sse.md
19232 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
19233 computation of the VEC_MERGE selector from mask.
19234 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
19235 Fix decoding of the VEC_MERGE selector into mask.
19237 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
19239 PR tree-optimization/85286
19240 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
19242 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
19244 * final.c (final_1): Set insn_last_address as well as
19245 insn_current_address.
19247 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19250 * explow.c (emit_stack_probe): Call validize_mem on memory location
19251 before passing it to gen_probe_stack. Create address operand and
19252 legitimize it for the probe_stack_address case.
19254 2018-04-09 Jan Hubicka <jh@suse.cz>
19257 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
19258 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
19259 * tree.c (free_lang_data_in_type): Fix handling of binfos;
19261 (free_lang_data): Rebuild type inheritance graph.
19263 2018-04-09 Martin Sebor <msebor@redhat.com>
19265 * invoke.texi (-finline-small-functions): Mention other optimization
19267 (-findirect-inlining, -fpartial-inlining): Same.
19268 (-finline-functions-called-once): Same.
19269 (-freorder-blocks-and-partition): Same.
19271 2018-04-09 Jan Hubicka <jh@suse.cz>
19274 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
19275 jumps; choose last target that matches the criteria (i.e.
19276 no partition changes for non-crossing jumps).
19277 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
19278 support for redirecting crossing jumps to non-crossing.
19280 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
19282 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
19283 also for naked functions.
19285 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
19287 * config/arc/arc.md (add_shift): New pattern.
19288 (add_shift2): Likewise.
19289 (sub_shift): Likewise.
19290 (sub_shift_cmp0_noout): Likewise.
19291 (compare_si_ashiftsi): Likewise.
19292 (xbfu_cmp0_noout): New combine pattern.
19293 (xbfu_cmp0"): Likewise.
19294 (movsi_set_cc_insn): Place the predicable variant first.
19295 (commutative_binary_cmp0_noout): Remove clobber.
19296 (commutative_binary_cmp0): New pattern.
19297 (noncommutative_binary_cmp0): Likewise.
19298 (noncommutative_binary_cmp0_noout): Likewise.
19299 (noncommutative_binary_comparison_result_used): Removed.
19300 (rsub_cmp0): New pattern.
19301 (rsub_cmp0_noout): Likewise.
19302 (extzvsi): Changed, keep only meaningful variants.
19303 (SQH, SEZ): New iterators.
19304 (SQH_postfix): New mode attribute.
19305 (SEZ_prefix): New code attribute.
19306 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
19307 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
19308 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
19309 of numerical value.
19310 (noncommutative_operator): Check the availability of barrel
19313 2018-04-09 Richard Biener <rguenther@suse.de>
19315 PR tree-optimization/85284
19316 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
19317 Only use the niter constraining form of simple_iv when the exit
19318 is always executed.
19320 2018-04-09 Tom de Vries <tom@codesourcery.com>
19323 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
19324 (define_expand "*memory_barrier"): New define_expand.
19325 (define_insn "memory_barrier"): New insn.
19327 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19329 PR rtl-optimization/80463
19330 PR rtl-optimization/83972
19331 PR rtl-optimization/83480
19333 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
19334 correct producer for the insn.
19335 (tidy_control_flow): Fixup seqnos in case of debug insns.
19337 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19339 PR rtl-optimization/83913
19341 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
19342 different sched-times when merging exprs.
19344 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19346 PR rtl-optimization/83962
19348 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
19349 tidy_fallthru_edge and tidy_control_flow.
19351 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19353 PR rtl-optimization/83530
19355 * sel-sched.c (force_next_insn): New global variable.
19356 (remove_insn_for_debug): When force_next_insn is true, also leave only
19357 next insn in the ready list.
19358 (sel_sched_region): When the region wasn't scheduled, make another pass
19359 over it with force_next_insn set to 1.
19361 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
19363 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
19365 * config/nds32/constants.md (unspec_volatile_element): Add enum values
19366 for interrupt control.
19367 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
19368 functions for interrupt control.
19369 * config/nds32/nds32-intrinsic.md: Likewise.
19370 * config/nds32/nds32_intrinsic.h: Likewise.
19371 * config/nds32/nds32.h (nds32_builtins): Likewise.
19373 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
19375 * config/nds32/nds32.c (nds32_init_machine_status,
19376 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
19377 strict_aligned_p field.
19378 (nds32_expand_to_rtl_hook): New function.
19379 (TARGET_EXPAND_TO_RTL_HOOK): Define.
19380 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
19382 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19383 Chung-Ju Wu <jasonwucj@gmail.com>
19385 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
19386 * config/nds32/nds32-n7.md: New file.
19387 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
19388 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
19390 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
19391 * config/nds32/nds32.md (pipeline_model): Add n7.
19392 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
19393 * config/nds32/pipelines.md: Include n7 settings.
19395 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19396 Chung-Ju Wu <jasonwucj@gmail.com>
19398 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
19399 * config/nds32/nds32-e8.md: New file.
19400 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
19401 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
19403 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
19404 * config/nds32/nds32.md (pipeline_model): Add e8.
19405 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
19406 * config/nds32/pipelines.md: Include e8 settings.
19408 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19409 Chung-Ju Wu <jasonwucj@gmail.com>
19411 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
19412 * config/nds32/nds32-n8.md: New file.
19413 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
19414 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
19416 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
19417 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
19418 * config/nds32/nds32.md (pipeline_model): Add n8.
19419 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
19420 * config/nds32/pipelines.md: Include n8 settings.
19422 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19423 Chung-Ju Wu <jasonwucj@gmail.com>
19425 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
19426 * config/nds32/nds32-n9-2r1w.md: New file.
19427 * config/nds32/nds32-n9-3r2w.md: New file.
19428 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
19429 nds32_register_ports): New or modify for cpu n9.
19430 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
19432 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
19433 * config/nds32/nds32-utils.c: New file.
19434 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
19435 TARGET_MUL_SLOW): Define.
19436 * config/nds32/nds32.md (pipeline_model): New attribute.
19437 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
19438 New options that support cpu n9.
19439 * config/nds32/pipelines.md: Include n9 settings.
19440 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
19442 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
19444 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
19445 information if necessary.
19446 (output_cond_branch_compare_zero): Likewise.
19447 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
19448 (nds32_target_alignment): Refine for alignment.
19449 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
19450 (FUNCTION_BOUNDARY): Modify.
19451 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
19453 * config/nds32/nds32.opt (malways-align, malign-functions): New.
19455 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
19457 * config/nds32/constants.md (unspec_volatile_element): Add values for
19458 TLB operation and data prefetch.
19459 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
19460 functions for TLB operation and data prefetch.
19461 * config/nds32/nds32-intrinsic.md: Likewise.
19462 * config/nds32/nds32_intrinsic.h: Likewise.
19463 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
19464 (nds32_print_operand): Likewise.
19465 * config/nds32/nds32.h (nds32_builtins): Likewise.
19467 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
19468 Andrew Pinski <pinsika@gcc.gnu.org>
19470 PR middle-end/82976
19471 * match.pd: Use constant_boolean_node of correct type instead of
19472 boolean_true_node or boolean_false_node for simplifying
19473 pointer comparisons to zero.
19475 2018-04-07 Jakub Jelinek <jakub@redhat.com>
19477 PR tree-optimization/80021
19478 * tree.c (verify_type_variant): Make error call in verify_variant_match
19479 translatable and remove final full stop.
19481 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19483 * config/nds32/constants.md (unspec_volatile_element): Add
19484 UNSPEC_VOLATILE_EH_RETURN.
19485 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
19486 nds32_output_stack_pop): Support dwarf exception handling process.
19487 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
19488 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
19489 exception handling process.
19490 (nds32_compute_stack_frame): Likewise.
19491 (nds32_return_addr_rtx): Likewise.
19492 (nds32_initial_elimination_offset): Likewise.
19493 (nds32_expand_prologue): Likewise.
19494 (nds32_expand_epilogue): Likewise.
19495 (nds32_dynamic_chain_address): New function.
19496 * config/nds32/nds32.h (machine_function): Add fields for dwarf
19497 exception handling.
19498 (DYNAMIC_CHAIN_ADDRESS): Define.
19499 (EH_RETURN_DATA_REGNO): Define.
19500 (EH_RETURN_STACKADJ_RTX): Define.
19501 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
19502 patterns for dwarf exception handling.
19504 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19506 * config/nds32/nds32.h: Clean up obsolete macros.
19508 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19510 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
19511 Add enum values for particular instructions.
19512 * config/nds32/nds32-intrinsic.c: Implementation of expanding
19513 particular intrinsic functions.
19514 * config/nds32/nds32-intrinsic.md: Likewise.
19515 * config/nds32/nds32_intrinsic.h: Likewise.
19516 * config/nds32/nds32.h (nds32_builtins): Likewise.
19517 * config/nds32/nds32.md (type): Add pbsad and pbsada.
19518 (btst, ave): New patterns for particular instructions.
19520 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19522 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
19523 Add enum values for atomic load/store and memory sync.
19524 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
19526 * config/nds32/nds32-intrinsic.md: Likewise.
19527 * config/nds32/nds32_intrinsic.h: Likewise.
19528 * config/nds32/nds32.h (nds32_builtins): Likewise.
19530 2018-04-07 Jakub Jelinek <jakub@redhat.com>
19532 PR tree-optimization/85257
19533 * fold-const.c (native_encode_vector): If not all elts could fit
19534 and off is -1, return 0 rather than offset.
19535 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
19536 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
19537 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
19538 adjust buffer in native_interpret_expr call.
19540 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19542 * config/nds32/constants.md (unspec_volatile_element): Add cache
19543 control enum values.
19544 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
19545 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
19546 * config/nds32/nds32.c (nds32_cctl_names): New.
19547 (nds32_print_operand): Handle cache control register names.
19548 * config/nds32/nds32.h (nds32_builtins): New enum values.
19549 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
19551 * config/nds32/nds32.md (type): Add mmu.
19552 * config/nds32/pipelines.md (simple_insn): Add mmu.
19554 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19556 * config/nds32/nds32.md (type): Remove call.
19557 * config/nds32/pipelines.md (simple_insn): Likewise.
19559 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19561 * config/nds32/constants.md (unspec_volatile_element): Add
19562 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
19563 UNSPEC_VOLATILE_FMFCFG.
19564 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
19565 description for fmfcfg and fmfcsr.
19566 (bdesc_1arg): Add fmtcsr.
19567 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
19568 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
19569 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
19570 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
19571 unspec_fmfcfg): New patterns.
19572 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
19573 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
19574 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
19575 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
19576 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
19577 __nds32__fmfcfg): Define.
19579 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19581 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
19582 intrinsic register names.
19583 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
19584 intrinsic register enum values and macros.
19586 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19588 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
19589 for load/store addressing form.
19590 (nds32_print_operand_address): Likewise.
19592 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
19595 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
19596 based on LABEL_REF. Remove useless assertion.
19597 (pic_address_needs_scratch): Fix formatting.
19598 (sparc_legitimize_pic_address): Minor tweaks.
19599 (sparc_delegitimize_address): Adjust assertion accordingly.
19600 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
19601 into symbolic_operand.
19602 (movsi_high_pic_label_ref): Likewise.
19603 (movsi_lo_sum_pic_label_ref): Likewise.
19604 (movdi_pic_label_ref): Likewise.
19605 (movdi_high_pic_label_ref): Likewise.
19606 (movdi_lo_sum_pic_label_ref): Likewise.
19608 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
19610 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
19611 custom LIB_SPEC setup.
19613 2018-04-06 Ruslan Bukin <br@bsdpad.com>
19614 Kito Cheng <kito.cheng@gmail.com>
19616 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
19617 * config/riscv/freebsd.h: New.
19619 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
19621 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
19622 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
19625 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
19626 Kito Cheng <kito.cheng@gmail.com>
19628 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
19629 nds32_output_call, nds32_symbol_binds_local_p): New functions.
19630 * config/nds32/nds32-protos.h (nds32_output_call,
19631 nds32_output_return): Declare.
19632 * config/nds32/nds32.md: Refine all the call and return patterns.
19634 2018-04-06 Jakub Jelinek <jakub@redhat.com>
19637 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
19638 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
19640 PR rtl-optimization/84872
19641 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
19642 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
19643 EDGE_CROSSING edge.
19645 2018-04-06 Tamar Christina <tamar.christina@arm.com>
19647 * expr.c (copy_blkmode_to_reg): Revert 254862.
19648 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
19650 2018-04-06 Richard Biener <rguenther@suse.de>
19652 PR middle-end/85244
19653 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
19654 after seeing a component reference with an adjacent field. Treat
19655 refs to arrays at struct end of external decls similar to
19656 refs to unconstrained commons.
19658 2018-04-06 Jakub Jelinek <jakub@redhat.com>
19661 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
19662 look through SAVE_EXPRs with non-side-effects argument. Adjust
19664 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
19667 2018-04-06 Richard Biener <rguenther@suse.de>
19669 PR middle-end/85180
19670 * alias.c (find_base_term): New wrapper around find_base_term
19671 unwinding CSELIB_VAL_PTR changes.
19672 (find_base_term): Do not restore CSELIB_VAL_PTR during the
19675 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19677 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
19679 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
19680 constant definitions.
19681 ("nop"): lr 0,0 -> nopr r0
19682 ("nop_lr0", "nop_lr1"): New insn definitions.
19684 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
19686 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
19687 NDS32_V3PUSH_AVAILABLE_P macro.
19689 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
19690 Chung-Ju Wu <jasonwucj@gmail.com>
19692 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
19693 (nds32*-*-*): Add float and fpu_config into supported_defaults.
19694 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
19695 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
19696 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
19697 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
19698 * config/nds32/constraints.md: New constraints and checking for hard
19699 float configuration.
19700 * config/nds32/iterators.md: New mode iterator and attribute for hard
19701 float configuration.
19702 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
19704 * config/nds32/nds32-fpu.md: New file.
19705 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
19706 deal with hard float code generation.
19707 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
19709 (abi_type, float_reg_number): New enum type.
19710 * config/nds32/nds32-predicates.c: New predicates for hard float.
19711 * config/nds32/nds32-protos.h: Declare functions for hard float.
19712 * config/nds32/nds32.c: Implementation for hard float configuration.
19713 * config/nds32/nds32.h: Definitions for hard float configuration.
19714 * config/nds32/nds32.md: Include hard float machine description and
19715 modify patterns for hard float configuration.
19716 * config/nds32/nds32.opt: New options for hard float configuration.
19717 * config/nds32/predicates.md: New predicates for hard float
19720 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
19722 * common/config/nds32/nds32-common.c
19723 (nds32_option_optimization_table): Enable -mreleax-hint by default.
19725 2018-04-05 Jakub Jelinek <jakub@redhat.com>
19727 PR middle-end/85195
19728 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
19729 CONSTRUCTOR_ELT (ctor, ...)->value.
19731 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
19734 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
19736 2018-04-05 Tom de Vries <tom@codesourcery.com>
19739 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
19742 2018-04-05 Shiva Chen <shiva0217@gmail.com>
19743 Kito Cheng <kito.cheng@gmail.com>
19745 * config/nds32/constraints.md (U33): Fine-tune checking condition.
19746 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
19747 * config/nds32/nds32.h (nds32_16bit_address_type): Add
19748 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
19750 2018-04-05 Shiva Chen <shiva0217@gmail.com>
19751 Kito Cheng <kito.cheng@gmail.com>
19753 * config/nds32/constraints.md (Ufe): New memory constraint.
19754 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
19755 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
19756 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
19758 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
19759 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
19761 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19763 * config/nds32/nds32.md: Use optimize_size in the condition for
19764 alu-shift instructions.
19766 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19768 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
19770 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19772 * config/nds32/nds32.md (negsi2): Refine pattern.
19774 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
19775 Chung-Ju Wu <jasonwucj@gmail.com>
19777 * config/nds32/iterators.md (shift_rotate): New code iterator.
19778 (shift): New code attribute.
19779 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
19780 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
19781 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
19782 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
19783 bit-wise operations.
19784 (andsi3, *andsi3): Ditto.
19785 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
19786 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
19787 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
19788 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
19789 nds32_ior_operand, nds32_xor_operand): New predicates.
19791 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19793 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
19794 (addsi3, subsi3): ... this.
19796 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19798 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
19800 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19802 * config/nds32/nds32.md: Adjust indention.
19804 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
19806 * config/nds32/nds32.md (feature): New attribute.
19808 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19810 * config/nds32/nds32.md (subtype): New attribute.
19812 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19815 * config/arm/arm-builtins.c (arm_expand_builtin): Change
19816 expansion to perform a bitwise AND of the argument followed by a
19817 boolean negation of the result.
19819 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
19821 PR rtl-optimization/84878
19822 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
19823 the basic block. Assert the use reference is not artificial and that
19824 it has an associated insn.
19826 2018-04-04 Michael Matz <matz@suse.de>
19828 * builtins.c (compute_objsize): Pass correct operand
19829 to array_at_struct_end_p.
19831 2018-04-04 Richard Biener <rguenther@suse.de>
19834 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
19835 from contexts for DINFO_LEVEL_TERSE and below.
19837 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
19839 * config/nds32/nds32-doubleword.md (move_<mode>): Require
19840 resiter_operand condition.
19841 * config/nds32/nds32.md (*move<mode>): Ditto.
19843 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
19844 Monk Chiang <sh.chiang04@gmail.com>
19846 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
19848 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
19850 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
19852 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
19853 Kito Cheng <kito.cheng@gmail.com>
19855 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
19856 nds32_cond_code_str, output_cond_branch,
19857 output_cond_branch_compare_zero, nds32_expand_cbranch,
19858 nds32_expand_cstore, nds32_expand_movcc,
19859 nds32_output_cbranchsi4_equality_zero,
19860 nds32_output_cbranchsi4_equality_reg,
19861 nds32_output_cbranchsi4_equality_reg_or_const_int,
19862 nds32_output_cbranchsi4_greater_less_zero: New functions.
19863 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
19864 nds32_expand_cstore, nds32_expand_movcc,
19865 nds32_output_cbranchsi4_equality_zero,
19866 nds32_output_cbranchsi4_equality_reg,
19867 nds32_output_cbranchsi4_equality_reg_or_const_int,
19868 nds32_output_cbranchsi4_greater_less_zero): Declare.
19869 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
19870 nds32_rimm11s_operand): New predicates.
19871 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
19872 * config/nds32/nds32.md: Rewrite all the branch and conditional move
19875 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
19877 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
19878 * config/nds32/nds32.md: Ditto.
19879 * config/nds32/pipelines.md: Ditto.
19881 2018-04-04 Richard Biener <rguenther@suse.de>
19883 PR tree-optimization/85168
19884 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
19885 propagating abnormals.
19887 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
19889 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
19891 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
19892 Kito Cheng <kito.cheng@gmail.com>
19894 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
19895 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
19896 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
19897 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
19898 * config/nds32/nds32.md (sibcall_internal): New.
19899 (sibcall_register): Remove.
19900 (sibcall_immediate): Remove.
19901 (sibcall_value_internal): New.
19902 (sibcall_value_register): Remove.
19903 (sibcall_value_immediate): Remove.
19904 * config/nds32/predicates.md (nds32_general_register_operand): New.
19905 (nds32_call_address_operand): New.
19907 2018-04-03 Jakub Jelinek <jakub@redhat.com>
19909 PR rtl-optimization/85167
19910 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
19911 bb_defs if *split_p, instead preinitialize it to NULL.
19913 PR tree-optimization/85156
19914 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
19915 evaluating the argument multiple times.
19917 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
19919 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
19921 (_mm_cvtpd_ps): Likewise.
19922 (_mm_cvttpd_epi32): Likewise.
19923 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
19924 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
19925 vector, pixel, and bool following altivec.h include.
19927 2018-04-03 Martin Sebor <msebor@redhat.com>
19929 * doc/extend.texi (Common Function Attributes): Clarify.
19930 (const attribute): Likewise.
19931 (pure attribute): Likewise.
19933 2018-04-03 Jakub Jelinek <jakub@redhat.com>
19936 * config/i386/i386.c (ix86_expand_vector_set): Use
19937 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
19939 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
19941 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
19942 instructions when changing rounding bits to preserve precision bits
19943 in the x87 control word.
19945 2018-04-03 Martin Liska <mliska@suse.cz>
19947 PR tree-optimization/82491
19948 * rtl.h (strip_offset_and_add): Replace += suboffset with
19949 poly_uint64 () + suboffset.
19951 2018-03-29 Martin Liska <mliska@suse.cz>
19952 Martin Jambor <mjambor@suse.cz>
19955 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
19956 param_type is not an integral or pointer type.
19958 2018-04-03 Richard Biener <rguenther@suse.de>
19960 * sese.h (recompute_all_dominators): Remove.
19962 2018-04-02 Martin Sebor <msebor@redhat.com>
19964 * doc/invoke.texi (-Wrestrict): Fix typos.
19966 2018-04-02 Jim Wilson <jimw@sifive.com>
19968 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
19969 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
19970 (<optab>di3, <optab>si3_extend): Likewise.
19971 (<optab>si3_mask, <optab>si3_mask_1): New.
19972 (<optab>di3_mask, <optab>di3_mask_1): New.
19973 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
19974 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
19975 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
19977 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
19979 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
19982 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
19984 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
19985 (nds32_canonicalize_comparison): New function.
19987 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
19988 Kito Cheng <kito.cheng@gmail.com>
19989 Kuan-Lin Chen <kuanlinchentw@gmail.com>
19991 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
19992 * config/nds32/constants.md (unspec_volatile_element): Add
19993 UNSPEC_VOLATILE_RELAX_GROUP.
19994 * config/nds32/nds32-relax-opt.c: New file.
19995 * config/nds32/nds32-predicates.c
19996 (nds32_symbol_load_store_p): New function.
19997 * config/nds32/nds32-protos.h
19998 (nds32_symbol_load_store_p): Declare function.
19999 (make_pass_nds32_relax_opt): Declare new rtl pass function.
20000 * config/nds32/nds32.c
20001 (nds32_register_pass): New function to register pass.
20002 (nds32_register_passes): New function to register passes.
20003 * config/nds32/nds32.md (relax_group): New pattern.
20004 * config/nds32/nds32.opt (mrelax-hint): New option.
20005 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
20007 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
20009 * config/nds32/t-nds32: Modify files dependency.
20011 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
20013 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
20014 (PROFILE_HOOK): Define its implementation.
20016 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
20018 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
20019 type and 32-bit size.
20021 2018-04-01 Jakub Jelinek <jakub@redhat.com>
20023 PR middle-end/85090
20024 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
20025 (V_128_256): New mode iterator.
20026 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
20027 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
20028 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
20030 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
20031 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
20033 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
20036 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
20037 NaN inputs correctly.
20039 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
20042 * config/rs6000/vsx.md (??r): New mode attribute.
20043 (*vsx_mov<mode>_64bit): Use it.
20044 (*vsx_mov<mode>_32bit): Likewise.
20046 2018-03-30 Martin Sebor <msebor@redhat.com>
20048 PR tree-optimization/84818
20049 * builtins.c (check_access): Use warning_n.
20051 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20054 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
20056 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
20059 2018-03-30 Julia Koval <julia.koval@intel.com>
20062 * x86-tune.def (movx, partial_reg_dependency): Enable for
20065 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
20067 PR inline-asm/84985
20068 * lra-constraints.c (process_alt_operands): Move setting
20069 this_alternative_matches below.
20071 2018-03-29 Martin Liska <mliska@suse.cz>
20074 * doc/invoke.texi: Document how LTO works with debug info.
20075 Describe auto-load support of binutils. Mention 'x86-64'
20076 as valid option value of -march option.
20078 2018-03-29 Jakub Jelinek <jakub@redhat.com>
20080 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
20083 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
20084 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
20085 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
20088 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
20091 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
20092 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
20093 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
20094 for RS6000_BTM_POWERPC64.
20095 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
20096 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
20097 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
20100 (DIVDEU): Likewise.
20102 2018-03-28 Carl Love <cel@us.ibm.com>
20105 2017-09-27 Carl Love <cel@us.ibm.com>
20107 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
20108 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
20109 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
20112 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20114 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
20115 instead of __vector bool.
20116 (_mm_max_pu8): Likewise.
20117 (_mm_min_pi16): Likewise.
20119 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
20122 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
20123 (DIVWEUO): Likewise.
20124 (DIVDEO): Likewise.
20125 (DIVDEUO): Likewise.
20126 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
20127 DIVWEUO and DIVDEUO.
20128 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
20129 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
20130 (div_extend): Likewise.
20131 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
20133 (__builtin_divweuo): Likewise.
20134 (__builtin_divdeo): Likewise.
20135 (__builtin_divdeuo): Likewise.
20137 2018-03-28 Jakub Jelinek <jakub@redhat.com>
20140 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
20141 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
20143 PR tree-optimization/82004
20144 * gimple-match-head.c (optimize_pow_to_exp): New function.
20145 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
20146 Don't fold to exp if optimize_pow_to_exp is false.
20148 2018-03-28 Martin Liska <mliska@suse.cz>
20151 * calls.c (initialize_argument_information): Fix trailing space.
20152 * common.opt: Fix typo and provide better explanation for
20153 -fsanitize-coverage option.
20154 * config/i386/i386.opt: Fix typo.
20156 2018-03-28 Jakub Jelinek <jakub@redhat.com>
20157 Martin Liska <mliska@suse.cz>
20160 * gimplify.c (asan_poison_variable): Don't do the check for
20161 gimplify_omp_ctxp here.
20162 (gimplify_decl_expr): Do it here.
20163 (gimplify_target_expr): Likewise.
20165 2018-03-28 Martin Liska <mliska@suse.cz>
20168 * config/i386/i386.c (ix86_function_arg_advance): Do not call
20169 chkp_type_bounds_count if MPX is not enabled.
20171 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
20173 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
20175 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20178 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
20179 function to create the function decl for complex long double
20180 multiply and divide for -mabi=ieeelongdouble.
20181 (init_float128_ieee): Call it.
20183 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
20186 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
20187 -fcf-protection=branch -mibt.
20188 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
20190 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20193 * config/arm/arm.c (arm_valid_symbolic_address): Handle
20194 arm_word_relocations.
20196 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
20199 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
20200 extern array declarations.
20202 2018-03-27 Richard Biener <rguenther@suse.de>
20204 PR middle-end/84067
20205 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
20206 explicit single_use checks.
20208 2018-03-27 Richard Biener <rguenther@suse.de>
20210 PR tree-optimization/85082
20211 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
20214 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20216 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
20217 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
20218 Turn on fasynchronous-unwind-tables and funwind-tables.
20220 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
20223 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
20224 (*bmi_blsr_<mode>_ccz): Ditto.
20226 2018-03-26 Tom de Vries <tom@codesourcery.com>
20228 PR tree-optimization/85063
20229 * omp-general.c (offloading_function_p): New function. Factor out
20231 * omp-offload.c (pass_omp_target_link::gate): ... here.
20232 * omp-general.h (offloading_function_p): Declare.
20233 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
20234 with attribute omp declare target for offloading functions.
20236 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
20238 PR tree-optimization/84005
20239 * tree-data-ref.h (get_base_for_alignment): Declare.
20240 * tree-data-ref.c (get_base_for_alignment_1): New function.
20241 (get_base_for_alignment): Likewise.
20242 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
20243 get_base_for_alignment to find a suitable base object, instead
20244 of always using drb->base_address.
20246 2018-03-23 Jakub Jelinek <jakub@redhat.com>
20248 PR inline-asm/85022
20249 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
20250 known size by default.
20252 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
20254 PR inline-asm/85030
20255 * lra-constraints.c (process_alt_operands): Don't match BLKmode
20256 and non BLKmode operands.
20258 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20261 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
20262 Clean up attributes.
20264 2018-03-23 Richard Biener <rguenther@suse.de>
20267 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
20268 we are going to emit early debug for LTO.
20270 2018-03-23 Jakub Jelinek <jakub@redhat.com>
20272 PR inline-asm/85034
20273 * function.c (match_asm_constraints_1): Don't optimize if input
20274 doesn't satisfy general_operand predicate for output's mode.
20276 PR inline-asm/85022
20277 * alias.c (write_dependence_p): Don't require for x_canonicalized
20278 non-VOIDmode if x has VOIDmode.
20281 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
20282 just don't try to optimize it rather than assert it never happens.
20284 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
20286 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
20287 macro expansions for definition of ST_INTERNAL_<mode> and
20288 LD_INTERNAL_<mode> builtins.
20289 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
20291 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
20293 (altivec_expand_st_builtin): Likewise.
20294 (altivec_expand_builtin): Remove calls to deleted functions.
20295 (rs6000_address_for_altivec): Delete this function.
20296 * config/rs6000/vector.md: Remove expands for
20297 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
20299 2018-03-22 Sudakshina Das <sudi.das@arm.com>
20302 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
20303 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
20304 re-computing once computed.
20305 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
20306 (arm_init_machine_status): Initialize
20307 machine->static_chain_stack_bytes.
20309 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
20312 * doc/extend.texi: Add four new prototypes for vec_ld.
20313 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
20314 definitions for more logical presentation.
20315 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
20316 entries for V1TI variants of __builtin_altivec_ld builtin.
20317 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
20318 handling of V1TI variant of LVX icode pattern.
20319 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
20320 (rs6000_gimple_fold_builtin): Likewise.
20321 (altivec_init_builtins): Add code to define
20322 __builtin_altivec_lvx_v1ti function.
20324 2018-03-22 Jakub Jelinek <jakub@redhat.com>
20326 PR inline-asm/84941
20327 * function.c (match_asm_constraints_1): Don't do the optimization
20328 if input isn't a REG, SUBREG, MEM or constant.
20330 2018-03-22 Tom de Vries <tom@codesourcery.com>
20332 PR tree-optimization/84956
20333 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
20334 bb_has_abnormal_pred.
20336 2018-03-22 Jakub Jelinek <jakub@redhat.com>
20339 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
20340 DECL_INITIAL (decl) to decl at the end.
20341 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
20342 adjust the comment.
20344 2018-03-21 Joseph Myers <joseph@codesourcery.com>
20346 * doc/extend.texi (__builtin_tgmath): Document when complex
20347 integer types are treated as _Complex _Float64.
20349 2018-03-21 Tom de Vries <tom@codesourcery.com>
20351 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
20353 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20355 PR tree-optimization/84960
20356 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
20357 if it is ENTRY block, move them into single succ of ENTRY in that case.
20359 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
20361 PR tree-optimization/84811
20362 * poly-int.h (poly_span_traits): Remove the T3 parameter and
20363 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
20364 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
20365 (known_subrange_p): Update accordingly. Cast each value involved
20366 in the size comparison, rather than casting the result of the
20369 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20371 PR tree-optimization/84982
20372 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
20373 by flipping the least significant bit rather than all bits from
20374 bitpos to bitpos + bitsize - 1.
20376 2018-03-21 Nathan Sidwell <nathan@acm.org>
20378 * doc/extend.texi (Deprecated Features): Remove mention of
20379 long-deleted deprecations.
20381 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20384 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
20385 * configure: Regenerate.
20387 2018-03-21 Tom de Vries <tom@codesourcery.com>
20389 PR tree-optimization/83126
20390 * tree-parloops.c (num_phis): New function.
20391 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
20393 2018-03-21 Nathan Sidwell <nathan@acm.org>
20395 * doc/extend.texi (Deprecated Features): Update deprecated flags,
20396 mention anon-struct/union members and trailing attributes.
20398 2018-03-21 Bin Cheng <bin.cheng@arm.com>
20400 PR tree-optimization/84969
20401 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
20402 builtin memset partitions if they set different rhs values.
20404 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20406 PR rtl-optimization/84989
20407 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
20408 VEC_DUPLICATE with scalar result mode.
20410 2018-03-21 Martin Liska <mliska@suse.cz>
20413 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
20414 not intended return statement.
20416 2018-03-21 Martin Liska <mliska@suse.cz>
20419 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
20420 (chkp_find_bound_slots_1): Limit number of iterations.
20422 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
20425 * Minor grammar fixes for x86 options.
20427 2018-03-20 Jakub Jelinek <jakub@redhat.com>
20430 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
20431 holding REG_CFA_RESTORE notes, instead turn them into a USE.
20433 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
20436 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
20437 (altivec_lvx_<mode>_1op): Likewise.
20438 (altivec_stvx_<mode>_2op): Likewise.
20439 (altivec_stvx_<mode>_1op): Likewise.
20440 (altivec_lvx_<VM2:mode>): New define_expand.
20441 (altivec_stvx_<VM2:mode>): Likewise.
20442 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
20443 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
20444 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
20445 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
20446 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
20447 (rs6000_gen_lvx): Likewise.
20448 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
20449 (altivec_expand_stv_builtin): Likewise.
20450 (altivec_expand_builtin): Likewise.
20451 * config/rs6000/vector.md: Likewise.
20453 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20456 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
20459 2018-03-20 Richard Biener <rguenther@suse.de>
20462 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
20463 sign-conversions as zero, fall back to standard scalar_stmt
20466 2018-03-20 Martin Liska <mliska@suse.cz>
20469 * predict.c (rebuild_frequencies): Handle case when we have
20470 PROFILE_ABSENT, but flag_guess_branch_prob is false.
20472 2018-03-20 Jakub Jelinek <jakub@redhat.com>
20475 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
20476 flag_section_anchors.
20477 * varasm.c (use_blocks_for_decl_p): Remove hack for
20478 dw2_force_const_mem.
20481 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
20483 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
20484 be created, use lowpart_subreg of operands[0] rather than operands[0]
20486 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
20487 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
20488 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
20489 and n constraint instead of aarch64_shift_imm_di and Usd.
20490 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
20491 (*aarch64_<optab>_reg_minus<mode>3): ... this.
20493 2018-03-20 Sudakshina Das <sudi.das@arm.com>
20496 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
20497 to favor GPR over NEON registers.
20498 (<shift>di3_neon): Likewise.
20500 2018-03-20 Tom de Vries <tom@codesourcery.com>
20503 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
20504 (nvptx_process_pars): Emit bar.sync asap and alap.
20506 2018-03-20 Tom de Vries <tom@codesourcery.com>
20509 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
20510 seen_label if seen_label is already set.
20512 2018-03-20 Jakub Jelinek <jakub@redhat.com>
20515 * config/i386/i386.c (fold_builtin_cpu): For features above 31
20516 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
20517 Use 1U instead of 1. Formatting fixes.
20520 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
20521 instead of TREE_TYPE (s1) for the return value.
20523 2018-03-19 Jakub Jelinek <jakub@redhat.com>
20525 PR tree-optimization/84946
20526 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
20527 bitsize + bitsize in poly_uint64 rather than poly_int64.
20530 * dwarf2asm.c: Include fold-const.c.
20531 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
20532 of decl rather than decl itself.
20534 PR rtl-optimization/84643
20535 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
20537 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
20540 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
20541 calling assemble_variable.
20543 2018-03-19 Sudakshina Das <sudi.das@arm.com>
20546 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
20547 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
20549 2018-03-19 Jim Wilson <jimw@sifive.com>
20552 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
20553 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
20554 (riscv_first_stack_step): Likewise.
20555 (riscv_option_override): Use STACK_BOUNDARY instead of
20556 MIN_STACK_BOUNDARY.
20557 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
20558 MIN_STACK_BOUNDARY.
20559 (BIGGEST_ALIGNMENT): Set to 128.
20560 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
20561 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
20564 2018-03-19 Richard Biener <rguenther@suse.de>
20566 PR tree-optimization/84933
20567 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
20568 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
20570 2018-03-19 Richard Biener <rguenther@suse.de>
20572 PR tree-optimization/84859
20573 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
20574 (cond_if_else_store_replacement): Perform sinking operation on
20575 single-store BBs regardless of MAX_STORES_TO_SINK setting.
20576 Generalize what a BB with a single eligible store is.
20578 2018-03-19 Richard Biener <rguenther@suse.de>
20580 PR tree-optimization/84929
20581 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
20582 chrec_is_positive against non-chrec arg.
20584 2018-03-19 Tamar Christina <tamar.christina@arm.com>
20587 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
20589 2018-03-18 Martin Liska <mliska@suse.cz>
20591 PR rtl-optimization/84635
20592 * regrename.c (build_def_use): Use matches_mode only when
20595 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
20597 PR tree-optimization/84913
20598 * tree-vect-loop.c (vectorizable_reduction): Don't try to
20599 vectorize chains of COND_EXPRs.
20601 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
20603 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
20605 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
20607 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
20609 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
20611 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
20613 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
20614 Kito Cheng <kito.cheng@gmail.com>
20616 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
20617 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
20618 (nds32_adjust_reg_alloc_order): New function.
20619 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
20621 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
20623 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
20624 nds32_print_operand, nds32_print_operand_address): Use
20625 HOST_WIDE_INT_PRINT_DEC instead.
20627 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
20629 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
20631 2018-03-17 Jakub Jelinek <jakub@redhat.com>
20634 * config/i386/i386.c (initial_ix86_tune_features,
20635 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
20636 unsigned long long.
20637 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
20638 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
20639 rather than 1u << ix86_tune. Formatting fix.
20640 (ix86_option_override_internal): Change ix86_arch_mask from
20641 unsigned int to unsigned HOST_WIDE_INT, initialize to
20642 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
20643 (ix86_function_specific_restore): Likewise.
20645 2018-03-16 Jakub Jelinek <jakub@redhat.com>
20648 * postreload.c (reload_combine_recognize_pattern): Perform
20649 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
20650 truncate_int_for_mode the result for the destination's mode.
20653 * hsa-gen.c (mem_type_for_type): Fix comment typo.
20654 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
20656 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
20659 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
20662 * lra-assigns.c (lra_split_hard_reg_for): Don't use
20663 regno_allocno_class_array and sorted_pseudos.
20664 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
20665 insns where regno is used.
20667 2018-03-16 Martin Liska <mliska@suse.cz>
20670 * multiple_target.c (create_dispatcher_calls): Redirect
20671 reference in the symbol table.
20673 2018-03-16 Martin Liska <mliska@suse.cz>
20676 * multiple_target.c (create_dispatcher_calls): Redirect also
20679 2018-03-16 Jakub Jelinek <jakub@redhat.com>
20683 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
20684 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
20685 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
20687 2018-03-16 Julia Koval <julia.koval@intel.com>
20689 * doc/invoke.texi (Skylake Server): Add CLWB.
20690 Cannonlake): Remove CLWB.
20692 2018-03-16 Jakub Jelinek <jakub@redhat.com>
20694 PR tree-optimization/84841
20695 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
20697 (FLOAT_ONE_CONST_TYPE): Define.
20698 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
20699 (sort_by_operand_rank): Put entries with higher constant_type last
20700 rather than first to match comments.
20702 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
20704 * config/nios2/nios2.md (movsi_internal): Fix thinko in
20707 2018-03-15 Jakub Jelinek <jakub@redhat.com>
20710 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
20711 check and use address of target always.
20713 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
20716 * config/i386/i386.c (indirect_thunk_needed): Update comments.
20717 (indirect_thunk_bnd_needed): Likewise.
20718 (indirect_thunks_used): Likewise.
20719 (indirect_thunks_bnd_used): Likewise.
20720 (indirect_return_needed): New.
20721 (indirect_return_bnd_needed): Likewise.
20722 (output_indirect_thunk_function): Add a bool argument for
20724 (output_indirect_thunk_function): Don't generate alias for
20725 function return thunk.
20726 (ix86_code_end): Call output_indirect_thunk_function to generate
20727 function return thunks.
20728 (ix86_output_function_return): Set indirect_return_bnd_needed
20729 and indirect_return_needed instead of indirect_thunk_bnd_needed
20730 and indirect_thunk_needed.
20732 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
20734 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
20735 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
20736 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
20738 2018-03-15 David Malcolm <dmalcolm@redhat.com>
20739 Paul Hua <paul.hua.gm@gmail.com>
20742 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
20744 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
20746 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
20747 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
20748 resp. SFmode cases.
20750 2018-03-15 Tamar Christina <tamar.christina@arm.com>
20753 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
20754 instead of GET_MODE_SIZE when comparing Units.
20756 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
20759 * varasm.c (hash_section): Return an unchangeble hash value
20760 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
20761 Return !aarch64_can_use_per_function_literal_pools_p ().
20763 2018-03-15 Jakub Jelinek <jakub@redhat.com>
20766 * optabs.c (emit_conditional_move): Pass address of cmode's copy
20767 rather than address of cmode as last argument to prepare_cmp_insn.
20769 2018-03-15 Julia Koval <julia.koval@intel.com>
20771 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
20772 F_AVX512VNNI, F_AVX512BITALG): New.
20774 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
20777 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
20778 insn for floating-point loads and stores.
20780 2018-03-14 Carl Love <cel@us.ibm.com>
20782 * config/rs6000/rs6000-c.c: Add macro definitions for
20783 ALTIVEC_BUILTIN_VEC_PERMXOR.
20784 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
20785 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
20786 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
20787 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
20789 * config/doc/extend.texi: Add prototypes for vec_permxor.
20791 2018-03-14 David Malcolm <dmalcolm@redhat.com>
20794 * diagnostic-show-locus.c (class layout_point): Convert m_line
20795 from int to linenum_type.
20796 (line_span::comparator): Use linenum "compare" function when
20797 comparing line numbers.
20798 (test_line_span): New function.
20799 (layout_range::contains_point): Convert param "row" from int to
20801 (layout_range::intersects_line_p): Likewise.
20802 (layout::will_show_line_p): Likewise.
20803 (layout::print_source_line): Likewise.
20804 (layout::should_print_annotation_line_p): Likewise.
20805 (layout::print_annotation_line): Likewise.
20806 (layout::print_leading_fixits): Likewise.
20807 (layout::annotation_line_showed_range_p): Likewise.
20808 (struct line_corrections): Likewise for field m_row.
20809 (line_corrections::line_corrections): Likewise for param "row".
20810 (layout::print_trailing_fixits): Likewise.
20811 (layout::get_state_at_point): Likewise.
20812 (layout::get_x_bound_for_row): Likewise.
20813 (layout::print_line): Likewise.
20814 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
20815 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
20816 * input.c (selftest::test_linenum_comparisons): New function.
20817 (selftest::input_c_tests): Call it.
20818 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
20819 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
20820 * selftest.h (ASSERT_GT): New macro.
20821 (ASSERT_GT_AT): New macro.
20822 (ASSERT_LT): New macro.
20823 (ASSERT_LT_AT): New macro.
20825 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
20827 PR rtl-optimization/84780
20828 * combine.c (distribute_links): Don't make a link based on pc_rtx.
20830 2018-03-14 Martin Liska <mliska@suse.cz>
20832 * tree.c (record_node_allocation_statistics): Use
20833 get_stats_node_kind.
20834 (get_stats_node_kind): New function extracted from
20835 record_node_allocation_statistics.
20836 (free_node): Use get_stats_node_kind.
20838 2018-03-14 Richard Biener <rguenther@suse.de>
20840 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
20841 that the value-set of ANTIC_IN doesn't grow.
20844 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
20846 (BB_VISITED_WITH_VISITED_SUCCS): New define.
20847 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
20849 2018-03-14 Julia Koval <julia.koval@intel.com>
20851 * config.gcc (icelake-client, icelake-server): New.
20853 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
20854 (initial_ix86_arch_features): Ditto.
20855 (PTA_SKYLAKE): Add SGX.
20856 (PTA_ICELAKE): Remove.
20857 (PTA_ICELAKE_CLIENT): New.
20858 (PTA_ICELAKE_SERVER): New.
20859 (ix86_option_override_internal): Split up icelake on icelake client and
20861 (get_builtin_code_for_version): Ditto.
20862 (fold_builtin_cpu): Ditto.
20863 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
20864 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
20865 * config/i386/i386.h (processor_type): Ditto.
20866 * doc/invoke.texi: Ditto.
20868 2018-03-14 Jakub Jelinek <jakub@redhat.com>
20871 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
20872 INTEGER_CST offset, add it together with bitpos / 8 and
20873 sign extend based on POINTER_SIZE.
20877 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20880 * config/i386/constraints.md (Yc): New register constraint.
20881 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
20882 Use Yc constraint for alternative 2 of operand 0. Remove
20883 preferred_for_speed attribute.
20885 2018-03-14 Richard Biener <rguenther@suse.de>
20887 PR tree-optimization/84830
20888 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
20889 with the old one to avoid oscillations.
20891 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
20894 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
20896 (assign_by_spills): Return a flag of reload assignment failure.
20897 Do not process the reload assignment failures. Do not spill other
20898 reload pseudos if they has the same reg class. Update n if
20900 (lra_assign): Add a return arg. Set up from the result of
20901 assign_by_spills call.
20902 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
20903 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
20904 usage_insns if it is not NULL.
20905 (spill_hard_reg_in_range): New function.
20906 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
20907 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
20908 function prototypes.
20909 (lra_assign): Change prototype.
20910 * lra.c (lra): Add code to deal with fails by splitting hard reg
20913 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
20915 * config/riscv/riscv.opt (mrelax): New option.
20916 * config/riscv/riscv.c (riscv_file_start): Emit ".option
20917 "norelax" when riscv_mrelax is disabled.
20918 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
20920 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20923 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
20924 reassociation for int modes.
20926 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
20928 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
20929 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
20931 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
20932 * config/aarch64/aarch64-sve.md
20933 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
20934 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
20935 (*extend<mode><Vwide>2): Rename to...
20936 (aarch64_sve_extend<mode><Vwide>2): ...this.
20937 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
20938 renaming the old pattern to...
20939 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
20941 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
20942 define_expand, renaming the old pattern to...
20943 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
20944 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
20945 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
20946 account when deciding which SVE instruction the optab should use.
20947 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
20949 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
20951 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
20952 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
20953 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
20954 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
20955 (tlsdesc_small_<mode>): Turn a define_expand and use
20956 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
20957 (tlsdesc_small_advsimd_<mode>): ...this.
20958 (tlsdesc_small_sve_<mode>): New pattern.
20960 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
20962 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
20963 (UNSPEC_UMUL_HIGHPART): New constants.
20964 (MUL_HIGHPART): New int iteraor.
20965 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
20966 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
20968 (*<su>mul<mode>3_highpart): New define_insn.
20970 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
20973 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
20976 2018-03-13 Martin Liska <mliska@suse.cz>
20979 * (sem_item_optimizer::sem_item_optimizer): Initialize new
20981 (sem_item_optimizer::~sem_item_optimizer): Release it.
20982 (sem_item_optimizer::merge_classes): Register variable aliases.
20983 (sem_item_optimizer::fixup_pt_set): New function.
20984 (sem_item_optimizer::fixup_points_to_sets): Likewise.
20985 * ipa-icf.h: Declare new variables and functions.
20987 2018-03-13 Jakub Jelinek <jakub@redhat.com>
20989 PR middle-end/84834
20990 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
20991 integer_pow2p@2 and test integer_pow2p in condition.
20992 (A < 0 ? C : 0): Similarly for @1.
20994 PR middle-end/84831
20995 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
20996 characters starting at p contain '\0' character, don't look beyond
21000 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
21001 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
21004 * reg-stack.c (change_stack): Change update_end var from int to
21005 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
21006 also call set_block_for_insn on the newly added insns and rescan.
21009 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
21010 on the last operand.
21013 * tree.c (stabilize_reference_1): Return save_expr (e) for
21014 STATEMENT_LIST even if it doesn't have side-effects.
21016 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
21018 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
21020 2018-03-12 Renlin Li <renlin.li@arm.com>
21022 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
21023 aarch64_output_scalar_simd_mov_immediate.
21025 2018-03-12 Martin Sebor <msebor@redhat.com>
21027 PR tree-optimization/83456
21028 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
21029 for perfectly overlapping calls to memcpy.
21030 (gimple_fold_builtin_memory_chk): Same.
21031 (gimple_fold_builtin_strcpy): Handle no-warning.
21032 (gimple_fold_builtin_stxcpy_chk): Same.
21033 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
21035 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
21037 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
21038 parameter. Use it for SFmode.
21039 (rs6000_function_arg_advance_1): Adjust.
21040 (rs6000_function_arg): Adjust.
21041 (rs6000_gimplify_va_arg): Pass false for that new parameter.
21043 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
21045 PR rtl-optimization/84169
21046 PR rtl-optimization/84780
21047 * combine.c (can_combine_p): Check for a 2-insn combination whether
21048 the destination register is used between the two insns, too.
21050 2018-03-12 Richard Biener <rguenther@suse.de>
21052 PR tree-optimization/84803
21053 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
21054 for refs DR analysis didn't process.
21056 2018-03-12 Richard Biener <rguenther@suse.de>
21058 PR tree-optimization/84777
21059 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
21060 force-vectorize loops ignore whether we are optimizing for size.
21062 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
21064 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
21065 (TARGET_MD_ASM_ADJUST): Define.
21067 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
21068 Kito Cheng <kito.cheng@gmail.com>
21069 Chung-Ju Wu <jasonwucj@gmail.com>
21071 * config/nds32/nds32.c (nds32_compute_stack_frame,
21072 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
21073 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
21074 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
21075 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
21076 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
21077 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
21078 * config/nds32/nds32.md (prologue, epilogue): Use macro
21079 NDS32_V3PUSH_AVAILABLE_P to do checking.
21081 2018-03-11 Jakub Jelinek <jakub@redhat.com>
21084 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
21085 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
21086 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
21087 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
21088 addition of most attributes on !orig_type_die or the attribute not
21089 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
21091 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21092 Chung-Ju Wu <jasonwucj@gmail.com>
21094 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
21095 __NDS32_VH__ macro.
21096 * config/nds32/nds32.opt (mvh): New option.
21098 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21099 Chung-Ju Wu <jasonwucj@gmail.com>
21101 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
21103 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
21104 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
21107 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21108 Chung-Ju Wu <jasonwucj@gmail.com>
21110 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
21112 * config/nds32/nds32-multiple.md (strlensi): New pattern.
21113 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
21115 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
21116 Kito Cheng <kito.cheng@gmail.com>
21117 Chung-Ju Wu <jasonwucj@gmail.com>
21119 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
21120 UNSPEC_FFMISM and UNSPEC_FLMISM.
21121 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
21122 for ffb, ffmism and flmism.
21123 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
21124 (unspec_ffmism): Ditto.
21125 (unspec_flmism): Ditto.
21126 (nds32_expand_builtin_impl): Check if string extension is available.
21127 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
21128 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
21130 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
21133 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
21136 * lra-assigns.c (assign_by_spills): Return a flag of reload
21137 assignment failure. Do not process the reload assignment
21138 failures. Do not spill other reload pseudos if they has the same
21140 (lra_assign): Add a return arg. Set up from the result of
21141 assign_by_spills call.
21142 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
21143 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
21144 usage_insns if it is not NULL.
21145 (spill_hard_reg_in_range): New function.
21146 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
21147 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
21148 function prototypes.
21149 (lra_assign): Change prototype.
21150 * lra.c (lra): Add code to deal with fails by splitting hard reg
21153 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
21156 * config/i386/i386.opt: Replace Enforcment with Enforcement.
21158 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
21161 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
21162 (dw_val_node): Add val_symbolic_view.
21163 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
21164 (symview_upper_bound): New.
21165 (new_line_info_table): Initialize symviews_since_reset.
21166 (dwarf2out_source_line): Count symviews_since_reset and set
21167 symview_upper_bound.
21168 (dw_val_equal_p): Handle symview.
21169 (add_AT_symview): New.
21170 (print_dw_val): Handle symview.
21171 (attr_checksum, attr_checksum_ordered): Likewise.
21172 (same_dw_val_p, size_of_die): Likewise.
21173 (value_format, output_die): Likewise.
21174 (add_high_low_attributes): Use add_AT_symview for entry_view.
21175 (dwarf2out_finish): Reset symview_upper_bound, clear
21178 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
21181 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
21182 Add strict argument and use it.
21183 (rs6000_split_multireg_move): Update for new strict argument.
21184 (mem_operand_gpr): Disallow all non-offsettable addresses.
21185 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
21187 2018-03-09 Jakub Jelinek <jakub@redhat.com>
21190 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
21191 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
21192 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
21195 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
21196 decl, use remap_type if we want to use the type.
21198 2018-03-09 Martin Sebor <msebor@redhat.com>
21200 PR tree-optimization/84526
21201 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
21203 (builtin_access::generic_overlap): Be prepared to handle non-array
21206 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
21208 PR rtl-optimization/84682
21209 * lra-constraints.c (process_address_1): Check is_address flag
21210 for address constraints.
21211 (process_alt_operands): Likewise.
21212 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
21213 preprocess_constraints.
21214 * recog.h (preprocess_constraints): Add oploc parameter.
21216 * recog.c (preprocess_constraints): Test address_operand for
21217 CT_ADDRESS constraints.
21219 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
21222 * lra-assigns.c (assign_by_spills): Return a flag of reload
21223 assignment failure. Do not process the reload assignment
21224 failures. Do not spill other reload pseudos if they has the same
21226 (lra_assign): Add a return arg. Set up from the result of
21227 assign_by_spills call.
21228 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
21229 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
21230 usage_insns if it is not NULL.
21231 (spill_hard_reg_in_range): New function.
21232 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
21233 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
21234 function prototypes.
21235 (lra_assign): Change prototype.
21236 * lra.c (lra): Add code to deal with fails by splitting hard reg
21239 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21242 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
21243 Accept complain bool parameter. Only emit errors if it is true.
21244 (arm_parse_cpu_option_name): Likewise.
21245 (arm_target_thumb_only): Adjust callers of the above.
21246 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
21247 prototype to take a default true bool parameter.
21248 (arm_parse_arch_option_name): Likewise.
21250 2018-03-09 David Malcolm <dmalcolm@redhat.com>
21251 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
21255 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
21256 * configure: Regenerate.
21257 * configure.ac ("linker --version-script option"): New.
21258 ("linker soname option"): New.
21260 2018-03-09 Richard Biener <rguenther@suse.de>
21262 PR tree-optimization/84775
21263 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
21264 immediate uses of predicate stmts and mark them modified.
21267 PR tree-optimization/84178
21268 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
21270 (version_loop_for_if_conversion): Delay update_ssa call.
21271 (tree_if_conversion): Delay update_ssa until after predicate
21274 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
21277 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
21278 when the function accesses prior frames.
21280 2018-03-08 Jakub Jelinek <jakub@redhat.com>
21283 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
21284 gen_llsym, otherwise call maybe_gen_llsym.
21286 PR inline-asm/84742
21287 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
21288 has ',' character inside of it.
21290 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21293 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
21294 as clobbering CC_REGNUM.
21296 2018-03-08 Richard Biener <rguenther@suse.de>
21298 PR middle-end/84552
21299 * tree-scalar-evolution.c: Include tree-into-ssa.h.
21300 (follow_copies_to_constant): Do not follow SSA names registered
21303 2018-03-08 Richard Biener <rguenther@suse.de>
21305 PR tree-optimization/84178
21306 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
21308 (version_loop_for_if_conversion): Delay update_ssa call.
21309 (tree_if_conversion): Delay update_ssa until after predicate
21312 2018-03-08 David Malcolm <dmalcolm@redhat.com>
21314 PR tree-optimization/84178
21315 * tree-if-conv.c (release_bb_predicate): Remove the
21316 the assertion that the stmts have NULL use_ops.
21317 Discard the statements, asserting that they haven't
21318 yet been added to a BB.
21320 2018-03-08 Richard Biener <rguenther@suse.de>
21322 PR tree-optimization/84746
21323 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
21324 (phi_translate): Pass in destination ANTIC_OUT set.
21325 (phi_translate_1): Likewise. For a simplified result lookup
21326 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
21327 (phi_translate_set): Adjust.
21328 (do_pre_regular_insertion): Likewise.
21329 (do_pre_partial_partial_insertion): Likewise.
21331 2018-03-08 Martin Liska <mliska@suse.cz>
21333 PR gcov-profile/84735
21334 * doc/gcov.texi: Document usage of profile files.
21335 * gcov-io.h: Document changes in the format.
21337 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
21341 * dwarf2out.c (struct dw_line_info_table): Update comments for
21343 (FORCE_RESET_NEXT_VIEW): New.
21344 (FORCE_RESETTING_VIEW_P): New.
21345 (RESETTING_VIEW_P): Check for -1 too.
21346 (ZERO_VIEW_P): Likewise.
21347 (new_line_info_table): Force-reset next view.
21348 (dwarf2out_begin_function): Likewise.
21349 (dwarf2out_source_line): Simplify zero_view_p initialization.
21350 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
21351 view directly. Omit view when omitting .loc at line 0.
21353 2018-03-08 Jakub Jelinek <jakub@redhat.com>
21355 PR tree-optimization/84740
21356 * tree-switch-conversion.c (process_switch): Call build_constructors
21357 only if info.phi_count is non-zero.
21359 PR tree-optimization/84739
21360 * tree-tailcall.c (find_tail_calls): Check call arguments against
21361 DECL_ARGUMENTS (current_function_decl) rather than
21362 DECL_ARGUMENTS (func) when checking for tail recursion.
21364 2018-03-07 Jakub Jelinek <jakub@redhat.com>
21366 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
21367 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
21368 Volker Reichelt's entry and add entries for people that perform
21369 GCC fuzzy testing and report numerous bugs.
21371 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21374 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
21375 readonly data in sdata, if that is disabled.
21376 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
21377 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
21378 -mreadonly-in-sdata option.
21380 2018-03-07 Martin Sebor <msebor@redhat.com>
21382 PR tree-optimization/84468
21383 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
21384 basic block when looking for nul assignment.
21386 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
21389 * except.h (output_function_exception_table): Adjust prototype.
21390 * except.c (output_function_exception_table): Remove FNNAME parameter
21391 and add SECTION parameter. Ouput one part of the table at a time.
21392 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
21393 the first part of the exception table and emit unwind directives.
21394 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
21395 (i386_pe_seh_cold_init): Likewise.
21396 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
21397 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
21398 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
21399 (ix86_output_call_insn): Emit a nop in one more case for SEH.
21400 * config/i386/winnt.c: Include except.h.
21401 (struct seh_frame_state): Add reg_offset, after_prologue and
21402 in_cold_section fields.
21403 (i386_pe_seh_end_prologue): Set seh->after_prologue.
21404 (i386_pe_seh_cold_init): New function.
21405 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
21406 to seh->in_cold_section.
21407 (seh_emit_push): Record the offset of the push.
21408 (seh_emit_save): Record the offet of the save.
21409 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
21410 Test seh->after_prologue to disregard the epilogue.
21411 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
21412 (i386_pe_end_cold_function): New function.
21414 2018-03-07 Jakub Jelinek <jakub@redhat.com>
21417 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
21418 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
21421 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
21423 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
21424 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
21426 PR middle-end/84723
21427 * multiple_target.c: Include tree-inline.h and intl.h.
21428 (expand_target_clones): Diagnose and fail if node->definition and
21429 !tree_versionable_function_p (node->decl).
21431 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
21433 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
21435 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
21436 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
21437 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
21439 2018-03-06 Jakub Jelinek <jakub@redhat.com>
21442 * combine.c (try_combine): Use reg_or_subregno instead of handling
21443 just paradoxical SUBREGs and REGs.
21445 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
21447 * config/arc/arc.c (arc_finalize_pic): Remove function.
21448 (arc_must_save_register): We use single base PIC register, remove
21449 checks to save/restore the PIC register.
21450 (arc_expand_prologue): Likewise.
21451 * config/arc/arc-protos.h (arc_set_default_type_attributes):
21453 (arc_verify_short): Likewise.
21454 (arc_attr_type): Likewise.
21455 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
21456 (walk_stores): Likewise.
21457 (arc_address_cost): Make it static.
21458 (arc_verify_short): Likewise.
21459 (branch_dest): Likewise.
21460 (arc_attr_type): Likewise.
21461 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
21462 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
21463 (arc_final_prescan_insn): Remove inserting the nops due to
21464 hardware hazards. It is done in reorg step.
21465 (insn_length_variant_t): Remove.
21466 (insn_length_parameters_t): Likewise.
21467 (arc_insn_length_parameters): Likewise.
21468 (arc_get_insn_variants): Likewise.
21469 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
21471 2018-03-06 Jakub Jelinek <jakub@redhat.com>
21473 PR inline-asm/84683
21474 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
21477 PR tree-optimization/84687
21478 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
21480 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
21482 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21484 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
21485 Rename to ppc_speculation_barrier.
21486 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
21487 __builtin_ppc_speculation_barrier.
21489 2018-03-05 Jakub Jelinek <jakub@redhat.com>
21492 * combine.c (combine_simplify_rtx): Don't try to simplify if
21493 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
21496 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
21498 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
21499 to 32 bytes when compiling for POWER9.
21501 2018-03-05 Jakub Jelinek <jakub@redhat.com>
21504 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
21505 regparm >= 3 with no arg reg available also for calls with
21506 flag_force_indirect_call. Pass decl to ix86_function_regparm.
21509 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
21511 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
21513 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
21516 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
21518 2018-03-05 Richard Biener <rguenther@suse.de>
21520 PR tree-optimization/84486
21521 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
21522 When inserting a __builtin_assume_aligned call set the LHS
21523 SSA name alignment info accordingly.
21525 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
21527 PR tree-optimization/84114
21528 * config/aarch64/aarch64.c (aarch64_reassociation_width)
21529 Avoid reassociation of FLOAT_MODE addition.
21531 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
21533 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
21534 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
21535 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
21536 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
21537 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
21538 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
21539 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
21541 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
21542 __builtin_ia32_wbinvd): New builtins.
21543 (SPECIAL_ARGS2): New.
21544 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
21545 (SPECIAL_ARGS2): New.
21546 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
21547 (ix86_valid_target_attribute_inner_p): Ditto.
21548 (ix86_init_mmx_sse_builtins): Add special_args2.
21549 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
21550 TARGET_WBNOINVD_P): New.
21551 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
21552 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
21553 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
21554 * config/i386/immintrin.h (_wbinvd): New intrinsic.
21555 * config/i386/pconfigintrin.h: New file.
21556 * config/i386/wbnoinvdintrin.h: Ditto.
21557 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
21559 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
21561 2018-03-05 Richard Biener <rguenther@suse.de>
21563 PR tree-optimization/84670
21564 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
21566 (BB_VISITED_WITH_VISITED_SUCCS): New define.
21567 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
21568 (compute_antic_aux): Only assert the number of values in ANTIC_IN
21569 doesn't grow if all successors (recursively) were visited at least
21572 2018-03-05 Richard Biener <rguenther@suse.de>
21574 PR tree-optimization/84650
21575 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
21576 if executed in the loop pipeline.
21578 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
21580 * doc/configfiles.texi (Configuration Files): Move info about
21581 conditionalizing $target-protos.h to...
21582 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
21583 differs from $target-protos.h.
21585 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
21586 Chung-Ju Wu <jasonwucj@gmail.com>
21588 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
21589 * config/nds32/nds32-multiple.md (setmemsi): Define.
21590 * config/nds32/nds32-memory-manipulation.c
21591 (nds32_gen_dup_4_byte_to_word_value): New.
21592 (emit_setmem_word_loop): New.
21593 (emit_setmem_byte_loop): New.
21594 (nds32_expand_setmem_loop): New.
21595 (nds32_expand_setmem_loop_v3m): New.
21596 (nds32_expand_setmem_unroll): New.
21597 (nds32_expand_setmem): New.
21599 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
21600 Chung-Ju Wu <jasonwucj@gmail.com>
21602 * config/nds32/nds32-memory-manipulation.c
21603 (nds32_emit_load_store): New.
21604 (nds32_emit_post_inc_load_store): New.
21605 (nds32_emit_mem_move): New.
21606 (nds32_emit_mem_move_block): New.
21607 (nds32_expand_movmemsi_loop_unknown_size): New.
21608 (nds32_expand_movmemsi_loop_known_size): New.
21609 (nds32_expand_movmemsi_loop): New.
21610 (nds32_expand_movmemsi_unroll): New.
21611 (nds32_expand_movmemqi): Rename ...
21612 (nds32_expand_movmemsi): ... to this.
21613 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
21614 (movmemsi): ... to this.
21615 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
21616 (nds32_expand_movmemsi): ... to this.
21618 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
21619 Monk Chiang <sh.chiang04@gmail.com>
21620 Chung-Ju Wu <jasonwucj@gmail.com>
21622 * config/nds32/nds32-protos.h
21623 (nds32_expand_load_multiple): New arguments.
21624 (nds32_expand_store_multiple): Ditto.
21625 (nds32_valid_multiple_load_store): Rename ...
21626 (nds32_valid_multiple_load_store_p): ... to this.
21627 * config/nds32/nds32-memory-manipulation.c
21628 (nds32_expand_load_multiple): Refine implementation.
21629 (nds32_expand_store_multiple): Ditto.
21630 * config/nds32/nds32-multiple.md
21631 (load_multiple): Update nds32_expand_load_multiple interface.
21632 (store_multiple): Update nds32_expand_store_multiple interface.
21633 * config/nds32/nds32-predicates.c
21634 (nds32_valid_multiple_load_store): Rename ...
21635 (nds32_valid_multiple_load_store_p): ... to this and refine
21637 * config/nds32/predicates.md
21638 (nds32_load_multiple_and_update_address_operation): New predicate.
21639 (nds32_store_multiple_and_update_address_operation): New predicate.
21641 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
21642 Chung-Ju Wu <jasonwucj@gmail.com>
21644 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
21645 (combo): New attribute.
21646 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
21648 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
21650 * config/nds32/nds32.opt: Change -mcmodel= default value.
21652 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
21653 Monk Chiang <sh.chiang04@gmail.com>
21654 Chung-Ju Wu <jasonwucj@gmail.com>
21656 * config/nds32/constants.md (unspec_element): New enum.
21657 * config/nds32/constraints.md (Umw): New constraint.
21658 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
21659 * config/nds32/nds32-intrinsic.md: Likewise.
21660 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
21661 (nds32_valid_smw_lwm_base_p): New.
21662 (nds32_output_smw_single_word): New.
21663 (nds32_output_lmw_single_word): New.
21664 (nds32_expand_unaligned_load): New.
21665 (nds32_expand_unaligned_store): New.
21666 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
21667 (nds32_output_smw_single_word): Declare.
21668 (nds32_output_lmw_single_word): Declare.
21669 (nds32_expand_unaligned_load): Declare.
21670 (nds32_expand_unaligned_store): Declare.
21671 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
21672 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
21673 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
21674 NDS32_BUILTIN_UASTORE_DW.
21675 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
21678 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
21679 Kito Cheng <kito.cheng@gmail.com>
21680 Chung-Ju Wu <jasonwucj@gmail.com>
21682 * config/nds32/nds32-intrinsic.c
21683 (nds32_expand_builtin_null_ftype_reg): Delete.
21684 (nds32_expand_builtin_reg_ftype_imm): Ditto.
21685 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
21686 (nds32_read_argument): New.
21687 (nds32_legitimize_target): Ditto.
21688 (nds32_legitimize_argument): Ditto.
21689 (nds32_check_constant_argument): Ditto.
21690 (nds32_expand_unop_builtin): Ditto.
21691 (nds32_expand_unopimm_builtin): Ditto.
21692 (nds32_expand_binop_builtin): Ditto.
21693 (nds32_builtin_decl_impl): Ditto.
21694 (builtin_description): Ditto.
21695 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
21696 (nds32_init_builtins_impl): Ditto.
21697 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
21698 (nds32_builtin_decl): New.
21699 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
21700 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
21702 2018-03-02 Jeff Law <law@redhat.com>
21704 * reorg.c (stop_search_p): Handle DEBUG_INSN.
21705 (redundant_insn, fill_simple_delay_slots): Likewise.
21706 (fill_slots_from_thread): Likewise.
21707 * resource.c (mark_referenced_resources): Likewise.
21708 (mark_set_resources, find_dead_or_set_registers): Likewise.
21710 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21712 * substring-locations.h (format_warning_va): Formatting fix for
21713 ATTRIBUTE_GCC_DIAG.
21714 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
21716 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
21717 * substring-locations.c: Include intl.h.
21718 (format_warning_va): Turned into small wrapper around
21719 format_warning_n_va, renamed to ...
21720 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
21721 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
21723 (format_warning_at_substring_n): New function.
21724 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
21725 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
21726 format_warning_at_substring with just a shorter name instead of
21727 const function pointer.
21728 (fmtwarn_n): New function.
21729 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
21730 appropriate, get rid of all the fmtstr temporaries, move conditionals
21731 with G_() wrapped string literals directly into fmtwarn arguments,
21732 cast dir.len to (int), formatting fixes.
21734 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
21736 * doc/invoke.texi: Remove "Cilk Plus" references.
21738 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21739 Richard Biener <rguenther@suse.de>
21742 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
21743 for error or warning attributes if CALL_FROM_THUNK_P is set.
21746 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21749 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
21750 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
21753 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
21754 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
21756 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
21757 instead of -1U in last predictors element's probability member.
21759 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
21762 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
21763 arguments if they are comparable.
21765 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
21767 PR tree-optimization/84634
21768 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
21769 masks and masked_loop_p with a single loop_masks, making sure it's
21770 null for bb vectorization.
21772 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
21774 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
21775 (vect_analyze_data_ref_access): Use loop->safe_len rather than
21776 loop->force_vectorize to check whether there is no alias.
21778 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21781 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
21783 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
21785 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
21786 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
21787 instead of a loop around prev_real_insn.
21788 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
21791 PR inline-asm/84625
21792 * config/i386/i386.c (ix86_print_operand): Use conditional
21793 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
21796 2018-03-02 Richard Biener <rguenther@suse.de>
21798 PR tree-optimization/84427
21799 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
21800 (bitmap_set_subtract_values): Rewrite to handle multiple
21803 (prune_clobbered_mems): Likewise.
21804 (phi_translate): Take edge instead of pred/phiblock.
21805 (phi_translate_1): Likewise.
21806 (phi_translate_set): Likewise. Insert all translated
21807 exprs for a value into the set, keeping possibly multiple
21808 expressions per value.
21809 (compute_antic_aux): Adjust for phi_translate changes.
21810 When intersecting union the expressions and prune those
21811 not in the final value set, keeping possibly multiple
21812 expressions per value. Do not use value-insertion
21813 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
21814 all expressions. Add verification that the value-sets
21815 only shrink during iteration.
21816 (compute_partial_antic_aux): Adjust for the phi_translate changes.
21817 (do_pre_regular_insertion): Likewise.
21818 (do_pre_partial_partial_insertion): Likewise.
21820 2018-03-02 Richard Biener <rguenther@suse.de>
21823 * config/darwin.c (saved_debug_info_level): New static global.
21824 (darwin_asm_lto_start): Disable debug info generation for LTO out.
21825 (darwin_asm_lto_end): Restore debug info generation settings.
21827 2018-03-01 Martin Liska <mliska@suse.cz>
21830 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
21831 volatile arguments.
21833 2018-03-01 Richard Biener <rguenther@suse.de>
21836 * dwarf2out.c (gen_variable_die): Properly handle late VLA
21837 type annotation with LTO when debug was disabled at compile-time.
21839 2018-03-01 Matthew Fortune <mfortune@gmail.com>
21841 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
21843 (mips_final_postscan_insn): Likewise.
21845 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
21847 PR rtl-optimization/84528
21848 * alias.c (init_alias_target): Add commentary.
21849 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
21850 a unique base value if the frame pointer is not eliminated
21851 to the stack pointer.
21853 2018-03-01 Tom de Vries <tom@codesourcery.com>
21855 PR rtl-optimization/83327
21856 * lra-int.h (hard_regs_spilled_into): Declare.
21857 * lra.c (hard_regs_spilled_into): Define.
21858 (init_reg_info): Init hard_regs_spilled_into.
21859 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
21860 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
21861 (process_bb_lives): Handle hard_regs_spilled_into.
21862 (lra_create_live_ranges_1): Before doing liveness propagation, clear
21863 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
21865 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
21867 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
21868 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
21869 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
21870 * config/rs6000/aix72.h: New file.
21872 2018-02-28 Jakub Jelinek <jakub@redhat.com>
21874 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
21875 instead of warning_at with conditional singular and plural messages
21879 * stor-layout.c (update_alignment_for_field): For
21880 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
21881 && !DECL_PACKED (field), do the alignment update, just use
21882 only desired_align instead of MAX (type_align, desired_align)
21884 (place_field): Don't do known_align < desired_align handling
21885 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
21886 is non-NULL, instead do it after rli->prev_field handling and
21887 only if not within a bitfield word. For DECL_PACKED (field)
21888 use type_align of BITS_PER_UNIT.
21890 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
21892 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
21893 superfluous parentheses and trailing spaces.
21895 2018-02-28 Richard Biener <rguenther@suse.de>
21897 PR tree-optimization/84584
21898 * graphite-scop-detection.c (scop_detection::add_scop): Discard
21899 SCoPs with fake exit edge.
21901 2018-02-28 Martin Liska <mliska@suse.cz>
21904 * timevar.c (timer::print): Fix format to properly print 100%
21907 2018-02-28 Richard Biener <rguenther@suse.de>
21909 PR middle-end/84607
21910 * genmatch.c (capture_info::walk_match): Do not mark
21911 captured expressions without operands as expr_p given
21912 they act more like predicates and should be subject to
21913 "lost tail" side-effect preserving.
21915 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
21917 PR rtl-optimization/81611
21918 * auto-inc-dec.c (attempt_change): Move dead note from
21919 mem_insn if it's the next use of regno
21920 (find_address): Take address use of reg holding
21921 non-incremented value. Add parm to limit search to the named
21923 (merge_in_block): Attempt to use a mem insn that is the next
21924 use of the original regno.
21926 2018-02-27 Martin Sebor <msebor@redhat.com>
21929 * doc/invoke.texi (-Wmissing-attributes): New option.
21930 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
21932 2018-02-27 Martin Sebor <msebor@redhat.com>
21934 PR translation/84207
21935 * diagnostic-core.h (warning_n, error_n, inform_n): Change
21936 n argument to unsigned HOST_WIDE_INT.
21937 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
21938 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
21939 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
21940 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
21942 2018-02-27 Richard Biener <rguenther@suse.de>
21944 PR tree-optimization/84512
21945 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
21946 Do not use the estimate returned from record_stmt_cost for
21947 the scalar iteration cost but sum properly using add_stmt_cost.
21949 2018-02-27 Richard Biener <rguenther@suse.de>
21951 PR tree-optimization/84466
21952 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
21953 Adjust last change to less strictly validate use operands.
21955 2018-02-27 Martin Liska <mliska@suse.cz>
21957 PR gcov-profile/84548
21958 * gcov.c (process_file): Allow partial overlap and consider it
21959 also as group functions.
21960 (output_lines): Properly calculate range of lines for a group.
21962 2018-02-27 Martin Liska <mliska@suse.cz>
21964 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
21965 'ggc' suffixes. Change first column width.
21966 (timer::print): Fix formatting of the column.
21968 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
21970 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
21971 preserve inline entry blocks for the sake of debug inline
21972 entry point markers alone.
21973 (remove_unused_locals): Suggest in comments a better place to
21974 force the preservation of inline entry blocks that are
21975 otherwise unused, but do not preserve them.
21977 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
21979 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
21981 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
21984 * config/i386/constraints.md (Bs): Replace
21985 ix86_indirect_branch_register with
21986 TARGET_INDIRECT_BRANCH_REGISTER.
21988 * config/i386/i386.md (indirect_jump): Likewise.
21989 (tablejump): Likewise.
21990 (*sibcall_memory): Likewise.
21991 (*sibcall_value_memory): Likewise.
21992 Peepholes of indirect call and jump via memory: Likewise.
21993 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
21994 (*sibcall_value_GOT_32): Likewise.
21995 * config/i386/predicates.md (indirect_branch_operand): Likewise.
21996 (GOT_memory_operand): Likewise.
21997 (call_insn_operand): Likewise.
21998 (sibcall_insn_operand): Likewise.
21999 (GOT32_symbol_operand): Likewise.
22000 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
22002 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
22004 PR rtl-optimization/83496
22005 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
22006 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
22007 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
22008 redundant insn, if any.
22009 (relax_delay_slots): Likewise.
22010 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
22012 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
22014 PR tree-optimization/83965
22015 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
22016 that grouped statements are part of a reduction chain. Return
22017 true if the statement is not marked as a reduction itself but
22018 is part of a group.
22019 (vect_recog_dot_prod_pattern): Don't check whether the statement
22020 is part of a group here.
22021 (vect_recog_sad_pattern): Likewise.
22022 (vect_recog_widen_sum_pattern): Likewise.
22024 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
22027 * final.c (rest_of_clean_state): Also look for calls inside sequences.
22029 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
22032 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
22034 (ix86_output_indirect_function_return): New prototype.
22035 (ix86_split_simple_return_pop_internal): Likewise.
22036 * config/i386/i386.c (indirect_return_via_cx): New.
22037 (indirect_return_via_cx_bnd): Likewise.
22038 (indirect_thunk_name): Handle return va CX_REG.
22039 (output_indirect_thunk_function): Create alias for
22040 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
22041 (ix86_output_indirect_jmp): Remove the bool argument.
22042 (ix86_output_indirect_function_return): New function.
22043 (ix86_split_simple_return_pop_internal): Likewise.
22044 * config/i386/i386.md (*indirect_jump): Don't pass false
22045 to ix86_output_indirect_jmp.
22046 (*tablejump_1): Likewise.
22047 (simple_return_pop_internal): Change it to define_insn_and_split.
22048 Call ix86_split_simple_return_pop_internal to split it for
22049 -mfunction-return=.
22050 (simple_return_indirect_internal): Call
22051 ix86_output_indirect_function_return instead of
22052 ix86_output_indirect_jmp.
22054 2018-02-26 Jakub Jelinek <jakub@redhat.com>
22057 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
22058 memset and value initialization afterwards.
22060 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
22062 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
22064 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22067 * common/config/aarch64/aarch64-common.c
22068 (aarch_option_optimization_table[]): Switch
22069 off fomit-frame-pointer
22071 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
22072 Chung-Ju Wu <jasonwucj@gmail.com>
22074 * config/nds32/nds32-multiple.md (load_multiple): Disallow
22076 (store_multiple): Ditto.
22078 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
22080 * config.gcc: Add --with-cpu support for nds32 target.
22081 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
22082 * config/nds32/nds32.opt: Add -mcpu= option.
22084 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
22086 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
22087 isel=yes): Warn for these deprecated options.
22089 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
22091 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
22092 ISA_2_5_MASKS_EMBEDDED.
22094 2018-02-23 Jakub Jelinek <jakub@redhat.com>
22096 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
22097 p->max as pointers rather than using iterative_hash_expr.
22099 2018-02-23 Carl Love <cel@us.ibm.com>
22101 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
22102 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
22104 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
22105 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
22106 P8V_BUILTIN_VEC_VUNSIGNED2.
22108 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
22111 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
22112 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
22114 * lra-constraints.c (curr_insn_transform): Set up
22115 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
22117 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
22118 * lra-eliminations.c (spill_pseudos): Ditto.
22119 (process_insn_for_elimination): Ditto.
22120 * lra-lives.c (reg_early_clobber_p): Use the new macros.
22121 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
22122 LRA_NON_CLOBBERED_ALT.
22124 2018-02-22 Martin Sebor <msebor@redhat.com>
22126 PR tree-optimization/84480
22127 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
22128 to maybe_diag_stxncpy_trunc. Call it.
22129 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
22130 from gimple_fold_builtin_strcpy. Print inlining stack.
22131 (handle_builtin_stxncpy): Print inlining stack.
22132 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
22134 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
22137 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
22138 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
22139 and -fcheck-pointer-bounds are used together.
22140 (indirect_thunk_prefix): New enum.
22141 (indirect_thunk_need_prefix): New function.
22142 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
22143 "_nt" instead of "_bnd" for NOTRACK prefix.
22144 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
22145 (output_indirect_thunk_function): Likewise.
22147 (ix86_code_end): Update output_indirect_thunk_function calls.
22148 (ix86_output_indirect_branch_via_reg): Replace
22149 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
22150 (ix86_output_indirect_branch_via_push): Likewise.
22151 (ix86_output_function_return): Likewise.
22152 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
22153 incompatible with -fcf-protection=branch and
22154 -fcheck-pointer-bounds.
22156 2018-02-22 Steve Ellcey <sellcey@cavium.com>
22159 * config/aarch64/aarch64.c (aarch64_print_address_internal):
22160 Change gcc_assert call to output_operand_lossage.
22162 2018-02-22 Steve Ellcey <sellcey@cavium.com>
22164 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
22166 2018-02-22 DJ Delorie <dj@redhat.com>
22167 Sebastian Perta <sebastian.perta@renesas.com>
22168 Oleg Endo <olegendo@gcc.gnu.org>
22170 * config/rx/rx.c (rx_rtx_costs): New function.
22171 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
22173 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
22175 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
22177 2018-02-22 Martin Liska <mliska@suse.cz>
22180 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
22181 Add "native" as a possible value.
22183 2018-02-22 Martin Liska <mliska@suse.cz>
22186 * config/i386/i386.c (ix86_option_override_internal):
22187 Add "native" as a possible value for -march and -mtune.
22189 2018-02-22 Jakub Jelinek <jakub@redhat.com>
22192 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
22193 to all type variants.
22195 PR tree-optimization/84503
22196 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
22197 width as info->bitpos + info->bitsize - start.
22198 (merged_store_group::merge_overlapping): Simplify width computation.
22199 (check_no_overlap): New function.
22200 (imm_store_chain_info::try_coalesce_bswap): Compute expected
22201 start + width and last_order of the group, fail if check_no_overlap
22203 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
22204 to group if check_no_overlap fails.
22206 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
22208 * config/rs6000/altivec.md: Delete contraint arguments to
22209 define_expand, define_split, and define_peephole2, and in
22210 define_insn_and_split if always unused.
22211 * config/rs6000/darwin.md: Ditto.
22212 * config/rs6000/dfp.md: Ditto.
22213 * config/rs6000/rs6000.md: Ditto.
22214 * config/rs6000/sync.md: Ditto.
22215 * config/rs6000/vector.md: Ditto.
22216 * config/rs6000/vsx.md: Ditto.
22218 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
22220 * config/rs6000/altivec.md: Write output control strings as braced
22221 blocks instead of double-quoted strings.
22222 * config/rs6000/darwin.md: Ditto.
22223 * config/rs6000/rs6000.md: Ditto.
22224 * config/rs6000/vector.md: Ditto.
22225 * config/rs6000/vsx.md: Ditto.
22227 2018-02-21 Jason Merrill <jason@redhat.com>
22229 PR c++/84314 - ICE with templates and fastcall attribute.
22230 * attribs.c (build_type_attribute_qual_variant): Remove assert.
22232 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
22234 * ipa-cp.c (determine_versionability): Fix comment typos.
22236 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
22239 * ipa-cp.c (determine_versionability): Do not version functions caling
22242 2018-02-21 Martin Liska <mliska@suse.cz>
22245 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
22246 Add "native" as a possible value.
22247 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
22248 the macro when native cpu detection is available.
22250 2018-02-21 Martin Liska <mliska@suse.cz>
22253 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
22254 Add "native" as a possible value.
22255 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
22256 when native cpu detection is available.
22258 2018-02-21 Jakub Jelinek <jakub@redhat.com>
22259 Martin Sebor <msebor@redhat.com>
22261 PR tree-optimization/84478
22262 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
22264 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
22265 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
22266 support which is conservatively correct, for 2 only stay conservative
22267 for maxlen. Formatting and comment capitalization fixes. Add STRICT
22268 argument to the 2 argument get_range_strlen, adjust 6 arg
22269 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
22271 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
22272 (gimple_fold_builtin_strlen): Pass true as last argument to
22275 2018-02-20 Martin Sebor <msebor@redhat.com>
22277 PR middle-end/84095
22278 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
22279 (builtin_memref::set_base_and_offset): Same. Handle inner references.
22280 (builtin_memref::builtin_memref): Factor out parts into
22281 set_base_and_offset and call it.
22283 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
22285 PR middle-end/84406
22286 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
22287 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
22288 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
22289 search at the associated MODE_INT.
22291 2018-02-20 Jeff Law <law@redhat.com>
22293 PR middle-end/82123
22294 PR tree-optimization/81592
22295 PR middle-end/79257
22296 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
22297 for range data rather than using global data.
22298 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
22299 range data rather than using global data.
22300 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
22301 pass it to children as needed.
22302 (struct directive::fmtresult): Similarly.
22303 (struct directive::set_width): Similarly.
22304 (struct directive::set_precision): Similarly.
22305 (format_integer, format_directive, parse_directive): Similarly.
22306 (format_none): Accept unnamed vr_values parameter.
22307 (format_percent, format_floating, format_character): Similarly.
22308 (format_string, format_plain): Similarly.
22309 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
22310 the EVRP range analyzer for range data rather than using global data.
22311 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
22312 gimple-ssa-evrp-analyze.h
22313 (class sprintf_dom_walker): Add after_dom_children member function.
22314 Add evrp_range_analyzer member.
22315 (sprintf_dom_walker::before_dom_children): Call into the EVRP
22316 range analyzer as needed.
22317 (sprintf_dom_walker::after_dom_children): New member function.
22318 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
22320 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
22321 (evrp_range_analyzer::pop_to_marker): Likewise.
22323 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
22325 PR tree-optimization/84419
22326 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
22327 with the required type if its current type is compatible but
22330 2018-02-20 Jakub Jelinek <jakub@redhat.com>
22332 PR middle-end/82004
22333 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
22334 after vectorization.
22336 2018-02-20 Martin Liska <mliska@suse.cz>
22339 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
22340 possible values if we don't have a hint.
22342 2018-02-20 Martin Liska <mliska@suse.cz>
22346 * final.c (shorten_branches): Build align_tab array with one
22348 * opts.c (finish_options): Add alignment option limit check.
22349 (MAX_CODE_ALIGN): Likewise.
22350 (MAX_CODE_ALIGN_VALUE): Likewise.
22351 * doc/invoke.texi: Document maximum allowed option value for
22352 all -falign-* options.
22354 2018-02-19 Jakub Jelinek <jakub@redhat.com>
22357 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
22358 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
22359 * var-tracking.c (emit_note_insn_var_location): Remove all references
22360 to NOTE_INSN_CALL_ARG_LOCATION.
22361 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
22362 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
22363 Use copy_rtx_if_shared.
22364 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
22365 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
22366 (dwarf2out_var_location): Remove handling of
22367 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
22369 * final.c (final_scan_insn): Remove all references to
22370 NOTE_INSN_CALL_ARG_LOCATION.
22371 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
22372 before dumping final insns.
22373 * except.c (emit_note_eh_region_end): Remove all references to
22374 NOTE_INSN_CALL_ARG_LOCATION.
22375 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
22376 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
22377 * config/arc/arc.c (hwloop_optimize): Likewise.
22378 * config/arm/arm.c (create_fix_barrier): Likewise.
22379 * config/s390/s390.c (s390_chunkify_start): Likewise.
22380 * config/sh/sh.c (find_barrier): Likewise.
22381 * config/i386/i386.c (rest_of_insert_endbranch,
22382 ix86_seh_fixup_eh_fallthru): Likewise.
22383 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
22384 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
22385 * config/frv/frv.c (frv_function_prologue): Likewise.
22386 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
22388 (note_outside_basic_block_p): Remove all references to
22389 NOTE_INSN_CALL_ARG_LOCATION.
22390 * gengtype.c (adjust_field_rtx_def): Likewise.
22391 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
22393 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
22394 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
22397 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
22400 PR tree-optimization/84452
22401 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
22402 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
22405 2018-02-19 Martin Liska <mliska@suse.cz>
22408 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
22410 2018-02-19 Martin Liska <mliska@suse.cz>
22411 Richard Sandiford <richard.sandiford@linaro.org>
22413 PR tree-optimization/82491
22414 * gimple-fold.c (get_base_constructor): Make earlier bail out
22417 2018-02-19 Carl Love <cel@us.ibm.com>
22419 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
22420 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
22422 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
22423 P8V_BUILTIN_VEC_NEG.
22425 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
22427 * config/rl78/rl78.md (movdf): New define expand.
22429 2018-02-19 Martin Liska <mliska@suse.cz>
22432 * doc/invoke.texi: Fix typo.
22433 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
22435 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
22437 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
22438 handle rs6000_single_float and rs6000_double_float specially for
22441 2018-02-16 Jeff Law <law@redhat.com>
22443 * config/rx/rx.c (add_pop_cfi_notes): New function.;
22444 (pop_regs): Use it.
22446 2018-02-16 Jakub Jelinek <jakub@redhat.com>
22449 * ipa-inline.c (inline_small_functions): Fix a typo.
22451 2018-02-16 Nathan Sidwell <nathan@acm.org>
22453 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
22455 2018-02-16 Carl Love <cel@us.ibm.com>
22457 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
22458 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
22459 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
22460 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
22461 expansion to P8V_BUILTIN_VEC_FLOAT2.
22463 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
22465 PR rtl-optimization/70023
22466 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
22467 src_regno into account.
22469 2018-02-16 Carl Love <cel@us.ibm.com>
22471 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
22472 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
22473 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
22474 * config/rs6000/rs6000.c: Remove case statements for
22475 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
22476 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
22477 and P9V_BUILTIN_VEC_VINSERT4B.
22478 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
22479 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
22480 * config/rs6000/vsx.md:
22481 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
22484 2018-02-16 Carl Love <cel@us.ibm.com>
22486 * config/rs6000/altivec.h: Add builtin names vec_extract4b
22488 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
22490 * config/rs6000/rs6000-c.c: Add the definitions for
22491 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
22492 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
22493 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
22494 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
22495 definition for insert4b and define insn *insert3b_internal.
22496 * doc/extend.texi: Add documentation for vec_extract4b.
22498 2018-02-16 Nathan Sidwell <nathan@acm.org>
22500 * doc/extend.texi (Backwards Compatibility): Mention friend
22501 injection. Note for-scope is deprecated.
22502 * doc/invoke.texi (-ffriend-injection): Deprecate.
22504 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
22506 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
22507 that moved to I2, also allow destinations that are a paradoxical
22508 subreg (instead of a normal reg).
22510 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
22513 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
22516 2018-02-16 Richard Biener <rguenther@suse.de>
22518 PR tree-optimization/84037
22519 PR tree-optimization/84016
22521 * config/i386/i386.c (ix86_builtin_vectorization_cost):
22522 Adjust vec_construct for the fact we need additional higher latency
22523 128bit inserts for AVX256 and AVX512 vector builds.
22524 (ix86_add_stmt_cost): Scale vector construction cost for
22527 2018-02-16 Richard Biener <rguenther@suse.de>
22529 PR tree-optimization/84417
22530 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
22531 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
22532 (non_rewritable_lvalue_p): Likewise, use poly-ints.
22534 2018-02-16 Martin Liska <mliska@suse.cz>
22537 * internal-fn.def (ASAN_CHECK): Set proper flags.
22538 (ASAN_MARK): Likewise.
22540 2018-02-16 Julia Koval <julia.koval@intel.com>
22542 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
22543 from PTA_CANNONLAKE.
22545 2018-02-16 Jakub Jelinek <jakub@redhat.com>
22548 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
22549 Use ++iter rather than iter++ for std::list iterators.
22550 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
22551 defer deleting them until all nodes in the forest are processed. Do
22552 free even leaf nodes. Change to_process into auto_vec.
22555 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
22556 * vec.h (vec_default_construct): Use memset instead of placement new
22557 if BROKEN_VALUE_INITIALIZATION is defined.
22558 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
22559 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
22562 PR rtl-optimization/83723
22563 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
22564 * lra.c (lra_substitute_pseudo): Likewise. If true, use
22565 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
22567 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
22569 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
22571 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
22573 PR rtl-optimization/81443
22574 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
22575 from inner REGs to paradoxical SUBREGs.
22577 2018-02-16 Richard Biener <rguenther@suse.de>
22579 PR tree-optimization/84399
22580 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
22581 For operands we can analyze at their definition make sure we can
22582 analyze them at each use as well.
22584 2018-02-16 Richard Biener <rguenther@suse.de>
22586 PR tree-optimization/84190
22587 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
22588 volatile accesses if the decl isn't volatile.
22590 2018-02-15 Jason Merrill <jason@redhat.com>
22592 PR c++/84314 - ICE with templates and fastcall attribute.
22593 * attribs.c (build_type_attribute_qual_variant): Don't clobber
22594 TYPE_CANONICAL on an existing type.
22596 2018-02-15 Jakub Jelinek <jakub@redhat.com>
22598 PR tree-optimization/84383
22599 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
22600 dstoff nor call operand_equal_p if dstbase is NULL.
22602 PR tree-optimization/84334
22603 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
22604 also a CONSTANT_CLASS_P, punt.
22606 2018-02-14 Jim Wilson <jimw@sifive.com>
22608 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
22609 first SMALL_OPERAND check. New local min_second_step. Move assert
22610 to where locals are set. Add TARGET_RVC support.
22611 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
22613 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
22615 * doc/invoke.texi: Correct -Wformat-overflow code sample.
22617 2018-02-14 Martin Sebor <msebor@redhat.com>
22619 PR tree-optimization/83698
22620 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
22621 arrays constrain the offset range to their bounds.
22622 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
22623 (builtin_access::overlap): Avoid setting the size of overlap if it's
22625 (maybe_diag_overlap): Also consider arrays when deciding what values
22626 of offsets to include in diagnostics.
22628 2018-02-14 Martin Sebor <msebor@redhat.com>
22631 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
22632 that correspond to the kind of a declaration.
22634 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
22637 * config/pa/pa.md: Load address of PIC label using the linkage table
22638 if the label is nonlocal.
22640 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
22642 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
22643 warning message if user requests -maltivec=be.
22644 * doc/invoke.texi: Document deprecation of -maltivec=be.
22646 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
22649 * config/rs6000/rs6000-c.c: Update definitions for
22650 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
22651 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
22653 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
22656 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
22657 add _get_ssp intrinsics. Remove argument from
22658 __builtin_ia32_rdssp[d|q].
22659 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
22660 * config/i386/i386-builtin.def: Remove argument from
22661 __builtin_ia32_rdssp[d|q].
22662 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
22663 ix86_expand_special_args_builtin for _rdssp[d|q].
22664 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
22665 Clear register before usage.
22666 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
22667 Add documentation for new _get_ssp and _inc_ssp intrinsics.
22669 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
22671 PR tree-optimization/84357
22672 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
22673 operand 1 of an ARRAY_REF too.
22675 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
22678 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
22679 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
22681 (set_of_reg): New struct.
22682 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
22683 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
22684 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
22686 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
22687 Split into bitclr, bitset, bitinvert patterns if appropriate.
22688 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
22689 use rx_fuse_in_memory_bitop.
22690 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
22691 to named insn, correct maximum insn length.
22693 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
22696 * machmode.def: Define a complex mode for PARTIAL_INT.
22697 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
22699 * doc/rtl.texi: Document CSPImode.
22700 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
22702 (msp430_hard_regno_nregs_with_padding): Likewise.
22704 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
22707 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
22709 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
22711 PR rtl-optimization/84169
22712 * combine.c (try_combine): New variable split_i2i3. Set it to true if
22713 we generated a parallel as new i3 and we split that to new i2 and i3
22714 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
22715 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
22716 those to i2, not i1. Partially rewrite this scan code.
22718 2018-02-13 Jakub Jelinek <jakub@redhat.com>
22721 * stor-layout.c (place_field): For variable length fields, adjust
22722 offset_align afterwards not just based on the field's alignment,
22723 but also on the size.
22725 PR middle-end/84309
22726 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
22727 of exps and logs in the use_exp2 case.
22729 2018-02-13 Jeff Law <law@redhat.com>
22731 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
22732 entry for "vector".
22734 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
22737 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
22741 * common.opt (gas-loc-support, gas-locview-support): New.
22742 (ginline-points, ginternal-reset-location-views): New.
22743 * doc/invoke.texi: Document them. Use @itemx where intended.
22744 (gvariable-location-views): Adjust.
22745 * target.def (reset_location_view): New.
22746 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
22747 (TARGET_RESET_LOCATION_VIEW): New.
22748 * doc/tm.texi: Rebuilt.
22749 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
22750 (dwarf2out_default_as_locview_support): New.
22751 (output_asm_line_debug_info): Use option variables.
22752 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
22753 (output_loc_list): Likewise.
22754 (add_high_low_attributes): Check option variables.
22755 Don't output entry view attribute in strict mode.
22756 (gen_inlined_subroutine_die): Check option variables.
22757 (dwarf2out_inline_entry): Likewise.
22758 (init_sections_and_labels): Likewise.
22759 (dwarf2out_early_finish): Likewise.
22760 (maybe_reset_location_view): New, from...
22761 (dwarf2out_var_location): ... here. Call it.
22762 * debug.h (dwarf2out_default_as_loc_support): Declare.
22763 (dwarf2out_default_as_locview_support): Declare.
22764 * hooks.c (hook_int_rtx_insn_0): New.
22765 * hooks.h (hook_int_rtx_insn_0): Declare.
22766 * toplev.c (process_options): Take -gas-loc-support and
22767 -gas-locview-support from dwarf2out. Enable
22768 -gvariable-location-views by default only with locview
22769 assembler support. Enable -ginternal-reset-location-views by
22770 default only if the target defines the corresponding hook.
22771 Enable -ginline-points by default if location views are
22772 enabled; force it disabled if statement frontiers are
22774 * tree-inline.c (expand_call_inline): Check option variables.
22775 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
22777 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
22779 PR tree-optimization/84321
22780 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
22781 handling. Also check whether the anti-range contains any values
22782 that satisfy the mask; switch to a VR_RANGE if not.
22784 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
22787 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
22789 2018-02-13 Martin Jambor <mjambor@suse.cz>
22792 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
22793 of call statements, also set location of a load to a temporary.
22795 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
22797 * config/rl78/rl78.c (add_vector_labels): New function.
22798 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
22799 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
22800 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
22801 which checks that no arguments are passed.
22802 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
22803 * doc/extend.texi: Documentation for the new attribute.
22805 2018-02-13 Andreas Schwab <schwab@suse.de>
22807 * config/riscv/linux.h (CPP_SPEC): Define.
22809 2018-02-13 Jakub Jelinek <jakub@redhat.com>
22812 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
22813 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
22814 OPTION_MASK_ISA_AES as first argument to def_builtin_const
22815 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
22816 instead of OPTION_MASK_ISA_PCLMUL as first argument to
22817 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
22818 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
22819 temporarily for AES and PCLMUL builtins.
22821 PR tree-optimization/84339
22822 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
22823 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
22826 PR middle-end/84309
22827 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
22828 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
22829 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
22831 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
22833 * omp-simd-clone.h: New file.
22834 * omp-simd-clone.c: Include omp-simd-clone.h.
22835 (expand_simd_clones): No longer static.
22836 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
22837 cgraph.h and omp-simd-clone.h.
22838 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
22839 (vect_recog_widen_shift_pattern): Formatting fix.
22840 (vect_pattern_recog_1): Don't check optab for calls.
22843 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
22844 operands[2] into a REG before using gen_lowpart on it.
22846 2018-02-12 Jeff Law <law@redhat.com>
22849 * config/sh/sh.c (find_barrier): Consider a sibling call
22852 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
22853 successfully back substituting a reg.
22855 2018-02-12 Richard Biener <rguenther@suse.de>
22857 PR tree-optimization/84037
22858 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
22859 parameter, move visited init to caller.
22860 (vect_slp_analyze_operations): Separate cost from validity
22861 check, initialize visited once for all instances.
22862 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
22864 * tree-vect-stmts.c (vect_model_simple_cost): Make early
22866 (vect_model_promotion_demotion_cost): Likewise.
22867 (vectorizable_bswap): Guard cost modeling with !slp_node
22868 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
22870 (vectorizable_call): Likewise.
22871 (vectorizable_conversion): Likewise.
22872 (vectorizable_assignment): Likewise.
22873 (vectorizable_shift): Likewise.
22874 (vectorizable_operation): Likewise.
22875 (vectorizable_store): Likewise.
22876 (vectorizable_load): Likewise.
22877 (vectorizable_condition): Likewise.
22878 (vectorizable_comparison): Likewise.
22880 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
22883 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
22884 (ASAN_MARK): Fix fnspec to account for return value, change pointer
22885 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
22887 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
22889 PR middle-end/83665
22890 * params.def (inline-min-speedup): Increase from 8 to 15.
22891 (max-inline-insns-auto): Decrease from 40 to 30.
22892 * ipa-split.c (consider_split): Add some buffer for function to
22893 be considered inlining candidate.
22894 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
22897 2018-02-12 Richard Biener <rguenther@suse.de>
22899 PR tree-optimization/84037
22900 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
22901 matched stmts if we cannot swap the non-matched ones.
22903 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
22905 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
22906 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
22907 _mm_maskz_scalef_round_ss): New intrinsics.
22908 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
22909 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
22910 __builtin_ia32_scalefss_round): Remove.
22911 (__builtin_ia32_scalefsd_mask_round,
22912 __builtin_ia32_scalefss_mask_round): New intrinsics.
22913 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
22914 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
22915 ((match_operand:VF_128 2 "<round_nimm_predicate>"
22916 "<round_constraint>")): Changed to ...
22917 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
22918 "<round_scalar_constraint>")): ... this.
22919 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
22920 %0, %1, %2<round_op3>}"): Changed to ...
22921 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
22922 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
22923 %2<round_scalar_mask_op3>}"): ... this.
22924 * config/i386/subst.md (round_scalar_nimm_predicate): New.
22926 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
22928 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
22929 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
22930 (_mm_maskz_sqrt_round_ss): New intrinsics.
22931 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
22932 (__builtin_ia32_sqrtsd_mask_round)
22933 (__builtin_ia32_sqrtss_mask_round): New builtins.
22934 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
22935 (__builtin_ia32_sqrtss_round): Remove.
22936 (__builtin_ia32_sqrtsd_mask_round)
22937 (__builtin_ia32_sqrtss_mask_round): New builtins.
22938 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
22939 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
22940 ((match_operand:VF_128 1 "vector_operand"
22941 "xBm,<round_constraint>")): Changed to ...
22942 ((match_operand:VF_128 1 "vector_operand"
22943 "xBm,<round_scalar_constraint>")): ... this.
22944 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
22945 %0, %2, %<iptr>1<round_op3>}): Changed to ...
22946 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
22947 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
22948 %<iptr>1<round_scalar_mask_op3>}): ... this.
22949 ((set_attr "prefix" "<round_prefix>")): Changed to ...
22950 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
22952 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
22955 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
22956 Cast vec_cmpeq result to correct type.
22957 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
22958 Cast vec_cmpgt result to correct type.
22960 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
22962 * final.c (final_scan_insn_1): Renamed from...
22963 (final_scan_insn): ... this. New wrapper, to recover
22964 seen from the outermost call in recursive ones.
22965 * config/sparc/sparc.c (output_return): Drop seen from call.
22966 (output_sibcall): Likewise.
22967 * config/visium/visium.c (output_branch): Likewise.
22969 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
22971 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
22974 2018-02-10 Alan Modra <amodra@gmail.com>
22977 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
22978 Specify LR as an input.
22980 2018-02-10 Jakub Jelinek <jakub@redhat.com>
22983 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
22984 remove_member_access_dummy_vars): New functions.
22985 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
22986 lower_omp_1, execute_lower_omp): Use them.
22988 PR rtl-optimization/84308
22989 * shrink-wrap.c (spread_components): Release todo vector.
22991 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
22993 PR rtl-optimization/57193
22994 * ira-color.c (struct allocno_color_data): Add member
22995 conflict_allocno_hard_prefs.
22996 (update_conflict_allocno_hard_prefs): New.
22997 (bucket_allocno_compare_func): Add a preference based on
22998 conflict_allocno_hard_prefs.
22999 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
23000 (color_allocnos): Remove a dead code. Initiate
23001 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
23003 2018-02-09 Jakub Jelinek <jakub@redhat.com>
23006 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
23007 constraint from =wa to wa. Avoid a subreg on the output operand,
23008 instead use a pseudo and subreg it in a move.
23009 (p9_xxbrd_<mode>): Changed to ...
23010 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
23011 (p9_xxbrd_v2df): New expander.
23012 (p9_xxbrw_<mode>): Changed to ...
23013 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
23014 (p9_xxbrw_v4sf): New expander.
23016 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
23018 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
23020 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
23023 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
23024 multiply in 32-bit mode.
23025 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
23026 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
23029 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
23031 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
23032 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
23033 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
23034 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
23036 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
23039 * dwarf2out.c (is_trivial_indirect_ref): New function.
23040 (dwarf2out_late_global_decl): Do not generate a location
23041 attribute for variables that have a non-trivial DECL_VALUE_EXPR
23042 and that are not defined in the current unit.
23044 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
23046 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
23047 instead of a libcall for UNORDERED.
23049 2018-02-09 Tamar Christina <tamar.christina@arm.com>
23052 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
23053 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
23055 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23058 * config/s390/s390.c (s390_set_current_function): Invoke
23059 s390_indirect_branch_settings also if fndecl didn't change.
23061 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
23063 * config/rs6000/rs6000.md (blockage): Set length to zero.
23065 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
23067 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
23069 2018-02-09 Jakub Jelinek <jakub@redhat.com>
23072 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
23073 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
23077 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
23078 PARALLEL incoming that failed vt_get_decl_and_offset check.
23080 PR middle-end/84237
23081 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
23082 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
23084 (get_variable_section): For decls in named .bss* sections pass true as
23085 second argument to bss_initializer_p.
23087 2018-02-09 Marek Polacek <polacek@redhat.com>
23088 Jakub Jelinek <jakub@redhat.com>
23091 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
23092 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
23093 Sync some changes from cxx_fold_indirect_ref.
23095 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
23097 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
23099 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
23100 (BLOCK_INLINE_ENTRY_LABEL): New.
23101 (dwarf2out_var_location): Disregard inline entry markers.
23102 (inline_entry_data): New struct.
23103 (inline_entry_data_hasher): New hashtable type.
23104 (inline_entry_data_hasher::hash): New.
23105 (inline_entry_data_hasher::equal): New.
23106 (inline_entry_data_table): New variable.
23107 (add_high_low_attributes): Add DW_AT_entry_pc and
23108 DW_AT_GNU_entry_view attributes if a pending entry is found
23109 in inline_entry_data_table. Add old entry_pc attribute only
23110 if debug nonbinding markers are disabled.
23111 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
23112 markers are enabled.
23113 (block_within_block_p, dwarf2out_inline_entry): New.
23114 (dwarf2out_finish): Check that no entries remained in
23115 inline_entry_data_table.
23116 * final.c (reemit_insn_block_notes): Handle inline entry notes.
23117 (final_scan_insn, notice_source_line): Likewise.
23118 (rest_of_clean_state): Skip inline entry markers.
23119 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
23121 * gimple.c (gimple_build_debug_inline_entry): New.
23122 * gimple.h (enum gimple_debug_subcode): Add
23123 GIMPLE_DEBUG_INLINE_ENTRY.
23124 (gimple_build_debug_inline_entry): Declare.
23125 (gimple_debug_inline_entry_p): New.
23126 (gimple_debug_nonbind_marker_p): Adjust.
23127 * insn-notes.def (INLINE_ENTRY): New.
23128 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
23129 inline entry marker notes.
23130 (print_insn): Likewise.
23131 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
23132 (INSN_DEBUG_MARKER_KIND): Likewise.
23133 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
23134 * tree-inline.c (expand_call_inline): Build and insert
23135 debug_inline_entry stmt.
23136 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
23137 inline entry blocks early, if nonbind markers are enabled.
23138 (dump_scope_block): Dump fragment info.
23139 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
23140 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
23141 (gimple_build_debug_inline_entry): New.
23142 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
23143 Enable/disable inline entry points too.
23144 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
23145 (DEBUG_INSN): Describe inline entry markers.
23147 * common.opt (gvariable-location-views): New.
23148 (gvariable-location-views=incompat5): New.
23149 * config.in: Rebuilt.
23150 * configure: Rebuilt.
23151 * configure.ac: Test assembler for view support.
23152 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
23153 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
23154 * dwarf2out.c (var_loc_view): New typedef.
23155 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
23156 (dwarf2out_locviews_in_attribute): New.
23157 (dwarf2out_locviews_in_loclist): New.
23158 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
23159 (enum dw_line_info_opcode): Add LI_adv_address.
23160 (struct dw_line_info_table): Add view.
23161 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
23162 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
23163 (zero_view_p): New variable.
23164 (ZERO_VIEW_P): New macro.
23165 (output_asm_line_debug_info): New.
23166 (struct var_loc_node): Add view.
23167 (add_AT_view_list, AT_loc_list): New.
23168 (add_var_loc_to_decl): Add view param. Test it against last.
23169 (new_loc_list): Add view params. Record them.
23170 (AT_loc_list_ptr): Handle loc and view lists.
23171 (view_list_to_loc_list_val_node): New.
23172 (print_dw_val): Handle dw_val_class_view_list.
23173 (size_of_die): Likewise.
23174 (value_format): Likewise.
23175 (loc_list_has_views): New.
23176 (gen_llsym): Set vl_symbol too.
23177 (maybe_gen_llsym, skip_loc_list_entry): New.
23178 (dwarf2out_maybe_output_loclist_view_pair): New.
23179 (output_loc_list): Output view list or entries too.
23180 (output_view_list_offset): New.
23181 (output_die): Handle dw_val_class_view_list.
23182 (output_dwarf_version): New.
23183 (output_compilation_unit_header): Use it.
23184 (output_skeleton_debug_sections): Likewise.
23185 (output_rnglists, output_line_info): Likewise.
23186 (output_pubnames, output_aranges): Update version comments.
23187 (output_one_line_info_table): Output view numbers in asm comments.
23188 (dw_loc_list): Determine current endview, pass it to new_loc_list.
23189 Call maybe_gen_llsym.
23190 (loc_list_from_tree_1): Adjust.
23191 (add_AT_location_description): Create view list attribute if
23192 needed, check it's absent otherwise.
23193 (convert_cfa_to_fb_loc_list): Adjust.
23194 (maybe_emit_file): Call output_asm_line_debug_info for test.
23195 (dwarf2out_var_location): Reset views as needed. Precompute
23196 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
23197 attribute. Set view.
23198 (new_line_info_table): Reset next view.
23199 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
23200 (dwarf2out_source_line): Likewise. Output view resets and labels to
23201 the assembler, or select appropriate line info opcodes.
23202 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
23203 (optimize_string_length): Catch it. Adjust.
23204 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
23205 dw_val_class_view_list, and remove it if no longer needed.
23206 (hash_loc_list): Hash view numbers.
23207 (loc_list_hasher::equal): Compare them.
23208 (optimize_location_lists): Check whether a view list symbol is
23209 needed, and whether the locview attribute is present, and
23210 whether they match. Remove the locview attribute if no longer
23212 (index_location_lists): Call skip_loc_list_entry for test.
23213 (dwarf2out_finish): Call output_asm_line_debug_info for test.
23214 Use output_dwarf_version.
23215 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
23216 (struct dw_val_node): Add val_view_list.
23217 * final.c (SEEN_NEXT_VIEW): New.
23218 (set_next_view_needed): New.
23219 (clear_next_view_needed): New.
23220 (maybe_output_next_view): New.
23221 (final_start_function): Rename to...
23222 (final_start_function_1): ... this. Take pointer to FIRST,
23223 add SEEN parameter. Emit param bindings in the initial view.
23224 (final_start_function): Reintroduce SEEN-less interface.
23225 (final): Rename to...
23226 (final_1): ... this. Take SEEN parameter. Output final pending
23227 next view at the end.
23228 (final): Reintroduce seen-less interface.
23229 (final_scan_insn): Output pending next view before switching
23230 sections or ending a block. Mark the next view as needed when
23231 outputting variable locations. Notify debug backend of section
23232 changes, and of location view changes.
23233 (rest_of_handle_final): Adjust.
23234 * toplev.c (process_options): Autodetect value for debug variable
23235 location views option. Warn on incompat5 without -gdwarf-5.
23236 * doc/invoke.texi (gvariable-location-views): New.
23237 (gvariable-location-views=incompat5): New.
23238 (gno-variable-location-views): New.
23240 2018-02-08 David Malcolm <dmalcolm@redhat.com>
23242 PR tree-optimization/84136
23243 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
23244 that the result of find_edge is non-NULL.
23246 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
23249 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
23250 storing integer register in SImode. Fix cost of 256 and 512
23251 byte aligned SSE register store.
23253 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
23255 * config/i386/i386.c (ix86_multiplication_cost): Fix
23256 multiplication cost for TARGET_AVX512DQ.
23258 2018-02-08 Marek Polacek <polacek@redhat.com>
23260 PR tree-optimization/84238
23261 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
23264 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23266 PR tree-optimization/84265
23267 * tree-vect-stmts.c (vectorizable_store): Don't treat
23268 VMAT_CONTIGUOUS accesses as grouped.
23269 (vectorizable_load): Likewise.
23271 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23273 PR tree-optimization/81635
23274 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
23275 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
23276 (test_round_for_mask): New functions.
23277 (wide_int_cc_tests): Call test_round_for_mask.
23278 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
23279 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
23280 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
23281 range returned by get_range_info.
23283 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
23286 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
23287 * symtab.c: Include builtins.h
23288 (symtab_node::output_to_lto_symbol_table_p): Move here
23289 from lto-streamer-out.c:output_symbol_p.
23290 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
23291 (output_symbol_p): Move all logic to symtab.c
23292 (produce_symtab): Update.
23294 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23296 * config/s390/s390-opts.h (enum indirect_branch): Define.
23297 * config/s390/s390-protos.h (s390_return_addr_from_memory)
23298 (s390_indirect_branch_via_thunk)
23299 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
23300 (enum s390_indirect_branch_type): Define.
23301 * config/s390/s390.c (struct s390_frame_layout, struct
23302 machine_function): Remove.
23303 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
23304 (indirect_branch_table_label_no, indirect_branch_table_name):
23306 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
23307 (enum s390_indirect_branch_option): Define.
23308 (s390_return_addr_from_memory): New function.
23309 (s390_handle_string_attribute): New function.
23310 (s390_attribute_table): Add new attribute handler.
23311 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
23312 (s390_indirect_branch_via_thunk): New function.
23313 (s390_indirect_branch_via_inline_thunk): New function.
23314 (s390_function_ok_for_sibcall): When jumping via thunk disallow
23315 sibling call optimization for non z10 compiles.
23316 (s390_emit_call): Force indirect branch target to be a single
23317 register. Add r1 clobber for non-z10 compiles.
23318 (s390_emit_epilogue): Emit return jump via return_use expander.
23319 (s390_reorg): Handle JUMP_INSNs as execute targets.
23320 (s390_option_override_internal): Perform validity checks for the
23321 new command line options.
23322 (s390_indirect_branch_attrvalue): New function.
23323 (s390_indirect_branch_settings): New function.
23324 (s390_set_current_function): Invoke s390_indirect_branch_settings.
23325 (s390_output_indirect_thunk_function): New function.
23326 (s390_code_end): Implement target hook.
23327 (s390_case_values_threshold): Implement target hook.
23328 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
23330 * config/s390/s390.h (struct s390_frame_layout)
23331 (struct machine_function): Move here from s390.c.
23332 (TARGET_INDIRECT_BRANCH_NOBP_RET)
23333 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
23334 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
23335 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
23336 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
23337 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
23338 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
23339 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
23340 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
23341 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
23342 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
23343 (mnemonic attribute): Add values which aren't recognized
23345 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
23346 pattern for branch conversion. Fix mnemonic attribute.
23347 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
23348 indirect branch via thunk if requested.
23349 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
23350 ("*indirect_jump"): Disable for branch conversion using out of
23352 ("indirect_jump_via_thunk<mode>_z10")
23353 ("indirect_jump_via_thunk<mode>")
23354 ("indirect_jump_via_inlinethunk<mode>_z10")
23355 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
23356 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
23357 ("casesi_jump_via_inlinethunk<mode>_z10")
23358 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
23359 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
23360 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
23361 ("*indirect2_jump"): Disable for branch conversion.
23362 ("casesi_jump"): Turn into expander and expand patterns for branch
23364 ("return_use"): New expander.
23365 ("*return"): Emit return via thunk and rename it to ...
23366 ("*return<mode>"): ... this one.
23367 * config/s390/s390.opt: Add new options and and enum for the
23370 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23372 * lra-constraints.c (match_reload): Unconditionally use
23373 gen_lowpart_SUBREG, rather than selecting between that
23374 and equivalent gen_rtx_SUBREG code.
23376 2018-02-08 Richard Biener <rguenther@suse.de>
23378 PR tree-optimization/84233
23379 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
23380 changed flag instead of boguously re-using phi_inserted.
23382 2018-02-08 Martin Jambor <mjambor@suse.cz>
23384 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
23385 static local variables.
23387 2018-02-08 Richard Biener <rguenther@suse.de>
23389 PR tree-optimization/84278
23390 * tree-vect-stmts.c (vectorizable_store): When looking for
23391 smaller vector types to perform grouped strided loads/stores
23392 make sure the mode is supported by the target.
23393 (vectorizable_load): Likewise.
23395 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
23397 * config/aarch64/aarch64.c (aarch64_components_for_bb):
23398 Increase LDP/STP opportunities by adding adjacent callee-saves.
23400 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
23402 PR rtl-optimization/84068
23403 PR rtl-optimization/83459
23404 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
23406 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
23408 PR tree-optimization/84224
23409 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
23410 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
23411 non-zero arguments.
23413 2018-02-07 Iain Sandoe <iain@codesourcery.com>
23416 * config/rs6000/altivec.md (*restore_world): Remove LR use.
23417 * config/rs6000/predicates.md (restore_world_operation): Adjust op
23418 count, remove one USE.
23420 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23422 * doc/install.texi (Configuration): Document the
23423 --with-long-double-format={ibm,ieee} PowerPC configuration
23427 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
23428 Convert from define_expand to be define_insn_and_split. Rework
23429 float/double/_Float128 conversions to QI/HI/SImode to work with
23430 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
23431 conversions to QI/HImode types did a store and then a load to
23432 truncate the value. For conversions to VSX registers, don't split
23433 the insn, instead emit the code directly. Use the code iterator
23434 any_fix to combine signed and unsigned conversions.
23435 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
23436 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
23437 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
23438 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
23439 (fix_<mode>di2_hw): Likewise.
23440 (fixuns_<mode>di2_hw): Likewise.
23441 (fix_<mode>si2_hw): Likewise.
23442 (fixuns_<mode>si2_hw): Likewise.
23443 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
23444 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
23445 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
23446 fix<uns>_trunc<SFDF:mode>si2_p8.
23447 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
23448 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23449 (fix<uns>_<mode>_mem): Likewise.
23450 (fctiw<u>z_<mode>_mem): Likewise.
23451 (fix<uns>_<mode>_mem): Likewise.
23452 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
23453 the register allocator from doing a direct move to the GPRs to do
23454 a store, and instead use the ISA 3.0 store byte/half-word from
23455 vector register instruction. For IEEE 128-bit floating point,
23456 also optimize stores of 32-bit ints.
23457 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
23459 2018-02-07 Alan Hayward <alan.hayward@arm.com>
23461 * genextract.c (push_pathstr_operand): New function to support
23463 (walk_rtx): Call push_pathstr_operand.
23464 (print_path): Support [a-zA-Z].
23466 2018-02-07 Richard Biener <rguenther@suse.de>
23468 PR tree-optimization/84037
23469 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
23470 (cse_and_gimplify_to_preheader): Declare.
23471 (vect_get_place_in_interleaving_chain): Likewise.
23472 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
23474 (_loop_vec_info::~_loop_vec_info): Delete it.
23475 (cse_and_gimplify_to_preheader): New function.
23476 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
23477 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
23478 (vectorizable_load): Likewise. For grouped stores always base
23479 the IV on the first element.
23480 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
23481 condition before gimplifying.
23483 2018-02-07 Jakub Jelinek <jakub@redhat.com>
23485 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
23486 *DIV_EXPR and *MOD_EXPR.
23488 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
23491 * config/i386/i386.c (ix86_option_override_internal): Mask out
23492 the CF_SET bit when checking -fcf-protection.
23494 2018-02-07 Tom de Vries <tom@codesourcery.com>
23497 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
23500 2018-02-07 Richard Biener <rguenther@suse.de>
23502 PR tree-optimization/84204
23503 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
23506 PR tree-optimization/84205
23507 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
23508 special-case isl_ast_op_zdiv_r.
23510 PR tree-optimization/84223
23511 * graphite-scop-detection.c (gather_bbs::before_dom_children):
23512 Only add conditions from within the region.
23513 (gather_bbs::after_dom_children): Adjust.
23515 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
23518 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
23519 * config/avr/avr.md: Only post-reload split REG-REG moves if
23520 either register is GENERAL_REG_P.
23522 2018-02-07 Jakub Jelinek <jakub@redhat.com>
23524 PR tree-optimization/84235
23525 * tree-ssa-scopedtables.c
23526 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
23527 if the subtraction is performed in floating point type where NaNs are
23528 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
23529 build 1. Formatting fix.
23531 2018-02-06 Jakub Jelinek <jakub@redhat.com>
23534 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
23535 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
23536 and skip it regardless of bb boundaries. Use CALL_P macro,
23537 don't test INSN_P (insn) together with CALL_P or JUMP_P check
23538 unnecessarily, formatting fix.
23540 2018-02-06 Michael Collison <michael.collison@arm.com>
23542 * config/arm/thumb2.md:
23543 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
23544 (*thumb_mov_notscc): Ditto.
23546 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
23549 * config/rs6000/rs6000.md (su code attribute): Use "u" for
23550 unsigned_fix, not "s".
23552 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23554 * configure.ac (gcc_fn_eh_frame_ro): New function.
23555 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
23556 correct .eh_frame permissions.
23557 * configure: Regenerate.
23559 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
23561 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
23562 irrelevant options.
23564 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23566 * config/rs6000/rs6000.c (rs6000_option_override_internal):
23567 Display warning message for -mno-speculate-indirect-jumps.
23569 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
23571 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
23573 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
23575 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
23577 PR tree-optimization/84225
23578 * tree-eh.c (find_trapping_overflow): Only call
23579 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
23581 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
23584 * config/i386/i386.c: Reimplement the check of possible options
23585 -mibt/-mshstk conbination. Change error messages.
23586 * doc/invoke.texi: Fix a typo: remove extra '='.
23588 2018-02-06 Marek Polacek <polacek@redhat.com>
23590 PR tree-optimization/84228
23591 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
23593 2018-02-06 Tamar Christina <tamar.christina@arm.com>
23596 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
23597 emitted arch directives.
23598 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
23599 __ARM_FEATURE_COPROC before changing architectures.
23601 2018-02-06 Richard Biener <rguenther@suse.de>
23603 * config/i386/i386.c (print_reg): Fix typo.
23604 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
23606 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
23608 * configure: Regenerate.
23610 2018-02-05 Martin Sebor <msebor@redhat.com>
23612 PR tree-optimization/83369
23613 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
23616 2018-02-05 Martin Liska <mliska@suse.cz>
23618 * doc/invoke.texi: Cherry-pick upstream r323995.
23620 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
23622 * ira.c (ira_init_register_move_cost): Adjust comment.
23624 2018-02-05 Martin Liska <mliska@suse.cz>
23626 PR gcov-profile/84137
23627 * doc/gcov.texi: Fix typo in documentation.
23629 2018-02-05 Martin Liska <mliska@suse.cz>
23631 PR gcov-profile/83879
23632 * doc/gcov.texi: Document necessity of --dynamic-list-data when
23633 using dlopen functionality.
23635 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
23637 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
23638 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
23639 _mm_maskz_range_ss, _mm_mask_range_round_ss,
23640 _mm_maskz_range_round_ss): New intrinsics.
23641 (__builtin_ia32_rangesd128_round)
23642 (__builtin_ia32_rangess128_round): Remove.
23643 (__builtin_ia32_rangesd128_mask_round,
23644 __builtin_ia32_rangess128_mask_round): New builtins.
23645 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
23646 __builtin_ia32_rangess128_round): Remove.
23647 (__builtin_ia32_rangesd128_mask_round,
23648 __builtin_ia32_rangess128_mask_round): New builtins.
23649 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
23650 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
23651 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
23652 "<round_saeonly_constraint>")): Changed to ...
23653 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
23654 "<round_saeonly_scalar_constraint>")): ... this.
23655 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
23656 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
23657 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
23658 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
23659 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
23661 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
23663 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
23665 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
23666 Remove all values except native, 8540 and 8548.
23668 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
23670 * config/i386/i386.c (ix86_output_function_return): Pass
23671 INVALID_REGNUM, instead of -1, as invalid register number to
23672 indirect_thunk_name and output_indirect_thunk.
23674 2018-02-02 Julia Koval <julia.koval@intel.com>
23676 * config.gcc: Add -march=icelake.
23677 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
23678 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
23679 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
23680 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
23681 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
23682 (processor_target_table): Add icelake.
23683 (ix86_option_override_internal): Handle new PTAs.
23684 (get_builtin_code_for_version): Handle icelake.
23685 (M_INTEL_COREI7_ICELAKE): New.
23686 (fold_builtin_cpu): Handle icelake.
23687 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
23688 * doc/invoke.texi: Add -march=icelake.
23690 2018-02-02 Julia Koval <julia.koval@intel.com>
23692 * config/i386/i386.c (ix86_option_override_internal): Change flags type
23693 to wide_int_bitmask.
23694 * wide-int-bitmask.h: New.
23696 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
23699 * config/i386/i386.md: Replace Pmode with word_mode in
23700 builtin_setjmp_setup and builtin_longjmp to support x32.
23702 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
23706 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
23708 (rs6000_supported_cpu_names): New static variable.
23709 (linux_cpu_translation_table): Likewise.
23710 (elf_platform) <cpu>: Define new static variable and use it.
23711 Translate kernel AT_PLATFORM name to canonical name if needed.
23712 Error if platform name is unknown.
23714 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
23717 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
23719 2018-02-01 Jeff Law <law@redhat.com>
23722 * config/i386/i386.c (release_scratch_register_on_entry): Add new
23723 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
23724 the scratch if RELEASE_VIA_POP is false.
23725 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
23726 If we have to save a temporary register, decrement SIZE appropriately.
23727 Pass new arguments to release_scratch_register_on_entry.
23728 (ix86_adjust_stack_and_probe): Likewise.
23729 (ix86_emit_probe_stack_range): Pass new arguments to
23730 release_scratch_register_on_entry.
23732 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
23734 PR rtl-optimization/84157
23735 * combine.c (change_zero_ext): Use REG_P predicate in
23736 front of HARD_REGISTER_P predicate.
23738 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
23740 * config/avr/avr.c (avr_option_override): Move disabling of
23741 -fdelete-null-pointer-checks to...
23742 * common/config/avr/avr-common.c (avr_option_optimization_table):
23745 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23747 PR tree-optimization/81635
23748 * tree-data-ref.c (split_constant_offset_1): For types that
23749 wrap on overflow, try to use range info to prove that wrapping
23752 2018-02-01 Renlin Li <renlin.li@arm.com>
23755 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
23756 TAILCALL_ADDR_REGS.
23757 (aarch64_register_move_cost): Likewise.
23758 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
23759 TAILCALL_ADDR_REGS.
23760 (REG_CLASS_NAMES): Likewise.
23761 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
23762 TAILCALL_ADDR_REGS. Remove IP registers.
23763 * config/aarch64/aarch64.md (Ucs): Update register constraint.
23765 2018-02-01 Richard Biener <rguenther@suse.de>
23767 * domwalk.h (dom_walker::dom_walker): Add additional constructor
23768 for specifying RPO order and allow NULL for that.
23769 * domwalk.c (dom_walker::dom_walker): Likewise.
23770 (dom_walker::walk): Handle NULL RPO order.
23771 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
23773 (rewrite_update_dom_walker): Likewise.
23774 (mark_def_dom_walker): Likewise.
23776 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23778 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
23779 (aarch64_maybe_expand_sve_subreg_move): Declare.
23780 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
23781 * config/aarch64/predicates.md (aarch64_any_register_operand): New
23783 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
23784 that are semantically a reverse operation.
23785 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
23786 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
23787 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
23789 (aarch64_can_change_mode_class): For big-endian, forbid changes
23790 between two SVE modes if they have different element sizes.
23792 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23794 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
23795 the TImode handling for big-endian targets.
23797 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23799 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
23800 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
23802 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
23803 Remove BSWAP handing for big-endian targets and use the form of
23804 LD1RQ appropariate for the mode.
23806 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23808 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
23809 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
23810 duplicated element.
23812 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23815 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
23816 check for operands that need to go through aarch64_sve_reload_be.
23818 2018-02-01 Jakub Jelinek <jakub@redhat.com>
23820 PR tree-optimization/81661
23821 PR tree-optimization/84117
23822 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
23823 * tree-eh.c: Include gimplify.h.
23824 (find_trapping_overflow, replace_trapping_overflow,
23825 rewrite_to_non_trapping_overflow): New functions.
23826 * tree-vect-loop.c: Include tree-eh.h.
23827 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
23828 * tree-data-ref.c: Include tree-eh.h.
23829 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
23831 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
23833 PR rtl-optimization/84123
23834 * combine.c (change_zero_ext): Check if hard register satisfies
23835 can_change_dest_mode before calling gen_lowpart_SUBREG.
23837 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
23840 * ira.c (ira_init_register_move_cost): Remove assert.
23842 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
23844 PR rtl-optimization/84071
23845 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
23846 * doc/tm.texi: Regenerate.
23848 2018-01-31 Richard Biener <rguenther@suse.de>
23850 PR tree-optimization/84132
23851 * tree-data-ref.c (analyze_miv_subscript): Properly
23852 check whether evolution_function_is_affine_multivariate_p
23853 before calling gcd_of_steps_may_divide_p.
23855 2018-01-31 Julia Koval <julia.koval@intel.com>
23858 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
23859 * config/i386/i386.md (rdpid_rex64) New.
23860 (rdpid): Make 32bit only.
23862 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
23865 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
23866 an IDENTIFIER_NODE for FUNCTION_TYPE's.
23868 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
23871 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
23873 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
23875 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
23877 PR rtl-optimization/84071
23878 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
23879 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
23881 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
23883 * config/arc/arc.c (arc_handle_aux_attribute): New function.
23884 (arc_attribute_table): Add 'aux' attribute.
23885 (arc_in_small_data_p): Consider aux like variables.
23886 (arc_is_aux_reg_p): New function.
23887 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
23888 (arc_get_aux_arg): New function.
23889 (prepare_move_operands): Handle aux-register access.
23890 (arc_handle_aux_attribute): New function.
23891 * doc/extend.texi (ARC Variable attributes): Add subsection.
23893 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
23895 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
23896 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
23897 (arc_attribute_table): Add 'uncached' attribute.
23898 (arc_print_operand): Print '.di' flag for uncached memory
23900 (arc_in_small_data_p): Do not consider for small data the uncached
23902 (arc_is_uncached_mem_p): New function.
23903 * config/arc/predicates.md (compact_store_memory_operand): Check
23904 for uncached memory accesses.
23905 (nonvol_nonimm_operand): Likewise.
23906 * doc/extend.texi (ARC Type Attribute): New subsection.
23908 2018-01-31 Jakub Jelinek <jakub@redhat.com>
23911 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
23912 falign-loops=): Add Optimization flag.
23914 2018-01-30 Jeff Law <law@redhat.com>
23917 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
23918 INT_REGISTERS_SAVED. Check it prior to calling
23919 get_scratch_register_on_entry.
23920 (ix86_adjust_stack_and_probe): Similarly.
23921 (ix86_emit_probe_stack_range): Similarly.
23922 (ix86_expand_prologue): Corresponding changes.
23924 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23927 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
23928 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
23930 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
23933 * lra-constraints.c (curr_insn_transform): Process AND in the
23936 2018-01-30 Jakub Jelinek <jakub@redhat.com>
23938 PR rtl-optimization/83986
23939 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
23940 dependence against last_pending_memory_flush in addition to
23941 pending_jump_insns.
23943 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
23945 PR tree-optimization/81611
23946 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
23949 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23952 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
23955 2018-01-30 Richard Biener <rguenther@suse.de>
23956 Jakub Jelinek <jakub@redhat.com>
23958 PR tree-optimization/84111
23959 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
23960 inner loops added during recursion, as they don't have up-to-date
23963 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
23966 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
23967 (can_inline_edge_by_limits_p): ... here.
23968 (can_early_inline_edge_p, check_callers,
23969 update_caller_keys, update_callee_keys, recursive_inlining,
23970 add_new_edges_to_heap, speculation_useful_p,
23971 inline_small_functions,
23972 inline_small_functions, flatten_function,
23973 inline_to_all_callers_1): Update.
23975 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
23977 * profile-count.c (profile_count::combine_with_ipa_count): Handle
23980 2018-01-30 Richard Biener <rguenther@suse.de>
23982 PR tree-optimization/83008
23983 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
23984 invariant and constant vector uses in stmts when they need
23985 more than one stmt.
23987 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23990 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
23991 * configure: Regenerate.
23993 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
23995 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
23997 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
23998 Use gen_rtx_REG rather than gen_lowpart.
24000 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24002 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
24003 rather than 0 when creating partial subregs.
24005 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24007 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
24010 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
24013 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
24014 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
24015 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
24016 flags. This restores the settings used before the 2017-07-24.
24017 Turning off pre increment/decrement/modify allows IVOPTS to
24018 optimize DF/SF loops where the index is an int.
24020 2018-01-29 Richard Biener <rguenther@suse.de>
24021 Kelvin Nilsen <kelvin@gcc.gnu.org>
24024 * tree-vect-stmts.c (vectorizable_call): Don't call
24025 targetm.vectorize_builtin_md_vectorized_function if callee is
24028 2018-01-22 Carl Love <cel@us.ibm.com>
24030 * doc/extend.tex: Fix typo in second arg in
24031 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
24033 2018-01-29 Richard Biener <rguenther@suse.de>
24035 PR tree-optimization/84086
24036 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
24037 (flush_ssaname_freelist): When SSA names were released reset
24038 the SCEV hash table.
24040 2018-01-29 Richard Biener <rguenther@suse.de>
24042 PR tree-optimization/84057
24043 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
24044 removed paths when removing edges.
24046 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
24048 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
24049 -mfunction-return=@var{choice}.
24051 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
24053 PR diagnostic/84034
24054 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
24055 Handle CR like TAB.
24056 (layout::print_source_line): Likewise.
24057 (test_get_line_width_without_trailing_whitespace): Add test cases.
24059 2018-01-27 Jakub Jelinek <jakub@redhat.com>
24061 PR middle-end/84040
24062 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
24065 2018-01-26 Jim Wilson <jimw@sifive.com>
24067 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
24069 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
24072 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24074 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
24075 and CMP + SUB-immediate -> SUBS.
24077 2018-01-26 Martin Sebor <msebor@redhat.com>
24079 PR tree-optimization/83896
24080 * tree-ssa-strlen.c (get_string_len): Rename...
24081 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
24082 Avoid assuming length is constant.
24083 (handle_char_store): Use HOST_WIDE_INT for string length.
24085 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
24088 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
24089 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
24091 2018-01-26 Richard Biener <rguenther@suse.de>
24093 PR rtl-optimization/84003
24094 * dse.c (record_store): Only record redundant stores when
24095 the earlier store aliases at least all accesses the later one does.
24097 2018-01-26 Jakub Jelinek <jakub@redhat.com>
24099 PR rtl-optimization/83985
24100 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
24101 REG_CFA_RESTORE insns.
24102 (delete_unmarked_insns): Don't ignore separate shrink wrapping
24103 REG_CFA_RESTORE insns here.
24106 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
24107 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
24109 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24111 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
24112 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
24113 (arc_init): Likewise.
24114 (arc_override_options): Likewise.
24115 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
24117 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
24119 * config/arc/arc.h (TARGET_DBNZ): Define.
24120 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
24121 properly set the tune attribute.
24122 (dbnz): Use TARGET_DBNZ guard.
24123 * config/arc/arc.opt (mtune): Add core3 option.
24125 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24127 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
24128 recognize new pic like addresses.
24129 (arc_delegitimize_address): Clean up.
24131 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24133 * config/arc/arc-arches.def: Option mrf16 valid for all
24135 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
24136 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
24137 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
24138 * config/arc/arc-tables.opt: Regenerate.
24139 * config/arc/arc.c (arc_conditional_register_usage): Handle
24140 reduced register file case.
24141 (arc_file_start): Set must have build attributes.
24142 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
24143 mrf16 option value.
24144 * config/arc/arc.opt (mrf16): Add new option.
24145 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
24146 * config/arc/genmultilib.awk: Handle new mrf16 option.
24147 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
24148 * config/arc/t-multilib: Regenerate.
24149 * doc/invoke.texi (ARC Options): Document mrf16 option.
24151 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24153 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
24154 * config/arc/arc.c (arc_handle_secure_attribute): New function.
24155 (arc_attribute_table): Add 'secure_call' attribute.
24156 (arc_print_operand): Print secure call operand.
24157 (arc_function_ok_for_sibcall): Don't optimize tail calls when
24159 (arc_is_secure_call_p): New function. * config/arc/arc.md
24160 (call_i): Add support for sjli instruction.
24161 (call_value_i): Likewise.
24162 * config/arc/constraints.md (Csc): New constraint.
24164 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24165 John Eric Martin <John.Martin@emmicro-us.com>
24167 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
24168 * config/arc/arc.c (_arc_jli_section): New struct.
24169 (arc_jli_section): New type.
24170 (rc_jli_sections): New static variable.
24171 (arc_handle_jli_attribute): New function.
24172 (arc_attribute_table): Add jli_always and jli_fixed attribute.
24173 (arc_file_end): New function.
24174 (TARGET_ASM_FILE_END): Define.
24175 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
24176 (arc_add_jli_section): New function.
24177 (jli_call_scan): Likewise.
24178 (arc_reorg): Call jli_call_scan.
24179 (arc_output_addsi): Remove 'S' from printing asm operand.
24180 (arc_is_jli_call_p): New function.
24181 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
24183 (movhi_insn): Likewise.
24184 (movsi_insn): Likewise.
24185 (movsi_set_cc_insn): Likewise.
24186 (loadqi_update): Likewise.
24187 (load_zeroextendqisi_update): Likewise.
24188 (load_signextendqisi_update): Likewise.
24189 (loadhi_update): Likewise.
24190 (load_zeroextendhisi_update): Likewise.
24191 (load_signextendhisi_update): Likewise.
24192 (loadsi_update): Likewise.
24193 (loadsf_update): Likewise.
24194 (movsicc_insn): Likewise.
24195 (bset_insn): Likewise.
24196 (bxor_insn): Likewise.
24197 (bclr_insn): Likewise.
24198 (bmsk_insn): Likewise.
24199 (bicsi3_insn): Likewise.
24200 (cmpsi_cc_c_insn): Likewise.
24201 (movsi_ne): Likewise.
24202 (movsi_cond_exec): Likewise.
24203 (clrsbsi2): Likewise.
24204 (norm_f): Likewise.
24214 (call_i): Remove 'S' asm letter, add jli instruction.
24215 (call_value_i): Likewise.
24216 * config/arc/arc.op (mjli-always): New option.
24217 * config/arc/constraints.md (Cji): New constraint.
24218 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
24220 (subsf3_fpx): Likewise.
24221 (mulsf3_fpx): Likewise.
24222 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
24224 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
24225 function attrbutes.
24226 * doc/invoke.texi (ARC): Document mjli-always option.
24228 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
24230 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
24231 avoid addition with 0 and use incw and decw where possible.
24233 2018-01-26 Richard Biener <rguenther@suse.de>
24235 PR tree-optimization/81082
24236 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
24237 association if it requires casting to unsigned.
24238 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
24239 from fold_plusminus_mult_expr to catch important cases late when
24240 range info is available.
24242 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24244 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
24245 * configure.ac (hidden_linkonce): New test.
24246 * configure: Regenerate.
24247 * config.in: Regenerate.
24249 2018-01-26 Julia Koval <julia.koval@intel.com>
24251 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
24252 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
24253 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
24254 _mm_mask_bitshuffle_epi64_mask): Fix type.
24255 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
24256 USI_FTYPE_V4DI_V4DI_USI): Remove.
24257 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
24258 __builtin_ia32_vpshufbitqmb256_mask,
24259 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
24260 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
24261 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
24263 2018-01-26 Alan Modra <amodra@gmail.com>
24266 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
24267 UNSPEC_VBPERMQ. Sort other unspecs.
24269 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
24271 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
24273 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
24275 PR middle-end/83055
24276 * predict.c (drop_profile): Do not push/pop cfun; update also
24278 (handle_missing_profiles): Fix logic looking for zero profiles.
24280 2018-01-25 Jakub Jelinek <jakub@redhat.com>
24282 PR middle-end/83977
24283 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
24284 on functions with #pragma omp declare simd or functions with simd
24286 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
24287 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
24288 Remove trailing \n from warning_at calls.
24290 2018-01-25 Tom de Vries <tom@codesourcery.com>
24293 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
24294 for neutered workers.
24296 2018-01-24 Joseph Myers <joseph@codesourcery.com>
24299 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
24300 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
24302 2018-01-24 Jeff Law <law@redhat.com>
24305 * i386.c (get_probe_interval): Move to earlier point.
24306 (ix86_compute_frame_layout): If -fstack-clash-protection and
24307 the frame is larger than the probe interval, then use pushes
24308 to save registers rather than reg->mem moves.
24309 (ix86_expand_prologue): Remove conditional for int_registers_saved
24312 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
24315 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
24316 min/max for never referenced object.
24318 2018-01-24 Jakub Jelinek <jakub@redhat.com>
24320 PR middle-end/83977
24321 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
24323 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
24324 attributes from DECL_ATTRIBUTES (decl) without affecting
24325 DECL_ATTRIBUTES (current_function_decl).
24326 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
24327 functions with non-NULL DECL_ABSTRACT_ORIGIN.
24329 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
24331 PR tree-optimization/83979
24332 * fold-const.c (fold_comparison): Use constant_boolean_node
24333 instead of boolean_{true,false}_node.
24335 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
24337 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
24340 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24342 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
24343 Simplify the clause that sets the length attribute.
24344 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
24345 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
24346 clause that sets the length attribute.
24347 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
24349 2018-01-24 Tom de Vries <tom@codesourcery.com>
24352 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
24353 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
24354 Add strict parameter.
24355 (prevent_branch_around_nothing): Insert dummy insn between branch to
24356 label and label with no ptx insn inbetween.
24357 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
24359 2018-01-24 Tom de Vries <tom@codesourcery.com>
24362 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
24363 for neutered threads in warp.
24364 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
24366 2018-01-24 Richard Biener <rguenther@suse.de>
24368 PR tree-optimization/83176
24369 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
24372 2018-01-24 Richard Biener <rguenther@suse.de>
24374 PR tree-optimization/82819
24375 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
24376 code generating pluses that are no-ops in the target precision.
24378 2018-01-24 Richard Biener <rguenther@suse.de>
24380 PR middle-end/84000
24381 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
24383 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
24385 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
24386 to merge probabilities.
24387 * predict.c (probably_never_executed): Also mark as cold functions
24388 with global 0 profile and guessed local profile.
24389 * profile-count.c (profile_probability::combine_with_count): New
24391 * profile-count.h (profile_probability::operator*,
24392 profile_probability::operator*=, profile_probability::operator/,
24393 profile_probability::operator/=): Reduce precision to adjusted
24394 and set value to guessed on contradictory divisions.
24395 (profile_probability::combine_with_freq): Remove.
24396 (profile_probability::combine_wiht_count): Declare.
24397 (profile_count::force_nonzero):: Set to adjusted.
24398 (profile_count::probability_in):: Set quality to adjusted.
24399 * tree-ssa-tail-merge.c (replace_block_by): Use
24400 combine_with_count.
24402 2018-01-23 Andrew Waterman <andrew@sifive.com>
24403 Jim Wilson <jimw@sifive.com>
24405 * config/riscv/riscv.c (riscv_stack_boundary): New.
24406 (riscv_option_override): Set riscv_stack_boundary. Handle
24407 riscv_preferred_stack_boundary_arg.
24408 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
24409 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
24410 (STACK_BOUNDARY): Set to riscv_stack_boundary.
24411 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
24412 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
24413 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
24415 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
24418 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
24419 of struct ix86_frame.
24420 (ix86_expand_epilogue): Likewise. Add a local variable for
24421 the reg_save_offset field in struct ix86_frame.
24423 2018-01-23 Bin Cheng <bin.cheng@arm.com>
24425 PR tree-optimization/82604
24426 * tree-loop-distribution.c (enum partition_kind): New enum item
24427 PKIND_PARTIAL_MEMSET.
24428 (partition_builtin_p): Support above new enum item.
24429 (generate_code_for_partition): Ditto.
24430 (compute_access_range): Differentiate cases that equality can be
24431 proven at all loops, the innermost loops or no loops.
24432 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
24433 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
24434 (finalize_partitions, distribute_loop): Don't fuse partition of
24435 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
24436 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
24437 parloop is enabled.
24439 2018-01-23 Martin Liska <mliska@suse.cz>
24441 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
24442 order to ignore the predictor.
24443 (PRED_POLYMORPHIC_CALL): Likewise.
24444 (PRED_RECURSIVE_CALL): Likewise.
24446 2018-01-23 Martin Liska <mliska@suse.cz>
24448 * tree-profile.c (tree_profiling): Print function header to
24449 aware reader which function we are working on.
24450 * value-prof.c (gimple_find_values_to_profile): Do not print
24451 not interesting value histograms.
24453 2018-01-23 Martin Liska <mliska@suse.cz>
24455 * profile-count.h (enum profile_quality): Add
24456 profile_uninitialized as the first value. Do not number values
24457 as they are zero based.
24458 (profile_count::verify): Update sanity check.
24459 (profile_probability::verify): Likewise.
24461 2018-01-23 Nathan Sidwell <nathan@acm.org>
24463 * doc/invoke.texi (ffor-scope): Deprecate.
24465 2018-01-23 David Malcolm <dmalcolm@redhat.com>
24467 PR tree-optimization/83510
24468 * domwalk.c (set_all_edges_as_executable): New function.
24469 (dom_walker::dom_walker): Convert bool param
24470 "skip_unreachable_blocks" to enum reachability. Move setup of
24471 edge flags to set_all_edges_as_executable and only do it when
24472 reachability is REACHABLE_BLOCKS.
24473 * domwalk.h (enum dom_walker::reachability): New enum.
24474 (dom_walker::dom_walker): Convert bool param
24475 "skip_unreachable_blocks" to enum reachability.
24476 (set_all_edges_as_executable): New decl.
24477 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
24478 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
24480 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
24481 but converting true to REACHABLE_BLOCKS.
24482 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
24484 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
24485 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
24486 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
24488 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
24489 if check_all_array_refs will be called.
24491 2018-01-23 David Malcolm <dmalcolm@redhat.com>
24493 * tree.c (selftest::test_location_wrappers): Add more test
24496 2018-01-23 David Malcolm <dmalcolm@redhat.com>
24498 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
24499 (selftest::test_bit_in_range): Likewise.
24501 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
24504 * doc/sourcebuild.texi (vect_float): Say that the selector
24505 only describes the situation when -funsafe-math-optimizations is on.
24506 (vect_float_strict): Document.
24508 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
24510 PR tree-optimization/83965
24511 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
24512 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
24513 instead of checking only for a reduction.
24514 (vect_recog_widen_sum_pattern): Likewise.
24516 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
24518 * predict.c (probably_never_executed): Only use precise profile info.
24519 (compute_function_frequency): Skip after inlining hack since we now
24520 have quality checking.
24522 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
24524 * profile-count.h (profile_probability::very_unlikely,
24525 profile_probability::unlikely, profile_probability::even): Set
24526 precision to guessed.
24528 2018-01-23 Richard Biener <rguenther@suse.de>
24530 PR tree-optimization/83963
24531 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
24532 Properly terminate dominator walk when crossing the exit edge not
24533 when visiting its source block.
24535 2018-01-23 Jakub Jelinek <jakub@redhat.com>
24538 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
24539 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
24541 2018-01-22 Jakub Jelinek <jakub@redhat.com>
24543 PR tree-optimization/83957
24544 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
24545 semicolon after for body surrounded by braces.
24547 PR tree-optimization/83081
24548 * profile-count.h (profile_probability::split): New method.
24549 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
24550 Use profile_probability::split.
24551 (do_compare_rtx_and_jump): Fix adjustment of probabilities
24552 when splitting a single conditional jump into 2.
24554 2018-01-22 David Malcolm <dmalcolm@redhat.com>
24556 PR tree-optimization/69452
24557 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
24560 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24562 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
24563 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
24564 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
24566 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24568 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
24570 * config/rl78/rl78.md (movdi): New define_expand.
24571 * config/rl78/rl78.c (rl78_split_movdi): New function.
24573 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
24576 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
24578 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
24579 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
24580 128-bit to produce an UNSPEC move to get the double word with the
24581 signbit and then a shift directly to do signbit.
24582 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
24583 implementation with a new version that just does either a direct
24584 move or a regular move. Move memory interface to separate insns.
24585 Move insns so they are next to the expander.
24586 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
24587 with signbit move. Split big and little endian case.
24588 (signbit<mode>2_dm_mem_le): Likewise.
24589 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
24590 (signbit<mode>2_dm2): Likewise.
24592 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24594 * config/rl78/rl78.md (anddi3): New define_expand.
24596 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24598 * config/rl78/rl78.md (umindi3): New define_expand.
24600 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24602 * config/rl78/rl78.md (smindi3): New define_expand.
24604 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24606 * config/rl78/rl78.md (smaxdi3): New define_expand.
24608 2018-01-22 Carl Love <cel@us.ibm.com>
24610 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
24611 LVX_V1TI): Add macro expansion.
24612 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
24613 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
24614 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
24615 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
24616 Change check to determine if the instruction is a byte reversing
24617 entry. Fix typo in comment.
24618 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
24619 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
24620 Add def_builtin calls for new builtins.
24621 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
24622 Add define_insn expansion.
24624 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24626 * config/rl78/rl78.md (umaxdi3): New define_expand.
24628 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24630 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
24631 for non-QImode registers.
24633 2018-01-22 Richard Biener <rguenther@suse.de>
24635 PR tree-optimization/83963
24636 * graphite-scop-detection.c (scop_detection::get_sese): Delay
24637 including the loop exit block.
24638 (scop_detection::merge_sese): Likewise.
24639 (scop_detection::add_scop): Do it here instead.
24641 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24643 * doc/sourcebuild.texi (arm_softfloat): Document.
24645 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
24648 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
24649 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
24650 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
24652 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24653 David Edelsohn <dje.gcc@gmail.com>
24656 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
24657 Change "crset eq" to "crset 2".
24658 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
24659 (*call_indirect_aix<mode>_nospec): Likewise.
24660 (*call_value_indirect_aix<mode>_nospec): Likewise.
24661 (*call_indirect_elfv2<mode>_nospec): Likewise.
24662 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
24663 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
24664 change assembly output from . to $.
24665 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
24666 (indirect_jump<mode>_nospec): Change assembly output from . to $.
24667 (*tablejump<mode>_internal1_nospec): Likewise.
24669 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
24672 * config/sh/sh_optimize_sett_clrt.cc:
24673 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
24675 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
24677 PR tree-optimization/83940
24678 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
24679 offset_dt to vect_constant_def rather than vect_unknown_def_type.
24680 (vect_check_load_store_mask): Add a mask_dt_out parameter and
24681 use it to pass back the definition type.
24682 (vect_check_store_rhs): Likewise rhs_dt_out.
24683 (vect_build_gather_load_calls): Add a mask_dt argument and use
24684 it instead of a call to vect_is_simple_use.
24685 (vectorizable_store): Update calls to vect_check_load_store_mask
24686 and vect_check_store_rhs. Use the dt returned by the latter instead
24687 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
24688 instead of calls to vect_is_simple_use. Pass the scalar rather
24689 than the vector operand to vect_is_simple_use when handling
24690 second and subsequent copies of an rhs value.
24691 (vectorizable_load): Update calls to vect_check_load_store_mask
24692 and vect_build_gather_load_calls. Use the cached mask_dt and
24693 gs_info.offset_dt instead of calls to vect_is_simple_use.
24695 2018-01-20 Jakub Jelinek <jakub@redhat.com>
24697 PR middle-end/83945
24698 * tree-emutls.c: Include gimplify.h.
24699 (lower_emutls_2): New function.
24700 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
24701 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
24702 it before further processing.
24705 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
24706 UINTVAL (trueop1) instead of INTVAL (op1).
24708 2018-01-19 Jakub Jelinek <jakub@redhat.com>
24712 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
24713 INCOMING_FRAME_SP_OFFSET if not defined.
24714 (scan_trace): Add ENTRY argument. If true and
24715 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
24716 emit a note to adjust the CFA offset.
24717 (create_cfi_notes): Adjust scan_trace callers.
24718 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
24719 INCOMING_FRAME_SP_OFFSET in the CIE.
24720 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
24721 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
24723 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
24724 * doc/tm.texi: Regenerated.
24726 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24728 PR rtl-optimization/83147
24729 * lra-constraints.c (remove_inheritance_pseudos): Use
24730 lra_substitute_pseudo_within_insn.
24732 2018-01-19 Tom de Vries <tom@codesourcery.com>
24733 Cesar Philippidis <cesar@codesourcery.com>
24736 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
24738 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
24741 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
24742 spaces for function labels.
24744 2018-01-19 Martin Liska <mliska@suse.cz>
24746 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
24747 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
24748 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
24749 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
24750 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
24751 (PRED_CONST_RETURN): Change from 69 to 65.
24752 (PRED_NULL_RETURN): Change from 91 to 71.
24753 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
24754 (PRED_LOOP_GUARD): Change from 66 to 73.
24756 2018-01-19 Martin Liska <mliska@suse.cz>
24758 * predict.c (predict_insn_def): Add new assert.
24759 (struct branch_predictor): Change type to signed integer.
24760 (test_prediction_value_range): Amend test to cover
24761 PROB_UNINITIALIZED.
24762 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
24763 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
24764 (PRED_LOOP_ITERATIONS_MAX): Likewise.
24765 (PRED_LOOP_IV_COMPARE): Likewise.
24766 * predict.h (PROB_UNINITIALIZED): Define new constant.
24768 2018-01-19 Martin Liska <mliska@suse.cz>
24770 * predict.c (dump_prediction): Add new format for
24771 analyze_brprob.py script which is enabled with -details
24773 * profile-count.h (precise_p): New function.
24775 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
24777 PR tree-optimization/83922
24778 * tree-vect-loop.c (vect_verify_full_masking): Return false if
24779 there are no statements that need masking.
24780 (vect_active_double_reduction_p): New function.
24781 (vect_analyze_loop_operations): Use it when handling phis that
24782 are not in the loop header.
24784 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
24786 PR tree-optimization/83914
24787 * tree-vect-loop.c (vectorizable_induction): Don't convert
24788 init_expr or apply the peeling adjustment for inductions
24789 that are nested within the vectorized loop.
24791 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24793 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
24796 2018-01-18 Jakub Jelinek <jakub@redhat.com>
24800 * function.h (gimplify_parameters): Add gimple_seq * argument.
24801 * function.c: Include gimple.h and options.h.
24802 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
24803 for the added local temporaries if needed.
24804 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
24805 if there are any parameter cleanups, wrap whole body into a
24806 try/finally with the cleanups.
24808 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
24811 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
24812 Use GET_MODE_CLASS for scalar floating point.
24814 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
24818 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
24819 Fix call of call_cgraph_insertion_hooks.
24821 2018-01-18 Martin Sebor <msebor@redhat.com>
24823 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
24825 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
24828 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
24831 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
24834 * common.opt: (-ffile-prefix-map): New option.
24835 * opts.c (common_handle_option): Defer it.
24836 * opts-global.c (handle_common_deferred_options): Handle it.
24837 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
24838 * file-prefix-map.h: New file.
24839 (remap_debug_filename, add_debug_prefix_map): ...here.
24840 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
24841 * final.c (debug_prefix_map, add_debug_prefix_map
24842 remap_debug_filename): Move to...
24843 * file-prefix-map.c: New file.
24844 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
24845 generalize, get rid of alloca(), use strrchr() instead of strchr().
24846 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
24847 Implement in terms of add_prefix_map().
24848 (remap_macro_filename, remap_debug_filename): Implement in term of
24850 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
24851 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
24852 * dbxout.c: Include file-prefix-map.h.
24853 * varasm.c: Likewise.
24854 * vmsdbgout.c: Likewise.
24855 * xcoffout.c: Likewise.
24856 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
24857 * doc/cppopts.texi (-fmacro-prefix-map): Document.
24858 * doc/invoke.texi (-ffile-prefix-map): Document.
24859 (-fdebug-prefix-map): Update description.
24861 2018-01-18 Martin Liska <mliska@suse.cz>
24863 * config/i386/i386.c (indirect_thunk_name): Document that also
24865 (output_indirect_thunk): Document why both instructions
24866 (pause and lfence) are generated.
24868 2018-01-18 Richard Biener <rguenther@suse.de>
24870 PR tree-optimization/83887
24871 * graphite-scop-detection.c
24872 (scop_detection::get_nearest_dom_with_single_entry): Remove.
24873 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
24874 (scop_detection::merge_sese): Re-implement with a flood-fill
24875 algorithm that properly finds a SESE region if it exists.
24877 2018-01-18 Jakub Jelinek <jakub@redhat.com>
24880 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
24881 pointer_diff optimizations use view_convert instead of convert.
24883 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24885 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
24886 Generate different code for -mno-speculate-indirect-jumps.
24887 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
24888 (*call_indirect_aix<mode>): Disable for
24889 -mno-speculate-indirect-jumps.
24890 (*call_indirect_aix<mode>_nospec): New define_insn.
24891 (*call_value_indirect_aix<mode>): Disable for
24892 -mno-speculate-indirect-jumps.
24893 (*call_value_indirect_aix<mode>_nospec): New define_insn.
24894 (*sibcall_nonlocal_sysv<mode>): Generate different code for
24895 -mno-speculate-indirect-jumps.
24896 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
24898 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
24900 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
24901 long double type, set the flags for noting the default long double
24902 type, even if we don't pass or return a long double type.
24904 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
24907 * ipa-inline.c (flatten_function): Do not overwrite final inlining
24910 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
24912 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
24913 support for merge[hl].
24914 (fold_mergehl_helper): New helper function.
24915 (tree-vector-builder.h): New #include for tree_vector_builder usage.
24916 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
24917 (altivec_vmrglw_direct): Add xxmrglw insn.
24919 2018-01-17 Andrew Waterman <andrew@sifive.com>
24921 * config/riscv/riscv.c (riscv_conditional_register_usage): If
24922 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
24924 2018-01-17 David Malcolm <dmalcolm@redhat.com>
24927 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
24928 call the lto_location_cache before reading the
24929 DECL_SOURCE_LOCATION of the types.
24931 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
24932 Richard Sandiford <richard.sandiford@linaro.org>
24934 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
24935 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
24936 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
24937 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
24938 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
24940 * config/aarch64/constraints.md (aarch64_movti_operand):
24942 * config/aarch64/predicates.md (Uti): Add new constraint.
24944 2018-01-17 Carl Love <cel@us.ibm.com>
24946 * config/rs6000/vsx.md (define_expand xl_len_r,
24947 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
24948 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
24950 (define_expand, define_insn): Move the shift left from the
24951 define_insn to the define_expand for lxvl and stxvl instructions.
24952 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
24953 and XL_LEN_R definitions to PURE.
24955 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
24957 * config/i386/i386.c (indirect_thunk_name): Declare regno
24958 as unsigned int. Compare regno with INVALID_REGNUM.
24959 (output_indirect_thunk): Ditto.
24960 (output_indirect_thunk_function): Ditto.
24961 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
24962 in the call to output_indirect_thunk_function.
24964 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
24966 PR middle-end/83884
24967 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
24968 rather than the size of inner_type to determine the stack slot size
24969 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
24971 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
24974 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
24977 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
24979 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
24980 endian Linux systems to optionally enable multilibs for selecting
24981 the long double type if the user configured an explicit type.
24982 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
24983 have no long double multilibs if not defined.
24984 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
24985 warn if the user used -mabi={ieee,ibm}longdouble and we built
24986 multilibs for long double.
24987 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
24988 appropriate multilib option.
24989 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
24991 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
24992 for building long double multilibs.
24993 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
24995 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
24997 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
25000 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
25002 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
25005 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
25008 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
25011 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
25013 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
25014 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
25016 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
25018 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
25019 different rtl trees depending on TARGET_64BIT.
25020 (rs6000_gen_lvx): Likewise.
25022 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
25024 * config/visium/visium.md (nop): Tweak comment.
25025 (hazard_nop): Likewise.
25027 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25029 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
25030 -mspeculate-indirect-jumps.
25031 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
25032 for -mno-speculate-indirect-jumps.
25033 (*call_indirect_elfv2<mode>_nospec): New define_insn.
25034 (*call_value_indirect_elfv2<mode>): Disable for
25035 -mno-speculate-indirect-jumps.
25036 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
25037 (indirect_jump): Emit different RTL for
25038 -mno-speculate-indirect-jumps.
25039 (*indirect_jump<mode>): Disable for
25040 -mno-speculate-indirect-jumps.
25041 (*indirect_jump<mode>_nospec): New define_insn.
25042 (tablejump): Emit different RTL for
25043 -mno-speculate-indirect-jumps.
25044 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
25045 (tablejumpsi_nospec): New define_expand.
25046 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
25047 (tablejumpdi_nospec): New define_expand.
25048 (*tablejump<mode>_internal1): Disable for
25049 -mno-speculate-indirect-jumps.
25050 (*tablejump<mode>_internal1_nospec): New define_insn.
25051 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
25054 2018-01-16 Artyom Skrobov tyomitch@gmail.com
25056 * caller-save.c (insert_save): Drop unnecessary parameter. All
25059 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25060 Richard Biener <rguenth@suse.de>
25063 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
25064 return early, inline manually is_gimple_sizepos. Make sure if we
25065 call gimplify_expr we don't end up with a gimple constant.
25066 * tree.c (variably_modified_type_p): Don't return true for
25067 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
25068 * gimplify.h (is_gimple_sizepos): Remove.
25070 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25072 PR tree-optimization/83857
25073 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
25074 vectorizable_live_operation for pure SLP statements.
25075 (vectorizable_live_operation): Handle PHIs.
25077 2018-01-16 Richard Biener <rguenther@suse.de>
25079 PR tree-optimization/83867
25080 * tree-vect-stmts.c (vect_transform_stmt): Precompute
25081 nested_in_vect_loop_p since the scalar stmt may get invalidated.
25083 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25086 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
25087 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
25088 If off is not INTEGER_CST, issue a may not be aligned warning
25089 rather than isn't aligned. Use isn%'t rather than isn't.
25090 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
25092 <case MULT_EXPR>: Improve the case when bottom and one of the
25093 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
25094 operand, in that case check if the other operand is multiple of
25095 bottom divided by the INTEGER_CST operand.
25097 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25100 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
25101 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
25102 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
25103 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
25104 * config/pa/pa.c (pa_function_arg_advance): Likewise.
25105 (pa_function_arg, pa_arg_partial_bytes): Likewise.
25106 (pa_function_arg_size): New function.
25108 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25110 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
25111 in a separate statement.
25113 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25115 PR tree-optimization/83847
25116 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
25117 group gathers and scatters.
25119 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25121 PR rtl-optimization/86620
25122 * params.def (max-sched-ready-insns): Bump minimum value to 1.
25124 PR rtl-optimization/83213
25125 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
25126 to last if both are JUMP_INSNs.
25128 PR tree-optimization/83843
25129 * gimple-ssa-store-merging.c
25130 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
25131 store_immediate_info for bswap/nop orig_stores.
25133 2018-01-15 Andrew Waterman <andrew@sifive.com>
25135 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
25137 <UDIV>: Increase cost if !TARGET_DIV.
25139 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25141 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
25142 (define_attr "cr_logical_3op"): New.
25143 (cceq_ior_compare): Adjust.
25144 (cceq_ior_compare_complement): Adjust.
25145 (*cceq_rev_compare): Adjust.
25146 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
25147 (is_cracked_insn): Adjust.
25148 (insn_must_be_first_in_group): Adjust.
25149 * config/rs6000/40x.md: Adjust.
25150 * config/rs6000/440.md: Adjust.
25151 * config/rs6000/476.md: Adjust.
25152 * config/rs6000/601.md: Adjust.
25153 * config/rs6000/603.md: Adjust.
25154 * config/rs6000/6xx.md: Adjust.
25155 * config/rs6000/7450.md: Adjust.
25156 * config/rs6000/7xx.md: Adjust.
25157 * config/rs6000/8540.md: Adjust.
25158 * config/rs6000/cell.md: Adjust.
25159 * config/rs6000/e300c2c3.md: Adjust.
25160 * config/rs6000/e500mc.md: Adjust.
25161 * config/rs6000/e500mc64.md: Adjust.
25162 * config/rs6000/e5500.md: Adjust.
25163 * config/rs6000/e6500.md: Adjust.
25164 * config/rs6000/mpc.md: Adjust.
25165 * config/rs6000/power4.md: Adjust.
25166 * config/rs6000/power5.md: Adjust.
25167 * config/rs6000/power6.md: Adjust.
25168 * config/rs6000/power7.md: Adjust.
25169 * config/rs6000/power8.md: Adjust.
25170 * config/rs6000/power9.md: Adjust.
25171 * config/rs6000/rs64.md: Adjust.
25172 * config/rs6000/titan.md: Adjust.
25174 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25176 * config/i386/predicates.md (indirect_branch_operand): Rewrite
25177 ix86_indirect_branch_register logic.
25179 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25181 * config/i386/constraints.md (Bs): Update
25182 ix86_indirect_branch_register check. Don't check
25183 ix86_indirect_branch_register with GOT_memory_operand.
25185 * config/i386/predicates.md (GOT_memory_operand): Don't check
25186 ix86_indirect_branch_register here.
25187 (GOT32_symbol_operand): Likewise.
25189 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25191 * config/i386/predicates.md (constant_call_address_operand):
25192 Rewrite ix86_indirect_branch_register logic.
25193 (sibcall_insn_operand): Likewise.
25195 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25197 * config/i386/constraints.md (Bs): Replace
25198 ix86_indirect_branch_thunk_register with
25199 ix86_indirect_branch_register.
25201 * config/i386/i386.md (indirect_jump): Likewise.
25202 (tablejump): Likewise.
25203 (*sibcall_memory): Likewise.
25204 (*sibcall_value_memory): Likewise.
25205 Peepholes of indirect call and jump via memory: Likewise.
25206 * config/i386/i386.opt: Likewise.
25207 * config/i386/predicates.md (indirect_branch_operand): Likewise.
25208 (GOT_memory_operand): Likewise.
25209 (call_insn_operand): Likewise.
25210 (sibcall_insn_operand): Likewise.
25211 (GOT32_symbol_operand): Likewise.
25213 2018-01-15 Jakub Jelinek <jakub@redhat.com>
25215 PR middle-end/83837
25216 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
25217 type rather than type addr's type points to.
25218 (expand_omp_atomic_mutex): Likewise.
25219 (expand_omp_atomic): Likewise.
25221 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25224 * config/i386/i386.c (output_indirect_thunk_function): Use
25225 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
25226 for __x86_return_thunk.
25228 2018-01-15 Richard Biener <rguenther@suse.de>
25230 PR middle-end/83850
25231 * expmed.c (extract_bit_field_1): Fix typo.
25233 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25236 * config/arm/iterators.md (VF): New mode iterator.
25237 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
25238 Remove integer-related logic from pattern.
25239 (neon_vabd<mode>_3): Likewise.
25241 2018-01-15 Jakub Jelinek <jakub@redhat.com>
25243 PR middle-end/82694
25244 * common.opt (fstrict-overflow): No longer an alias.
25245 (fwrapv-pointer): New option.
25246 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
25247 also for pointer types based on flag_wrapv_pointer.
25248 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
25249 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
25250 opts->x_flag_wrapv got set.
25251 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
25252 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
25253 POINTER_TYPE_OVERFLOW_UNDEFINED.
25254 * match.pd: Likewise in address comparison pattern.
25255 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
25257 2018-01-15 Richard Biener <rguenther@suse.de>
25260 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
25261 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
25262 Reset type names to their identifier if their TYPE_DECL doesn't
25263 have linkage (and thus is used for ODR and devirt).
25264 (save_debug_info_for_decl): Remove.
25265 (save_debug_info_for_type): Likewise.
25266 (add_tree_to_fld_list): Adjust.
25267 * tree-pretty-print.c (dump_generic_node): Make dumping of
25268 type names more robust.
25270 2018-01-15 Richard Biener <rguenther@suse.de>
25272 * BASE-VER: Bump to 8.0.1.
25274 2018-01-14 Martin Sebor <msebor@redhat.com>
25277 * builtins.c (check_access): Avoid warning when the no-warning bit
25280 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
25282 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
25283 * ira-color (allocno_hard_regs_compare): Likewise.
25285 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
25288 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
25289 Use .pushsection/.popsection.
25291 2018-01-14 Martin Sebor <msebor@redhat.com>
25294 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
25296 2018-01-14 Jakub Jelinek <jakub@redhat.com>
25298 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
25299 entry from extra_headers.
25300 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
25301 extra_headers, make the list bitwise identical to the i?86-*-* one.
25303 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25305 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
25306 -mcmodel=large with -mindirect-branch=thunk,
25307 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
25308 -mfunction-return=thunk-extern.
25309 * doc/invoke.texi: Document -mcmodel=large is incompatible with
25310 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
25311 -mfunction-return=thunk and -mfunction-return=thunk-extern.
25313 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25315 * config/i386/i386.c (print_reg): Print the name of the full
25316 integer register without '%'.
25317 (ix86_print_operand): Handle 'V'.
25318 * doc/extend.texi: Document 'V' modifier.
25320 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25322 * config/i386/constraints.md (Bs): Disallow memory operand for
25323 -mindirect-branch-register.
25325 * config/i386/predicates.md (indirect_branch_operand): Likewise.
25326 (GOT_memory_operand): Likewise.
25327 (call_insn_operand): Likewise.
25328 (sibcall_insn_operand): Likewise.
25329 (GOT32_symbol_operand): Likewise.
25330 * config/i386/i386.md (indirect_jump): Call convert_memory_address
25331 for -mindirect-branch-register.
25332 (tablejump): Likewise.
25333 (*sibcall_memory): Likewise.
25334 (*sibcall_value_memory): Likewise.
25335 Disallow peepholes of indirect call and jump via memory for
25336 -mindirect-branch-register.
25337 (*call_pop): Replace m with Bw.
25338 (*call_value_pop): Likewise.
25339 (*sibcall_pop_memory): Replace m with Bs.
25340 * config/i386/i386.opt (mindirect-branch-register): New option.
25341 * doc/invoke.texi: Document -mindirect-branch-register option.
25343 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25345 * config/i386/i386-protos.h (ix86_output_function_return): New.
25346 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
25347 set function_return_type.
25348 (indirect_thunk_name): Add ret_p to indicate thunk for function
25350 (output_indirect_thunk_function): Pass false to
25351 indirect_thunk_name.
25352 (ix86_output_indirect_branch_via_reg): Likewise.
25353 (ix86_output_indirect_branch_via_push): Likewise.
25354 (output_indirect_thunk_function): Create alias for function
25355 return thunk if regno < 0.
25356 (ix86_output_function_return): New function.
25357 (ix86_handle_fndecl_attribute): Handle function_return.
25358 (ix86_attribute_table): Add function_return.
25359 * config/i386/i386.h (machine_function): Add
25360 function_return_type.
25361 * config/i386/i386.md (simple_return_internal): Use
25362 ix86_output_function_return.
25363 (simple_return_internal_long): Likewise.
25364 * config/i386/i386.opt (mfunction-return=): New option.
25365 (indirect_branch): Mention -mfunction-return=.
25366 * doc/extend.texi: Document function_return function attribute.
25367 * doc/invoke.texi: Document -mfunction-return= option.
25369 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25371 * config/i386/i386-opts.h (indirect_branch): New.
25372 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
25373 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
25374 with local indirect jump when converting indirect call and jump.
25375 (ix86_set_indirect_branch_type): New.
25376 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
25377 (indirectlabelno): New.
25378 (indirect_thunk_needed): Likewise.
25379 (indirect_thunk_bnd_needed): Likewise.
25380 (indirect_thunks_used): Likewise.
25381 (indirect_thunks_bnd_used): Likewise.
25382 (INDIRECT_LABEL): Likewise.
25383 (indirect_thunk_name): Likewise.
25384 (output_indirect_thunk): Likewise.
25385 (output_indirect_thunk_function): Likewise.
25386 (ix86_output_indirect_branch_via_reg): Likewise.
25387 (ix86_output_indirect_branch_via_push): Likewise.
25388 (ix86_output_indirect_branch): Likewise.
25389 (ix86_output_indirect_jmp): Likewise.
25390 (ix86_code_end): Call output_indirect_thunk_function if needed.
25391 (ix86_output_call_insn): Call ix86_output_indirect_branch if
25393 (ix86_handle_fndecl_attribute): Handle indirect_branch.
25394 (ix86_attribute_table): Add indirect_branch.
25395 * config/i386/i386.h (machine_function): Add indirect_branch_type
25396 and has_local_indirect_jump.
25397 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
25399 (tablejump): Likewise.
25400 (*indirect_jump): Use ix86_output_indirect_jmp.
25401 (*tablejump_1): Likewise.
25402 (simple_return_indirect_internal): Likewise.
25403 * config/i386/i386.opt (mindirect-branch=): New option.
25404 (indirect_branch): New.
25407 (thunk-inline): Likewise.
25408 (thunk-extern): Likewise.
25409 * doc/extend.texi: Document indirect_branch function attribute.
25410 * doc/invoke.texi: Document -mindirect-branch= option.
25412 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
25415 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
25417 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
25419 * ipa-inline.c (want_inline_small_function_p): Return false if
25420 inlining has already failed with CIF_FINAL_ERROR.
25421 (update_caller_keys): Call want_inline_small_function_p before
25423 (update_callee_keys): Likewise.
25425 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25427 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
25429 (rs6000_quadword_masked_address_p): Likewise.
25430 (quad_aligned_load_p): Likewise.
25431 (quad_aligned_store_p): Likewise.
25432 (const_load_sequence_p): Add comment to describe the outer-most loop.
25433 (mimic_memory_attributes_and_flags): New function.
25434 (rs6000_gen_stvx): Likewise.
25435 (replace_swapped_aligned_store): Likewise.
25436 (rs6000_gen_lvx): Likewise.
25437 (replace_swapped_aligned_load): Likewise.
25438 (replace_swapped_load_constant): Capitalize argument name in
25439 comment describing this function.
25440 (rs6000_analyze_swaps): Add a third pass to search for vector loads
25441 and stores that access quad-word aligned addresses and replace
25442 with stvx or lvx instructions when appropriate.
25443 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
25444 New function prototype.
25445 (rs6000_quadword_masked_address_p): Likewise.
25446 (rs6000_gen_lvx): Likewise.
25447 (rs6000_gen_stvx): Likewise.
25448 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
25449 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
25450 when memory address is aligned.
25451 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
25452 this split to select lvx instruction when memory address is aligned.
25453 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
25454 instruction when memory address is aligned.
25455 (*vsx_le_perm_load_v16qi): Likewise.
25456 (four unnamed splitters): Modify to select the stvx instruction
25457 when memory is aligned.
25459 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
25461 * predict.c (determine_unlikely_bbs): Handle correctly BBs
25462 which appears in the queue multiple times.
25464 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25465 Alan Hayward <alan.hayward@arm.com>
25466 David Sherwood <david.sherwood@arm.com>
25468 * tree-vectorizer.h (vec_lower_bound): New structure.
25469 (_loop_vec_info): Add check_nonzero and lower_bounds.
25470 (LOOP_VINFO_CHECK_NONZERO): New macro.
25471 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
25472 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
25473 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
25474 fields. Make seg_len the distance travelled, not including the
25476 (dr_direction_indicator): Declare.
25477 (dr_zero_step_indicator): Likewise.
25478 (dr_known_forward_stride_p): Likewise.
25479 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
25481 (runtime_alias_check_p): Allow runtime alias checks with
25483 (operator ==): Compare access_size and align.
25484 (prune_runtime_alias_test_list): Rework for new distinction between
25485 the access_size and seg_len.
25486 (create_intersect_range_checks_index): Likewise. Cope with polynomial
25488 (get_segment_min_max): New function.
25489 (create_intersect_range_checks): Use it.
25490 (dr_step_indicator): New function.
25491 (dr_direction_indicator): Likewise.
25492 (dr_zero_step_indicator): Likewise.
25493 (dr_known_forward_stride_p): Likewise.
25494 * tree-loop-distribution.c (data_ref_segment_size): Return
25495 DR_STEP * (niters - 1).
25496 (compute_alias_check_pairs): Update call to the dr_with_seg_len
25498 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
25499 (vect_preserves_scalar_order_p): New function, split out from...
25500 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
25501 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
25502 (vect_vfa_access_size): New function.
25503 (vect_vfa_align): Likewise.
25504 (vect_compile_time_alias): Take access_size_a and access_b arguments.
25505 (dump_lower_bound): New function.
25506 (vect_check_lower_bound): Likewise.
25507 (vect_small_gap_p): Likewise.
25508 (vectorizable_with_step_bound_p): Likewise.
25509 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
25510 depencies if the vectorization factor is 1. Convert the checks
25511 for nonzero steps into checks on the bounds of DR_STEP. Try using
25512 a bunds check for variable steps if the minimum required step is
25513 relatively small. Update calls to the dr_with_seg_len
25514 constructor and to vect_compile_time_alias.
25515 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
25517 (vect_loop_versioning): Call it.
25518 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
25520 (vect_estimate_min_profitable_iters): Account for any bounds checks.
25522 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25523 Alan Hayward <alan.hayward@arm.com>
25524 David Sherwood <david.sherwood@arm.com>
25526 * doc/sourcebuild.texi (vect_scatter_store): Document.
25527 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
25529 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
25531 * genopinit.c (main): Add supports_vec_scatter_store and
25532 supports_vec_scatter_store_cached to target_optabs.
25533 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
25534 IFN_MASK_SCATTER_STORE.
25535 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
25537 * internal-fn.h (internal_store_fn_p): Declare.
25538 (internal_fn_stored_value_index): Likewise.
25539 * internal-fn.c (scatter_store_direct): New macro.
25540 (expand_scatter_store_optab_fn): New function.
25541 (direct_scatter_store_optab_supported_p): New macro.
25542 (internal_store_fn_p): New function.
25543 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
25544 IFN_MASK_SCATTER_STORE.
25545 (internal_fn_mask_index): Likewise.
25546 (internal_fn_stored_value_index): New function.
25547 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
25548 for scatter stores.
25549 * optabs-query.h (supports_vec_scatter_store_p): Declare.
25550 * optabs-query.c (supports_vec_scatter_store_p): New function.
25551 * tree-vectorizer.h (vect_get_store_rhs): Declare.
25552 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
25553 true for scatter stores.
25554 (vect_gather_scatter_fn_p): Handle scatter stores too.
25555 (vect_check_gather_scatter): Consider using scatter stores if
25556 supports_vec_scatter_store_p.
25557 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
25558 scatter stores too.
25559 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
25560 internal_fn_stored_value_index.
25561 (check_load_store_masking): Handle scatter stores too.
25562 (vect_get_store_rhs): Make public.
25563 (vectorizable_call): Use internal_store_fn_p.
25564 (vectorizable_store): Handle scatter store internal functions.
25565 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
25566 when deciding whether the end of the group has been reached.
25567 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
25568 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
25569 (mask_scatter_store<mode>): New insns.
25571 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25572 Alan Hayward <alan.hayward@arm.com>
25573 David Sherwood <david.sherwood@arm.com>
25575 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
25576 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
25577 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
25579 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
25580 Use vect_truncate_gather_scatter_offset if we can't treat the
25581 operation as a normal gather load or scatter store.
25582 (get_group_load_store_type): Take the gather_scatter_info
25583 as argument. Try using a gather load or scatter store for
25584 single-element groups.
25585 (get_load_store_type): Update calls to get_group_load_store_type
25586 and vect_use_strided_gather_scatters_p.
25588 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25589 Alan Hayward <alan.hayward@arm.com>
25590 David Sherwood <david.sherwood@arm.com>
25592 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
25593 optional tree argument.
25594 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
25596 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
25597 but continue to use the current value as a fallback.
25598 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
25599 to compare the updates.
25600 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
25601 (get_load_store_type): Use it when handling a strided access.
25602 (vect_get_strided_load_store_ops): New function.
25603 (vect_get_data_ptr_increment): Likewise.
25604 (vectorizable_load): Handle strided gather loads. Always pass
25605 a step to vect_create_data_ref_ptr and bump_vector_ptr.
25607 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25608 Alan Hayward <alan.hayward@arm.com>
25609 David Sherwood <david.sherwood@arm.com>
25611 * doc/md.texi (gather_load@var{m}): Document.
25612 (mask_gather_load@var{m}): Likewise.
25613 * genopinit.c (main): Add supports_vec_gather_load and
25614 supports_vec_gather_load_cached to target_optabs.
25615 * optabs-tree.c (init_tree_optimization_optabs): Use
25616 ggc_cleared_alloc to allocate target_optabs.
25617 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
25618 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
25620 * internal-fn.h (internal_load_fn_p): Declare.
25621 (internal_gather_scatter_fn_p): Likewise.
25622 (internal_fn_mask_index): Likewise.
25623 (internal_gather_scatter_fn_supported_p): Likewise.
25624 * internal-fn.c (gather_load_direct): New macro.
25625 (expand_gather_load_optab_fn): New function.
25626 (direct_gather_load_optab_supported_p): New macro.
25627 (direct_internal_fn_optab): New function.
25628 (internal_load_fn_p): Likewise.
25629 (internal_gather_scatter_fn_p): Likewise.
25630 (internal_fn_mask_index): Likewise.
25631 (internal_gather_scatter_fn_supported_p): Likewise.
25632 * optabs-query.c (supports_at_least_one_mode_p): New function.
25633 (supports_vec_gather_load_p): Likewise.
25634 * optabs-query.h (supports_vec_gather_load_p): Declare.
25635 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
25636 and memory_type field.
25637 (NUM_PATTERNS): Bump to 15.
25638 * tree-vect-data-refs.c: Include internal-fn.h.
25639 (vect_gather_scatter_fn_p): New function.
25640 (vect_describe_gather_scatter_call): Likewise.
25641 (vect_check_gather_scatter): Try using internal functions for
25642 gather loads. Recognize existing calls to a gather load function.
25643 (vect_analyze_data_refs): Consider using gather loads if
25644 supports_vec_gather_load_p.
25645 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
25646 (vect_get_gather_scatter_offset_type): Likewise.
25647 (vect_convert_mask_for_vectype): Likewise.
25648 (vect_add_conversion_to_patterm): Likewise.
25649 (vect_try_gather_scatter_pattern): Likewise.
25650 (vect_recog_gather_scatter_pattern): New pattern recognizer.
25651 (vect_vect_recog_func_ptrs): Add it.
25652 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
25653 internal_fn_mask_index and internal_gather_scatter_fn_p.
25654 (check_load_store_masking): Take the gather_scatter_info as an
25655 argument and handle gather loads.
25656 (vect_get_gather_scatter_ops): New function.
25657 (vectorizable_call): Check internal_load_fn_p.
25658 (vectorizable_load): Likewise. Handle gather load internal
25660 (vectorizable_store): Update call to check_load_store_masking.
25661 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
25662 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
25663 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
25664 (aarch64_gather_scale_operand_d): New predicates.
25665 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
25666 (mask_gather_load<mode>): New insns.
25668 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25669 Alan Hayward <alan.hayward@arm.com>
25670 David Sherwood <david.sherwood@arm.com>
25672 * optabs.def (fold_left_plus_optab): New optab.
25673 * doc/md.texi (fold_left_plus_@var{m}): Document.
25674 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
25675 * internal-fn.c (fold_left_direct): Define.
25676 (expand_fold_left_optab_fn): Likewise.
25677 (direct_fold_left_optab_supported_p): Likewise.
25678 * fold-const-call.c (fold_const_fold_left): New function.
25679 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
25680 * tree-parloops.c (valid_reduction_p): New function.
25681 (gather_scalar_reductions): Use it.
25682 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
25683 (vect_finish_replace_stmt): Declare.
25684 * tree-vect-loop.c (fold_left_reduction_fn): New function.
25685 (needs_fold_left_reduction_p): New function, split out from...
25686 (vect_is_simple_reduction): ...here. Accept reductions that
25687 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
25688 (vect_force_simple_reduction): Also store the reduction type in
25689 the assignment's STMT_VINFO_REDUC_TYPE.
25690 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
25691 (merge_with_identity): New function.
25692 (vect_expand_fold_left): Likewise.
25693 (vectorize_fold_left_reduction): Likewise.
25694 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
25695 scalar phi in place for it. Check for target support and reject
25696 cases that would reassociate the operation. Defer the transform
25697 phase to vectorize_fold_left_reduction.
25698 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
25699 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
25700 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
25702 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25704 * tree-if-conv.c (predicate_mem_writes): Remove redundant
25705 call to ifc_temp_var.
25707 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25708 Alan Hayward <alan.hayward@arm.com>
25709 David Sherwood <david.sherwood@arm.com>
25711 * target.def (legitimize_address_displacement): Take the original
25712 offset as a poly_int.
25713 * targhooks.h (default_legitimize_address_displacement): Update
25715 * targhooks.c (default_legitimize_address_displacement): Likewise.
25716 * doc/tm.texi: Regenerate.
25717 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
25718 as an argument, moving assert of ad->disp == ad->disp_term to...
25719 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
25720 Try calling targetm.legitimize_address_displacement before expanding
25721 the address rather than afterwards, and adjust for the new interface.
25722 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
25723 Match the new hook interface. Handle SVE addresses.
25724 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
25725 new hook interface.
25727 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25729 * Makefile.in (OBJS): Add early-remat.o.
25730 * target.def (select_early_remat_modes): New hook.
25731 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
25732 * doc/tm.texi: Regenerate.
25733 * targhooks.h (default_select_early_remat_modes): Declare.
25734 * targhooks.c (default_select_early_remat_modes): New function.
25735 * timevar.def (TV_EARLY_REMAT): New timevar.
25736 * passes.def (pass_early_remat): New pass.
25737 * tree-pass.h (make_pass_early_remat): Declare.
25738 * early-remat.c: New file.
25739 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
25741 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
25743 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25744 Alan Hayward <alan.hayward@arm.com>
25745 David Sherwood <david.sherwood@arm.com>
25747 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
25748 vfm1 with a bound_epilog parameter.
25749 (vect_do_peeling): Update calls accordingly, and move the prologue
25750 call earlier in the function. Treat the base bound_epilog as 0 for
25751 fully-masked loops and retain vf - 1 for other loops. Add 1 to
25752 this base when peeling for gaps.
25753 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
25754 with fully-masked loops.
25755 (vect_estimate_min_profitable_iters): Handle the single peeled
25756 iteration in that case.
25758 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25759 Alan Hayward <alan.hayward@arm.com>
25760 David Sherwood <david.sherwood@arm.com>
25762 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
25763 single-element interleaving even if the size is not a power of 2.
25764 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
25765 accesses for single-element interleaving if the group size is
25768 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25769 Alan Hayward <alan.hayward@arm.com>
25770 David Sherwood <david.sherwood@arm.com>
25772 * doc/md.texi (fold_extract_last_@var{m}): Document.
25773 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
25774 * optabs.def (fold_extract_last_optab): New optab.
25775 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
25776 * internal-fn.c (fold_extract_direct): New macro.
25777 (expand_fold_extract_optab_fn): Likewise.
25778 (direct_fold_extract_optab_supported_p): Likewise.
25779 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
25780 * tree-vect-loop.c (vect_model_reduction_cost): Handle
25781 EXTRACT_LAST_REDUCTION.
25782 (get_initial_def_for_reduction): Do not create an initial vector
25783 for EXTRACT_LAST_REDUCTION reductions.
25784 (vectorizable_reduction): Leave the scalar phi in place for
25785 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
25786 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
25787 epilogue code for EXTRACT_LAST_REDUCTION and defer the
25788 transform phase to vectorizable_condition.
25789 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
25791 (vect_finish_stmt_generation): ...here.
25792 (vect_finish_replace_stmt): New function.
25793 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
25794 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
25796 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
25798 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25799 Alan Hayward <alan.hayward@arm.com>
25800 David Sherwood <david.sherwood@arm.com>
25802 * doc/md.texi (extract_last_@var{m}): Document.
25803 * optabs.def (extract_last_optab): New optab.
25804 * internal-fn.def (EXTRACT_LAST): New internal function.
25805 * internal-fn.c (cond_unary_direct): New macro.
25806 (expand_cond_unary_optab_fn): Likewise.
25807 (direct_cond_unary_optab_supported_p): Likewise.
25808 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
25809 loops using EXTRACT_LAST.
25810 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
25811 (extract_last_<mode>): ...this optab.
25812 (vec_extract<mode><Vel>): Update accordingly.
25814 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25815 Alan Hayward <alan.hayward@arm.com>
25816 David Sherwood <david.sherwood@arm.com>
25818 * target.def (empty_mask_is_expensive): New hook.
25819 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
25820 * doc/tm.texi: Regenerate.
25821 * targhooks.h (default_empty_mask_is_expensive): Declare.
25822 * targhooks.c (default_empty_mask_is_expensive): New function.
25823 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
25824 if the target says that empty masks are expensive.
25825 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
25827 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
25829 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25830 Alan Hayward <alan.hayward@arm.com>
25831 David Sherwood <david.sherwood@arm.com>
25833 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
25834 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
25835 (vect_use_loop_mask_for_alignment_p): New function.
25836 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
25837 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
25838 niters_skip argument. Make sure that the first niters_skip elements
25839 of the first iteration are inactive.
25840 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
25841 Update call to vect_set_loop_masks_directly.
25842 (get_misalign_in_elems): New function, split out from...
25843 (vect_gen_prolog_loop_niters): ...here.
25844 (vect_update_init_of_dr): Take a code argument that specifies whether
25845 the adjustment should be added or subtracted.
25846 (vect_update_init_of_drs): Likewise.
25847 (vect_prepare_for_masked_peels): New function.
25848 (vect_do_peeling): Skip prologue peeling if we're using a mask
25849 instead. Update call to vect_update_inits_of_drs.
25850 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
25852 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
25853 alignment. Do not include the number of peeled iterations in
25854 the minimum threshold in that case.
25855 (vectorizable_induction): Adjust the start value down by
25856 LOOP_VINFO_MASK_SKIP_NITERS iterations.
25857 (vect_transform_loop): Call vect_prepare_for_masked_peels.
25858 Take the number of skipped iterations into account when calculating
25860 * tree-vect-stmts.c (vect_gen_while_not): New function.
25862 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25863 Alan Hayward <alan.hayward@arm.com>
25864 David Sherwood <david.sherwood@arm.com>
25866 * doc/sourcebuild.texi (vect_fully_masked): Document.
25867 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
25868 default value to 0.
25869 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
25871 (vect_analyze_loop_2): ...here. Don't check the vectorization
25872 factor against the number of loop iterations if the loop is
25875 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25876 Alan Hayward <alan.hayward@arm.com>
25877 David Sherwood <david.sherwood@arm.com>
25879 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
25880 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
25881 (dump_groups): Update accordingly.
25882 (iv_use::mem_type): New member variable.
25883 (address_p): New function.
25884 (record_use): Add a mem_type argument and initialize the new
25886 (record_group_use): Add a mem_type argument. Use address_p.
25887 Remove obsolete null checks of base_object. Update call to record_use.
25888 (find_interesting_uses_op): Update call to record_group_use.
25889 (find_interesting_uses_cond): Likewise.
25890 (find_interesting_uses_address): Likewise.
25891 (get_mem_type_for_internal_fn): New function.
25892 (find_address_like_use): Likewise.
25893 (find_interesting_uses_stmt): Try find_address_like_use before
25894 calling find_interesting_uses_op.
25895 (addr_offset_valid_p): Use the iv mem_type field as the type
25896 of the addressed memory.
25897 (add_autoinc_candidates): Likewise.
25898 (get_address_cost): Likewise.
25899 (split_small_address_groups_p): Use address_p.
25900 (split_address_groups): Likewise.
25901 (add_iv_candidate_for_use): Likewise.
25902 (autoinc_possible_for_pair): Likewise.
25903 (rewrite_groups): Likewise.
25904 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
25905 (determine_group_iv_cost): Update after split of USE_ADDRESS.
25906 (get_alias_ptr_type_for_ptr_address): New function.
25907 (rewrite_use_address): Rewrite address uses in calls that were
25908 identified by find_address_like_use.
25910 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25911 Alan Hayward <alan.hayward@arm.com>
25912 David Sherwood <david.sherwood@arm.com>
25914 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
25916 * gimple-expr.h (is_gimple_addressable: Likewise.
25917 * gimple-expr.c (is_gimple_address): Likewise.
25918 * internal-fn.c (expand_call_mem_ref): New function.
25919 (expand_mask_load_optab_fn): Use it.
25920 (expand_mask_store_optab_fn): Likewise.
25922 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25923 Alan Hayward <alan.hayward@arm.com>
25924 David Sherwood <david.sherwood@arm.com>
25926 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
25927 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
25928 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
25929 (cond_umax@var{mode}): Document.
25930 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
25931 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
25932 (cond_umin_optab, cond_umax_optab): New optabs.
25933 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
25934 (COND_IOR, COND_XOR): New internal functions.
25935 * internal-fn.h (get_conditional_internal_fn): Declare.
25936 * internal-fn.c (cond_binary_direct): New macro.
25937 (expand_cond_binary_optab_fn): Likewise.
25938 (direct_cond_binary_optab_supported_p): Likewise.
25939 (get_conditional_internal_fn): New function.
25940 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
25941 Cope with reduction statements that are vectorized as calls rather
25943 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
25944 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
25945 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
25946 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
25947 (UNSPEC_COND_EOR): New unspecs.
25948 (optab): Add mappings for them.
25949 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
25950 (sve_int_op, sve_fp_op): New int attributes.
25952 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25953 Alan Hayward <alan.hayward@arm.com>
25954 David Sherwood <david.sherwood@arm.com>
25956 * optabs.def (while_ult_optab): New optab.
25957 * doc/md.texi (while_ult@var{m}@var{n}): Document.
25958 * internal-fn.def (WHILE_ULT): New internal function.
25959 * internal-fn.h (direct_internal_fn_supported_p): New override
25960 that takes two types as argument.
25961 * internal-fn.c (while_direct): New macro.
25962 (expand_while_optab_fn): New function.
25963 (convert_optab_supported_p): Likewise.
25964 (direct_while_optab_supported_p): New macro.
25965 * wide-int.h (wi::udiv_ceil): New function.
25966 * tree-vectorizer.h (rgroup_masks): New structure.
25967 (vec_loop_masks): New typedef.
25968 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
25969 and fully_masked_p.
25970 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
25971 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
25972 (vect_max_vf): New function.
25973 (slpeel_make_loop_iterate_ntimes): Delete.
25974 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
25975 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
25976 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
25977 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
25978 internal-fn.h, stor-layout.h and optabs-query.h.
25979 (vect_set_loop_mask): New function.
25980 (add_preheader_seq): Likewise.
25981 (add_header_seq): Likewise.
25982 (interleave_supported_p): Likewise.
25983 (vect_maybe_permute_loop_masks): Likewise.
25984 (vect_set_loop_masks_directly): Likewise.
25985 (vect_set_loop_condition_masked): Likewise.
25986 (vect_set_loop_condition_unmasked): New function, split out from
25987 slpeel_make_loop_iterate_ntimes.
25988 (slpeel_make_loop_iterate_ntimes): Rename to..
25989 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
25990 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
25991 (vect_do_peeling): Update call accordingly.
25992 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
25994 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
25995 mask_compare_type, can_fully_mask_p and fully_masked_p.
25996 (release_vec_loop_masks): New function.
25997 (_loop_vec_info): Use it to free the loop masks.
25998 (can_produce_all_loop_masks_p): New function.
25999 (vect_get_max_nscalars_per_iter): Likewise.
26000 (vect_verify_full_masking): Likewise.
26001 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
26002 retries, and free the mask rgroups before retrying. Check loop-wide
26003 reasons for disallowing fully-masked loops. Make the final decision
26004 about whether use a fully-masked loop or not.
26005 (vect_estimate_min_profitable_iters): Do not assume that peeling
26006 for the number of iterations will be needed for fully-masked loops.
26007 (vectorizable_reduction): Disable fully-masked loops.
26008 (vectorizable_live_operation): Likewise.
26009 (vect_halve_mask_nunits): New function.
26010 (vect_double_mask_nunits): Likewise.
26011 (vect_record_loop_mask): Likewise.
26012 (vect_get_loop_mask): Likewise.
26013 (vect_transform_loop): Handle the case in which the final loop
26014 iteration might handle a partial vector. Call vect_set_loop_condition
26015 instead of slpeel_make_loop_iterate_ntimes.
26016 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
26017 (check_load_store_masking): New function.
26018 (prepare_load_store_mask): Likewise.
26019 (vectorizable_store): Handle fully-masked loops.
26020 (vectorizable_load): Likewise.
26021 (supportable_widening_operation): Use vect_halve_mask_nunits for
26023 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
26024 (vect_gen_while): New function.
26025 * config/aarch64/aarch64.md (umax<mode>3): New expander.
26026 (aarch64_uqdec<mode>): New insn.
26028 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26029 Alan Hayward <alan.hayward@arm.com>
26030 David Sherwood <david.sherwood@arm.com>
26032 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
26033 (reduc_xor_scal_optab): New optabs.
26034 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
26035 (reduc_xor_scal_@var{m}): Document.
26036 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
26037 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
26038 internal functions.
26039 * fold-const-call.c (fold_const_call): Handle them.
26040 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
26041 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
26042 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
26043 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
26044 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
26045 (UNSPEC_XORV): New unspecs.
26046 (optab): Add entries for them.
26047 (BITWISEV): New int iterator.
26048 (bit_reduc_op): New int attributes.
26050 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26051 Alan Hayward <alan.hayward@arm.com>
26052 David Sherwood <david.sherwood@arm.com>
26054 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
26055 * internal-fn.def (VEC_SHL_INSERT): New internal function.
26056 * optabs.def (vec_shl_insert_optab): New optab.
26057 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
26058 (duplicate_and_interleave): Likewise.
26059 * tree-vect-loop.c: Include internal-fn.h.
26060 (neutral_op_for_slp_reduction): New function, split out from
26061 get_initial_defs_for_reduction.
26062 (get_initial_def_for_reduction): Handle option 2 for variable-length
26063 vectors by loading the neutral value into a vector and then shifting
26064 the initial value into element 0.
26065 (get_initial_defs_for_reduction): Replace the code argument with
26066 the neutral value calculated by neutral_op_for_slp_reduction.
26067 Use gimple_build_vector for constant-length vectors.
26068 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
26069 but the first group_size elements have a neutral value.
26070 Use duplicate_and_interleave otherwise.
26071 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
26072 Update call to get_initial_defs_for_reduction. Handle SLP
26073 reductions for variable-length vectors by creating one vector
26074 result for each scalar result, with the elements associated
26075 with other scalar results stubbed out with the neutral value.
26076 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
26077 Require IFN_VEC_SHL_INSERT for double reductions on
26078 variable-length vectors, or SLP reductions that have
26079 a neutral value. Require can_duplicate_and_interleave_p
26080 support for variable-length unchained SLP reductions if there
26081 is no neutral value, such as for MIN/MAX reductions. Also require
26082 the number of vector elements to be a multiple of the number of
26083 SLP statements when doing variable-length unchained SLP reductions.
26084 Update call to vect_create_epilog_for_reduction.
26085 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
26086 and remove initial values.
26087 (duplicate_and_interleave): Make public.
26088 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
26089 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
26091 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26092 Alan Hayward <alan.hayward@arm.com>
26093 David Sherwood <david.sherwood@arm.com>
26095 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
26096 (can_duplicate_and_interleave_p): New function.
26097 (vect_get_and_check_slp_defs): Take the vector of statements
26098 rather than just the current one. Remove excess parentheses.
26099 Restriction rejectinon of vect_constant_def and vect_external_def
26100 for variable-length vectors to boolean types, or types for which
26101 can_duplicate_and_interleave_p is false.
26102 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
26103 (duplicate_and_interleave): New function.
26104 (vect_get_constant_vectors): Use gimple_build_vector for
26105 constant-length vectors and suitable variable-length constant
26106 vectors. Use duplicate_and_interleave for other variable-length
26107 vectors. Don't defer the update when inserting new statements.
26109 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26110 Alan Hayward <alan.hayward@arm.com>
26111 David Sherwood <david.sherwood@arm.com>
26113 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
26114 min_profitable_iters doesn't go negative.
26116 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26117 Alan Hayward <alan.hayward@arm.com>
26118 David Sherwood <david.sherwood@arm.com>
26120 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
26121 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
26122 * optabs.def (vec_mask_load_lanes_optab): New optab.
26123 (vec_mask_store_lanes_optab): Likewise.
26124 * internal-fn.def (MASK_LOAD_LANES): New internal function.
26125 (MASK_STORE_LANES): Likewise.
26126 * internal-fn.c (mask_load_lanes_direct): New macro.
26127 (mask_store_lanes_direct): Likewise.
26128 (expand_mask_load_optab_fn): Handle masked operations.
26129 (expand_mask_load_lanes_optab_fn): New macro.
26130 (expand_mask_store_optab_fn): Handle masked operations.
26131 (expand_mask_store_lanes_optab_fn): New macro.
26132 (direct_mask_load_lanes_optab_supported_p): Likewise.
26133 (direct_mask_store_lanes_optab_supported_p): Likewise.
26134 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
26136 (vect_load_lanes_supported): Likewise.
26137 * tree-vect-data-refs.c (strip_conversion): New function.
26138 (can_group_stmts_p): Likewise.
26139 (vect_analyze_data_ref_accesses): Use it instead of checking
26140 for a pair of assignments.
26141 (vect_store_lanes_supported): Take a masked_p parameter.
26142 (vect_load_lanes_supported): Likewise.
26143 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
26144 vect_store_lanes_supported and vect_load_lanes_supported.
26145 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
26146 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
26147 parameter. Don't allow gaps for masked accesses.
26148 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
26149 and vect_load_lanes_supported.
26150 (get_load_store_type): Take a masked_p parameter and update
26151 call to get_group_load_store_type.
26152 (vectorizable_store): Update call to get_load_store_type.
26153 Handle IFN_MASK_STORE_LANES.
26154 (vectorizable_load): Update call to get_load_store_type.
26155 Handle IFN_MASK_LOAD_LANES.
26157 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26158 Alan Hayward <alan.hayward@arm.com>
26159 David Sherwood <david.sherwood@arm.com>
26161 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
26163 * config/aarch64/aarch64-protos.h
26164 (aarch64_sve_struct_memory_operand_p): Declare.
26165 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
26166 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
26167 (VPRED, vpred): Handle SVE structure modes.
26168 * config/aarch64/constraints.md (Utx): New constraint.
26169 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
26170 (aarch64_sve_struct_nonimmediate_operand): New predicates.
26171 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
26172 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
26173 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
26174 structure modes. Split into pieces after RA.
26175 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
26176 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
26178 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
26179 SVE structure modes.
26180 (aarch64_classify_address): Likewise.
26181 (sizetochar): Move earlier in file.
26182 (aarch64_print_operand): Handle SVE register lists.
26183 (aarch64_array_mode): New function.
26184 (aarch64_sve_struct_memory_operand_p): Likewise.
26185 (TARGET_ARRAY_MODE): Redefine.
26187 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26188 Alan Hayward <alan.hayward@arm.com>
26189 David Sherwood <david.sherwood@arm.com>
26191 * target.def (array_mode): New target hook.
26192 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
26193 * doc/tm.texi: Regenerate.
26194 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
26195 * hooks.c (hook_optmode_mode_uhwi_none): New function.
26196 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
26197 targetm.array_mode.
26198 * stor-layout.c (mode_for_array): Likewise. Support polynomial
26201 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26202 Alan Hayward <alan.hayward@arm.com>
26203 David Sherwood <david.sherwood@arm.com>
26205 * fold-const.c (fold_binary_loc): Check the argument types
26206 rather than the result type when testing for a vector operation.
26208 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26210 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
26211 * doc/tm.texi: Regenerate.
26213 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26214 Alan Hayward <alan.hayward@arm.com>
26215 David Sherwood <david.sherwood@arm.com>
26217 * doc/invoke.texi (-msve-vector-bits=): Document new option.
26218 (sve): Document new AArch64 extension.
26219 * doc/md.texi (w): Extend the description of the AArch64
26220 constraint to include SVE vectors.
26221 (Upl, Upa): Document new AArch64 predicate constraints.
26222 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
26224 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
26225 (msve-vector-bits=): New option.
26226 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
26227 SVE when these are disabled.
26228 (sve): New extension.
26229 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
26230 modes. Adjust their number of units based on aarch64_sve_vg.
26231 (MAX_BITSIZE_MODE_ANY_MODE): Define.
26232 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
26233 aarch64_addr_query_type.
26234 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
26235 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
26236 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
26237 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
26238 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
26239 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
26240 (aarch64_simd_imm_zero_p): Delete.
26241 (aarch64_check_zero_based_sve_index_immediate): Declare.
26242 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
26243 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
26244 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
26245 (aarch64_sve_float_mul_immediate_p): Likewise.
26246 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
26247 rather than an rtx.
26248 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
26249 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
26250 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
26251 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
26252 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
26253 (aarch64_regmode_natural_size): Likewise.
26254 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
26255 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
26257 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
26258 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
26259 for VG and the SVE predicate registers.
26260 (V_ALIASES): Add a "z"-prefixed alias.
26261 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
26262 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
26263 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
26264 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
26265 (REG_CLASS_NAMES): Add entries for them.
26266 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
26267 and the predicate registers.
26268 (aarch64_sve_vg): Declare.
26269 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
26270 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
26271 (REGMODE_NATURAL_SIZE): Define.
26272 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
26274 * config/aarch64/aarch64.c: Include cfgrtl.h.
26275 (simd_immediate_info): Add a constructor for series vectors,
26276 and an associated step field.
26277 (aarch64_sve_vg): New variable.
26278 (aarch64_dbx_register_number): Handle VG and the predicate registers.
26279 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
26280 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
26281 (VEC_ANY_DATA, VEC_STRUCT): New constants.
26282 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
26283 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
26284 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
26285 (aarch64_get_mask_mode): New functions.
26286 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
26287 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
26288 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
26289 predicate modes and predicate registers. Explicitly restrict
26290 GPRs to modes of 16 bytes or smaller. Only allow FP registers
26291 to store a vector mode if it is recognized by
26292 aarch64_classify_vector_mode.
26293 (aarch64_regmode_natural_size): New function.
26294 (aarch64_hard_regno_caller_save_mode): Return the original mode
26296 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
26297 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
26298 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
26299 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
26301 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
26302 does not overlap dest if the function is frame-related. Handle
26304 (aarch64_split_add_offset): New function.
26305 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
26306 them aarch64_add_offset.
26307 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
26308 and update call to aarch64_sub_sp.
26309 (aarch64_add_cfa_expression): New function.
26310 (aarch64_expand_prologue): Pass extra temporary registers to the
26311 functions above. Handle the case in which we need to emit new
26312 DW_CFA_expressions for registers that were originally saved
26313 relative to the stack pointer, but now have to be expressed
26314 relative to the frame pointer.
26315 (aarch64_output_mi_thunk): Pass extra temporary registers to the
26317 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
26318 IP0 and IP1 values for SVE frames.
26319 (aarch64_expand_vec_series): New function.
26320 (aarch64_expand_sve_widened_duplicate): Likewise.
26321 (aarch64_expand_sve_const_vector): Likewise.
26322 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
26323 Handle SVE constants. Use emit_move_insn to move a force_const_mem
26324 into the register, rather than emitting a SET directly.
26325 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
26326 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
26327 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
26328 (offset_9bit_signed_scaled_p): New functions.
26329 (aarch64_replicate_bitmask_imm): New function.
26330 (aarch64_bitmask_imm): Use it.
26331 (aarch64_cannot_force_const_mem): Reject expressions involving
26332 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
26333 (aarch64_classify_index): Handle SVE indices, by requiring
26334 a plain register index with a scale that matches the element size.
26335 (aarch64_classify_address): Handle SVE addresses. Assert that
26336 the mode of the address is VOIDmode or an integer mode.
26337 Update call to aarch64_classify_symbol.
26338 (aarch64_classify_symbolic_expression): Update call to
26339 aarch64_classify_symbol.
26340 (aarch64_const_vec_all_in_range_p): New function.
26341 (aarch64_print_vector_float_operand): Likewise.
26342 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
26343 "vN" for FP registers with SVE modes. Handle (const ...) vectors
26344 and the FP immediates 1.0 and 0.5.
26345 (aarch64_print_address_internal): Handle SVE addresses.
26346 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
26347 (aarch64_regno_regclass): Handle predicate registers.
26348 (aarch64_secondary_reload): Handle big-endian reloads of SVE
26350 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
26351 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
26352 (aarch64_convert_sve_vector_bits): New function.
26353 (aarch64_override_options): Use it to handle -msve-vector-bits=.
26354 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
26355 rather than an rtx.
26356 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
26357 Handle SVE vector and predicate modes. Accept VL-based constants
26358 that need only one temporary register, and VL offsets that require
26359 no temporary registers.
26360 (aarch64_conditional_register_usage): Mark the predicate registers
26361 as fixed if SVE isn't available.
26362 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
26363 Return true for SVE vector and predicate modes.
26364 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
26365 rather than an unsigned int. Handle SVE modes.
26366 (aarch64_preferred_simd_mode): Update call accordingly. Handle
26368 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
26370 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
26371 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
26372 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
26373 (aarch64_sve_float_mul_immediate_p): New functions.
26374 (aarch64_sve_valid_immediate): New function.
26375 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
26376 Explicitly reject structure modes. Check for INDEX constants.
26377 Handle PTRUE and PFALSE constants.
26378 (aarch64_check_zero_based_sve_index_immediate): New function.
26379 (aarch64_simd_imm_zero_p): Delete.
26380 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
26381 vector modes. Accept constants in the range of CNT[BHWD].
26382 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
26383 ask for an Advanced SIMD mode.
26384 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
26385 (aarch64_simd_vector_alignment): Handle SVE predicates.
26386 (aarch64_vectorize_preferred_vector_alignment): New function.
26387 (aarch64_simd_vector_alignment_reachable): Use it instead of
26389 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
26390 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
26392 (MAX_VECT_LEN): Delete.
26393 (expand_vec_perm_d): Add a vec_flags field.
26394 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
26395 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
26396 (aarch64_evpc_ext): Don't apply a big-endian lane correction
26398 (aarch64_evpc_rev): Rename to...
26399 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
26400 (aarch64_evpc_rev_global): New function.
26401 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
26402 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
26404 (aarch64_evpc_sve_tbl): New function.
26405 (aarch64_expand_vec_perm_const_1): Update after rename of
26406 aarch64_evpc_rev. Handle SVE permutes too, trying
26407 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
26408 than aarch64_evpc_tbl.
26409 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
26410 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
26411 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
26412 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
26413 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
26414 (aarch64_expand_sve_vcond): New functions.
26415 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
26416 of aarch64_vector_mode_p.
26417 (aarch64_dwarf_poly_indeterminate_value): New function.
26418 (aarch64_compute_pressure_classes): Likewise.
26419 (aarch64_can_change_mode_class): Likewise.
26420 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
26421 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
26422 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
26423 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
26424 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
26425 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
26426 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
26427 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
26429 (Dn, Dl, Dr): Accept const as well as const_vector.
26430 (Dz): Likewise. Compare against CONST0_RTX.
26431 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
26432 of "vector" where appropriate.
26433 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
26434 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
26435 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
26436 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
26437 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
26438 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
26439 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
26440 (v_int_equiv): Extend to SVE modes.
26441 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
26443 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
26444 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
26445 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
26446 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
26447 (SVE_COND_FP_CMP): New int iterators.
26448 (perm_hilo): Handle the new unpack unspecs.
26449 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
26451 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
26452 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
26453 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
26454 (aarch64_equality_operator, aarch64_constant_vector_operand)
26455 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
26456 (aarch64_sve_nonimmediate_operand): Likewise.
26457 (aarch64_sve_general_operand): Likewise.
26458 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
26459 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
26460 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
26461 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
26462 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
26463 (aarch64_sve_float_arith_immediate): Likewise.
26464 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
26465 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
26466 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
26467 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
26468 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
26469 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
26470 (aarch64_sve_float_arith_operand): Likewise.
26471 (aarch64_sve_float_arith_with_sub_operand): Likewise.
26472 (aarch64_sve_float_mul_operand): Likewise.
26473 (aarch64_sve_vec_perm_operand): Likewise.
26474 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
26475 (aarch64_mov_operand): Accept const_poly_int and const_vector.
26476 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
26477 as well as const_vector.
26478 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
26479 in file. Use CONST0_RTX and CONSTM1_RTX.
26480 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
26481 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
26482 Use aarch64_simd_imm_zero.
26483 * config/aarch64/aarch64-sve.md: New file.
26484 * config/aarch64/aarch64.md: Include it.
26485 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
26486 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
26487 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
26488 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
26489 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
26490 (sve): New attribute.
26491 (enabled): Disable instructions with the sve attribute unless
26493 (movqi, movhi): Pass CONST_POLY_INT operaneds through
26494 aarch64_expand_mov_immediate.
26495 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
26496 CNT[BHSD] immediates.
26497 (movti): Split CONST_POLY_INT moves into two halves.
26498 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
26499 Split additions that need a temporary here if the destination
26500 is the stack pointer.
26501 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
26502 (*add<mode>3_poly_1): New instruction.
26503 (set_clobber_cc): New expander.
26505 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26507 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
26508 parameter and use it instead of GET_MODE_SIZE (innermode). Use
26509 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
26510 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
26511 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
26512 Change innermode from fixed_mode_size to machine_mode.
26513 (simplify_subreg): Update call accordingly. Handle a constant-sized
26514 subreg of a variable-length CONST_VECTOR.
26516 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26517 Alan Hayward <alan.hayward@arm.com>
26518 David Sherwood <david.sherwood@arm.com>
26520 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
26521 (add_offset_to_base): New function, split out from...
26522 (create_mem_ref): ...here. When handling a scale other than 1,
26523 check first whether the address is valid without the offset.
26524 Add it into the base if so, leaving the index and scale as-is.
26526 2018-01-12 Jakub Jelinek <jakub@redhat.com>
26529 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
26530 fold_for_warn before checking if arg2 is INTEGER_CST.
26532 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
26534 * config/rs6000/predicates.md (load_multiple_operation): Delete.
26535 (store_multiple_operation): Delete.
26536 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
26537 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
26538 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
26539 guarded by TARGET_STRING.
26540 (rs6000_output_load_multiple): Delete.
26541 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26542 OPTION_MASK_STRING / TARGET_STRING handling.
26543 (print_operand) <'N', 'O'>: Add comment that these are unused now.
26544 (const rs6000_opt_masks) <"string">: Change mask to 0.
26545 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
26546 (MASK_STRING): Delete.
26547 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
26549 (load_multiple): Delete.
26556 (store_multiple): Delete.
26563 (movmemsi_8reg): Delete.
26564 (corresponding unnamed define_insn): Delete.
26565 (movmemsi_6reg): Delete.
26566 (corresponding unnamed define_insn): Delete.
26567 (movmemsi_4reg): Delete.
26568 (corresponding unnamed define_insn): Delete.
26569 (movmemsi_2reg): Delete.
26570 (corresponding unnamed define_insn): Delete.
26571 (movmemsi_1reg): Delete.
26572 (corresponding unnamed define_insn): Delete.
26573 * config/rs6000/rs6000.opt (mno-string): New.
26574 (mstring): Replace by deprecation warning stub.
26575 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
26577 2018-01-12 Jakub Jelinek <jakub@redhat.com>
26579 * regrename.c (regrename_do_replace): If replacing the same
26580 reg multiple times, try to reuse last created gen_raw_REG.
26583 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
26584 main to workaround a bug in GDB.
26586 2018-01-12 Tom de Vries <tom@codesourcery.com>
26589 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
26591 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
26593 PR rtl-optimization/80481
26594 * ira-color.c (get_cap_member): New function.
26595 (allocnos_conflict_by_live_ranges_p): Use it.
26596 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
26597 (setup_slot_coalesced_allocno_live_ranges): Ditto.
26599 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
26602 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
26603 (*saddl_se_1): Ditto.
26604 (*ssubsi_1): Ditto.
26605 (*ssubl_se_1): Ditto.
26607 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
26609 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
26610 rather than wi::to_widest for DR_INITs.
26611 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
26612 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
26613 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
26615 (vect_analyze_group_access_1): Note that here.
26617 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
26619 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
26620 polynomial type sizes.
26622 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
26624 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
26625 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
26626 (gimple_add_tmp_var): Likewise.
26628 2018-01-12 Martin Liska <mliska@suse.cz>
26630 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
26631 (gimple_alloc_sizes): Likewise.
26632 (dump_gimple_statistics): Use PRIu64 in printf format.
26633 * gimple.h: Change uint64_t to int.
26635 2018-01-12 Martin Liska <mliska@suse.cz>
26637 * tree-core.h: Use uint64_t instead of int.
26638 * tree.c (tree_node_counts): Likewise.
26639 (tree_node_sizes): Likewise.
26640 (dump_tree_statistics): Use PRIu64 in printf format.
26642 2018-01-12 Martin Liska <mliska@suse.cz>
26644 * Makefile.in: As qsort_chk is implemented in vec.c, add
26645 vec.o to linkage of gencfn-macros.
26646 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
26647 passing the info to record_node_allocation_statistics.
26648 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
26650 * ggc-common.c (struct ggc_usage): Add operator== and use
26651 it in operator< and compare function.
26652 * mem-stats.h (struct mem_usage): Likewise.
26653 * vec.c (struct vec_usage): Remove operator< and compare
26654 function. Can be simply inherited.
26656 2018-01-12 Martin Jambor <mjambor@suse.cz>
26659 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
26660 * tree-ssa-math-opts.c: Include domwalk.h.
26661 (convert_mult_to_fma_1): New function.
26662 (fma_transformation_info): New type.
26663 (fma_deferring_state): Likewise.
26664 (cancel_fma_deferring): New function.
26665 (result_of_phi): Likewise.
26666 (last_fma_candidate_feeds_initial_phi): Likewise.
26667 (convert_mult_to_fma): Added deferring logic, split actual
26668 transformation to convert_mult_to_fma_1.
26669 (math_opts_dom_walker): New type.
26670 (math_opts_dom_walker::after_dom_children): New method, body moved
26671 here from pass_optimize_widening_mul::execute, added deferring logic
26673 (pass_optimize_widening_mul::execute): Moved most of code to
26674 math_opts_dom_walker::after_dom_children.
26675 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
26676 * config/i386/i386.c (ix86_option_override_internal): Added
26677 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
26679 2018-01-12 Richard Biener <rguenther@suse.de>
26682 * dwarf2out.c (gen_variable_die): Do not reset old_die for
26683 inline instance vars.
26685 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
26688 * config/rx/rx.c (rx_is_restricted_memory_address):
26689 Handle SUBREG case.
26691 2018-01-12 Richard Biener <rguenther@suse.de>
26693 PR tree-optimization/80846
26694 * target.def (split_reduction): New target hook.
26695 * targhooks.c (default_split_reduction): New function.
26696 * targhooks.h (default_split_reduction): Declare.
26697 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
26698 target requests first reduce vectors by combining low and high
26700 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
26701 (get_vectype_for_scalar_type_and_size): Export.
26702 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
26703 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
26704 * doc/tm.texi: Regenerate.
26705 * config/i386/i386.c (ix86_split_reduction): Implement
26706 TARGET_VECTORIZE_SPLIT_REDUCTION.
26708 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
26711 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
26712 in PIC mode except for TARGET_VXWORKS_RTP.
26713 * config/sparc/sparc.c: Include cfgrtl.h.
26714 (TARGET_INIT_PIC_REG): Define.
26715 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
26716 (sparc_pic_register_p): New predicate.
26717 (sparc_legitimate_address_p): Use it.
26718 (sparc_legitimize_pic_address): Likewise.
26719 (sparc_delegitimize_address): Likewise.
26720 (sparc_mode_dependent_address_p): Likewise.
26721 (gen_load_pcrel_sym): Remove 4th parameter.
26722 (load_got_register): Adjust call to above. Remove obsolete stuff.
26723 (sparc_expand_prologue): Do not call load_got_register here.
26724 (sparc_flat_expand_prologue): Likewise.
26725 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
26726 (sparc_use_pseudo_pic_reg): New function.
26727 (sparc_init_pic_reg): Likewise.
26728 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
26729 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
26731 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
26733 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26734 Add item for branch_cost.
26736 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
26738 PR rtl-optimization/83565
26739 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
26740 not extend the result to a larger mode for rotate operations.
26741 (num_sign_bit_copies1): Likewise.
26743 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26746 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
26748 Use values-Xc.o for -pedantic.
26749 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
26751 2018-01-12 Martin Liska <mliska@suse.cz>
26754 * ipa-devirt.c (final_warning_record::grow_type_warnings):
26756 (possible_polymorphic_call_targets): Use it.
26757 (ipa_devirt): Likewise.
26759 2018-01-12 Martin Liska <mliska@suse.cz>
26761 * profile-count.h (enum profile_quality): Use 0 as invalid
26762 enum value of profile_quality.
26764 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
26766 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
26767 -mext-string options.
26769 2018-01-12 Richard Biener <rguenther@suse.de>
26771 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
26772 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
26773 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
26775 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
26777 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
26779 * configure.ac (--with-long-double-format): Add support for the
26780 configuration option to change the default long double format on
26782 * config.gcc (powerpc*-linux*-*): Likewise.
26783 * configure: Regenerate.
26784 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
26785 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
26786 used without modification.
26788 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26790 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
26791 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
26792 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
26793 MISC_BUILTIN_SPEC_BARRIER.
26794 (rs6000_init_builtins): Likewise.
26795 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
26797 (speculation_barrier): New define_insn.
26798 * doc/extend.texi: Document __builtin_speculation_barrier.
26800 2018-01-11 Jakub Jelinek <jakub@redhat.com>
26803 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
26804 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
26805 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
26807 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
26808 integral modes instead of "ss" and "sd".
26809 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
26810 vectors with 32-bit and 64-bit elements.
26811 (vecdupssescalarmodesuffix): New mode attribute.
26812 (vec_dup<mode>): Use it.
26814 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
26817 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
26818 frame if argument is passed on stack.
26820 2018-01-11 Jakub Jelinek <jakub@redhat.com>
26823 * ree.c (combine_reaching_defs): Optimize also
26824 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
26825 reg2=any_extend(exp); reg1=reg2;, formatting fix.
26827 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
26829 PR middle-end/83189
26830 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
26832 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
26834 PR middle-end/83718
26835 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
26836 after they are computed.
26838 2018-01-11 Bin Cheng <bin.cheng@arm.com>
26840 PR tree-optimization/83695
26841 * gimple-loop-linterchange.cc
26842 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
26843 reset cached scev information after interchange.
26844 (pass_linterchange::execute): Remove call to scev_reset_htab.
26846 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26848 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
26849 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
26850 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
26851 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
26852 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
26853 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
26854 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
26855 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
26856 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
26857 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
26858 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
26859 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
26860 (V_lane_reg): Likewise.
26861 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
26863 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
26864 (vfmal_lane_low<mode>_intrinsic,
26865 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
26866 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
26867 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
26868 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
26869 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
26870 vfmsl_lane_high<mode>_intrinsic): New define_insns.
26872 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26874 * config/arm/arm-cpus.in (fp16fml): New feature.
26875 (ALL_SIMD): Add fp16fml.
26876 (armv8.2-a): Add fp16fml as an option.
26877 (armv8.3-a): Likewise.
26878 (armv8.4-a): Add fp16fml as part of fp16.
26879 * config/arm/arm.h (TARGET_FP16FML): Define.
26880 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
26882 * config/arm/arm-modes.def (V2HF): Define.
26883 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
26884 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
26885 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
26886 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
26887 vfmsl_low, vfmsl_high): New set of builtins.
26888 * config/arm/iterators.md (PLUSMINUS): New code iterator.
26889 (vfml_op): New code attribute.
26890 (VFMLHALVES): New int iterator.
26891 (VFML, VFMLSEL): New mode attributes.
26892 (V_reg): Define mapping for V2HF.
26893 (V_hi, V_lo): New mode attributes.
26894 (VF_constraint): Likewise.
26895 (vfml_half, vfml_half_selector): New int attributes.
26896 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
26898 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
26899 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
26901 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
26902 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
26903 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
26904 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
26906 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
26907 Document new effective target and option set.
26909 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26911 * config/arm/arm-cpus.in (armv8_4): New feature.
26912 (ARMv8_4a): New fgroup.
26913 (armv8.4-a): New arch.
26914 * config/arm/arm-tables.opt: Regenerate.
26915 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
26916 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
26917 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
26918 Add matching rules for -march=armv8.4-a and extensions.
26919 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
26921 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
26924 * config/rx/rx.md (BW): New mode attribute.
26925 (sync_lock_test_and_setsi): Add mode suffix to insn output.
26927 2018-01-11 Richard Biener <rguenther@suse.de>
26929 PR tree-optimization/83435
26930 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
26931 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
26932 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
26934 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
26935 Alan Hayward <alan.hayward@arm.com>
26936 David Sherwood <david.sherwood@arm.com>
26938 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
26940 (aarch64_classify_address): Initialize it. Track polynomial offsets.
26941 (aarch64_print_address_internal): Use it to check for a zero offset.
26943 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
26944 Alan Hayward <alan.hayward@arm.com>
26945 David Sherwood <david.sherwood@arm.com>
26947 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
26948 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
26949 Return a poly_int64 rather than a HOST_WIDE_INT.
26950 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
26951 rather than a HOST_WIDE_INT.
26952 * config/aarch64/aarch64.h (aarch64_frame): Protect with
26953 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
26954 hard_fp_offset, frame_size, initial_adjust, callee_offset and
26955 final_offset from HOST_WIDE_INT to poly_int64.
26956 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
26957 to_constant when getting the number of units in an Advanced SIMD
26959 (aarch64_builtin_vectorized_function): Check for a constant number
26961 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
26963 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
26964 attribute instead of GET_MODE_NUNITS.
26965 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
26966 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
26967 GET_MODE_SIZE for fixed-size registers.
26968 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
26969 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
26970 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
26971 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
26972 (aarch64_print_operand, aarch64_print_address_internal)
26973 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
26974 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
26975 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
26976 Handle polynomial GET_MODE_SIZE.
26977 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
26978 wider than SImode without modification.
26979 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
26980 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
26981 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
26982 passing and returning SVE modes.
26983 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
26984 rather than GEN_INT.
26985 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
26986 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
26987 (aarch64_allocate_and_probe_stack_space): Likewise.
26988 (aarch64_layout_frame): Cope with polynomial offsets.
26989 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
26990 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
26991 polynomial offsets.
26992 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
26993 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
26994 poly_int64 rather than a HOST_WIDE_INT.
26995 (aarch64_get_separate_components, aarch64_process_components)
26996 (aarch64_expand_prologue, aarch64_expand_epilogue)
26997 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
26998 (aarch64_anchor_offset): New function, split out from...
26999 (aarch64_legitimize_address): ...here.
27000 (aarch64_builtin_vectorization_cost): Handle polynomial
27001 TYPE_VECTOR_SUBPARTS.
27002 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
27004 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
27005 number of elements from the PARALLEL rather than the mode.
27006 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
27007 rather than GET_MODE_BITSIZE.
27008 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
27009 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
27010 (aarch64_expand_vec_perm_const_1): Handle polynomial
27011 d->perm.length () and d->perm elements.
27012 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
27013 Apply to_constant to d->perm elements.
27014 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
27015 polynomial CONST_VECTOR_NUNITS.
27016 (aarch64_move_pointer): Take amount as a poly_int64 rather
27018 (aarch64_progress_pointer): Avoid temporary variable.
27019 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
27020 the mode attribute instead of GET_MODE.
27022 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27023 Alan Hayward <alan.hayward@arm.com>
27024 David Sherwood <david.sherwood@arm.com>
27026 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
27027 x exists before using it.
27028 (aarch64_add_constant_internal): Rename to...
27029 (aarch64_add_offset_1): ...this. Replace regnum with separate
27030 src and dest rtxes. Handle the case in which they're different,
27031 including when the offset is zero. Replace scratchreg with an rtx.
27032 Use 2 additions if there is no spare register into which we can
27033 move a 16-bit constant.
27034 (aarch64_add_constant): Delete.
27035 (aarch64_add_offset): Replace reg with separate src and dest
27036 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
27037 Use aarch64_add_offset_1.
27038 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
27039 an rtx rather than an int. Take the delta as a poly_int64
27040 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
27041 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
27042 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
27043 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
27044 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
27045 and aarch64_add_sp.
27046 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
27047 aarch64_add_constant.
27049 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27051 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
27052 Use scalar_float_mode.
27054 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27056 * config/aarch64/aarch64-simd.md
27057 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
27058 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
27059 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
27060 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
27061 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
27062 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
27063 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
27064 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
27065 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
27066 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
27068 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27071 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
27072 targ_options->x_arm_arch_string is non NULL.
27074 2018-01-11 Tamar Christina <tamar.christina@arm.com>
27076 * config/aarch64/aarch64.h
27077 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
27079 2018-01-11 Sudakshina Das <sudi.das@arm.com>
27082 * expmed.c (emit_store_flag_force): Swap if const op0
27083 and change VOIDmode to mode of op0.
27085 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27087 PR rtl-optimization/83761
27088 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
27089 than bytes to mode_for_size.
27091 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
27093 PR middle-end/83189
27094 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
27095 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
27098 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
27100 PR middle-end/83575
27101 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
27102 when in layout mode.
27103 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
27104 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
27107 2018-01-10 Michael Collison <michael.collison@arm.com>
27109 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
27110 * config/aarch64/aarch64-option-extension.def: Add
27111 AARCH64_OPT_EXTENSION of 'fp16fml'.
27112 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27113 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
27114 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
27115 * config/aarch64/constraints.md (Ui7): New constraint.
27116 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
27117 (VFMLA_SEL_W): Ditto.
27120 (VFMLA16_LOW): New int iterator.
27121 (VFMLA16_HIGH): Ditto.
27122 (UNSPEC_FMLAL): New unspec.
27123 (UNSPEC_FMLSL): Ditto.
27124 (UNSPEC_FMLAL2): Ditto.
27125 (UNSPEC_FMLSL2): Ditto.
27126 (f16mac): New code attribute.
27127 * config/aarch64/aarch64-simd-builtins.def
27128 (aarch64_fmlal_lowv2sf): Ditto.
27129 (aarch64_fmlsl_lowv2sf): Ditto.
27130 (aarch64_fmlalq_lowv4sf): Ditto.
27131 (aarch64_fmlslq_lowv4sf): Ditto.
27132 (aarch64_fmlal_highv2sf): Ditto.
27133 (aarch64_fmlsl_highv2sf): Ditto.
27134 (aarch64_fmlalq_highv4sf): Ditto.
27135 (aarch64_fmlslq_highv4sf): Ditto.
27136 (aarch64_fmlal_lane_lowv2sf): Ditto.
27137 (aarch64_fmlsl_lane_lowv2sf): Ditto.
27138 (aarch64_fmlal_laneq_lowv2sf): Ditto.
27139 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
27140 (aarch64_fmlalq_lane_lowv4sf): Ditto.
27141 (aarch64_fmlsl_lane_lowv4sf): Ditto.
27142 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
27143 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
27144 (aarch64_fmlal_lane_highv2sf): Ditto.
27145 (aarch64_fmlsl_lane_highv2sf): Ditto.
27146 (aarch64_fmlal_laneq_highv2sf): Ditto.
27147 (aarch64_fmlsl_laneq_highv2sf): Ditto.
27148 (aarch64_fmlalq_lane_highv4sf): Ditto.
27149 (aarch64_fmlsl_lane_highv4sf): Ditto.
27150 (aarch64_fmlalq_laneq_highv4sf): Ditto.
27151 (aarch64_fmlsl_laneq_highv4sf): Ditto.
27152 * config/aarch64/aarch64-simd.md:
27153 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
27154 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
27155 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
27156 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
27157 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
27158 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
27159 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
27160 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
27161 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
27162 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
27163 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
27164 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
27165 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
27166 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
27167 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
27168 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
27169 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
27170 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
27171 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
27172 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
27173 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
27174 (vfmlsl_low_u32): Ditto.
27175 (vfmlalq_low_u32): Ditto.
27176 (vfmlslq_low_u32): Ditto.
27177 (vfmlal_high_u32): Ditto.
27178 (vfmlsl_high_u32): Ditto.
27179 (vfmlalq_high_u32): Ditto.
27180 (vfmlslq_high_u32): Ditto.
27181 (vfmlal_lane_low_u32): Ditto.
27182 (vfmlsl_lane_low_u32): Ditto.
27183 (vfmlal_laneq_low_u32): Ditto.
27184 (vfmlsl_laneq_low_u32): Ditto.
27185 (vfmlalq_lane_low_u32): Ditto.
27186 (vfmlslq_lane_low_u32): Ditto.
27187 (vfmlalq_laneq_low_u32): Ditto.
27188 (vfmlslq_laneq_low_u32): Ditto.
27189 (vfmlal_lane_high_u32): Ditto.
27190 (vfmlsl_lane_high_u32): Ditto.
27191 (vfmlal_laneq_high_u32): Ditto.
27192 (vfmlsl_laneq_high_u32): Ditto.
27193 (vfmlalq_lane_high_u32): Ditto.
27194 (vfmlslq_lane_high_u32): Ditto.
27195 (vfmlalq_laneq_high_u32): Ditto.
27196 (vfmlslq_laneq_high_u32): Ditto.
27197 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
27198 (AARCH64_FL_FOR_ARCH8_4): New.
27199 (AARCH64_ISA_F16FML): New ISA flag.
27200 (TARGET_F16FML): New feature flag for fp16fml.
27201 (doc/invoke.texi): Document new fp16fml option.
27203 2018-01-10 Michael Collison <michael.collison@arm.com>
27205 * config/aarch64/aarch64-builtins.c:
27206 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
27207 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27208 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
27209 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
27210 (AARCH64_ISA_SHA3): New ISA flag.
27211 (TARGET_SHA3): New feature flag for sha3.
27212 * config/aarch64/iterators.md (sha512_op): New int attribute.
27213 (CRYPTO_SHA512): New int iterator.
27214 (UNSPEC_SHA512H): New unspec.
27215 (UNSPEC_SHA512H2): Ditto.
27216 (UNSPEC_SHA512SU0): Ditto.
27217 (UNSPEC_SHA512SU1): Ditto.
27218 * config/aarch64/aarch64-simd-builtins.def
27219 (aarch64_crypto_sha512hqv2di): New builtin.
27220 (aarch64_crypto_sha512h2qv2di): Ditto.
27221 (aarch64_crypto_sha512su0qv2di): Ditto.
27222 (aarch64_crypto_sha512su1qv2di): Ditto.
27223 (aarch64_eor3qv8hi): Ditto.
27224 (aarch64_rax1qv2di): Ditto.
27225 (aarch64_xarqv2di): Ditto.
27226 (aarch64_bcaxqv8hi): Ditto.
27227 * config/aarch64/aarch64-simd.md:
27228 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
27229 (aarch64_crypto_sha512su0qv2di): Ditto.
27230 (aarch64_crypto_sha512su1qv2di): Ditto.
27231 (aarch64_eor3qv8hi): Ditto.
27232 (aarch64_rax1qv2di): Ditto.
27233 (aarch64_xarqv2di): Ditto.
27234 (aarch64_bcaxqv8hi): Ditto.
27235 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
27236 (vsha512h2q_u64): Ditto.
27237 (vsha512su0q_u64): Ditto.
27238 (vsha512su1q_u64): Ditto.
27239 (veor3q_u16): Ditto.
27240 (vrax1q_u64): Ditto.
27241 (vxarq_u64): Ditto.
27242 (vbcaxq_u16): Ditto.
27243 * config/arm/types.md (crypto_sha512): New type attribute.
27244 (crypto_sha3): Ditto.
27245 (doc/invoke.texi): Document new sha3 option.
27247 2018-01-10 Michael Collison <michael.collison@arm.com>
27249 * config/aarch64/aarch64-builtins.c:
27250 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
27251 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27252 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
27253 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
27254 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
27255 (AARCH64_ISA_SM4): New ISA flag.
27256 (TARGET_SM4): New feature flag for sm4.
27257 * config/aarch64/aarch64-simd-builtins.def
27258 (aarch64_sm3ss1qv4si): Ditto.
27259 (aarch64_sm3tt1aq4si): Ditto.
27260 (aarch64_sm3tt1bq4si): Ditto.
27261 (aarch64_sm3tt2aq4si): Ditto.
27262 (aarch64_sm3tt2bq4si): Ditto.
27263 (aarch64_sm3partw1qv4si): Ditto.
27264 (aarch64_sm3partw2qv4si): Ditto.
27265 (aarch64_sm4eqv4si): Ditto.
27266 (aarch64_sm4ekeyqv4si): Ditto.
27267 * config/aarch64/aarch64-simd.md:
27268 (aarch64_sm3ss1qv4si): Ditto.
27269 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
27270 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
27271 (aarch64_sm4eqv4si): Ditto.
27272 (aarch64_sm4ekeyqv4si): Ditto.
27273 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
27274 (sm3part_op): Ditto.
27275 (CRYPTO_SM3TT): Ditto.
27276 (CRYPTO_SM3PART): Ditto.
27277 (UNSPEC_SM3SS1): New unspec.
27278 (UNSPEC_SM3TT1A): Ditto.
27279 (UNSPEC_SM3TT1B): Ditto.
27280 (UNSPEC_SM3TT2A): Ditto.
27281 (UNSPEC_SM3TT2B): Ditto.
27282 (UNSPEC_SM3PARTW1): Ditto.
27283 (UNSPEC_SM3PARTW2): Ditto.
27284 (UNSPEC_SM4E): Ditto.
27285 (UNSPEC_SM4EKEY): Ditto.
27286 * config/aarch64/constraints.md (Ui2): New constraint.
27287 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
27288 * config/arm/types.md (crypto_sm3): New type attribute.
27289 (crypto_sm4): Ditto.
27290 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
27291 (vsm3tt1aq_u32): Ditto.
27292 (vsm3tt1bq_u32): Ditto.
27293 (vsm3tt2aq_u32): Ditto.
27294 (vsm3tt2bq_u32): Ditto.
27295 (vsm3partw1q_u32): Ditto.
27296 (vsm3partw2q_u32): Ditto.
27297 (vsm4eq_u32): Ditto.
27298 (vsm4ekeyq_u32): Ditto.
27299 (doc/invoke.texi): Document new sm4 option.
27301 2018-01-10 Michael Collison <michael.collison@arm.com>
27303 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
27304 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
27305 (AARCH64_FL_FOR_ARCH8_4): New.
27306 (AARCH64_FL_V8_4): New flag.
27307 (doc/invoke.texi): Document new armv8.4-a option.
27309 2018-01-10 Michael Collison <michael.collison@arm.com>
27311 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27312 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
27313 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
27314 * config/aarch64/aarch64-option-extension.def: Add
27315 AARCH64_OPT_EXTENSION of 'sha2'.
27316 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
27317 (crypto): Disable sha2 and aes if crypto disabled.
27318 (crypto): Enable aes and sha2 if enabled.
27319 (simd): Disable sha2 and aes if simd disabled.
27320 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
27322 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
27323 (TARGET_SHA2): New feature flag for sha2.
27324 (TARGET_AES): New feature flag for aes.
27325 * config/aarch64/aarch64-simd.md:
27326 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
27327 conditional on TARGET_AES.
27328 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
27329 (aarch64_crypto_sha1hsi): Make pattern conditional
27331 (aarch64_crypto_sha1hv4si): Ditto.
27332 (aarch64_be_crypto_sha1hv4si): Ditto.
27333 (aarch64_crypto_sha1su1v4si): Ditto.
27334 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
27335 (aarch64_crypto_sha1su0v4si): Ditto.
27336 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
27337 (aarch64_crypto_sha256su0v4si): Ditto.
27338 (aarch64_crypto_sha256su1v4si): Ditto.
27339 (doc/invoke.texi): Document new aes and sha2 options.
27341 2018-01-10 Martin Sebor <msebor@redhat.com>
27343 PR tree-optimization/83781
27344 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
27347 2018-01-11 Martin Sebor <msebor@gmail.com>
27348 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27350 PR tree-optimization/83501
27351 PR tree-optimization/81703
27353 * tree-ssa-strlen.c (get_string_cst): Rename...
27354 (get_string_len): ...to this. Handle global constants.
27355 (handle_char_store): Adjust.
27357 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
27358 Jim Wilson <jimw@sifive.com>
27360 * config/riscv/riscv-protos.h (riscv_output_return): New.
27361 * config/riscv/riscv.c (struct machine_function): New naked_p field.
27362 (riscv_attribute_table, riscv_output_return),
27363 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
27364 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
27365 (riscv_compute_frame_info): Only compute frame->mask if not a naked
27367 (riscv_expand_prologue): Add early return for naked function.
27368 (riscv_expand_epilogue): Likewise.
27369 (riscv_function_ok_for_sibcall): Return false for naked function.
27370 (riscv_set_current_function): New.
27371 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
27372 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
27373 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
27374 * doc/extend.texi (RISC-V Function Attributes): New.
27376 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
27378 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
27379 check for 128-bit long double before checking TCmode.
27380 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
27381 128-bit long doubles before checking TFmode or TCmode.
27382 (FLOAT128_IBM_P): Likewise.
27384 2018-01-10 Martin Sebor <msebor@redhat.com>
27386 PR tree-optimization/83671
27387 * builtins.c (c_strlen): Unconditionally return zero for the empty
27389 Use -Warray-bounds for warnings.
27390 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
27391 for non-constant array indices with COMPONENT_REF, arrays of
27392 arrays, and pointers to arrays.
27393 (gimple_fold_builtin_strlen): Determine and set length range for
27394 non-constant character arrays.
27396 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
27398 PR middle-end/81897
27399 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
27402 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
27404 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
27406 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
27409 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
27410 VECTOR_MEM_ALTIVEC_OR_VSX_P.
27411 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
27412 indexed_or_indirect_operand predicate.
27413 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
27414 (*vsx_le_perm_load_v8hi): Likewise.
27415 (*vsx_le_perm_load_v16qi): Likewise.
27416 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
27417 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
27418 (*vsx_le_perm_store_v8hi): Likewise.
27419 (*vsx_le_perm_store_v16qi): Likewise.
27420 (eight unnamed splitters): Likewise.
27422 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
27424 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
27425 * config/rs6000/emmintrin.h: Likewise.
27426 * config/rs6000/mmintrin.h: Likewise.
27427 * config/rs6000/xmmintrin.h: Likewise.
27429 2018-01-10 David Malcolm <dmalcolm@redhat.com>
27432 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
27434 * tree.c (tree_nop_conversion): Return true for location wrapper
27436 (maybe_wrap_with_location): New function.
27437 (selftest::check_strip_nops): New function.
27438 (selftest::test_location_wrappers): New function.
27439 (selftest::tree_c_tests): Call it.
27440 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
27441 (maybe_wrap_with_location): New decl.
27442 (EXPR_LOCATION_WRAPPER_P): New macro.
27443 (location_wrapper_p): New inline function.
27444 (tree_strip_any_location_wrapper): New inline function.
27446 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
27449 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
27450 stack_realign_offset for the largest alignment of stack slot
27452 (ix86_find_max_used_stack_alignment): New function.
27453 (ix86_finalize_stack_frame_flags): Use it. Set
27454 max_used_stack_alignment if we don't realign stack.
27455 * config/i386/i386.h (machine_function): Add
27456 max_used_stack_alignment.
27458 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
27460 * config/arm/arm.opt (-mbranch-cost): New option.
27461 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
27464 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
27467 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
27468 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
27470 2018-01-10 Richard Biener <rguenther@suse.de>
27473 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
27474 early out so it also covers the case where we have a non-NULL
27477 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
27479 PR tree-optimization/83753
27480 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
27481 for non-strided grouped accesses if the number of elements is 1.
27483 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
27486 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
27487 * i386.h (TARGET_USE_GATHER): Define.
27488 * x86-tune.def (X86_TUNE_USE_GATHER): New.
27490 2018-01-10 Martin Liska <mliska@suse.cz>
27493 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
27494 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
27496 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
27497 CLEANUP_NO_PARTITIONING is not set.
27499 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
27501 * doc/rtl.texi: Remove documentation of (const ...) wrappers
27502 for vectors, as a partial revert of r254296.
27503 * rtl.h (const_vec_p): Delete.
27504 (const_vec_duplicate_p): Don't test for vector CONSTs.
27505 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
27506 * expmed.c (make_tree): Likewise.
27509 * common.md (E, F): Use CONSTANT_P instead of checking for
27511 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
27512 checking for CONST_VECTOR.
27514 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
27516 PR middle-end/83575
27517 * predict.c (force_edge_cold): Handle in more sane way edges
27518 with no prediction.
27520 2018-01-09 Carl Love <cel@us.ibm.com>
27522 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
27524 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
27525 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
27526 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
27527 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
27528 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
27529 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
27530 * config/rs6000/rs6000-protos.h: Add extern defition for
27531 rs6000_generate_float2_double_code.
27532 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
27534 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
27535 (float2_v2df): Add define_expand.
27537 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
27540 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
27541 op_mode in the force_to_mode call.
27543 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
27545 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
27546 instead of checking each element individually.
27547 (aarch64_evpc_uzp): Likewise.
27548 (aarch64_evpc_zip): Likewise.
27549 (aarch64_evpc_ext): Likewise.
27550 (aarch64_evpc_rev): Likewise.
27551 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
27552 instead of checking each element individually. Return true without
27554 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
27555 whether all selected elements come from the same input, instead of
27556 checking each element individually. Remove calls to gen_rtx_REG,
27557 start_sequence and end_sequence and instead assert that no rtl is
27560 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
27562 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
27563 order of HIGH and CONST checks.
27565 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
27567 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
27568 if the destination isn't an SSA_NAME.
27570 2018-01-09 Richard Biener <rguenther@suse.de>
27572 PR tree-optimization/83668
27573 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
27575 (canonicalize_loop_form): ... here, renamed from ...
27576 (canonicalize_loop_closed_ssa_form): ... this and amended to
27577 swap successor edges for loop exit blocks to make us use
27578 the RPO order we need for initial schedule generation.
27580 2018-01-09 Joseph Myers <joseph@codesourcery.com>
27582 PR tree-optimization/64811
27583 * match.pd: When optimizing comparisons with Inf, avoid
27584 introducing or losing exceptions from comparisons with NaN.
27586 2018-01-09 Martin Liska <mliska@suse.cz>
27589 * asan.c (shadow_mem_size): Add gcc_assert.
27591 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
27593 Don't save registers in main().
27596 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
27597 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
27598 * config/avr/avr.c (avr_set_current_function): Don't error if
27599 naked, OS_task or OS_main are specified at the same time.
27600 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
27602 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
27604 * common/config/avr/avr-common.c (avr_option_optimization_table):
27605 Switch on -mmain-is-OS_task for optimizing compilations.
27607 2018-01-09 Richard Biener <rguenther@suse.de>
27609 PR tree-optimization/83572
27610 * graphite.c: Include cfganal.h.
27611 (graphite_transform_loops): Connect infinite loops to exit
27612 and remove fake edges at the end.
27614 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
27616 * ipa-inline.c (edge_badness): Revert accidental checkin.
27618 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
27621 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
27622 symbols; not inline clones.
27624 2018-01-09 Jakub Jelinek <jakub@redhat.com>
27627 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
27628 hard registers. Formatting fixes.
27630 PR preprocessor/83722
27631 * gcc.c (try_generate_repro): Pass
27632 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
27633 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
27636 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
27637 Kito Cheng <kito.cheng@gmail.com>
27639 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
27640 (riscv_leaf_function_p): Delete.
27641 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
27643 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27645 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
27647 (do_ifelse): New function.
27648 (do_isel): New function.
27649 (do_sub3): New function.
27650 (do_add3): New function.
27651 (do_load_mask_compare): New function.
27652 (do_overlap_load_compare): New function.
27653 (expand_compare_loop): New function.
27654 (expand_block_compare): Call expand_compare_loop() when appropriate.
27655 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
27656 option description.
27657 (-mblock-compare-inline-loop-limit): New option.
27659 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27662 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
27663 Reverse order of second and third operands in first alternative.
27664 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
27665 of first and second elements in UNSPEC_VPERMR vector.
27666 (altivec_expand_vec_perm_le): Likewise.
27668 2018-01-08 Jeff Law <law@redhat.com>
27670 PR rtl-optimizatin/81308
27671 * tree-switch-conversion.c (cfg_altered): New file scoped static.
27672 (process_switch): If group_case_labels makes a change, then set
27674 (pass_convert_switch::execute): If a switch is converted, then
27675 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
27677 PR rtl-optimization/81308
27678 * recog.c (split_all_insns): Conditionally cleanup the CFG after
27681 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
27683 PR target/83663 - Revert r255946
27684 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
27685 generation for cases where splatting a value is not useful.
27686 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
27687 across a vec_duplicate and a paradoxical subreg forming a vector
27688 mode to a vec_concat.
27690 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27692 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
27693 -march=armv8.3-a variants.
27694 * config/arm/t-multilib: Likewise.
27695 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
27697 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27699 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
27701 (cceq_ior_compare_complement): Give it a name so I can use it, and
27702 change boolean_or_operator predicate to boolean_operator so it can
27703 be used to generate a crand.
27704 (eqne): New code iterator.
27705 (bd/bd_neg): New code_attrs.
27706 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
27707 a single define_insn.
27708 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
27709 decrement (bdnzt/bdnzf/bdzt/bdzf).
27710 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
27711 with the new names of the branch decrement patterns, and added the
27712 names of the branch decrement conditional patterns.
27714 2018-01-08 Richard Biener <rguenther@suse.de>
27716 PR tree-optimization/83563
27717 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
27720 2018-01-08 Richard Biener <rguenther@suse.de>
27722 PR middle-end/83713
27723 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
27725 2018-01-08 Richard Biener <rguenther@suse.de>
27727 PR tree-optimization/83685
27728 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
27729 references to abnormals.
27731 2018-01-08 Richard Biener <rguenther@suse.de>
27734 * dwarf2out.c (output_indirect_strings): Handle empty
27735 skeleton_debug_str_hash.
27736 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
27738 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
27740 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
27741 (emit_store_direct): Likewise.
27742 (arc_trampoline_adjust_address): Likewise.
27743 (arc_asm_trampoline_template): New function.
27744 (arc_initialize_trampoline): Use asm_trampoline_template.
27745 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
27746 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
27747 * config/arc/arc.md (flush_icache): Delete pattern.
27749 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
27751 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
27752 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
27755 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
27758 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
27759 by not USED_FOR_TARGET.
27760 (make_pass_resolve_sw_modes): Likewise.
27762 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
27764 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
27767 2018-01-08 Richard Biener <rguenther@suse.de>
27769 PR middle-end/83580
27770 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
27772 2018-01-08 Richard Biener <rguenther@suse.de>
27774 PR middle-end/83517
27775 * match.pd ((t * 2) / 2) -> t): Add missing :c.
27777 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
27779 PR middle-end/81897
27780 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
27781 basic blocks with a small number of successors.
27782 (convert_control_dep_chain_into_preds): Improve handling of
27784 (dump_predicates): Split apart into...
27785 (dump_pred_chain): ...here...
27786 (dump_pred_info): ...and here.
27787 (can_one_predicate_be_invalidated_p): Add debugging printfs.
27788 (can_chain_union_be_invalidated_p): Improve check for invalidation
27790 (uninit_uses_cannot_happen): Avoid unnecessary if
27791 convert_control_dep_chain_into_preds yielded nothing.
27793 2018-01-06 Martin Sebor <msebor@redhat.com>
27795 PR tree-optimization/83640
27796 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
27797 subtracting negative offset from size.
27798 (builtin_access::overlap): Adjust offset bounds of the access to fall
27799 within the size of the object if possible.
27801 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
27803 PR rtl-optimization/83699
27804 * expmed.c (extract_bit_field_1): Restrict the vector usage of
27805 extract_bit_field_as_subreg to cases in which the extracted
27806 value is also a vector.
27808 * lra-constraints.c (process_alt_operands): Test for the equivalence
27809 substitutions when detecting a possible reload cycle.
27811 2018-01-06 Jakub Jelinek <jakub@redhat.com>
27814 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
27815 by default if flag_selective_schedling{,2}. Formatting fixes.
27817 PR rtl-optimization/83682
27818 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
27819 if it has non-VECTOR_MODE element mode.
27820 (vec_duplicate_p): Likewise.
27822 PR middle-end/83694
27823 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
27824 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
27826 2018-01-05 Jakub Jelinek <jakub@redhat.com>
27829 * config/i386/i386-builtin.def
27830 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
27831 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
27832 Require also OPTION_MASK_ISA_AVX512F in addition to
27833 OPTION_MASK_ISA_GFNI.
27834 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
27835 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
27836 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
27837 to OPTION_MASK_ISA_GFNI.
27838 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
27839 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
27840 OPTION_MASK_ISA_AVX512BW.
27841 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
27842 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
27843 addition to OPTION_MASK_ISA_GFNI.
27844 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
27845 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
27846 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
27847 to OPTION_MASK_ISA_GFNI.
27848 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
27849 a requirement for all ISAs rather than any of them with a few
27851 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
27853 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
27854 bitmasks to be enabled with 3 exceptions, instead of requiring any
27855 enabled ISA with lots of exceptions.
27856 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
27857 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
27858 Change avx512bw in isa attribute to avx512f.
27859 * config/i386/sgxintrin.h: Add license boilerplate.
27860 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
27861 to __AVX512F__ and __AVX512VL to __AVX512VL__.
27862 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
27863 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
27865 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
27866 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
27867 temporarily sse2 rather than sse if not enabled already.
27870 * config/i386/sse.md (VI248_VLBW): Rename to ...
27871 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
27872 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
27873 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
27874 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
27875 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
27876 mode iterator instead of VI248_VLBW.
27878 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
27880 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
27881 (record_modified): Skip clobbers; add debug output.
27882 (param_change_prob): Use sreal frequencies.
27884 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
27886 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
27887 punt for user-aligned variables.
27889 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
27891 * tree-chrec.c (chrec_contains_symbols): Return true for
27894 2018-01-05 Sudakshina Das <sudi.das@arm.com>
27897 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
27898 of (x|y) == x for BICS pattern.
27900 2018-01-05 Jakub Jelinek <jakub@redhat.com>
27902 PR tree-optimization/83605
27903 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
27904 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
27907 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
27909 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
27910 * config/epiphany/rtems.h: New file.
27912 2018-01-04 Jakub Jelinek <jakub@redhat.com>
27913 Uros Bizjak <ubizjak@gmail.com>
27916 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
27917 QIreg_operand instead of register_operand predicate.
27918 * config/i386/i386.c (ix86_rop_should_change_byte_p,
27919 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
27920 comments instead of -fmitigate[-_]rop.
27922 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27925 * cgraphunit.c (symbol_table::compile): Switch to text_section
27926 before calling assembly_start debug hook.
27927 * run-rtl-passes.c (run_rtl_passes): Likewise.
27930 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27932 * tree-vrp.c (extract_range_from_binary_expr_1): Check
27933 range_int_cst_p rather than !symbolic_range_p before calling
27934 extract_range_from_multiplicative_op_1.
27936 2018-01-04 Jeff Law <law@redhat.com>
27938 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
27939 redundant test in assertion.
27941 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27943 * doc/rtl.texi: Document machine_mode wrapper classes.
27945 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27947 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
27948 using tree_to_uhwi.
27950 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27952 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
27953 the VEC_PERM_EXPR fold to fail.
27955 2018-01-04 Jakub Jelinek <jakub@redhat.com>
27958 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
27959 to switched_sections.
27961 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27964 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
27965 test for d.testing.
27967 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
27970 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
27971 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
27973 2018-01-04 Jakub Jelinek <jakub@redhat.com>
27976 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
27977 is BLKmode and bitpos not zero or mode change is needed.
27979 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
27982 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
27985 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
27988 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
27989 instead of MULT rtx. Update all corresponding splitters.
27990 (*saddl_se): Ditto.
27991 (*ssub<modesuffix>): Ditto.
27992 (*ssubl_se): Ditto.
27993 (*cmp_sadd_di): Update split patterns.
27994 (*cmp_sadd_si): Ditto.
27995 (*cmp_sadd_sidi): Ditto.
27996 (*cmp_ssub_di): Ditto.
27997 (*cmp_ssub_si): Ditto.
27998 (*cmp_ssub_sidi): Ditto.
27999 * config/alpha/predicates.md (const23_operand): New predicate.
28000 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
28001 Look for ASHIFT, not MULT inner operand.
28002 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
28004 2018-01-04 Martin Liska <mliska@suse.cz>
28006 PR gcov-profile/83669
28007 * gcov.c (output_intermediate_file): Add version to intermediate
28009 * doc/gcov.texi: Document new field 'version' in intermediate
28010 file format. Fix location of '-k' option of gcov command.
28012 2018-01-04 Martin Liska <mliska@suse.cz>
28015 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
28017 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28019 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
28021 2018-01-03 Martin Sebor <msebor@redhat.com>
28023 PR tree-optimization/83655
28024 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
28025 checking calls with invalid arguments.
28027 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28029 * tree-vect-stmts.c (vect_get_store_rhs): New function.
28030 (vectorizable_mask_load_store): Delete.
28031 (vectorizable_call): Return false for masked loads and stores.
28032 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
28033 instead of gimple_assign_rhs1.
28034 (vectorizable_load): Handle IFN_MASK_LOAD.
28035 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
28037 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28039 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
28041 (vectorizable_mask_load_store): ...here.
28042 (vectorizable_load): ...and here.
28044 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28046 * tree-vect-stmts.c (vect_build_all_ones_mask)
28047 (vect_build_zero_merge_argument): New functions, split out from...
28048 (vectorizable_load): ...here.
28050 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28052 * tree-vect-stmts.c (vect_check_store_rhs): New function,
28054 (vectorizable_mask_load_store): ...here.
28055 (vectorizable_store): ...and here.
28057 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28059 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
28061 (vectorizable_mask_load_store): ...here.
28063 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28065 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
28066 (vect_model_store_cost): Take a vec_load_store_type instead of a
28068 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
28069 (vect_model_store_cost): Take a vec_load_store_type instead of a
28071 (vectorizable_mask_load_store): Update accordingly.
28072 (vectorizable_store): Likewise.
28073 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
28075 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28077 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
28078 IFN_MASK_LOAD calls here rather than...
28079 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
28081 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28082 Alan Hayward <alan.hayward@arm.com>
28083 David Sherwood <david.sherwood@arm.com>
28085 * expmed.c (extract_bit_field_1): For vector extracts,
28086 fall back to extract_bit_field_as_subreg if vec_extract
28089 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28090 Alan Hayward <alan.hayward@arm.com>
28091 David Sherwood <david.sherwood@arm.com>
28093 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
28094 they are variable or constant sized.
28095 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
28096 slots for constant-sized data.
28098 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28099 Alan Hayward <alan.hayward@arm.com>
28100 David Sherwood <david.sherwood@arm.com>
28102 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
28103 handling COND_EXPRs with boolean comparisons, try to find a better
28104 basis for the mask type than the boolean itself.
28106 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28108 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
28109 is calculated and how it can be overridden.
28110 * genmodes.c (max_bitsize_mode_any_mode): New variable.
28111 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
28113 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
28116 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28117 Alan Hayward <alan.hayward@arm.com>
28118 David Sherwood <david.sherwood@arm.com>
28120 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
28121 Remove the mode argument.
28122 (aarch64_simd_valid_immediate): Remove the mode and inverse
28124 * config/aarch64/iterators.md (bitsize): New iterator.
28125 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
28126 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
28127 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
28128 aarch64_simd_valid_immediate.
28129 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
28130 (aarch64_reg_or_bic_imm): Likewise.
28131 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
28132 with an insn_type enum and msl with a modifier_type enum.
28133 Replace element_width with a scalar_mode. Change the shift
28134 to unsigned int. Add constructors for scalar_float_mode and
28135 scalar_int_mode elements.
28136 (aarch64_vect_float_const_representable_p): Delete.
28137 (aarch64_can_const_movi_rtx_p)
28138 (aarch64_simd_scalar_immediate_valid_for_move)
28139 (aarch64_simd_make_constant): Update call to
28140 aarch64_simd_valid_immediate.
28141 (aarch64_advsimd_valid_immediate_hs): New function.
28142 (aarch64_advsimd_valid_immediate): Likewise.
28143 (aarch64_simd_valid_immediate): Remove mode and inverse
28144 arguments. Rewrite to use the above. Use const_vec_duplicate_p
28145 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
28146 and aarch64_float_const_representable_p on the result.
28147 (aarch64_output_simd_mov_immediate): Remove mode argument.
28148 Update call to aarch64_simd_valid_immediate and use of
28149 simd_immediate_info.
28150 (aarch64_output_scalar_simd_mov_immediate): Update call
28153 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28154 Alan Hayward <alan.hayward@arm.com>
28155 David Sherwood <david.sherwood@arm.com>
28157 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
28158 (mode_nunits): Likewise CONST_MODE_NUNITS.
28159 * machmode.def (ADJUST_NUNITS): Document.
28160 * genmodes.c (mode_data::need_nunits_adj): New field.
28161 (blank_mode): Update accordingly.
28162 (adj_nunits): New variable.
28163 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
28165 (emit_mode_size_inline): Set need_bytesize_adj for all modes
28166 listed in adj_nunits.
28167 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
28168 listed in adj_nunits. Don't emit case statements for such modes.
28169 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
28170 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
28171 nothing if adj_nunits is nonnull.
28172 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
28173 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
28174 (emit_mode_fbit): Update use of print_maybe_const_decl.
28175 (emit_move_size): Likewise. Treat the array as non-const
28177 (emit_mode_adjustments): Handle adj_nunits.
28179 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28181 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
28182 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
28183 (VECTOR_MODES): Use it.
28184 (make_vector_modes): Take the prefix as an argument.
28186 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28187 Alan Hayward <alan.hayward@arm.com>
28188 David Sherwood <david.sherwood@arm.com>
28190 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
28191 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
28192 for MODE_VECTOR_BOOL.
28193 * machmode.def (VECTOR_BOOL_MODE): Document.
28194 * genmodes.c (VECTOR_BOOL_MODE): New macro.
28195 (make_vector_bool_mode): New function.
28196 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
28198 * lto-streamer-in.c (lto_input_mode_table): Likewise.
28199 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
28201 * stor-layout.c (int_mode_for_mode): Likewise.
28202 * tree.c (build_vector_type_for_mode): Likewise.
28203 * varasm.c (output_constant_pool_2): Likewise.
28204 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
28205 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
28206 for MODE_VECTOR_BOOL.
28207 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
28208 of mode class checks.
28209 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
28210 instead of a list of mode class checks.
28211 (expand_vector_scalar_condition): Likewise.
28212 (type_for_widest_vector_mode): Handle BImode as an inner mode.
28214 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28215 Alan Hayward <alan.hayward@arm.com>
28216 David Sherwood <david.sherwood@arm.com>
28218 * machmode.h (mode_size): Change from unsigned short to
28220 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
28221 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
28222 or if measurement_type is not polynomial.
28223 (fixed_size_mode::includes_p): Check for constant-sized modes.
28224 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
28225 return a poly_uint16 rather than an unsigned short.
28226 (emit_mode_size): Change the type of mode_size from unsigned short
28227 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
28228 (emit_mode_adjustments): Cope with polynomial vector sizes.
28229 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
28231 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
28233 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
28234 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
28235 * caller-save.c (setup_save_areas): Likewise.
28236 (replace_reg_with_saved_mem): Likewise.
28237 * calls.c (emit_library_call_value_1): Likewise.
28238 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
28239 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
28240 (gen_lowpart_for_combine): Likewise.
28241 * convert.c (convert_to_integer_1): Likewise.
28242 * cse.c (equiv_constant, cse_insn): Likewise.
28243 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
28244 (cselib_subst_to_values): Likewise.
28245 * dce.c (word_dce_process_block): Likewise.
28246 * df-problems.c (df_word_lr_mark_ref): Likewise.
28247 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
28248 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
28249 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
28250 (rtl_for_decl_location): Likewise.
28251 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
28252 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
28253 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
28254 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
28255 (expand_expr_real_1): Likewise.
28256 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
28257 (pad_below): Likewise.
28258 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
28259 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
28260 * ira.c (get_subreg_tracking_sizes): Likewise.
28261 * ira-build.c (ira_create_allocno_objects): Likewise.
28262 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
28263 (ira_sort_regnos_for_alter_reg): Likewise.
28264 * ira-costs.c (record_operand_costs): Likewise.
28265 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
28266 (resolve_simple_move): Likewise.
28267 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
28268 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
28269 (lra_constraints): Likewise.
28270 (CONST_POOL_OK_P): Reject variable-sized modes.
28271 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
28272 (add_pseudo_to_slot, lra_spill): Likewise.
28273 * omp-low.c (omp_clause_aligned_alignment): Likewise.
28274 * optabs-query.c (get_best_extraction_insn): Likewise.
28275 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
28276 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
28277 (expand_mult_highpart, valid_multiword_target_p): Likewise.
28278 * recog.c (offsettable_address_addr_space_p): Likewise.
28279 * regcprop.c (maybe_mode_change): Likewise.
28280 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
28281 * regrename.c (build_def_use): Likewise.
28282 * regstat.c (dump_reg_info): Likewise.
28283 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
28284 (find_reloads, find_reloads_subreg_address): Likewise.
28285 * reload1.c (eliminate_regs_1): Likewise.
28286 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
28287 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
28288 (simplify_binary_operation_1, simplify_subreg): Likewise.
28289 * targhooks.c (default_function_arg_padding): Likewise.
28290 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
28291 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
28292 (verify_gimple_assign_ternary): Likewise.
28293 * tree-inline.c (estimate_move_cost): Likewise.
28294 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
28295 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
28296 (get_address_cost_ainc): Likewise.
28297 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28298 (vect_supportable_dr_alignment): Likewise.
28299 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28300 (vectorizable_reduction): Likewise.
28301 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
28302 (vectorizable_operation, vectorizable_load): Likewise.
28303 * tree.c (build_same_sized_truth_vector_type): Likewise.
28304 * valtrack.c (cleanup_auto_inc_dec): Likewise.
28305 * var-tracking.c (emit_note_insn_var_location): Likewise.
28306 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
28307 (ADDR_VEC_ALIGN): Likewise.
28309 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28310 Alan Hayward <alan.hayward@arm.com>
28311 David Sherwood <david.sherwood@arm.com>
28313 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
28315 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
28316 or if measurement_type is polynomial.
28317 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
28318 * combine.c (make_extraction): Likewise.
28319 * dse.c (find_shift_sequence): Likewise.
28320 * dwarf2out.c (mem_loc_descriptor): Likewise.
28321 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
28322 (extract_bit_field, extract_low_bits): Likewise.
28323 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
28324 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
28325 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
28326 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
28327 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
28328 * reload.c (find_reloads): Likewise.
28329 * reload1.c (alter_reg): Likewise.
28330 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
28331 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
28332 * tree-if-conv.c (predicate_mem_writes): Likewise.
28333 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
28334 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
28335 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
28336 * valtrack.c (dead_debug_insert_temp): Likewise.
28337 * varasm.c (mergeable_constant_section): Likewise.
28338 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
28340 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28341 Alan Hayward <alan.hayward@arm.com>
28342 David Sherwood <david.sherwood@arm.com>
28344 * expr.c (expand_assignment): Cope with polynomial mode sizes
28345 when assigning to a CONCAT.
28347 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28348 Alan Hayward <alan.hayward@arm.com>
28349 David Sherwood <david.sherwood@arm.com>
28351 * machmode.h (mode_precision): Change from unsigned short to
28353 (mode_to_precision): Return a poly_uint16 rather than an unsigned
28355 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
28356 or if measurement_type is not polynomial.
28357 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
28358 in which the mode is already known to be a scalar_int_mode.
28359 * genmodes.c (emit_mode_precision): Change the type of mode_precision
28360 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
28362 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
28363 for GET_MODE_PRECISION.
28364 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
28365 for GET_MODE_PRECISION.
28366 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
28368 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
28369 (expand_field_assignment, make_extraction): Likewise.
28370 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
28371 (get_last_value): Likewise.
28372 * convert.c (convert_to_integer_1): Likewise.
28373 * cse.c (cse_insn): Likewise.
28374 * expr.c (expand_expr_real_1): Likewise.
28375 * lra-constraints.c (simplify_operand_subreg): Likewise.
28376 * optabs-query.c (can_atomic_load_p): Likewise.
28377 * optabs.c (expand_atomic_load): Likewise.
28378 (expand_atomic_store): Likewise.
28379 * ree.c (combine_reaching_defs): Likewise.
28380 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
28381 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
28382 * tree.h (type_has_mode_precision_p): Likewise.
28383 * ubsan.c (instrument_si_overflow): Likewise.
28385 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28386 Alan Hayward <alan.hayward@arm.com>
28387 David Sherwood <david.sherwood@arm.com>
28389 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
28390 polynomial numbers of units.
28391 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
28392 (valid_vector_subparts_p): New function.
28393 (build_vector_type): Remove temporary shim and take the number
28394 of units as a poly_uint64 rather than an int.
28395 (build_opaque_vector_type): Take the number of units as a
28396 poly_uint64 rather than an int.
28397 * tree.c (build_vector_from_ctor): Handle polynomial
28398 TYPE_VECTOR_SUBPARTS.
28399 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
28400 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
28401 (build_vector_from_val): If the number of units is variable,
28402 use build_vec_duplicate_cst for constant operands and
28403 VEC_DUPLICATE_EXPR otherwise.
28404 (make_vector_type): Remove temporary is_constant ().
28405 (build_vector_type, build_opaque_vector_type): Take the number of
28406 units as a poly_uint64 rather than an int.
28407 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
28409 * cfgexpand.c (expand_debug_expr): Likewise.
28410 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
28411 (store_constructor, expand_expr_real_1): Likewise.
28412 (const_scalar_mask_from_tree): Likewise.
28413 * fold-const-call.c (fold_const_reduction): Likewise.
28414 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
28415 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
28416 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
28417 (fold_relational_const): Likewise.
28418 (native_interpret_vector): Likewise. Change the size from an
28419 int to an unsigned int.
28420 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
28421 TYPE_VECTOR_SUBPARTS.
28422 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
28423 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
28424 duplicating a non-constant operand into a variable-length vector.
28425 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
28426 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
28427 * ipa-icf.c (sem_variable::equals): Likewise.
28428 * match.pd: Likewise.
28429 * omp-simd-clone.c (simd_clone_subparts): Likewise.
28430 * print-tree.c (print_node): Likewise.
28431 * stor-layout.c (layout_type): Likewise.
28432 * targhooks.c (default_builtin_vectorization_cost): Likewise.
28433 * tree-cfg.c (verify_gimple_comparison): Likewise.
28434 (verify_gimple_assign_binary): Likewise.
28435 (verify_gimple_assign_ternary): Likewise.
28436 (verify_gimple_assign_single): Likewise.
28437 * tree-pretty-print.c (dump_generic_node): Likewise.
28438 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
28439 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
28440 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
28441 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
28442 (vect_shift_permute_load_chain): Likewise.
28443 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
28444 (expand_vector_condition, optimize_vector_constructor): Likewise.
28445 (lower_vec_perm, get_compute_type): Likewise.
28446 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28447 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
28448 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
28449 (vect_recog_mask_conversion_pattern): Likewise.
28450 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
28451 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
28452 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
28453 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
28454 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
28455 (vectorizable_shift, vectorizable_operation, vectorizable_store)
28456 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
28457 (supportable_widening_operation): Likewise.
28458 (supportable_narrowing_operation): Likewise.
28459 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
28461 * varasm.c (output_constant): Likewise.
28463 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28464 Alan Hayward <alan.hayward@arm.com>
28465 David Sherwood <david.sherwood@arm.com>
28467 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
28468 so that both the length == 3 and length != 3 cases set up their
28469 own permute vectors. Add comments explaining why we know the
28470 number of elements is constant.
28471 (vect_permute_load_chain): Likewise.
28473 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28474 Alan Hayward <alan.hayward@arm.com>
28475 David Sherwood <david.sherwood@arm.com>
28477 * machmode.h (mode_nunits): Change from unsigned char to
28479 (ONLY_FIXED_SIZE_MODES): New macro.
28480 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
28481 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
28482 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
28484 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
28485 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
28486 or if measurement_type is not polynomial.
28487 * genmodes.c (ZERO_COEFFS): New macro.
28488 (emit_mode_nunits_inline): Make mode_nunits_inline return a
28490 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
28491 Use ZERO_COEFFS when emitting initializers.
28492 * data-streamer.h (bp_pack_poly_value): New function.
28493 (bp_unpack_poly_value): Likewise.
28494 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
28495 for GET_MODE_NUNITS.
28496 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
28497 for GET_MODE_NUNITS.
28498 * tree.c (make_vector_type): Remove temporary shim and make
28499 the real function take the number of units as a poly_uint64
28500 rather than an int.
28501 (build_vector_type_for_mode): Handle polynomial nunits.
28502 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
28503 * emit-rtl.c (const_vec_series_p_1): Likewise.
28504 (gen_rtx_CONST_VECTOR): Likewise.
28505 * fold-const.c (test_vec_duplicate_folding): Likewise.
28506 * genrecog.c (validate_pattern): Likewise.
28507 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
28508 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
28509 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
28510 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
28511 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
28512 * rtlanal.c (subreg_get_info): Likewise.
28513 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
28514 (vect_grouped_load_supported): Likewise.
28515 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
28516 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
28517 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
28518 (simplify_const_unary_operation, simplify_binary_operation_1)
28519 (simplify_const_binary_operation, simplify_ternary_operation)
28520 (test_vector_ops_duplicate, test_vector_ops): Likewise.
28521 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
28522 instead of CONST_VECTOR_NUNITS.
28523 * varasm.c (output_constant_pool_2): Likewise.
28524 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
28525 explicit-encoded elements in the XVEC for variable-length vectors.
28527 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28529 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
28531 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28532 Alan Hayward <alan.hayward@arm.com>
28533 David Sherwood <david.sherwood@arm.com>
28535 * coretypes.h (fixed_size_mode): Declare.
28536 (fixed_size_mode_pod): New typedef.
28537 * builtins.h (target_builtins::x_apply_args_mode)
28538 (target_builtins::x_apply_result_mode): Change type to
28539 fixed_size_mode_pod.
28540 * builtins.c (apply_args_size, apply_result_size, result_vector)
28541 (expand_builtin_apply_args_1, expand_builtin_apply)
28542 (expand_builtin_return): Update accordingly.
28544 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28546 * cse.c (hash_rtx_cb): Hash only the encoded elements.
28547 * cselib.c (cselib_hash_rtx): Likewise.
28548 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
28549 CONST_VECTOR encoding.
28551 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28552 Jeff Law <law@redhat.com>
28555 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
28556 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
28557 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
28558 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
28561 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
28562 explicitly probe *sp in a noreturn function if there were any callee
28563 register saves or frame pointer is needed.
28565 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28568 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
28569 BLKmode for ternary, binary or unary expressions.
28572 * var-tracking.c (delete_vta_debug_insn): New inline function.
28573 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
28574 insns from get_insns () to NULL instead of each bb separately.
28575 Use delete_vta_debug_insn. No longer static.
28576 (vt_debug_insns_local, variable_tracking_main_1): Adjust
28577 delete_vta_debug_insns callers.
28578 * rtl.h (delete_vta_debug_insns): Declare.
28579 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
28580 instead of variable_tracking_main.
28582 2018-01-03 Martin Sebor <msebor@redhat.com>
28584 PR tree-optimization/83603
28585 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
28586 arguments past the endof the argument list in functions declared
28587 without a prototype.
28588 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
28589 Avoid checking when arguments are null.
28591 2018-01-03 Martin Sebor <msebor@redhat.com>
28594 * doc/extend.texi (attribute const): Fix a typo.
28595 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
28596 issuing -Wsuggest-attribute for void functions.
28598 2018-01-03 Martin Sebor <msebor@redhat.com>
28600 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
28601 offset_int::from instead of wide_int::to_shwi.
28602 (maybe_diag_overlap): Remove assertion.
28603 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
28604 * gimple-ssa-sprintf.c (format_directive): Same.
28605 (parse_directive): Same.
28606 (sprintf_dom_walker::compute_format_length): Same.
28607 (try_substitute_return_value): Same.
28609 2018-01-03 Jeff Law <law@redhat.com>
28611 PR middle-end/83654
28612 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
28613 non-constant residual for zero at runtime and avoid probing in
28614 that case. Reorganize code for trailing problem to mirror handling
28617 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28619 PR tree-optimization/83501
28620 * tree-ssa-strlen.c (get_string_cst): New.
28621 (handle_char_store): Call get_string_cst.
28623 2018-01-03 Martin Liska <mliska@suse.cz>
28625 PR tree-optimization/83593
28626 * tree-ssa-strlen.c: Include tree-cfg.h.
28627 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
28628 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
28629 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
28631 (strlen_dom_walker::before_dom_children): Call
28632 gimple_purge_dead_eh_edges. Dump tranformation with details
28634 (strlen_dom_walker::before_dom_children): Update call by adding
28635 new argument cleanup_eh.
28636 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
28638 2018-01-03 Martin Liska <mliska@suse.cz>
28641 * cif-code.def (VARIADIC_THUNK): New enum value.
28642 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
28645 2018-01-03 Jan Beulich <jbeulich@suse.com>
28647 * sse.md (mov<mode>_internal): Tighten condition for when to use
28648 vmovdqu<ssescalarsize> for TI and OI modes.
28650 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28652 Update copyright years.
28654 2018-01-03 Martin Liska <mliska@suse.cz>
28657 * ipa-visibility.c (function_and_variable_visibility): Skip
28658 functions with noipa attribure.
28660 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28662 * gcc.c (process_command): Update copyright notice dates.
28663 * gcov-dump.c (print_version): Ditto.
28664 * gcov.c (print_version): Ditto.
28665 * gcov-tool.c (print_version): Ditto.
28666 * gengtype.c (create_file): Ditto.
28667 * doc/cpp.texi: Bump @copying's copyright year.
28668 * doc/cppinternals.texi: Ditto.
28669 * doc/gcc.texi: Ditto.
28670 * doc/gccint.texi: Ditto.
28671 * doc/gcov.texi: Ditto.
28672 * doc/install.texi: Ditto.
28673 * doc/invoke.texi: Ditto.
28675 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28677 * vector-builder.h (vector_builder::m_full_nelts): Change from
28678 unsigned int to poly_uint64.
28679 (vector_builder::full_nelts): Update prototype accordingly.
28680 (vector_builder::new_vector): Likewise.
28681 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
28682 (vector_builder::operator ==): Likewise.
28683 (vector_builder::finalize): Likewise.
28684 * int-vector-builder.h (int_vector_builder::int_vector_builder):
28685 Take the number of elements as a poly_uint64 rather than an
28687 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
28688 from unsigned int to poly_uint64.
28689 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
28690 (vec_perm_indices::new_vector): Likewise.
28691 (vec_perm_indices::length): Likewise.
28692 (vec_perm_indices::nelts_per_input): Likewise.
28693 (vec_perm_indices::input_nelts): Likewise.
28694 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
28695 number of elements per input as a poly_uint64 rather than an
28696 unsigned int. Use the original encoding for variable-length
28697 vectors, rather than clamping each individual element.
28698 For the second and subsequent elements in each pattern,
28699 clamp the step and base before clamping their sum.
28700 (vec_perm_indices::series_p): Handle polynomial element counts.
28701 (vec_perm_indices::all_in_range_p): Likewise.
28702 (vec_perm_indices_to_tree): Likewise.
28703 (vec_perm_indices_to_rtx): Likewise.
28704 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
28705 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
28706 (tree_vector_builder::new_binary_operation): Handle polynomial
28707 element counts. Return false if we need to know the number
28708 of elements at compile time.
28709 * fold-const.c (fold_vec_perm): Punt if the number of elements
28710 isn't known at compile time.
28712 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28714 * vec-perm-indices.h (vec_perm_builder): Change element type
28715 from HOST_WIDE_INT to poly_int64.
28716 (vec_perm_indices::element_type): Update accordingly.
28717 (vec_perm_indices::clamp): Handle polynomial element_types.
28718 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
28719 (vec_perm_indices::all_in_range_p): Likewise.
28720 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
28722 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
28723 polynomial vec_perm_indices element types.
28724 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
28725 * fold-const.c (fold_vec_perm): Likewise.
28726 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
28727 * tree-vect-generic.c (lower_vec_perm): Likewise.
28728 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
28729 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
28730 element type to HOST_WIDE_INT.
28732 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28733 Alan Hayward <alan.hayward@arm.com>
28734 David Sherwood <david.sherwood@arm.com>
28736 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
28737 rather than an int. Use plus_constant.
28738 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
28739 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
28741 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28742 Alan Hayward <alan.hayward@arm.com>
28743 David Sherwood <david.sherwood@arm.com>
28745 * calls.c (emit_call_1, expand_call): Change struct_value_size from
28746 a HOST_WIDE_INT to a poly_int64.
28748 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28749 Alan Hayward <alan.hayward@arm.com>
28750 David Sherwood <david.sherwood@arm.com>
28752 * calls.c (load_register_parameters): Cope with polynomial
28753 mode sizes. Require a constant size for BLKmode parameters
28754 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
28755 forces a parameter to be padded at the lsb end in order to
28756 fill a complete number of words, require the parameter size
28757 to be ordered wrt UNITS_PER_WORD.
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 * reload1.c (spill_stack_slot_width): Change element type
28764 from unsigned int to poly_uint64_pod.
28765 (alter_reg): Treat mode sizes as polynomial.
28767 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28768 Alan Hayward <alan.hayward@arm.com>
28769 David Sherwood <david.sherwood@arm.com>
28771 * reload.c (complex_word_subreg_p): New function.
28772 (reload_inner_reg_of_subreg, push_reload): Use it.
28774 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28775 Alan Hayward <alan.hayward@arm.com>
28776 David Sherwood <david.sherwood@arm.com>
28778 * lra-constraints.c (process_alt_operands): Reject matched
28779 operands whose sizes aren't ordered.
28780 (match_reload): Refer to this check here.
28782 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28783 Alan Hayward <alan.hayward@arm.com>
28784 David Sherwood <david.sherwood@arm.com>
28786 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
28787 that the mode size is in the set {1, 2, 4, 8, 16}.
28789 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28790 Alan Hayward <alan.hayward@arm.com>
28791 David Sherwood <david.sherwood@arm.com>
28793 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
28794 Use plus_constant instead of gen_rtx_PLUS.
28796 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28797 Alan Hayward <alan.hayward@arm.com>
28798 David Sherwood <david.sherwood@arm.com>
28800 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
28801 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
28802 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
28803 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
28804 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
28805 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
28806 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
28807 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
28808 * config/i386/i386.c (ix86_push_rounding): ...this new function.
28809 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
28811 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
28812 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
28813 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
28814 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
28815 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
28816 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
28817 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
28818 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
28819 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
28820 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
28822 * expr.c (emit_move_resolve_push): Treat the input and result
28823 of PUSH_ROUNDING as a poly_int64.
28824 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
28825 (emit_push_insn): Likewise.
28826 * lra-eliminations.c (mark_not_eliminable): Likewise.
28827 * recog.c (push_operand): Likewise.
28828 * reload1.c (elimination_effects): Likewise.
28829 * rtlanal.c (nonzero_bits1): Likewise.
28830 * calls.c (store_one_arg): Likewise. Require the padding to be
28831 known at compile time.
28833 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28834 Alan Hayward <alan.hayward@arm.com>
28835 David Sherwood <david.sherwood@arm.com>
28837 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
28838 Use plus_constant instead of gen_rtx_PLUS.
28840 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28841 Alan Hayward <alan.hayward@arm.com>
28842 David Sherwood <david.sherwood@arm.com>
28844 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
28845 rather than an int.
28847 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28848 Alan Hayward <alan.hayward@arm.com>
28849 David Sherwood <david.sherwood@arm.com>
28851 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
28852 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
28853 via stack temporaries. Treat the mode size as polynomial too.
28855 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28856 Alan Hayward <alan.hayward@arm.com>
28857 David Sherwood <david.sherwood@arm.com>
28859 * expr.c (expand_expr_real_2): When handling conversions involving
28860 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
28861 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
28862 as a poly_uint64 too.
28864 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28865 Alan Hayward <alan.hayward@arm.com>
28866 David Sherwood <david.sherwood@arm.com>
28868 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
28870 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28871 Alan Hayward <alan.hayward@arm.com>
28872 David Sherwood <david.sherwood@arm.com>
28874 * combine.c (can_change_dest_mode): Handle polynomial
28875 REGMODE_NATURAL_SIZE.
28876 * expmed.c (store_bit_field_1): Likewise.
28877 * expr.c (store_constructor): Likewise.
28878 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
28879 and polynomial REGMODE_NATURAL_SIZE.
28880 (gen_lowpart_common): Likewise.
28881 * reginfo.c (record_subregs_of_mode): Likewise.
28882 * rtlanal.c (read_modify_subreg_p): Likewise.
28884 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28885 Alan Hayward <alan.hayward@arm.com>
28886 David Sherwood <david.sherwood@arm.com>
28888 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
28889 numbers of elements.
28891 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28892 Alan Hayward <alan.hayward@arm.com>
28893 David Sherwood <david.sherwood@arm.com>
28895 * match.pd: Cope with polynomial numbers of vector elements.
28897 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28898 Alan Hayward <alan.hayward@arm.com>
28899 David Sherwood <david.sherwood@arm.com>
28901 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
28902 in a POINTER_PLUS_EXPR.
28904 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28905 Alan Hayward <alan.hayward@arm.com>
28906 David Sherwood <david.sherwood@arm.com>
28908 * omp-simd-clone.c (simd_clone_subparts): New function.
28909 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
28910 (ipa_simd_modify_function_body): Likewise.
28912 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28913 Alan Hayward <alan.hayward@arm.com>
28914 David Sherwood <david.sherwood@arm.com>
28916 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
28917 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
28918 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
28919 (expand_vector_condition, vector_element): Likewise.
28920 (subparts_gt): New function.
28921 (get_compute_type): Use subparts_gt.
28922 (count_type_subparts): Delete.
28923 (expand_vector_operations_1): Use subparts_gt instead of
28924 count_type_subparts.
28926 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28927 Alan Hayward <alan.hayward@arm.com>
28928 David Sherwood <david.sherwood@arm.com>
28930 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
28931 (vect_compile_time_alias): ...this new function. Do the calculation
28932 on poly_ints rather than trees.
28933 (vect_prune_runtime_alias_test_list): Update call accordingly.
28935 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28936 Alan Hayward <alan.hayward@arm.com>
28937 David Sherwood <david.sherwood@arm.com>
28939 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
28941 (vect_schedule_slp_instance): Likewise.
28943 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28944 Alan Hayward <alan.hayward@arm.com>
28945 David Sherwood <david.sherwood@arm.com>
28947 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
28948 constant and extern definitions for variable-length vectors.
28949 (vect_get_constant_vectors): Note that the number of units
28950 is known to be constant.
28952 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28953 Alan Hayward <alan.hayward@arm.com>
28954 David Sherwood <david.sherwood@arm.com>
28956 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
28957 of units as polynomial. Choose between WIDE and NARROW based
28960 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28961 Alan Hayward <alan.hayward@arm.com>
28962 David Sherwood <david.sherwood@arm.com>
28964 * tree-vect-stmts.c (simd_clone_subparts): New function.
28965 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
28967 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28968 Alan Hayward <alan.hayward@arm.com>
28969 David Sherwood <david.sherwood@arm.com>
28971 * tree-vect-stmts.c (vectorizable_call): Treat the number of
28972 vectors as polynomial. Use build_index_vector for
28973 IFN_GOMP_SIMD_LANE.
28975 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28976 Alan Hayward <alan.hayward@arm.com>
28977 David Sherwood <david.sherwood@arm.com>
28979 * tree-vect-stmts.c (get_load_store_type): Treat the number of
28980 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
28981 for variable-length vectors.
28982 (vectorizable_mask_load_store): Treat the number of units as
28983 polynomial, asserting that it is constant if the condition has
28984 already been enforced.
28985 (vectorizable_store, vectorizable_load): Likewise.
28987 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28988 Alan Hayward <alan.hayward@arm.com>
28989 David Sherwood <david.sherwood@arm.com>
28991 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
28992 of units as polynomial. Punt if we can't tell at compile time
28993 which vector contains the final result.
28995 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28996 Alan Hayward <alan.hayward@arm.com>
28997 David Sherwood <david.sherwood@arm.com>
28999 * tree-vect-loop.c (vectorizable_induction): Treat the number
29000 of units as polynomial. Punt on SLP inductions. Use an integer
29001 VEC_SERIES_EXPR for variable-length integer reductions. Use a
29002 cast of such a series for variable-length floating-point
29005 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29006 Alan Hayward <alan.hayward@arm.com>
29007 David Sherwood <david.sherwood@arm.com>
29009 * tree.h (build_index_vector): Declare.
29010 * tree.c (build_index_vector): New function.
29011 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
29012 of units as polynomial, forcibly converting it to a constant if
29013 vectorizable_reduction has already enforced the condition.
29014 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
29015 to create a {1,2,3,...} vector.
29016 (vectorizable_reduction): Treat the number of units as polynomial.
29017 Choose vectype_in based on the largest scalar element size rather
29018 than the smallest number of units. Enforce the restrictions
29021 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29022 Alan Hayward <alan.hayward@arm.com>
29023 David Sherwood <david.sherwood@arm.com>
29025 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
29026 number of units as polynomial.
29028 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29029 Alan Hayward <alan.hayward@arm.com>
29030 David Sherwood <david.sherwood@arm.com>
29032 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
29033 * target.def (autovectorize_vector_sizes): Return the vector sizes
29034 by pointer, using vector_sizes rather than a bitmask.
29035 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
29036 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
29037 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
29039 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
29040 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
29041 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
29042 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
29043 * omp-general.c (omp_max_vf): Likewise.
29044 * omp-low.c (omp_clause_aligned_alignment): Likewise.
29045 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
29046 * tree-vect-loop.c (vect_analyze_loop): Likewise.
29047 * tree-vect-slp.c (vect_slp_bb): Likewise.
29048 * doc/tm.texi: Regenerate.
29049 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
29051 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
29052 the vector size as a poly_uint64 rather than an unsigned int.
29053 (current_vector_size): Change from an unsigned int to a poly_uint64.
29054 (get_vectype_for_scalar_type): Update accordingly.
29055 * tree.h (build_truth_vector_type): Take the size and number of
29056 units as a poly_uint64 rather than an unsigned int.
29057 (build_vector_type): Add a temporary overload that takes
29058 the number of units as a poly_uint64 rather than an unsigned int.
29059 * tree.c (make_vector_type): Likewise.
29060 (build_truth_vector_type): Take the number of units as a poly_uint64
29061 rather than an unsigned int.
29063 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29064 Alan Hayward <alan.hayward@arm.com>
29065 David Sherwood <david.sherwood@arm.com>
29067 * target.def (get_mask_mode): Take the number of units and length
29068 as poly_uint64s rather than unsigned ints.
29069 * targhooks.h (default_get_mask_mode): Update accordingly.
29070 * targhooks.c (default_get_mask_mode): Likewise.
29071 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
29072 * doc/tm.texi: Regenerate.
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 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
29079 * omp-general.c (omp_max_vf): Likewise.
29080 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
29081 (expand_omp_simd): Handle polynomial safelen.
29082 * omp-low.c (omplow_simd_context): Add a default constructor.
29083 (omplow_simd_context::max_vf): Change from int to poly_uint64.
29084 (lower_rec_simd_input_clauses): Update accordingly.
29085 (lower_rec_input_clauses): Likewise.
29087 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29088 Alan Hayward <alan.hayward@arm.com>
29089 David Sherwood <david.sherwood@arm.com>
29091 * tree-vectorizer.h (vect_nunits_for_cost): New function.
29092 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
29093 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
29094 (vect_analyze_slp_cost): Likewise.
29095 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
29096 (vect_model_load_cost): 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-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
29103 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
29104 from an unsigned int * to a poly_uint64_pod *.
29105 (calculate_unrolling_factor): New function.
29106 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
29108 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29109 Alan Hayward <alan.hayward@arm.com>
29110 David Sherwood <david.sherwood@arm.com>
29112 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
29113 from an unsigned int to a poly_uint64.
29114 (_loop_vec_info::slp_unrolling_factor): Likewise.
29115 (_loop_vec_info::vectorization_factor): Change from an int
29117 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
29118 (vect_get_num_vectors): New function.
29119 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
29120 (vect_get_num_copies): Use vect_get_num_vectors.
29121 (vect_analyze_data_ref_dependences): Change max_vf from an int *
29122 to an unsigned int *.
29123 (vect_analyze_data_refs): Change min_vf from an int * to a
29125 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
29126 than an unsigned HOST_WIDE_INT.
29127 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
29128 (vect_analyze_data_ref_dependence): Change max_vf from an int *
29129 to an unsigned int *.
29130 (vect_analyze_data_ref_dependences): Likewise.
29131 (vect_compute_data_ref_alignment): Handle polynomial vf.
29132 (vect_enhance_data_refs_alignment): Likewise.
29133 (vect_prune_runtime_alias_test_list): Likewise.
29134 (vect_shift_permute_load_chain): Likewise.
29135 (vect_supportable_dr_alignment): Likewise.
29136 (dependence_distance_ge_vf): Take the vectorization factor as a
29137 poly_uint64 rather than an unsigned HOST_WIDE_INT.
29138 (vect_analyze_data_refs): Change min_vf from an int * to a
29140 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
29141 vfm1 as a poly_uint64 rather than an int. Make the same change
29142 for the returned bound_scalar.
29143 (vect_gen_vector_loop_niters): Handle polynomial vf.
29144 (vect_do_peeling): Likewise. Update call to
29145 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
29146 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
29148 * tree-vect-loop.c (vect_determine_vectorization_factor)
29149 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
29150 (vect_get_known_peeling_cost): Likewise.
29151 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
29152 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
29153 (vect_transform_loop): Likewise. Use the lowest possible VF when
29154 updating the upper bounds of the loop.
29155 (vect_min_worthwhile_factor): Make static. Return an unsigned int
29156 rather than an int.
29157 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
29158 polynomial unroll factors.
29159 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
29160 (vect_make_slp_decision): Likewise.
29161 (vect_supported_load_permutation_p): Likewise, and polynomial
29163 (vect_analyze_slp_cost): Handle polynomial vf.
29164 (vect_slp_analyze_node_operations): Likewise.
29165 (vect_slp_analyze_bb_1): Likewise.
29166 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
29167 than an unsigned HOST_WIDE_INT.
29168 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
29169 (vectorizable_load): Handle polynomial vf.
29170 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
29172 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
29174 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29175 Alan Hayward <alan.hayward@arm.com>
29176 David Sherwood <david.sherwood@arm.com>
29178 * match.pd: Handle bit operations involving three constants
29179 and try to fold one pair.
29181 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29183 * tree-vect-loop-manip.c: Include gimple-fold.h.
29184 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
29185 niters_maybe_zero parameters. Handle other cases besides a step of 1.
29186 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
29187 Add a path that uses a step of VF instead of 1, but disable it
29189 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
29190 and niters_no_overflow parameters. Update calls to
29191 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
29192 Create a new SSA name if the latter choses to use a ste other
29193 than zero, and return it via niters_vector_mult_vf_var.
29194 * tree-vect-loop.c (vect_transform_loop): Update calls to
29195 vect_do_peeling, vect_gen_vector_loop_niters and
29196 slpeel_make_loop_iterate_ntimes.
29197 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
29198 (vect_gen_vector_loop_niters): Update declarations after above changes.
29200 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
29202 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
29203 128-bit round to integer instructions.
29204 (ceil<mode>2): Likewise.
29205 (btrunc<mode>2): Likewise.
29206 (round<mode>2): Likewise.
29208 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29210 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
29211 unaligned VSX load/store on P8/P9.
29212 (expand_block_clear): Allow the use of unaligned VSX
29213 load/store on P8/P9.
29215 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29217 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
29219 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
29220 swap associated with both a load and a store.
29222 2018-01-02 Andrew Waterman <andrew@sifive.com>
29224 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
29225 * config/riscv/riscv.md (clear_cache): Use it.
29227 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
29229 * web.c: Remove out-of-date comment.
29231 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29233 * expr.c (fixup_args_size_notes): Check that any existing
29234 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
29235 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
29236 (emit_single_push_insn): ...here.
29238 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29240 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
29241 (const_vector_encoded_nelts): New function.
29242 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
29243 (const_vector_int_elt, const_vector_elt): Declare.
29244 * emit-rtl.c (const_vector_int_elt_1): New function.
29245 (const_vector_elt): Likewise.
29246 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
29247 of CONST_VECTOR_ELT.
29249 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29251 * expr.c: Include rtx-vector-builder.h.
29252 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
29253 directly on the tree encoding.
29254 (const_vector_from_tree): Likewise.
29255 * optabs.c: Include rtx-vector-builder.h.
29256 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
29257 sequence of "u" values.
29258 * vec-perm-indices.c: Include rtx-vector-builder.h.
29259 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
29260 directly on the vec_perm_indices encoding.
29262 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29264 * doc/rtl.texi (const_vector): Describe new encoding scheme.
29265 * Makefile.in (OBJS): Add rtx-vector-builder.o.
29266 * rtx-vector-builder.h: New file.
29267 * rtx-vector-builder.c: Likewise.
29268 * rtl.h (rtx_def::u2): Add a const_vector field.
29269 (CONST_VECTOR_NPATTERNS): New macro.
29270 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
29271 (CONST_VECTOR_DUPLICATE_P): Likewise.
29272 (CONST_VECTOR_STEPPED_P): Likewise.
29273 (CONST_VECTOR_ENCODED_ELT): Likewise.
29274 (const_vec_duplicate_p): Check for a duplicated vector encoding.
29275 (unwrap_const_vec_duplicate): Likewise.
29276 (const_vec_series_p): Check for a non-duplicated vector encoding.
29277 Say that the function only returns true for integer vectors.
29278 * emit-rtl.c: Include rtx-vector-builder.h.
29279 (gen_const_vec_duplicate_1): Delete.
29280 (gen_const_vector): Call gen_const_vec_duplicate instead of
29281 gen_const_vec_duplicate_1.
29282 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
29283 (gen_const_vec_duplicate): Use rtx_vector_builder.
29284 (gen_const_vec_series): Likewise.
29285 (gen_rtx_CONST_VECTOR): Likewise.
29286 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
29287 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
29288 Build a new vector rather than modifying a CONST_VECTOR in-place.
29289 (handle_special_swappables): Update call accordingly.
29290 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
29291 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
29292 Build a new vector rather than modifying a CONST_VECTOR in-place.
29293 (handle_special_swappables): Update call accordingly.
29295 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29297 * simplify-rtx.c (simplify_const_binary_operation): Use
29298 CONST_VECTOR_ELT instead of XVECEXP.
29300 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29302 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
29303 the selector elements to be different from the data elements
29304 if the selector is a VECTOR_CST.
29305 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
29306 ssizetype for the selector.
29308 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29310 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
29311 before testing each element individually.
29312 * tree-vect-generic.c (lower_vec_perm): Likewise.
29314 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29316 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
29317 * selftest-run-tests.c (selftest::run_tests): Call it.
29318 * vector-builder.h (vector_builder::operator ==): New function.
29319 (vector_builder::operator !=): Likewise.
29320 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
29321 (vec_perm_indices::all_from_input_p): New function.
29322 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
29323 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
29324 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
29325 instead of reading the VECTOR_CST directly. Detect whether both
29326 vector inputs are the same before constructing the vec_perm_indices,
29327 and update the number of inputs argument accordingly. Use the
29328 utility functions added above. Only construct sel2 if we need to.
29330 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29332 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
29333 the broadcast of the low byte.
29334 (expand_mult_highpart): Use an explicit encoding for the permutes.
29335 * optabs-query.c (can_mult_highpart_p): Likewise.
29336 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
29337 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29338 (vectorizable_bswap): Likewise.
29339 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
29340 explicit encoding for the power-of-2 permutes.
29341 (vect_permute_store_chain): Likewise.
29342 (vect_grouped_load_supported): Likewise.
29343 (vect_permute_load_chain): Likewise.
29345 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29347 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
29348 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
29349 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
29350 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
29351 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
29352 (vect_gen_perm_mask_any): Likewise.
29354 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29356 * int-vector-builder.h: New file.
29357 * vec-perm-indices.h: Include int-vector-builder.h.
29358 (vec_perm_indices): Redefine as an int_vector_builder.
29359 (auto_vec_perm_indices): Delete.
29360 (vec_perm_builder): Redefine as a stand-alone class.
29361 (vec_perm_indices::vec_perm_indices): New function.
29362 (vec_perm_indices::clamp): Likewise.
29363 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
29364 (vec_perm_indices::new_vector): New function.
29365 (vec_perm_indices::new_expanded_vector): Update for new
29366 vec_perm_indices class.
29367 (vec_perm_indices::rotate_inputs): New function.
29368 (vec_perm_indices::all_in_range_p): Operate directly on the
29369 encoded form, without computing elided elements.
29370 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
29371 encoding. Update for new vec_perm_indices class.
29372 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
29373 the given vec_perm_builder.
29374 (expand_vec_perm_var): Update vec_perm_builder constructor.
29375 (expand_mult_highpart): Use vec_perm_builder instead of
29376 auto_vec_perm_indices.
29377 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
29378 vec_perm_indices instead of auto_vec_perm_indices. Use a single
29379 or double series encoding as appropriate.
29380 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
29381 vec_perm_indices instead of auto_vec_perm_indices.
29382 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
29383 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
29384 (vect_permute_store_chain): Likewise.
29385 (vect_grouped_load_supported): Likewise.
29386 (vect_permute_load_chain): Likewise.
29387 (vect_shift_permute_load_chain): Likewise.
29388 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
29389 (vect_transform_slp_perm_load): Likewise.
29390 (vect_schedule_slp_instance): Likewise.
29391 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29392 (vectorizable_mask_load_store): Likewise.
29393 (vectorizable_bswap): Likewise.
29394 (vectorizable_store): Likewise.
29395 (vectorizable_load): Likewise.
29396 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
29397 vec_perm_indices instead of auto_vec_perm_indices. Use
29398 tree_to_vec_perm_builder to read the vector from a tree.
29399 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
29400 vec_perm_builder instead of a vec_perm_indices.
29401 (have_whole_vector_shift): Use vec_perm_builder and
29402 vec_perm_indices instead of auto_vec_perm_indices. Leave the
29403 truncation to calc_vec_perm_mask_for_shift.
29404 (vect_create_epilog_for_reduction): Likewise.
29405 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
29406 from auto_vec_perm_indices to vec_perm_indices.
29407 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
29408 instead of changing individual elements.
29409 (aarch64_vectorize_vec_perm_const): Use new_vector to install
29410 the vector in d.perm.
29411 * config/arm/arm.c (expand_vec_perm_d::perm): Change
29412 from auto_vec_perm_indices to vec_perm_indices.
29413 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
29414 instead of changing individual elements.
29415 (arm_vectorize_vec_perm_const): Use new_vector to install
29416 the vector in d.perm.
29417 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
29418 Update vec_perm_builder constructor.
29419 (rs6000_expand_interleave): Likewise.
29420 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
29421 (rs6000_expand_interleave): Likewise.
29423 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29425 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
29426 to qimode could truncate the indices.
29427 * optabs.c (expand_vec_perm_var): Likewise.
29429 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29431 * Makefile.in (OBJS): Add vec-perm-indices.o.
29432 * vec-perm-indices.h: New file.
29433 * vec-perm-indices.c: Likewise.
29434 * target.h (vec_perm_indices): Replace with a forward class
29436 (auto_vec_perm_indices): Move to vec-perm-indices.h.
29437 * optabs.h: Include vec-perm-indices.h.
29438 (expand_vec_perm): Delete.
29439 (selector_fits_mode_p, expand_vec_perm_var): Declare.
29440 (expand_vec_perm_const): Declare.
29441 * target.def (vec_perm_const_ok): Replace with...
29442 (vec_perm_const): ...this new hook.
29443 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
29444 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
29445 * doc/tm.texi: Regenerate.
29446 * optabs.def (vec_perm_const): Delete.
29447 * doc/md.texi (vec_perm_const): Likewise.
29448 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
29449 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
29450 expand_vec_perm for constant permutation vectors. Assert that
29451 the mode of variable permutation vectors is the integer equivalent
29452 of the mode that is being permuted.
29453 * optabs-query.h (selector_fits_mode_p): Declare.
29454 * optabs-query.c: Include vec-perm-indices.h.
29455 (selector_fits_mode_p): New function.
29456 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
29457 is defined, instead of checking whether the vec_perm_const_optab
29458 exists. Use targetm.vectorize.vec_perm_const instead of
29459 targetm.vectorize.vec_perm_const_ok. Check whether the indices
29460 fit in the vector mode before using a variable permute.
29461 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
29462 vec_perm_indices instead of an rtx.
29463 (expand_vec_perm): Replace with...
29464 (expand_vec_perm_const): ...this new function. Take the selector
29465 as a vec_perm_indices rather than an rtx. Also take the mode of
29466 the selector. Update call to shift_amt_for_vec_perm_mask.
29467 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
29468 Use vec_perm_indices::new_expanded_vector to expand the original
29469 selector into bytes. Check whether the indices fit in the vector
29470 mode before using a variable permute.
29471 (expand_vec_perm_var): Make global.
29472 (expand_mult_highpart): Use expand_vec_perm_const.
29473 * fold-const.c: Includes vec-perm-indices.h.
29474 * tree-ssa-forwprop.c: Likewise.
29475 * tree-vect-data-refs.c: Likewise.
29476 * tree-vect-generic.c: Likewise.
29477 * tree-vect-loop.c: Likewise.
29478 * tree-vect-slp.c: Likewise.
29479 * tree-vect-stmts.c: Likewise.
29480 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
29482 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
29483 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
29484 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
29485 (aarch64_vectorize_vec_perm_const): ...this new function.
29486 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
29487 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29488 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
29489 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
29490 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
29491 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29492 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
29494 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
29495 check for NEON modes.
29496 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
29497 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
29498 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
29499 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
29501 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
29502 the old VEC_PERM_CONST conditions.
29503 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
29504 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
29505 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
29506 (ia64_vectorize_vec_perm_const_ok): Merge into...
29507 (ia64_vectorize_vec_perm_const): ...this new function.
29508 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
29509 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
29510 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
29511 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
29512 * config/mips/mips.c (mips_expand_vec_perm_const)
29513 (mips_vectorize_vec_perm_const_ok): Merge into...
29514 (mips_vectorize_vec_perm_const): ...this new function.
29515 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
29516 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
29517 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
29518 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
29519 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
29520 (rs6000_expand_vec_perm_const): Delete.
29521 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
29523 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29524 (altivec_expand_vec_perm_const_le): Take each operand individually.
29525 Operate on constant selectors rather than rtxes.
29526 (altivec_expand_vec_perm_const): Likewise. Update call to
29527 altivec_expand_vec_perm_const_le.
29528 (rs6000_expand_vec_perm_const): Delete.
29529 (rs6000_vectorize_vec_perm_const_ok): Delete.
29530 (rs6000_vectorize_vec_perm_const): New function.
29531 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
29532 an element count and rtx array.
29533 (rs6000_expand_extract_even): Update call accordingly.
29534 (rs6000_expand_interleave): Likewise.
29535 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
29536 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
29537 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
29538 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
29539 (rs6000_expand_vec_perm_const): Delete.
29540 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
29541 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29542 (altivec_expand_vec_perm_const_le): Take each operand individually.
29543 Operate on constant selectors rather than rtxes.
29544 (altivec_expand_vec_perm_const): Likewise. Update call to
29545 altivec_expand_vec_perm_const_le.
29546 (rs6000_expand_vec_perm_const): Delete.
29547 (rs6000_vectorize_vec_perm_const_ok): Delete.
29548 (rs6000_vectorize_vec_perm_const): New function. Remove stray
29549 reference to the SPE evmerge intructions.
29550 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
29551 an element count and rtx array.
29552 (rs6000_expand_extract_even): Update call accordingly.
29553 (rs6000_expand_interleave): Likewise.
29554 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
29555 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
29557 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29559 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29561 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
29562 vector mode and that that mode matches the mode of the data
29564 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
29565 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
29566 directly using expand_vec_perm_1 when forcing selectors into
29568 (expand_vec_perm_var): New function, split out from expand_vec_perm.
29570 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29572 * optabs-query.h (can_vec_perm_p): Delete.
29573 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
29574 * optabs-query.c (can_vec_perm_p): Split into...
29575 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
29576 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
29577 particular selector is valid.
29578 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
29579 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
29580 (vect_grouped_load_supported): Likewise.
29581 (vect_shift_permute_load_chain): Likewise.
29582 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
29583 (vect_transform_slp_perm_load): Likewise.
29584 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29585 (vectorizable_bswap): Likewise.
29586 (vect_gen_perm_mask_checked): Likewise.
29587 * fold-const.c (fold_ternary_loc): Likewise. Don't take
29588 implementations of variable permutation vectors into account
29589 when deciding which selector to use.
29590 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
29591 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
29592 with a false third argument.
29593 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
29594 to test whether the constant selector is valid and can_vec_perm_var_p
29595 to test whether a variable selector is valid.
29597 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29599 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
29600 * optabs-query.c (can_vec_perm_p): Likewise.
29601 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
29602 instead of vec_perm_indices.
29603 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
29604 (vect_gen_perm_mask_checked): Likewise,
29605 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
29606 (vect_gen_perm_mask_checked): Likewise,
29608 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29610 * optabs-query.h (qimode_for_vec_perm): Declare.
29611 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
29612 (qimode_for_vec_perm): ...this new function.
29613 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
29615 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29617 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
29618 does not have a conditional at the top.
29620 2018-01-02 Richard Biener <rguenther@suse.de>
29622 * ipa-inline.c (big_speedup_p): Fix expression.
29624 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
29627 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
29630 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
29634 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
29635 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
29636 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
29637 cond_taken_branch_cost 3->4.
29639 2018-01-01 Jakub Jelinek <jakub@redhat.com>
29641 PR tree-optimization/83581
29642 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
29643 TODO_cleanup_cfg if any changes have been made.
29645 PR middle-end/83608
29646 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
29647 convert_modes if target mode has the right side, but different mode
29650 PR middle-end/83609
29651 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
29652 last argument when extracting from CONCAT. If either from_real or
29653 from_imag is NULL, use expansion through memory. If result is not
29654 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
29655 the parts directly to inner mode, if even that fails, use expansion
29658 PR middle-end/83623
29659 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
29660 check for bswap in mode rather than HImode and use that in expand_unop
29663 Copyright (C) 2018 Free Software Foundation, Inc.
29665 Copying and distribution of this file, with or without modification,
29666 are permitted in any medium without royalty provided the copyright
29667 notice and this notice are preserved.