1 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
3 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Force
4 16-byte modes held in GP registers to use an even regno.
6 * config/aarch64/aarch64.c (aarch64_emit_bic): Remove.
7 (aarch64_atomic_ldop_supported_p): Remove.
8 (aarch64_gen_atomic_ldop): Remove.
9 * config/aarch64/atomic.md (atomic_<atomic_optab><ALLI>):
10 Fully expand LSE operations here.
11 (atomic_fetch_<atomic_optab><ALLI>): Likewise.
12 (atomic_<atomic_optab>_fetch<ALLI>): Likewise.
13 (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse): Drop atomic_op iterator
14 and use ATOMIC_LDOP instead; use register_operand for the input;
15 drop the split and emit insns directly.
16 (aarch64_atomic_fetch_<ATOMIC_LDOP><ALLI>_lse): Likewise.
17 (aarch64_atomic_<atomic_op>_fetch<ALLI>_lse): Remove.
18 (@aarch64_atomic_load<ATOMIC_LDOP><ALLI>): Remove.
20 * config/aarch64/aarch64.c (aarch64_emit_atomic_swap): Remove.
21 (aarch64_gen_atomic_ldop): Don't call it.
22 * config/aarch64/atomics.md (atomic_exchange<ALLI>):
23 Use aarch64_reg_or_zero.
24 (aarch64_atomic_exchange<ALLI>): Likewise.
25 (aarch64_atomic_exchange<ALLI>_lse): Remove split; remove & from
26 operand 0; use aarch64_reg_or_zero for input; merge ...
27 (@aarch64_atomic_swp<ALLI>): ... this and remove.
29 * config/aarch64/aarch64.c (aarch64_gen_compare_reg_maybe_ze): New.
30 (aarch64_split_compare_and_swap): Use it.
31 (aarch64_expand_compare_and_swap): Likewise. Remove convert_modes;
32 test oldval against the proper predicate.
33 * config/aarch64/atomics.md (@atomic_compare_and_swap<ALLI>):
34 Use nonmemory_operand for expected.
35 (cas_short_expected_pred): New.
36 (@aarch64_compare_and_swap<SHORT>): Use it; use "rn" not "rI" to match.
37 (@aarch64_compare_and_swap<GPI>): Use "rn" not "rI" for expected.
38 * config/aarch64/predicates.md (aarch64_plushi_immediate): New.
39 (aarch64_plushi_operand): New.
41 * config/aarch64/aarch64.c (aarch64_expand_compare_and_swap):
42 Force oldval into the rval register for TARGET_LSE; emit the compare
43 during initial expansion so that it may be deleted if unused.
44 (aarch64_gen_atomic_cas): Remove.
45 * config/aarch64/atomics.md (@aarch64_compare_and_swap<SHORT>_lse):
46 Change =&r to +r for operand 0; use match_dup for operand 2;
47 remove is_weak and mod_f operands as unused. Drop the split
49 (@aarch64_atomic_cas<SHORT>): ... this pattern's output; remove.
50 (@aarch64_compare_and_swap<GPI>_lse): Similarly.
51 (@aarch64_atomic_cas<GPI>): Similarly.
53 2018-10-31 Richard Biener <rguenther@suse.de>
55 * tree-eh.c (replace_trapping_overflow): Simplify ABS_EXPR case
58 2018-10-31 Sameera Deshpande <sameera.deshpande@linaro.org>
60 * config/aarch64/aarch64-cores.def (saphira): Use saphira pipeline.
61 * config/aarch64/aarch64.md: Include saphira.md
62 * config/aarch64/saphira.md: New file for pipeline description.
64 2018-10-30 Martin Sebor <msebor@redhat.com>
67 * gimple-ssa-sprintf.c (format_directive): Use %G to include
69 (sprintf_dom_walker::compute_format_length):
70 Avoid setting POSUNDER4K here.
71 (get_destination_size): Handle null argument values.
72 (get_user_idx_format): New function.
73 (sprintf_dom_walker::handle_gimple_call): Handle all printf-like
74 functions, including user-defined with attribute format printf.
75 Use %G to include inlining context.
78 2018-10-30 Jan Hubicka <jh@suse.cz>
80 * params.def (lto-partitions): Bump from 32 to 128.
82 2018-10-30 Jan Hubicka <jh@suse.cz>
85 (free_lang_data_d, add_tree_to_fld_list, fld_worklist_push): Move
87 (free_lang_data_in_type): Forward declare.
88 (fld_type_variant_equal_p): New function.
89 (fld_type_variant): New function
90 (fld_incomplete_types): New hash.
91 (fld_incomplete_type_of): New function
92 (fld_simplfied-type): New function.
93 (free_lang_data_in_decl): Add fld parameter; simplify type of FIELD_DECL
94 (free_lang_data): Allocate and free fld_incomplete_type; update call
95 of free_lang_data_in_decl.
97 2018-10-30 Eric Botcazou <ebotcazou@adacore.com>
99 * gcov.c (output_lines): Remove duplicate line.
101 2018-10-30 Aaron Sawdey <acsawdey@linux.ibm.com>
103 * config/rs6000/rs6000.md (bswapdi2): Force address into register
104 if not in indexed or indirect form.
105 (bswapdi2_load): Change predicate to indexed_or_indirect_operand.
106 (bswapdi2_store): Ditto.
107 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem): New
109 * config/rs6000/rs6000-protos.h (rs6000_force_indexed_or_indirect_mem):
110 Prototype for helper function.
112 2018-10-30 Martin Sebor <msebor@redhat.com>
114 * doc/extend.texi (optimize): Clarify/expand attribute documentation.
115 (target, pragma GCC optimize, pragma GCC target): Ditto.
117 2018-10-30 Jonathan Wakely <jwakely@redhat.com>
119 * doc/extend.texi: Fix prototype and description of
120 __builtin_expect_with_probability.
122 2018-10-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
124 * cgraph.h (clone_function_name_1): Replaced by new
125 clone_function_name_numbered that takes name as string; for
126 privatize_symbol_name_1 use only.
127 (clone_function_name): Renamed to
128 clone_function_name_numbered to be explicit about numbering.
129 (clone_function_name): New two-argument function that does
130 not number its output.
131 (clone_function_name): New three-argument function that
132 takes a number to append to its output.
133 * cgraphclones.c (duplicate_thunk_for_node):
134 (clone_function_name_1): Renamed.
135 (clone_function_name_numbered): Two new functions.
136 (clone_function_name): Improved documentation.
137 (cgraph_node::create_virtual_clone): Use clone_function_name_numbered.
138 * config/rs6000/rs6000.c (make_resolver_func): Ditto.
139 * final.c (final_scan_insn_1): Use the new clone_function_name
141 * multiple_target.c (create_dispatcher_calls): Ditto.
142 (create_target_clone): Ditto.
143 * omp-expand.c (grid_expand_target_grid_body): Ditto.
144 * omp-low.c (create_omp_child_function_name): Ditto.
145 * omp-simd-clone.c (simd_clone_create): Ditto.
146 * symtab.c (simd_symtab_node::noninterposable_alias): Use the
147 new clone_function_name without numbering.
149 2018-10-30 Richard Earnshaw <rearnsha@arm.com>
151 * alloc-pool.h (base_pool_allocator <TBlockAllocator>::initialize):
152 Assert that the allocation size is not zero.
154 2018-10-30 Richard Biener <rguenther@suse.de>
156 PR tree-optimization/87800
157 * tree-vect-slp.c (vect_build_slp_tree_2): Reject any
158 non-induction or reduction PHIs.
160 2018-10-30 Sameera Deshpande <sameera.deshpande@linaro.org
162 * config/aarch64/falkor-tag-collision-avoidance.c
163 (execute_tag_collision_avoidance): Call df_note_add_problem.
165 2018-10-30 Martin Liska <mliska@suse.cz>
167 * doc/extend.texi: Fix typo in documentation
168 of __builtin_expect_with_probability.
170 2018-10-29 David Malcolm <dmalcolm@redhat.com>
173 * input.c (get_substring_ranges_for_loc): Detect if
174 linemap_resolve_location gives us a NULL map, and reject
177 2018-10-29 Iain Buclaw <ibuclaw@gdcproject.org>
179 * config.gcc (xstormy16-*-elf): Set tm_d_file.
181 2018-10-29 David Malcolm <dmalcolm@redhat.com>
182 Martin Sebor <msebor@redhat.com>
183 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
185 * doc/ux.texi (Quoting): New subsection, adapted from material at
186 https://gcc.gnu.org/wiki/DiagnosticsGuidelines written by
187 MartinSebor and ManuelLopezIbanez.
188 (Fix-it hints): Note that fix-it hints shouldn't be marked for
191 2018-10-29 Kugan Vivekanandarajah <kuganv@linaro.org>
194 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Fix niter
197 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
199 * gcc.target/powerpc/ssse3-check.h: Remove inaccurate comment.
201 2018-10-29 Paul Koning <ni1d@arrl.net>
203 * doc/rtl.texi (CONST_WIDE_INT_ELT): Give correct macro name.
205 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
207 * gcc/config/rs6000/mmintrin.h (_mm_packs_pi16, _mm_packs_pi32,
208 _mm_packs_pu16, _mm_unpackhi_pi8, _mm_unpacklo_pi8, _mm_add_pi8,
209 _mm_add_pi16, _mm_add_pi32, _mm_sub_pi8, _mm_sub_pi16, _mm_sub_pi32,
210 _mm_cmpgt_pi8, _mm_cmpeq_pi16, _mm_cmpgt_pi16, _mm_cmpeq_pi32,
211 _mm_cmpgt_pi32, _mm_adds_pi8, _mm_adds_pi16, _mm_adds_pu8,
212 _mm_adds_pu16, _mm_subs_pi8, _mm_subs_pi16, _mm_subs_pu8,
213 _mm_subs_pu16, _mm_madd_pi16, _mm_mulhi_pi16, _mm_mullo_pi16,
214 _mm_sll_pi16, _mm_sra_pi16, _mm_srl_pi16, _mm_set1_pi16, _mm_set1_pi8):
215 Change 'vector' to '__vector'.
216 * gcc/config/rs6000/xmmintrin.h (_mm_cvtps_pi32, _mm_cvttps_pi32,
217 _mm_cvtps_pi16, _mm_cvtps_pi8, _mm_max_pi16, _mm_max_pu8, _mm_min_pi16,
218 _mm_min_pu8, _mm_mulhi_pu16, _mm_shuffle_pi16, _mm_avg_pu8,
219 _mm_avg_pu16): Likewise. And, whitespace corrections.
221 2018-10-29 Richard Biener <rguenther@suse.de>
223 PR tree-optimization/87785
224 * tree-vect-slp.c (vect_gather_slp_loads): Only gather
227 2018-10-29 Olivier Hainque <hainque@adacore.com>
229 * VXWORKS_NET_LIBS_RTP: New macro, network part of VXWORKS_LIBS_RTP.
231 2018-10-29 Olivier Hainque <hainque@adacore.com>
233 * config/vxworks.h (VXWORKS_PERSONALITY): New VxWorks
234 ports configuration macro, defaults to "gnu".
235 (VXWORKS_OS_CPP_BUILTINS): Use it. Feed TOOL and
236 TOOL_FAMILY instead of _VX_TOOL/_VXTOOL_FAMILY.
238 2018-10-29 Olivier Hainque <hainque@adacore.com>
240 * config/vxworks.h (CLEAR_INSN_CACHE): #undef before #define.
242 2018-10-29 Richard Biener <rguenther@suse.de>
244 PR tree-optimization/87790
245 * tree-vect-slp.c (vect_mark_slp_stmts): Simplify.
246 (vect_make_slp_decision): Adjust.
247 (vect_slp_analyze_bb_1): Likewise.
248 (vect_detect_hybrid_slp_stmts): Properly union SLP type over
251 2018-10-29 Richard Biener <rguenther@suse.de>
253 PR tree-optimization/87785
254 * tree-vect-slp.c (vect_build_slp_tree_2): Remove loads argument
256 (vect_build_slp_tree): Likewise.
257 (vect_gather_slp_loads): New function.
258 (vect_analyze_slp_instance): Gather loads separately from the
261 2018-10-29 Martin Liska <mliska@suse.cz>
263 * Makefile.in: Make dependency to json.o.
264 * doc/gcov.texi: Document new JSON format, remove
265 old intermediate format documentation.
266 * gcov.c (struct function_info): Come up with m_name and
268 (function_info::function_info): Initialize it.
269 (function_info::~function_info): Release it.
270 (main): Rename flag_intermediate_format to flag_json_format.
271 (print_usage): Describe --json-format.
272 (process_args): Set flag_json_format.
273 (output_intermediate_line): Remove.
274 (output_intermediate_json_line): Likewise.
275 (get_gcov_intermediate_filename): Return new extension
277 (output_intermediate_file): Implement JSON emission.
278 (output_json_intermediate_file): Implement JSON emission.
279 (generate_results): Use ::get_name for function name.
280 Handle JSON output file.
281 (read_graph_file): Use ::get_name instead of cplus_demangle.
282 (read_count_file): Likewise.
283 (solve_flow_graph): Likewise.
284 (add_line_counts): Likewise.
285 (accumulate_line_counts): Use new flag_json_format.
286 (output_function_details): Use ::get_name instead of cplus_demangle.
287 (output_lines): Likewise.
288 * json.cc (test_writing_literals): Add new tests.
289 * json.h (class literal): Add new boolean constructor.
291 2018-10-29 Segher Boessenkool <segher@kernel.crashing.org>
293 PR rtl-optimization/87701
294 PR rtl-optimization/87780
295 * combine.c (make_more_copies): Rewrite.
297 2018-10-28 Kugan Vivekanandarajah <kuganv@linaro.org>
299 * doc/generic.texi (ABSU_EXPR): Document.
300 * match.pd (absu(x)*absu(x) -> x*x): Handle.
301 (absu(absu(X)) -> absu(X)): Likewise.
302 (absu(-X) -> absu(X)): Likewise.
303 (absu(X) where X is nonnegative -> X): Likewise.
305 2018-10-28 Iain Buclaw <ibuclaw@gdcproject.org>
307 * Makefile.in (tm_d_file_list, tm_d_include_list): New variables.
308 (TM_D_H, D_TARGET_DEF, D_TARGET_H, D_TARGET_OBJS): New variables.
309 (tm_d.h, cs-tm_d.h, default-d.o): New rules.
310 (d/d-target-hooks-def.h, s-d-target-hooks-def-h): New rules.
311 (s-tm-texi): Also check timestamp on d-target.def.
312 (generated_files): Add TM_D_H and d-target-hooks-def.h.
313 (build/genhooks.o): Also depend on D_TARGET_DEF.
314 * config.gcc (tm_d_file, d_target_objs, target_has_targetdm): New
316 * config/aarch64/aarch64-d.c: New file.
317 * config/aarch64/aarch64-linux.h (GNU_USER_TARGET_D_CRITSEC_SIZE):
319 * config/aarch64/aarch64-protos.h (aarch64_d_target_versions): New
321 * config/aarch64/aarch64.h (TARGET_D_CPU_VERSIONS): Define.
322 * config/aarch64/t-aarch64 (aarch64-d.o): New rule.
323 * config/arm/arm-d.c: New file.
324 * config/arm/arm-protos.h (arm_d_target_versions): New prototype.
325 * config/arm/arm.h (TARGET_D_CPU_VERSIONS): Define.
326 * config/arm/linux-eabi.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
327 * config/arm/t-arm (arm-d.o): New rule.
328 * config/default-d.c: New file.
329 * config/glibc-d.c: New file.
330 * config/gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
331 * config/i386/i386-d.c: New file.
332 * config/i386/i386-protos.h (ix86_d_target_versions): New prototype.
333 * config/i386/i386.h (TARGET_D_CPU_VERSIONS): Define.
334 * config/i386/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
335 (GNU_USER_TARGET_D_CRITSEC_SIZE): Define.
336 * config/i386/t-i386 (i386-d.o): New rule.
337 * config/kfreebsd-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
338 * config/kopensolaris-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
339 * config/linux-android.h (ANDROID_TARGET_D_OS_VERSIONS): Define.
340 * config/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
341 * config/mips/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
342 * config/mips/mips-d.c: New file.
343 * config/mips/mips-protos.h (mips_d_target_versions): New prototype.
344 * config/mips/mips.h (TARGET_D_CPU_VERSIONS): Define.
345 * config/mips/t-mips (mips-d.o): New rule.
346 * config/powerpcspe/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
347 * config/powerpcspe/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
348 * config/powerpcspe/powerpcspe-d.c: New file.
349 * config/powerpcspe/powerpcspe-protos.h (rs6000_d_target_versions):
351 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
352 Support GNU D by using 0 as the language type.
353 * config/powerpcspe/powerpcspe.h (TARGET_D_CPU_VERSIONS): Define.
354 * config/powerpcspe/t-powerpcspe (powerpcspe-d.o): New rule.
355 * config/riscv/riscv-d.c: New file.
356 * config/riscv/riscv-protos.h (riscv_d_target_versions): New
358 * config/riscv/riscv.h (TARGET_D_CPU_VERSIONS): Define.
359 * config/riscv/t-riscv (riscv-d.o): New rule.
360 * config/rs6000/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
361 * config/rs6000/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
362 * config/rs6000/rs6000-d.c: New file.
363 * config/rs6000/rs6000-protos.h (rs6000_d_target_versions): New
365 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
366 Support GNU D by using 0 as the language type.
367 * config/rs6000/rs6000.h (TARGET_D_CPU_VERSIONS): Define.
368 * config/rs6000/t-rs6000 (rs6000-d.o): New rule.
369 * config/s390/s390-d.c: New file.
370 * config/s390/s390-protos.h (s390_d_target_versions): New prototype.
371 * config/s390/s390.h (TARGET_D_CPU_VERSIONS): Define.
372 * config/s390/t-s390 (s390-d.o): New rule.
373 * config/sparc/sparc-d.c: New file.
374 * config/sparc/sparc-protos.h (sparc_d_target_versions): New
376 * config/sparc/sparc.h (TARGET_D_CPU_VERSIONS): Define.
377 * config/sparc/t-sparc (sparc-d.o): New rule.
378 * config/t-glibc (glibc-d.o): New rule.
379 * configure: Regenerated.
380 * configure.ac (tm_d_file): New variable.
381 (tm_d_file_list, tm_d_include_list, d_target_objs): Add substitutes.
382 * doc/contrib.texi (Contributors): Add self for the D frontend.
383 * doc/frontends.texi (G++ and GCC): Mention D as a supported language.
384 * doc/install.texi (Configuration): Mention libphobos as an option for
385 --enable-shared. Mention d as an option for --enable-languages.
386 (Testing): Mention check-d as a target.
387 * doc/invoke.texi (Overall Options): Mention .d, .dd, and .di as file
388 name suffixes. Mention d as a -x option.
389 * doc/sourcebuild.texi (Top Level): Mention libphobos.
390 * doc/standards.texi (Standards): Add section on D language.
391 * doc/tm.texi: Regenerated.
392 * doc/tm.texi.in: Add @node for D language and ABI, and @hook for
393 TARGET_CPU_VERSIONS, TARGET_D_OS_VERSIONS, and TARGET_D_CRITSEC_SIZE.
394 * dwarf2out.c (is_dlang): New function.
395 (gen_compile_unit_die): Use DW_LANG_D for D.
396 (declare_in_namespace): Return module die for D, instead of adding
397 extra declarations into the namespace.
398 (gen_namespace_die): Generate DW_TAG_module for D.
399 (gen_decl_die): Handle CONST_DECLSs for D.
400 (dwarf2out_decl): Likewise.
401 (prune_unused_types_walk_local_classes): Handle DW_tag_interface_type.
402 (prune_unused_types_walk): Handle DW_tag_interface_type same as other
404 * gcc.c (default_compilers): Add entries for .d, .dd and .di.
405 * genhooks.c: Include d/d-target.def.
407 2018-10-28 Iain Sandoe <iain@sandoe.co.uk>
410 * config/rs6000/darwin.h (STACK_BOUNDARY): New.
411 (RS6000_STARTING_FRAME_OFFSET): Adjust to preserve 16byte alignment.
412 (STACK_DYNAMIC_OFFSET): Likewise.
414 2018-10-27 Sandra Loosemore <sandra@codesourcery.com>
417 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Fix
420 2018-10-26 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
422 * doc/extend.texi (PowerPC builtins): Fix __builtin_unpack_ibm128
423 return type and other typos.
425 2018-10-26 Aaron Sawdey <acsawdey@linux.ibm.com>
427 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Change to
428 a shorter sequence with fewer branches.
429 (emit_final_str_compare_gpr): Ditto.
431 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
433 * config/rs6000/tmmintrin.h: New file.
434 * config.gcc (powerpc*-*-*): Add tmmintrin.h to extra_headers.
436 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
438 * config/rs6000/mmintrin.h: Enable 32bit compilation.
439 * config/rs6000/xmmintrin.h: Likewise.
441 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
443 * config/rs6000/xmmintrin.h (_mm_extract_pi16): Fix for big-endian.
445 2018-10-26 Richard Biener <rguenther@suse.de>
447 * tree-vect-slp.c (vect_mark_slp_stmts): Add visited hash_set
449 (vect_mark_slp_stmts_relevant): Likewise.
450 (vect_detect_hybrid_slp_stmts): Likewise.
451 (vect_bb_slp_scalar_cost): Likewise.
452 (vect_remove_slp_scalar_calls): Likewise.
454 2018-10-26 Jan Hubicka <jh@suse.cz>
456 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix recursion.
457 (warn_types_mismatch): Fix walk of DECL_NAME.
458 (odr_types_equivalent_p): Fix overactive assert.
460 2018-10-26 Richard Biener <rguenther@suse.de>
462 PR tree-optimization/87105
463 * tree-vectorizer.h (_slp_tree::refcnt): New member.
464 * tree-vect-slp.c (vect_free_slp_tree): Decrement and honor
466 (vect_create_new_slp_node): Initialize refcnt to one.
468 (scalar_stmts_set_t, bst_fail): Remove.
469 (vect_build_slp_tree_2): Add bst_map argument and adjust calls.
470 (vect_build_slp_tree): Add bst_map argument and lookup
471 already created SLP nodes.
472 (vect_print_slp_tree): Handle a SLP graph, print SLP node
474 (vect_slp_rearrange_stmts): Handle a SLP graph.
475 (vect_analyze_slp_instance): Adjust and free SLP nodes from
476 the CSE map. Fix indenting.
477 (vect_schedule_slp_instance): Add short-cut.
479 2018-10-26 Martin Liska <mliska@suse.cz>
482 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Hash with
483 addr_expr and not with pointers.
485 2018-10-26 Jan Hubicka <jh@suse.cz>
487 * tree.c (free_lang_data_in_type): Only check main variants.
488 * ipa-devirt.c (warn_odr): Make static.
489 (types_same_for_odr): Drop strict variant.
490 (types_odr_comparable): Likewise.
491 (odr_or_derived_type_p): Look for main variants.
492 (odr_name_hasher::equal): Cleanup comment.
493 (odr_subtypes_equivalent): Add warn and warned arguments; check main
495 (type_variants_equivalent_p): break out from ...
496 (odr_types_equivalent): ... here; go for main variants where needed.
497 (warn_odr): ... here; turn static.
498 (warn_types_mismatch): Compare mangled names of main variants.
499 * ipa-utils.h (types_odr_comparable): Drop strict parameter.
500 (type_with_linkage_p): Sanity check that we look at main variant.
501 * lto.c (lto_read_decls): Only consider main variant to be ODR type.
502 * tree.h (types_same_for_odr): Drop strict argument.
504 2018-10-26 Richard Biener <rguenther@suse.de>
506 PR tree-optimization/87746
507 * tree-vect-data-refs.c (vect_update_misalignment_for_peel):
508 Simplify and fix WRT strided store groups with size not
509 equal to step in element count.
510 (vect_analyze_group_access_1): Dump the whole group.
512 2018-10-25 Carl Love <cel@us.ibm.com>
514 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_VSCEDPGT,
515 P9V_BUILTIN_VEC_VSCEDPLT, P9V_BUILTIN_VEC_VSCEDPEQ,
516 P9V_BUILTIN_VEC_VSCEDPUO): Rename base overloaded name. Add quad
517 precicion entry for each overloaded builtin.
518 * config/rs6000/rs6000-builtin.def (VSCEDPGT, VSCEDPLT, VSCEDPEQ,
519 VSCEDPUO): Rename overloaded name.
520 (VSCEDPGT, VSCEQPGT, VSCEDPLT, VSCEQPLT, VSCEDPEQ, VSCEQPEQ,
521 VSCEDPUO, VSCEQPUO): Add defitions for overloaded builtins.
522 * config/rs6000/vsx.md (xscmpexpqp_<code>_<mode>): Add
523 define_expand for xscmpexqp instruction.
524 (*xscmpexpqp): Add define_insn for the xscmpexqp instruction.
526 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
527 Jinsong Ji <jji@us.ibm.com>
529 * config/rs6000/emmintrin.h (_mm_slli_epi16): Replace deprecated
530 function with vec_sl.
531 (_mm_slli_epi32): Likewise.
532 (_mm_slli_epi64): Likewise.
533 (_mm_srai_epi16): Replace deprecated function with vec_sra.
534 (_mm_srai_epi32): Likewise.
535 (_mm_srli_epi16): Replace deprecated function with vec_sr.
536 (_mm_srli_epi32): Likewise.
537 (_mm_srli_epi64): Likewise.
538 (_mm_sll_epi16): Replace deprecated function with vec_sl.
539 (_mm_sll_epi32): Likewise.
540 (_mm_sll_epi64): Likewise.
541 (_mm_sra_epi16): Replace deprecated function with vec_sra.
542 (_mm_sra_epi32): Likewise.
543 (_mm_srl_epi16): Replace deprecated function with vec_sr.
544 (_mm_srl_epi32): Likewise.
545 (_mm_srl_epi64): Likewise.
547 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
548 Jinsong Ji <jji@us.ibm.com>
550 * gcc/config/rs6000/emmintrin.h (_mm_sll_epi16): Replace
551 comparison operators with vec_cmp* for compatibility due to
552 unfortunate history; clean up formatting and use types more
554 (_mm_sll_epi32): Likewise.
555 (_mm_sll_epi64): Likewise.
556 (_mm_srl_epi16): Likewise.
557 (_mm_srl_epi32): Likewise.
558 (_mm_srl_epi64): Likewise.
560 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
561 Jinsong Ji <jji@us.ibm.com>
563 * config/rs6000/emmintrin.h (_mm_sll_epi64): Remove wrong cast.
564 * config/rs6000/xmmintrin.h (_mm_min_ps): Change m's type to
565 __vector __bool int. Use vec_cmpgt in preference to deprecated
566 function vec_vcmpgtfp.
567 (_mm_max_ps): Likewise.
569 2018-10-25 Jeff Law <law@redhat.com>
571 * config/rl78/rl78.c (insn_ok_now): Always re-recognize the insn
574 2018-10-25 Martin Sebor <msebor@redhat.com>
576 * doc/extend.texi (aligned): Expand attribute description.
577 (Alignment): Rename section. Discuss function arguments.
579 2018-10-25 Jan Hubicka <jh@suse.cz>
581 * ipa-devirt.c (main_odr_variant): Remove.
582 (hash_odr_name, types_same_for_odr, types_odr_comparable,
583 odr_name_hasher::equal, odr_subtypes_equivalent_p):
584 Drop use of main_odr_variant.
585 (add_type_duplicate): Silence confused warnings on integer types.
586 (get_odr_type): Always look for main variant.
587 (register_odr_type): Simplify.
589 2018-10-25 Richard Biener <rguenther@suse.de>
591 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses):
592 Initialize ng to silence error with release checking bootstrap.
594 2018-10-25 Richard Biener <rguenther@suse.de>
596 * tree-if-conv.c: Include tree-ssa-sccvn.h.
597 (tree_if_conversion): Run CSE on the if-converted loop body.
599 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
601 * config/s390/constraints.md (ZL): New constraint.
602 * config/s390/s390.c (legitimate_pic_operand_p): Accept LARL
604 * config/s390/s390.md (movdi_larl): Remove.
605 (movdi_64): Add the LARL alternative.
607 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
610 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
611 (RTX_CODE_HWINT_P): New macro.
612 (rtx_code_size): Use RTX_CODE_HWINT_P ().
614 2018-10-25 Jan Hubicka <jh@suse.cz>
616 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
619 2018-10-25 Richard Biener <rguenther@suse.de>
621 PR tree-optimization/87665
622 PR tree-optimization/87745
623 * tree-vectorizer.h (get_earlier_stmt): Remove.
624 (get_later_stmt): Pick up UID from the original non-pattern stmt.
626 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
628 * options.texi (Deprecated): Move list to Var section.
630 2018-10-24 Bill Schmidt <wschmidt@linux.ibm.com>
631 Jinsong Ji <jji@us.ibm.com>
633 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Change deprecated
634 __vector long to __vector long long.
635 (_mm_cvtpd_ps): Likewise.
636 (_mm_cvttpd_epi32): Likewise.
637 (_mm_cvtpi32_pd): Likewise.
638 (_mm_unpackhi_epi64): Likewise.
639 (_mm_unpacklo_epi64): Likewise.
641 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
643 PR rtl-optimization/87720
644 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
646 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
648 * gimple-ssa-isolate-paths.c
649 (find_implicit_erroneous_behavior): Do not change code if the
650 pass is running for warnings only.
651 (find_explicit_erroneous_behavior): Likewise.
653 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
655 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
656 Define as rs6000_mangle_decl_assembler_name.
657 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
658 long double to IEEE long double, switch the names of the long
659 double built-in functions to be <func>f128 instead of <func>l.
661 2018-10-24 Martin Sebor <msebor@redhat.com>
663 * doc/extend.texi (nonnull): List no-argument form. Reference
664 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
666 2018-10-24 Richard Biener <rguenther@suse.de>
668 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
670 2018-10-24 Martin Liska <mliska@suse.cz>
672 PR tree-optimization/84436
673 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
675 (switch_conversion::contains_linear_function_p): New.
676 (switch_conversion::build_one_array): Support linear
677 transformation on input.
678 * tree-switch-conversion.h (struct switch_conversion): Add
679 contains_linear_function_p declaration.
681 2018-10-24 Richard Biener <rguenther@suse.de>
683 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
684 if its argument is CONSTANT_CLASS_P.
686 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
688 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
689 it is wrong for forward declarations.
691 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
693 * config/s390/s390.c (s390_check_qrst_address): Add the missing
694 SYMBOL_REF_P () check.
696 2018-10-24 Richard Biener <rguenther@suse.de>
698 PR tree-optimization/87105
699 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
701 (vect_analyze_data_ref_accesses): Handle duplicate loads and
702 stores by splitting the affected group after the fact.
703 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
704 fail the SLP build because of size constraints.
706 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
708 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
709 * configure: Regenerate.
710 * config.in: Regenerate.
711 * varasm.c (mergeable_string_section): Use readonly_data_section
712 if linker doesn't support SHF_MERGE with alignment > 8.
713 (mergeable_constant_section): Likewise.
715 2018-10-24 Richard Biener <rguenther@suse.de>
717 PR tree-optimization/84013
718 * tree-ssa-structalias.c (struct msdi_data): New struct for
719 marshalling data to walk_stmt_load_store_ops.
720 (maybe_set_dependence_info): Refactor as callback for
721 walk_stmt_load_store_ops.
722 (compute_dependence_clique): Set restrict info on all stmt kinds.
724 2018-10-24 Martin Liska <mliska@suse.cz>
726 * cgraph.c (cgraph_node::dump):
727 Remove reduntant dumps and make tp_first_run dump more compact.
729 2018-10-24 Richard Biener <rguenther@suse.de>
731 PR tree-optimization/87665
732 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
735 2018-10-12 Jeff Law <law@redhat.com>
737 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
740 2018-10-23 Richard Biener <rguenther@suse.de>
742 * tree-vrp.c (add_assert_info): Guard dump_printf with
744 * gimple-ssa-evrp-analyze.c
745 (evrp_range_analyzer::record_ranges_from_incoming_edge):
746 Use value_range::ignore_equivs_equal_p.
748 2018-10-23 Richard Biener <rguenther@suse.de>
750 PR tree-optimization/87105
751 PR tree-optimization/87608
752 * passes.def (pass_all_early_optimizations): Add early phi-opt
754 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
755 addition to debug stmts.
756 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
757 and abs replacement early.
758 * tree-cfg.c (gimple_empty_block_p): Likewise.
760 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
763 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
764 specified to configure.
765 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
767 2018-10-23 Richard Biener <rguenther@suse.de>
769 PR tree-optimization/87700
770 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
772 2018-10-23 Jakub Jelinek <jakub@redhat.com>
775 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
776 second argument from __mmask16 to __mmask8.
777 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
778 _mm_mask_packs_epi32): Likewise.
779 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
781 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
783 2018-10-23 Richard Biener <rguenther@suse.de>
785 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
787 2018-10-23 Richard Biener <rguenther@suse.de>
789 PR tree-optimization/86144
790 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
793 2018-10-23 Richard Biener <rguenther@suse.de>
795 PR tree-optimization/87693
796 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
797 the case we do not find the taken edge.
799 2018-10-22 Bill Schmidt <wschmidt@linux.ibm.com>
800 Jinsong Ji <jji@us.ibm.com>
802 * config/rs6000/emmintrin.h (_MM_SHUFFLE2): Comment cleanup.
803 (_mm_store_pd): Use unaligned vector type for pointer cast.
804 (_mm_maskmoveu_si128): Likewise.
805 * config/rs6000/xmmintrin.h (__m128_u): New typedef.
806 (_mm_store_ps): Use unaligned vector type for pointer cast.
808 2018-10-22 Paul Koning <ni1d@arrl.net>
810 * symtab.c (symtab_node::increase_alignment): Correct max
813 2018-10-22 Yury Gribov <tetra2005@gmail.com>
815 PR tree-optimization/87633
816 * match.pd: Do not generate unordered integer comparisons.
818 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
820 PR rtl-optimization/87600
821 * combine.c: Add include of expr.h.
822 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
823 register to a pseudo.
824 (make_more_copies): New function, add a copy to a new pseudo after
825 the moves from hard registers into pseudos.
826 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
827 later. Call make_more_copies.
829 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
831 * lra-constraints.c (process_alt_operands): New local array,
832 matching_early_clobber. Check matching_early_clobber before
833 decrementing reject, and set matching_early_clobber after.
835 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
838 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
839 call output_addr_const and hope for the best.
841 2018-10-22 Richard Biener <rguenther@suse.de>
843 * gimple-ssa-evrp-analyze.c
844 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
845 smarter about what ranges to use.
846 * tree-vrp.c (add_assert_info): Dump here.
847 (register_edge_assert_for_2): Instead of here at multiple but
850 * gcc.dg/tree-ssa/evrp12.c: New testcase.
851 * gcc.dg/predict-6.c: Adjust.
852 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
853 * gcc.dg/tree-ssa/vrp02.c: Likewise.
854 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
856 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
857 Richard Biener <rguenther@suse.de>
859 * bitmap.h: Update data structure documentation, including a
860 description of bitmap views as either linked-lists or splay trees.
861 (struct bitmap_element_def): Update comments for splay tree bitmaps.
862 (struct bitmap_head_def): Likewise.
863 (bitmap_list_view, bitmap_tree_view): New prototypes.
864 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
866 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
867 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
868 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
869 released bitmap element here.
870 (bitmap_element_free): Remove.
871 (bitmap_elt_clear_from): Work on splay tree bitmaps.
872 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
873 this function similar ones such that linked-list bitmap implementation
874 functions are grouped.
875 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
876 and moved for grouping.
877 (bitmap_list_insert_element_after): Renamed from
878 bitmap_elt_insert_after, and moved for grouping.
879 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
880 (bitmap_tree_link_left, bitmap_tree_link_right,
881 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
882 bitmap_tree_link_element, bitmap_tree_unlink_element,
883 bitmap_tree_find_element): New functions for splay-tree bitmap
885 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
886 Renamed and moved, see above entries.
887 (bitmap_tree_listify_from): New function to convert part of a splay
888 tree bitmap to a linked-list bitmap.
889 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
890 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
891 (bitmap_find_bit): Remove.
892 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
893 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
894 Handle splay tree bitmaps.
895 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
896 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
897 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
898 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
899 bitmap_intersect_compl_p, bitmap_ior_and_compl,
900 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
901 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
902 corresponding changes to use linked-list specific bitmap_element
903 manipulation functions as applicable for efficiency.
904 (bitmap_tree_to_vec): New function.
905 (debug_bitmap_elt_file): New function split out from ...
906 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
907 (bitmap_print): Likewise.
909 PR tree-optimization/63155
910 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
912 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
915 2018-10-22 Martin Liska <mliska@suse.cz>
917 PR tree-optimization/87686
919 2018-08-29 Martin Liska <mliska@suse.cz>
921 * tree-switch-conversion.c (switch_conversion::expand):
922 Strenghten assumption about gswitch statements.
924 2018-10-22 Martin Liska <mliska@suse.cz>
926 * ipa-icf.c (sem_item::compare_attributes): Remove.
927 (sem_item::compare_referenced_symbol_properties): Use
928 attribute_list_equal instead.
929 (sem_function::equals_wpa): Likewise.
930 * ipa-icf.h: Remove compare_attributes.
932 2018-10-22 Richard Biener <rguenther@suse.de>
935 * mem-stats.h (mem_usage::operator==): Fix pasto.
937 2018-10-22 Richard Biener <rguenther@suse.de>
939 PR tree-optimization/87640
940 * tree-vrp.c (set_value_range_with_overflow): Decompose
942 (extract_range_from_binary_expr_1): Adjust.
944 2018-10-22 Martin Jambor <mjambor@suse.cz>
946 * tree-eh.h (stmt_could_throw_p): Add function parameter.
947 (stmt_can_throw_external): Likewise.
948 (stmt_can_throw_internal): Likewise.
949 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
950 (lower_eh_constructs_2): Likewise.
951 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
952 (stmt_can_throw_external): Likewise.
953 (stmt_can_throw_internal): Likewise.
954 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
955 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
956 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
957 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
958 (pass_lower_eh_dispatch::execute): Pass cfun to
959 stmt_can_throw_external.
960 (cleanup_empty_eh): Likewise.
961 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
962 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
963 stmt_can_throw_external instead of pushing it to cfun.
964 (symbol_table::create_edge): Likewise.
965 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
966 stmt_can_throw_internal.
967 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
968 to stmt_could_throw_p.
969 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
970 stmt_can_throw_internal.
971 (pass_store_merging::execute): Likewise.
972 * gimple-ssa-strength-reduction.c
973 (find_candidates_dom_walker::before_dom_children): Pass cfun to
975 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
976 stmt_can_throw_internal.
977 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
978 to stmt_can_throw_external.
979 (check_stmt): Pass cfun to stmt_could_throw_p.
980 (check_stmt): Pass cfun to stmt_can_throw_external.
981 (pass_nothrow::execute): Likewise.
982 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
983 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
984 stmt_can_throw_internal.
985 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
986 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
987 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
988 * tree-complex.c (expand_complex_libcall): Pass cfun to
989 stmt_could_throw_p and to stmt_can_throw_internal.
990 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
991 * tree-inline.c (copy_edges_for_bb): Likewise.
992 (maybe_move_debug_stmts_to_successors): Likewise.
993 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
995 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
996 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
997 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
998 stmt_can_throw_internal.
999 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
1000 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
1002 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
1003 stmt_can_throw_internal.
1004 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
1005 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
1007 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
1008 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
1009 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1010 (convert_mult_to_fma_1): Likewise.
1011 (convert_to_divmod): Likewise.
1012 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
1013 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
1014 * tree-ssa-propagate.c
1015 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
1016 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
1017 (maybe_optimize_range_tests): Likewise.
1018 (linearize_expr_tree): Likewise.
1019 (reassociate_bb): Likewise.
1020 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
1021 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
1022 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
1023 (handle_char_store): Likewise.
1024 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
1025 stmt_can_throw_internal.
1026 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
1028 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
1029 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
1030 (vectorizable_simd_clone_call): Likewise.
1031 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
1032 (gimple_stringop_fixed_value): Likewise.
1034 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
1036 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
1037 literal pool references.
1038 (s390_check_qrst_address): Adapt to the new behavior of
1039 s390_loadrelative_operand_p ().
1041 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
1044 * config/i386/sse.md (*andnot<mode>3_bcst): New.
1046 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
1049 * config/i386/sse.md (*<code><mode>3_bcst): New.
1051 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
1054 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
1055 V4DI, V16SI and V8DI.
1056 (*sub<mode>3<mask_name>_bcst): New.
1057 (*add<mode>3<mask_name>_bcst): Likewise.
1059 2018-10-21 Bill Schmidt <wschmidt@linux.ibm.com>
1060 Jinsong Ji <jji@us.ibm.com>
1062 * config/rs6000/emmintrin.h (_mm_movemask_pd): Replace __vector
1063 __m64 with __vector unsigned long long for compatibility.
1064 (_mm_movemask_epi8): Likewise.
1065 * config/rs6000/xmmintrin.h (_mm_cvtps_pi32): Likewise.
1066 (_mm_cvttps_pi32): Likewise.
1067 (_mm_cvtpi32_ps): Likewise.
1068 (_mm_cvtps_pi16): Likewise.
1069 (_mm_loadh_pi): Likewise.
1070 (_mm_storeh_pi): Likewise.
1071 (_mm_movehl_ps): Likewise.
1072 (_mm_movelh_ps): Likewise.
1073 (_mm_loadl_pi): Likewise.
1074 (_mm_storel_pi): Likewise.
1075 (_mm_movemask_ps): Likewise.
1076 (_mm_shuffle_pi16): Likewise.
1078 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1081 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
1082 __builtin_ia32_vfnmsubpd512_mask.
1083 (_mm512_mask_fnmsub_round_pd): Likewise.
1084 (_mm512_fnmsub_pd): Likewise.
1085 (_mm512_mask_fnmsub_pd): Likewise.
1086 (_mm512_maskz_fnmsub_round_pd): Use
1087 __builtin_ia32_vfnmsubpd512_maskz.
1088 (_mm512_maskz_fnmsub_pd): Likewise.
1089 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
1090 (_mm512_mask_fnmsub_round_ps): Likewise.
1091 (_mm512_fnmsub_ps): Likewise.
1092 (_mm512_mask_fnmsub_ps): Likewise.
1093 (_mm512_maskz_fnmsub_round_ps): Use
1094 __builtin_ia32_vfnmsubps512_maskz.
1095 (_mm512_maskz_fnmsub_ps): Likewise.
1096 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
1097 __builtin_ia32_vfnmsubpd256_mask.
1098 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
1099 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
1100 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
1101 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
1102 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
1103 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
1104 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
1105 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
1106 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
1107 __builtin_ia32_vfnmsubpd.
1108 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
1109 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
1110 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
1111 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
1112 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
1113 * config/i386/i386-builtin.def: Add
1114 __builtin_ia32_vfnmsubpd256_mask,
1115 __builtin_ia32_vfnmsubpd256_maskz,
1116 __builtin_ia32_vfnmsubpd128_mask,
1117 __builtin_ia32_vfnmsubpd128_maskz,
1118 __builtin_ia32_vfnmsubps256_mask,
1119 __builtin_ia32_vfnmsubps256_maskz,
1120 __builtin_ia32_vfnmsubps128_mask,
1121 __builtin_ia32_vfnmsubps128_maskz,
1122 __builtin_ia32_vfnmsubpd512_mask,
1123 __builtin_ia32_vfnmsubpd512_maskz,
1124 __builtin_ia32_vfnmsubps512_mask,
1125 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
1126 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
1127 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
1128 __builtin_ia32_vfnmsubpd256.
1129 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
1130 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
1131 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
1133 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
1135 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
1137 (fmai_vmfnmsub_<mode><round_name>): Likewise.
1139 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1142 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
1143 __builtin_ia32_vfnmaddpd512_mask.
1144 (_mm512_mask_fnmadd_round_pd): Likewise.
1145 (_mm512_fnmadd_pd): Likewise.
1146 (_mm512_mask_fnmadd_pd): Likewise.
1147 (_mm512_maskz_fnmadd_round_pd): Use
1148 __builtin_ia32_vfnmaddpd512_maskz.
1149 (_mm512_maskz_fnmadd_pd): Likewise.
1150 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
1151 (_mm512_mask_fnmadd_round_ps): Likewise.
1152 (_mm512_fnmadd_ps): Likewise.
1153 (_mm512_mask_fnmadd_ps): Likewise.
1154 (_mm512_maskz_fnmadd_round_ps): Use
1155 __builtin_ia32_vfnmaddps512_maskz.
1156 (_mm512_maskz_fnmadd_ps): Likewise.
1157 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
1158 __builtin_ia32_vfnmaddpd256_mask.
1159 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
1160 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
1161 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
1162 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
1163 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
1164 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
1165 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
1166 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
1167 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
1168 __builtin_ia32_vfnmaddpd.
1169 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
1170 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
1171 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
1172 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
1173 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
1174 * config/i386/i386-builtin.def: Add
1175 __builtin_ia32_vfnmaddpd256_mask,
1176 __builtin_ia32_vfnmaddpd256_maskz,
1177 __builtin_ia32_vfnmaddpd128_mask,
1178 __builtin_ia32_vfnmaddpd128_maskz,
1179 __builtin_ia32_vfnmaddps256_mask,
1180 __builtin_ia32_vfnmaddps256_maskz,
1181 __builtin_ia32_vfnmaddps128_mask,
1182 __builtin_ia32_vfnmaddps128_maskz,
1183 __builtin_ia32_vfnmaddpd512_mask,
1184 __builtin_ia32_vfnmaddpd512_maskz,
1185 __builtin_ia32_vfnmaddps512_mask,
1186 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
1187 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
1188 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
1189 __builtin_ia32_vfnmaddpd256.
1190 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
1191 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
1192 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
1194 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
1196 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
1198 (fmai_vmfnmadd_<mode><round_name>): Likewise.
1200 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1203 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
1204 __builtin_ia32_vfmsubpd512_mask.
1205 (_mm512_mask_fmsub_round_pd): Likewise.
1206 (_mm512_fmsub_pd): Likewise.
1207 (_mm512_mask_fmsub_pd): Likewise.
1208 (_mm512_maskz_fmsub_round_pd): Use
1209 __builtin_ia32_vfmsubpd512_maskz.
1210 (_mm512_maskz_fmsub_pd): Likewise.
1211 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
1212 (_mm512_mask_fmsub_round_ps): Likewise.
1213 (_mm512_fmsub_ps): Likewise.
1214 (_mm512_mask_fmsub_ps): Likewise.
1215 (_mm512_maskz_fmsub_round_ps): Use
1216 __builtin_ia32_vfmsubps512_maskz.
1217 (_mm512_maskz_fmsub_ps): Likewise.
1218 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
1219 __builtin_ia32_vfmsubpd256_mask.
1220 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
1221 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
1222 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
1223 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
1224 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
1225 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
1226 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
1227 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
1228 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
1229 __builtin_ia32_vfmsubpd.
1230 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
1231 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
1232 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
1233 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
1234 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
1235 * config/i386/i386-builtin.def: Add
1236 __builtin_ia32_vfmsubpd256_mask,
1237 __builtin_ia32_vfmsubpd256_maskz,
1238 __builtin_ia32_vfmsubpd128_mask,
1239 __builtin_ia32_vfmsubpd128_maskz,
1240 __builtin_ia32_vfmsubps256_mask,
1241 __builtin_ia32_vfmsubps256_maskz,
1242 __builtin_ia32_vfmsubps128_mask,
1243 __builtin_ia32_vfmsubps128_maskz,
1244 __builtin_ia32_vfmsubpd512_mask,
1245 __builtin_ia32_vfmsubpd512_maskz,
1246 __builtin_ia32_vfmsubps512_mask,
1247 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
1248 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
1249 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
1250 __builtin_ia32_vfmsubpd256.
1251 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
1252 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
1253 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
1255 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
1257 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
1259 (fmai_vmfmsub_<mode><round_name>): Likewise.
1261 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1263 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
1264 Remove plus. Renamed to ...
1265 (*sub<mode>3<mask_name>_bcst): This.
1266 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
1267 (*add<mode>3<mask_name>_bcst): This.
1269 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1272 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
1274 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1277 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
1278 (_mm_or_epi32): Likewise.
1279 (_mm256_xor_epi32): Likewise.
1280 (_mm_xor_epi32): Likewise.
1281 (_mm256_or_epi64): Likewise.
1282 (_mm_or_epi64): Likewise.
1283 (_mm256_xor_epi64): Likewise.
1284 (_mm_xor_epi64): Likewise.
1286 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
1289 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
1291 2018-10-20 Jakub Jelinek <jakub@redhat.com>
1294 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
1296 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
1298 * doc/ux.texi: Move @section directly after @node.
1300 2018-10-19 Jakub Jelinek <jakub@redhat.com>
1304 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
1305 depend closely nested inside of loop with ordered clause with
1308 2018-10-19 David Malcolm <dmalcolm@redhat.com>
1310 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
1311 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
1312 * doc/ux.texi: New file.
1314 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
1316 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
1317 be the first CR field allocated.
1319 2018-10-19 Richard Biener <rguenther@suse.de>
1322 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
1323 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
1325 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
1328 * config/i386/sse.md
1329 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
1330 (*add<mode>3<mask_name>_bcst_2): Likewise.
1332 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
1334 * config/i386/sse.md
1335 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
1336 Replace nonimmediate_operand with register_operand.
1337 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
1339 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
1342 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
1344 PR rtl-optimization/87596
1345 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
1346 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
1347 for instructions in FROM..TO range.
1349 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
1351 * cfgexpand.c (expand_one_var): Use specific wording in error message
1352 for non-local frame variables.
1353 * stor-layout.c (layout_decl): Do not issue a warning for them.
1355 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
1357 * haifa-sched.c (priority): Add force_recompute parameter.
1358 (apply_replacement): Call priority () with force_recompute = true.
1359 (restore_pattern): Likewise.
1361 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1363 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
1364 HOST_BITS_PER_WIDE_INT.
1365 (test_vector_ops_duplicate): Likewise.
1367 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1370 * config/i386/sse.md (VF_AVX512): New.
1371 (avx512bcst): Likewise.
1372 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
1374 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
1376 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
1379 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
1381 * doc/invoke.texi (-dumpversion): Improve grammar.
1382 (-dumpfullversion): Make more consistent with -dumpversion.
1384 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
1386 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
1387 Set JUMP_LABEL to the jump insn.
1388 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
1389 Predict emitted jump and add label to jump insn.
1391 2018-10-18 David Malcolm <dmalcolm@redhat.com>
1393 PR tree-optimization/87562
1394 * input.c (get_substring_ranges_for_loc): Use
1395 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
1396 getting the linemap for the endpoint. Verify that it's either
1397 in the same linemap as the start point's spelling location, or
1398 at least in the same file.
1400 2018-10-18 Richard Biener <rguenther@suse.de>
1402 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
1403 feed width-specific load/store costs through ix86_vec_cost.
1404 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
1405 (k8_cost): Likewise.
1406 (bdver_cost): Likewise.
1407 (znver1_cost): Likewise.
1408 (btver1_cost): Likewise.
1409 (btver2_cost): Likewise.
1411 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1413 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
1414 to simplify subreg of vec_merge.
1416 2018-10-18 Richard Biener <rguenther@suse.de>
1418 * config/i386/i386.c: Fix costing of vector FMA.
1420 2018-10-18 Richard Biener <rguenther@suse.de>
1422 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
1424 (ix86_builtin_vectorization_cost): For vec_construct properly
1425 cost insertion into SSE regs.
1426 (...): Adjust calls to ix86_vec_cost.
1428 2018-10-18 Richard Biener <rguenther@suse.de>
1432 2018-02-07 Richard Biener <rguenther@suse.de>
1434 PR tree-optimization/84204
1435 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
1438 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1441 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
1443 (test_vector_ops_duplicate): Add test for a scalar subreg of a
1444 VEC_MERGE of a VEC_DUPLICATE.
1446 2018-10-17 Joseph Myers <joseph@codesourcery.com>
1448 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
1449 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
1450 * doc/standards.texi (C Language): Document C2X.
1451 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
1452 (rl78_option_override): Handle "GNU C2X" language name.
1454 2018-10-17 Joseph Myers <joseph@codesourcery.com>
1456 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
1457 Document C17 as published in 2018.
1459 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
1462 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
1463 bail out if both sides do not have the same storage order.
1465 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
1467 * bitmap.c (bitmap_head::dump): New.
1468 * bitmap.h (bitmap_head): Add dump().
1469 * gimple-ssa-evrp-analyze.c
1470 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
1471 (evrp_range_analyzer::set_ssa_range_info): Same.
1472 (evrp_range_analyzer::record_ranges_from_phis): Same.
1473 (evrp_range_analyzer::record_ranges_from_stmt): Same.
1474 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
1475 * gimple-ssa-sprintf.c (get_int_range): Same.
1476 (format_integer): Same.
1477 (sprintf_dom_walker::handle_gimple_call): Same.
1478 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
1479 (ipcp_vr_lattice::top_p): Same.
1480 (ipcp_vr_lattice::bottom_p): Same.
1481 (ipcp_vr_lattice::set_to_bottom): Same.
1482 (ipa_vr_operation_and_type_effects): Same.
1483 (propagate_vr_across_jump_function): Same.
1484 (ipcp_store_vr_results): Same.
1485 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
1486 (ipa_print_node_jump_functions_for_edge): Same.
1487 (ipa_get_value_range): Same.
1488 (ipa_compute_jump_functions_for_edge): Same.
1489 (ipa_write_jump_function): Same.
1490 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
1491 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
1493 * vr-values.c (set_value_range_to_nonnegative): Same.
1494 (set_value_range_to_truthvalue): Same.
1495 (vr_values::get_value_range): Same.
1496 (vr_values::set_defs_to_varying): Same.
1497 (vr_values::update_value_range): Same.
1498 (symbolic_range_based_on_p): Same.
1499 (vr_values::op_with_boolean_value_range_p): Same.
1500 (vr_values::extract_range_for_var_from_comparison_expr): Same.
1501 (vr_values::extract_range_from_ssa_name): Same.
1502 (vr_values::extract_range_from_binary_expr): Same.
1503 (vr_values::extract_range_from_unary_expr): Same.
1504 (vr_values::extract_range_from_cond_expr): Same.
1505 (vr_values::extract_range_from_comparison): Same.
1506 (vr_values::check_for_binary_op_overflow): Same.
1507 (vr_values::extract_range_basic): Same.
1508 (vr_values::extract_range_from_assignment): Same.
1509 (compare_ranges): Same.
1510 (compare_range_with_value): Same.
1511 (vr_values::adjust_range_with_scev): Same.
1512 (vrp_valueize): Same.
1513 (vrp_valueize_1): Same.
1514 (vr_values::get_vr_for_comparison): Same.
1515 (vr_values::compare_name_with_value): Same.
1516 (vr_values::compare_names): Same.
1517 (vr_values::vrp_evaluate_conditional): Same.
1518 (find_case_label_ranges): Same.
1519 (vr_values::vrp_visit_switch_stmt): Same.
1520 (vr_values::extract_range_from_phi_node): Same.
1521 (vr_values::simplify_div_or_mod_using_ranges): Same.
1522 (vr_values::simplify_bit_ops_using_ranges): Same.
1523 (test_for_singularity): Same.
1524 (range_fits_type_p): Same.
1525 (vr_values::simplify_cond_using_ranges_1): Same.
1526 (vr_values::simplify_switch_using_ranges): Same.
1527 (vr_values::simplify_float_conversion_using_ranges): Same.
1528 (vr_values::two_valued_val_range_p): Same.
1529 (vr_values::add_equivalence): Move to value_range::equiv_add.
1530 * vr-values.h (vr_values::add_equivalence): Remove.
1531 (VR_INITIALIZER): Remove.
1532 * tree-vrp.c (value_range::set): New.
1533 (value_range::equiv_add): New.
1534 (value_range::value_range): New.
1535 (value_range::deep_copy): New.
1536 (value_range::check): New.
1537 (value_range::equal_p): New.
1538 (value_range::ignore_equivs_equal_p): New.
1539 (value_range::operator==): New.
1540 (value_range::operator!=): New.
1541 (value_range::symbolic_p): New.
1542 (value_range::numeric_p): New.
1543 (value_range::set_undefined): New.
1544 (value_range::set_varying): New.
1545 (value_range::may_contain_p): New.
1546 (value_range::equiv_clear): New.
1547 (value_range::singleton_p): New.
1548 (value_range::intersect): New.
1549 (value_range::dump): New.
1550 (value_range::set_and_canonicalize): New.
1551 (set_value_range): Adjust for value_range API.
1552 (set_value_range_to_undefined): Same.
1553 (set_value_range_to_varying): Same.
1554 (set_and_canonicalize_value_range): Same.
1555 (set_value_range_to_nonnull): Same.
1556 (set_value_range_to_null): Same.
1557 (range_is_null): Same.
1558 (range_is_nonnull): Same.
1559 (range_int_cst_p): Same.
1560 (range_int_cst_singleton_p): Same.
1561 (symbolic_range_p): Same.
1562 (range_includes_zero_p): Same.
1563 (value_range_constant_singleton): Same.
1564 (vrp_set_zero_nonzero_bits): Same.
1565 (ranges_from_anti_range): Same.
1566 (extract_range_into_wide_ints): Same.
1567 (extract_range_from_multiplicative_op): Same.
1568 (set_value_range_with_overflow): Same.
1569 (extract_range_from_binary_expr_1): Same.
1570 (extract_range_from_unary_expr): Same.
1571 (dump_value_range): Same.
1572 (debug_value_range): Same.
1573 (vrp_prop::check_array_ref): Same.
1574 (vrp_prop::check_mem_ref): Same.
1575 (vrp_prop::vrp_initialize): Same.
1576 (vrp_prop::visit_stmt): Same.
1577 (intersect_ranges): Same.
1578 (vrp_prop::visit_phi): Same.
1579 (vrp_prop::vrp_finalize): Same.
1580 (determine_value_range_1): Same.
1581 (determine_value_range): Same.
1582 (vrp_intersect_ranges_1): Rename to...
1583 (vrp_intersect_1): this.
1584 (vrp_intersect_ranges): Rename to...
1585 (value_range::intersect_helper): ...this.
1586 (vrp_meet_1): Rename to...
1587 (value_range::union_helper): ...this.
1588 (vrp_meet): Rename to...
1589 (value_range::union_): ...this.
1590 (copy_value_range): Remove.
1591 * tree-vrp.h (struct value_range): Rewrite into a proper class.
1592 (value_range::vrtype): New.
1593 (value_range::type): New.
1594 (value_range::equiv): New.
1595 (value_range::min): New.
1596 (value_range::max): New.
1597 (value_range::varying_p): New.
1598 (value_range::undefined_p): New.
1599 (value_range::null_p): New.
1600 (value_range::equiv_add): New.
1601 (copy_value_range): Remove.
1603 2018-10-17 David Malcolm <dmalcolm@redhat.com>
1605 * Makefile.in (SELFTEST_TARGETS): New.
1606 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
1607 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
1608 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
1610 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
1611 (selftest-c++-gdb, selftest-c++-valgrind): Move to
1613 * configure: Regenerate.
1614 * configure.ac (selftest_languages): New.
1616 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
1618 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
1619 overflow wraps argument.
1620 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
1621 wide_int_range_multiplicative_op.
1622 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
1623 overflow wraps argument.
1624 (wide_int_range_multiplicative_op): Same.
1625 (wide_int_range_lshift): Same.
1626 (wide_int_range_div): Same.
1627 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
1628 (wide_int_range_lshift): Same.
1629 (wide_int_range_div): Same.
1631 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
1633 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
1634 use sign as argument.
1635 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
1636 wide_int_range_shift_undefined_p.
1638 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1640 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
1642 (@despeculate_copy<ALLI_TI:mode>): ... This.
1643 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
1646 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1648 * config.gcc: Obsolete *-*-solaris2.10*.
1649 * doc/install.texi (Specific, *-*-solaris2*): Document it.
1651 2018-10-12 Jeff Law <law@redhat.com>
1653 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
1654 reg + sym +- const_int addressing modes.
1656 2018-10-15 David Malcolm <dmalcolm@redhat.com>
1658 * common.opt (fdiagnostics-minimum-margin-width=): New option.
1659 * diagnostic-show-locus.c (layout::layout): Apply the minimum
1661 (layout::start_annotation_line): Only print up to 3 of the
1662 margin character, to avoid touching the left-hand side.
1663 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
1664 minimum margin width, as set by test_diagnostic_context's ctor.
1665 (selftest::test_fixit_insert_containing_newline): Likewise.
1666 (selftest::test_fixit_insert_containing_newline_2): Likewise.
1667 (selftest::test_line_numbers_multiline_range): Clear
1668 dc.min_margin_width.
1669 * diagnostic.c (diagnostic_initialize): Initialize
1671 * diagnostic.h (struct diagnostic_context): Add field
1673 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
1674 * opts.c (common_handle_option): Handle
1675 OPT_fdiagnostics_minimum_margin_width_.
1676 * selftest-diagnostic.c
1677 (selftest::test_diagnostic_context::test_diagnostic_context):
1678 Initialize min_margin_width to 6.
1679 * toplev.c (general_init): Initialize global_dc->min_margin_width.
1681 2018-10-15 David Malcolm <dmalcolm@redhat.com>
1683 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1684 Fix usage of "error_at_rich_loc" in the comment.
1686 2018-10-15 Renlin Li <renlin.li@arm.com>
1689 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
1690 if-conversioned loop when it contains ifn with types not
1691 supported by backend.
1692 * internal-fn.c (expand_direct_optab_fn): Add an assert.
1693 (direct_internal_fn_supported_p): New helper function.
1694 * internal-fn.h (direct_internal_fn_supported_p): Declare.
1696 2018-10-15 Jakub Jelinek <jakub@redhat.com>
1699 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
1700 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
1701 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
1703 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
1705 PR tree-optimization/87022
1706 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
1707 bits in dist vector rather than the first one.
1709 2018-10-15 Richard Biener <rguenther@suse.de>
1712 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
1713 (visit_loadstore): When a used restrict tag escaped verify that
1714 the points-to solution of "other" pointers do not include
1716 (compute_dependence_clique): If a used restrict tag escaped
1717 communicated that down to visit_loadstore.
1719 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
1721 * config/s390/s390.c (s390_expand_vec_init): Force vector element
1722 into reg if it isn't a general operand.
1724 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
1727 * config/i386/sse.md (*vec_dupv2di): Add register source to
1730 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
1733 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
1734 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
1735 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
1736 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
1738 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
1740 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
1741 (notice_args_size): Set it in the current trace if no insn that can
1742 throw internally has been seen yet.
1743 (connect_traces): When connecting args_size between traces, allow the
1744 incoming values not to match if there is an insn setting it before the
1745 first insn that can throw internally; in that case, force the creation
1746 of a CFI note on this latter insn.
1748 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
1750 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
1751 the base class, because using the injected-class-name was not clearly
1752 specified until DR 176.
1754 2018-10-12 Paul Koning <ni1d@arrl.net>
1756 * config/pdp11/pdp11.md (doloop_end): New expander.
1757 (doloop_end_insn): renamed from "doloop_end".
1758 (addqi3): New pattern.
1759 (subqi3): New pattern.
1760 * config/pdp11/predicates.md (incdec_operand): New predicate.
1762 2018-10-12 Yury Gribov <tetra2005@gmail.com>
1765 * real.c (format_helper::can_represent_integral_type_p): New function
1766 * real.h (format_helper::can_represent_integral_type_p): Ditto.
1767 * match.pd: New pattern.
1769 2018-10-12 Alexandre Oliva <oliva@adacore.com>
1771 * configure.ac: Introduce --enable-large-address-aware
1772 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
1773 * doc/install.texi: Document it.
1774 * configure, config.in: Rebuilt.
1775 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
1776 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
1777 (LINK_SPEC): Insert it.
1778 * config/i386/mingw-264.h: Likewise.
1780 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
1782 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
1784 PR rtl-optimization/87600
1785 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
1787 2018-10-12 Paul Koning <ni1d@arrl.net>
1789 * doc/md.texi (doloop_end): Document that the pattern code may
1790 need to check operand mode.
1792 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
1794 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
1795 to zero-extend between int and floating-point registers.
1796 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
1797 ldp into floating-point registers. Add type and arch attributes.
1798 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
1799 Use f_loads for type attribute.
1801 2018-10-11 Martin Sebor <msebor@redhat.com>
1803 * doc/extend.texi (attribute packed): Correct typos.
1805 2018-10-11 Martin Sebor <msebor@redhat.com>
1807 * doc/extend.texi (attribute flatten): Mention interaction with
1810 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
1813 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
1814 Set new_decl virtual flag to zero.
1816 2018-10-11 Martin Sebor <msebor@redhat.com>
1819 * doc/extend.texi (attribute format_arg): Discuss using multiple
1820 attributes on a single function.
1822 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
1824 PR tree-optimization/86829
1825 * match.pd (sin (atan (x))): New simplification rules.
1826 (cos (atan (x))): Likewise.
1827 * real.c (build_sinatan_real): New function.
1828 * real.h (build_sinatan_real): Prototype.
1830 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
1832 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
1834 (fold_mergeeo_helper): New helper function.
1835 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
1836 intrinsics. Correct some whitespace indentation issues.
1838 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
1841 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
1842 Use HOST_WIDE_INT_1U for shift.
1844 2018-10-11 Doug Rupp <rupp@adacore.com>
1845 Olivier Hainque <hainque@adacore.com>
1847 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
1848 Pass --relax to the linker for RTPs.
1849 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
1851 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
1852 Jan Hubicka <jh@suse.cz>
1853 Martin Jambor <mjambor@suse.cz>
1855 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
1856 the same elements are repeated rather than printing all of them.
1857 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
1858 "repeated" elements.
1859 * read-rtl-function.c (test_loading_repeat): New function.
1860 (read_rtl_function_c_tests): Call test_loading_repeat.
1861 * rtl-tests.c (test_dumping_repeat): New function.
1862 (rtl_tests_c_tests): Call test_dumping_repeat.
1864 2018-10-11 Richard Biener <rguenther@suse.de>
1866 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
1867 bdver?_cost): Unify to ...
1868 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
1869 * config/i386/i386.c (processor_cost_table): Adjust.
1871 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
1874 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
1875 the thunk when expanding to GIMPLE.
1877 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1879 * varasm.c (mergeable_string_section): Don't try to move zero-length
1880 strings to the merge section.
1882 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
1885 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
1887 2018-10-10 Jakub Jelinek <jakub@redhat.com>
1890 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
1891 to special_args set.
1893 2018-10-10 Richard Biener <rguenther@suse.de>
1895 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
1896 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
1897 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
1898 and recursing and pattern terminating the recursion on SSE
1899 vector width using ix86_expand_reduc.
1900 (reduc_sminmax_scal_<mode>): Split into part reducing to half
1901 width and recursing and SSE2 vector variant doing the final
1902 reduction with ix86_expand_reduc.
1903 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
1904 with terminating the recursion at AVX level, splitting that
1907 2018-10-09 David Malcolm <dmalcolm@redhat.com>
1909 * genmatch.c (error_cb): Rename to...
1910 (diagnostic_cb): ...this, converting int params to enums.
1911 (fatal_at): Update for renaming.
1912 (warning_at): Likewise.
1914 * input.c (selftest::ebcdic_execution_charset::apply):
1915 Update for renaming of...
1916 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
1918 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
1919 converting level and reason to enums.
1920 (class selftest::lexer_error_sink): Rename to...
1921 (class selftest::lexer_test_options): ...this, renaming field
1922 "m_errors" to "m_diagnostics".
1923 (selftest::lexer_test_options::apply): Update for renaming of...
1924 (selftest::lexer_test_options::on_error): ...this, renaming to...
1925 (selftest::lexer_test_options::on_diagnostic): ...this
1926 converting level and reason to enums.
1927 (selftest::test_lexer_string_locations_raw_string_unterminated):
1928 Update for renamings.
1929 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
1932 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
1934 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
1935 * config/rs6000/pmmintrin.h: New file.
1937 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
1939 PR tree-optimization/86659
1940 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
1942 2018-10-09 Richard Biener <rguenther@suse.de>
1944 PR tree-optimization/63155
1945 * tree-ssa-structalias.c: Include tree-ssa.h.
1946 (get_constraint_for_ssa_var): For undefs return nothing_id.
1947 (find_func_aliases): Cleanup PHI handling.
1949 2018-10-09 Richard Biener <rguenther@suse.de>
1951 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
1954 2018-10-09 Martin Liska <mliska@suse.cz>
1956 * asan.c (asan_emit_stack_protection): If a stack variable
1957 is located in a same file as current function, then emit
1958 line info into variable definition string.
1960 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
1962 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
1965 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
1967 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
1970 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
1972 PR tree-optimization/86659
1973 * gimple-match.h (struct gimple_match_op): Add reverse field.
1974 (gimple_match_op::set_op): New overloaded method.
1975 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
1976 the REF_REVERSE_STORAGE_ORDER flag on the value.
1977 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
1978 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
1980 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
1983 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
1984 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
1986 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
1989 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
1990 Defined with __builtin_ia32_vfmaddsubpd512_mask.
1992 2018-10-08 Richard Biener <rguenther@suse.de>
1994 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
1995 cost the same as AVX128 ones.
1997 2018-10-08 Paul Koning <ni1d@arrl.net>
1999 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
2000 (expand_block_move): New function.
2001 * config/pdp11/pdp11.c (output_block_move): Remove.
2002 (expand_block_move): New function.
2003 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
2004 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
2005 (*movmemhi1): Remove.
2007 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
2009 * config/s390/2827.md: Increase latencies for some FP instructions.
2011 2018-10-08 Richard Biener <rguenther@suse.de>
2013 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
2015 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
2016 * tree-vectorizer.h (dump_stmt_cost): Adjust.
2017 (add_stmt_cost): Dump return value of the hook.
2019 2018-10-08 Richard Biener <rguenther@suse.de>
2021 PR tree-optimization/63155
2022 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
2023 (ssa_propagation_engine::ssa_propagate): Remove redundant
2024 bitmap bit clearing.
2026 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
2028 PR rtl-optimization/86939
2029 PR rtl-optimization/87479
2030 * ira.h (non_conflicting_reg_copy_p): New prototype.
2031 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
2032 (make_hard_regno_dead): Don't add conflicts for register
2033 ignore_reg_for_conflicts.
2034 (make_object_dead): Likewise.
2035 (non_conflicting_reg_copy_p): New function.
2036 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
2037 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
2038 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
2039 (make_hard_regno_dead): Don't add conflicts for register
2040 ignore_reg_for_conflicts. Remove special conflict handling of
2041 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
2042 check_pic_pseudo_p and update callers.
2043 (mark_pseudo_dead): Don't add conflicts for register
2044 ignore_reg_for_conflicts.
2045 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
2047 2018-10-05 Andrew Waterman <andrew@sifive.com>
2048 Jim Wilson <jimw@sifive.com>
2050 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
2051 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
2052 new pattern using HONOR_SNANS that emits one extra instruction.
2054 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2056 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
2057 patterns): Merge SI and DI patterns to a GPR pattern.
2058 (unnamed define_insn and define_split for record form of that): Merge
2059 to a single define_insn_and_split pattern.
2061 2018-10-05 David Malcolm <dmalcolm@redhat.com>
2064 * input.c (expand_location_to_spelling_point): Add param "aspect"
2065 and use rather than hardcoding LOCATION_ASPECT_CARET.
2066 (get_substring_ranges_for_loc): Handle the case of a single token
2067 within a macro expansion.
2068 * input.h (expand_location_to_spelling_point): Add "aspect" param,
2069 defaulting to LOCATION_ASPECT_CARET.
2071 2018-10-05 Paul Koning <ni1d@arrl.net>
2073 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
2074 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
2075 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
2076 (pdp11_guard_type): New function.
2078 2018-10-05 Paul Koning <ni1d@arrl.net>
2080 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
2081 * config/pdp11/pdp11.opt (mfloat32): Remove.
2083 * doc/invoke.texi (pdp11 -mfloat32): Remove:
2084 (pdp11 -mfloat64): Remove.
2086 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
2088 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
2089 (*cmp<mode>_cc_i387): Ditto.
2090 (*cmpu<mode>_cc_i387): Ditto.
2091 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2092 * config/i386/i386.c (ix86_expand_fp_compare): Remove
2094 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
2095 Emit x86_sahf_1 pattern.
2096 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
2097 (ix86_expand_carry_flag_compare): Ditto.
2099 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
2101 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
2102 to reg_or_0_operand. Add "C" constraint.
2103 (*cmpxf_cc_i387): Ditto.
2104 (*cmp<mode>_i387): Change operand 2 predicate
2105 to nonimm_or_0_operand. Add "C" constraint.
2106 (*cmp<mode>_cc_i387): Ditto.
2107 (*cmp<mode>_0_i387): Remove insn pattern.
2108 (*cmp<mode>_0_cc_i387): Ditto.
2110 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
2112 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
2113 * config/i386/predicates.md (nonimm_or_0_operand): Rename
2114 from vector_move_operand. Update all uses.
2116 2018-10-05 Martin Sebor <msebor@redhat.com>
2118 PR tree-optimization/87490
2119 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
2122 2018-10-05 Richard Biener <rguenther@suse.de>
2124 PR tree-optimization/63155
2125 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
2126 vertical space in dumpfiles.
2127 * tree-ssa-propagate.h
2128 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
2129 * tree-ssa-propagate.c (cfg_blocks_back): New global.
2130 (ssa_edge_worklist_back): Likewise.
2131 (curr_order): Likewise.
2132 (cfg_blocks_get): Remove abstraction.
2133 (cfg_blocks_add): Likewise.
2134 (cfg_blocks_empty_p): Likewise.
2135 (add_ssa_edge): Add to current or next worklist based on
2137 (add_control_edge): Likewise.
2138 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
2140 (ssa_propagation_engine::ssa_propagate): ... here. Unify
2141 iteration from CFG and SSA edge worklist so we process
2142 everything in RPO order, prioritizing forward progress
2144 (ssa_prop_init): Allocate new worklists, do not dump
2146 (ssa_prop_fini): Free new worklists.
2148 2018-10-05 Richard Biener <rguenther@suse.de>
2150 * tree-core.h (tree_block::abstract_flag): Remove.
2151 (tree_block::block_num): Make full 32bits.
2152 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
2153 * tree.h (BLOCK_ABSTRACT): Remove.
2154 * dwarf2out.c (gen_lexical_block_die): Remove dead code
2155 resulting from BLOCK_ABSTRACT being always false.
2156 (gen_inlined_subroutine_die): Likewise.
2157 (gen_block_die): Likewise.
2158 * tree.c (block_ultimate_origin): Likewise.
2159 * tree-pretty-print.c (dump_block_node): Remove code dealing
2160 with BLOCK_ABSTRACT.
2161 * tree-ssa-live.c (dump_scope_block): Likewise.
2162 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
2163 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
2165 2018-10-05 Richard Biener <rguenther@suse.de>
2167 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
2168 is asked for initialize mode to the component mode of the
2171 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
2174 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
2175 assembler for -mavx.
2176 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
2178 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2181 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
2183 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
2184 to be DImode. When using mffscrn, force the operand to a register.
2186 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
2188 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
2189 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
2190 X87MODEF mode iterator.
2191 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
2192 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
2193 X87MODEF mode iterator.
2195 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
2197 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
2198 -Wno-prio-ctor-dtor.
2200 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2202 * Makefile.in (OBJS): Add opt-problem.o.
2203 * dump-context.h: Include "selftest.h.
2204 (selftest::temp_dump_context): New forward decl.
2205 (class dump_context): Make friend of class
2206 selftest::temp_dump_context.
2207 (dump_context::dump_loc_immediate): New decl.
2208 (class dump_pretty_printer): Move here from dumpfile.c.
2209 (class temp_dump_context): Move to namespace selftest.
2210 (temp_dump_context::temp_dump_context): Add param
2211 "forcibly_enable_dumping".
2212 (selftest::verify_dumped_text):
2213 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
2214 (selftest::verify_item):
2215 (ASSERT_IS_TEXT): Move here from dumpfile.c.
2216 (ASSERT_IS_TREE): Likewise.
2217 (ASSERT_IS_GIMPLE): Likewise.
2218 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
2220 (dump_context::dump_loc_immediate): ...this new function.
2221 (class dump_pretty_printer): Move to dump-context.h.
2222 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
2223 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
2224 (temp_dump_context::temp_dump_context): Move to "selftest"
2225 namespace. Add param "forcibly_enable_dumping", and use it to
2226 conditionalize the use of m_pp;
2227 (selftest::verify_dumped_text): Make non-static.
2228 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
2229 (selftest::verify_item): Make non-static.
2230 (ASSERT_IS_TEXT): Move to dump-context.h.
2231 (ASSERT_IS_TREE): Likewise.
2232 (ASSERT_IS_GIMPLE): Likewise.
2233 (selftest::test_capture_of_dump_calls): Pass "true" for new
2234 param of temp_dump_context.
2235 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
2236 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
2238 * opt-problem.cc: New file.
2239 * opt-problem.h: New file.
2240 * optinfo-emit-json.cc
2241 (selftest::test_building_json_from_dump_calls): Pass "true" for
2242 new param of temp_dump_context.
2243 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
2244 (optinfo::emit_for_opt_problem): New function.
2245 (optinfo::emit): Clarity which emit_item is used.
2246 * optinfo.h (optinfo::get_dump_location): New accessor.
2247 (optinfo::emit_for_opt_problem): New decl.
2248 (optinfo::emit): Make const.
2249 * selftest-run-tests.c (selftest::run_tests): Call
2250 selftest::opt_problem_cc_tests.
2251 * selftest.h (selftest::opt_problem_cc_tests): New decl.
2252 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
2253 bool to opt_result, converting fprintf messages to
2254 opt_result::failure_at calls. Add "stmt" param for use by the
2256 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
2257 (runtime_alias_check_p): Convert return type from bool to
2258 opt_result, converting dump_printf calls to
2259 opt_result::failure_at, using the statement DDR_A for their
2261 (find_data_references_in_stmt): Convert return type from bool to
2262 opt_result, converting "return false" to opt_result::failure_at
2264 * tree-data-ref.h: Include "opt-problem.h".
2265 (dr_analyze_innermost): Convert return type from bool to opt_result,
2266 and add a const gimple * param.
2267 (find_data_references_in_stmt): Convert return type from bool to
2269 (runtime_alias_check_p): Likewise.
2270 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
2271 dr_analyze_innermost.
2272 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
2273 Convert return type from bool to opt_result, adding a message for
2274 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
2275 (vect_analyze_data_ref_dependence): Convert return type from bool
2276 to opt_result. Change sense of return type from "false"
2277 effectively meaning "no problems" to "false" meaning a problem,
2278 so that "return false" becomes "return opt_result::success".
2279 Convert "return true" calls to opt_result::failure_at, using
2280 the location of statement A rather than vect_location.
2281 (vect_analyze_data_ref_dependences): Convert return type from bool
2283 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
2284 calls to opt_result::failure_at, using the stmt location rather
2286 (vect_verify_datarefs_alignment): Convert return type from bool
2288 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
2289 into multiple more-tightly-scoped copies.
2290 (vect_analyze_data_refs_alignment): Convert return type from bool
2292 (vect_analyze_data_ref_accesses): Likewise, converting a
2293 "return false" to a "return opt_result::failure_at", adding a
2295 (vect_prune_runtime_alias_test_list): Convert return type from
2296 bool to opt_result, converting dump_printf_loc to
2297 opt_result::failure_at. Add a %G to show the pertinent statement,
2298 and use the stmt's location rather than vect_location.
2299 (vect_find_stmt_data_reference): Convert return type from
2300 bool to opt_result, converting dump_printf_loc to
2301 opt_result::failure_at, using stmt's location.
2302 (vect_analyze_data_refs): Convert return type from bool to
2303 opt_result. Convert "return false" to "return
2304 opt_result::failure_at", adding messages as needed.
2305 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
2306 type from bool to opt_result.
2307 (vect_determine_vf_for_stmt): Likewise.
2308 (vect_determine_vectorization_factor): Likewise, converting
2309 dump_printf_loc to opt_result::failure_at, using location of phi
2310 rather than vect_location.
2311 (vect_analyze_loop_form_1): Convert return type from bool to
2312 opt_result, converting dump_printf_loc calls, retaining the use of
2314 (vect_analyze_loop_form): Convert return type from loop_vec_info
2315 to opt_loop_vec_info.
2316 (vect_analyze_loop_operations): Convert return type from bool to
2317 opt_result, converting dump_printf_loc calls, using the location
2318 of phi/stmt rather than vect_location where available. Convert
2319 various "return false" to "return opt_result::failure_at" with
2320 "unsupported phi" messages.
2321 (vect_get_datarefs_in_loop): Convert return type from bool to
2322 opt_result. Add a message for the
2323 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
2324 (vect_analyze_loop_2): Convert return type from bool to
2325 opt_result. Ensure "ok" is set to a opt_result::failure_at before
2326 each "goto again;", adding new messages where needed.
2327 Add "unsupported grouped {store|load}" messages.
2328 (vect_analyze_loop): Convert return type from loop_vec_info to
2330 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
2332 * tree-vect-stmts.c (process_use): Likewise, converting
2333 dump_printf_loc call and using stmt location, rather than
2335 (vect_mark_stmts_to_be_vectorized): Likeise.
2336 (vect_analyze_stmt): Likewise, adding a %G.
2337 (vect_get_vector_types_for_stmt): Convert return type from bool to
2338 opt_result, converting dump_printf_loc calls and using stmt
2339 location, rather than vect_location.
2340 (vect_get_mask_type_for_stmt): Convert return type from tree to
2341 opt_tree, converting dump_printf_loc calls and using stmt location.
2342 * tree-vectorizer.c: Include "opt-problem.h.
2343 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
2344 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
2345 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
2346 enabled, use it to report at the top level "couldn't vectorize
2347 loop" followed by the problem.
2348 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
2349 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
2351 (vect_analyze_stmt): Likewise.
2352 (vect_get_vector_types_for_stmt): Likewise.
2353 (tree vect_get_mask_type_for_stmt): Likewise.
2354 (vect_analyze_data_ref_dependences): Likewise.
2355 (vect_enhance_data_refs_alignment): Likewise.
2356 (vect_analyze_data_refs_alignment): Likewise.
2357 (vect_verify_datarefs_alignment): Likewise.
2358 (vect_analyze_data_ref_accesses): Likewise.
2359 (vect_prune_runtime_alias_test_list): Likewise.
2360 (vect_find_stmt_data_reference): Likewise.
2361 (vect_analyze_data_refs): Likewise.
2362 (vect_analyze_loop): Convert return type from loop_vec_info to
2364 (vect_analyze_loop_form): Likewise.
2365 (vect_analyze_slp): Convert return type from bool to opt_result.
2367 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2369 * doc/invoke.texi (-fopt-info): Document new "internals"
2371 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
2372 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
2374 (optinfo_verbosity_options): Add "internals".
2375 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
2376 (dump_context::apply_dump_filter_p): New member function.
2377 (dump_context::dump_loc): Use apply_dump_filter_p rather than
2378 explicitly masking the dump_kind.
2379 (dump_context::begin_scope): Increment the scope depth first. Use
2380 apply_dump_filter_p rather than explicitly masking the dump_kind.
2381 (dump_context::emit_item): Use apply_dump_filter_p rather than
2382 explicitly masking the dump_kind.
2383 (dump_dec): Likewise.
2384 (dump_hex): Likewise.
2385 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
2386 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
2387 (opt_info_switch_p): Update handling of default
2388 MSG_OPTIMIZED_LOCATIONS to cope with default of
2389 MSG_PRIORITY_USER_FACING.
2390 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
2391 masking the dump_kind.
2392 (selftest::test_capture_of_dump_calls): Update test_dump_context
2393 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
2394 than MSG_ALL. Generalize scope test to be run at all four
2395 combinations of with/without MSG_PRIORITY_USER_FACING and
2396 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
2397 for each of the two values.
2398 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
2399 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
2400 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
2401 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
2402 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
2403 with MSG_PRIORITY_*.
2404 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
2405 dump messages as MSG_PRIORITY_USER_FACING.
2406 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
2407 about the interaction with MSG_PRIORITY_*.
2409 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2411 * varasm.c (output_constant): Add new parameter merge_strings.
2412 Make strings properly zero terminated in merge string sections.
2413 (mergeable_string_section): Don't fail if the last char is non-zero.
2414 (assemble_variable_contents): Handle merge string sections.
2415 (assemble_variable): Likewise.
2416 (assemble_constant_contents): Likewise.
2417 (output_constant_def_contents): Likewise.
2418 (output_constructor_array_range,
2419 output_constructor_regular_field): Adjust call to output_constant.
2420 (output_object_block): Adjust call to assemble_constant_contents
2421 and assemble_variable_contents.
2423 2018-10-04 Martin Liska <mliska@suse.cz>
2426 * cgraphunit.c (process_function_and_variable_attributes):
2427 Warn about a function with alias attribute and a body.
2429 2018-10-04 Martin Liska <mliska@suse.cz>
2432 * multiple_target.c (redirect_to_specific_clone): New function.
2433 (ipa_target_clone): Use it.
2434 * tree-inline.c: Fix comment.
2436 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2438 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
2440 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
2441 (gcc::dump_manager::register_pass): New member function, adapted
2442 from loop body in gcc::pass_manager::register_pass, adding a
2443 call to update_dfi_for_opt_info.
2444 (gcc::dump_manager::opt_info_enable_passes): Store the
2445 -fopt-info options into the new fields. Move the loop
2447 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
2449 * dumpfile.h (struct opt_pass): New forward decl.
2450 (gcc::dump_manager::register_pass): New decl.
2451 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
2452 (class gcc::dump_manager): Add fields "m_optgroup_flags",
2453 "m_optinfo_flags", and "m_optinfo_filename".
2454 * passes.c (gcc::pass_manager::register_pass): Move all of the
2455 dump-handling code to gcc::dump_manager::register_pass.
2457 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
2459 PR rtl-optimization/87466
2460 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
2461 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
2462 * doc/tm.texi: Regenerate.
2463 * ira-lives.c (process_bb_node_lives): Use the new target hook.
2464 * lra-lives.c (process_bb_lives): Likewise.
2465 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
2468 2018-10-04 Tamar Christina <tamar.christina@arm.com>
2470 * params.c (add_params): Fix initialization.
2472 2018-10-04 Martin Liska <mliska@suse.cz>
2474 PR gcov-profile/84107
2475 * tree-profile.c (init_ic_make_global_vars):
2476 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
2477 Come up with new ic_tuple* variables. Emit
2478 __gcov_indirect_call{,_topn} variables.
2479 (gimple_gen_ic_profiler): Access the variable
2481 (gimple_gen_ic_func_profiler): Access
2482 __gcov_indirect_call.callee field.
2483 (gimple_init_gcov_profiler): Use ptr_type_node.
2484 * value-prof.c (gimple_ic): Use ptr_type_node.
2486 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2488 PR tree-optimization/85787
2489 * ipa-pure-const.c (malloc_candidate_p_1): Move most of malloc_candidate_p
2490 into this function and add support for detecting multiple phis.
2491 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
2493 2018-10-04 Martin Liska <mliska@suse.cz>
2496 * ipa-inline.c (inline_to_all_callers_1):
2497 Call ultimate_alias_target for node being inlined.
2499 2018-10-03 Jeff Law <law@redhat.com>
2501 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
2503 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
2504 * tree.h (get_typenode_from_name): Prototype.
2506 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
2508 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
2509 Change operand 2 predicate to nonimmediate_operand.
2510 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2512 2018-10-03 Martin Sebor <msebor@redhat.com>
2513 Jeff Law <law@redhat.com>
2515 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
2517 (get_string_length): Detect unterminated arrays.
2518 (format_string): Same.
2519 (format_directive): Warn about unterminated arrays.
2520 (handle_gimple_call): Mark statements with no_warning as needed.
2522 2018-10-03 Jim Wilson <jimw@sifive.com>
2524 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
2525 also define __riscv_abi_rve. Delete trailing white space.
2527 2018-10-03 Paul Koning <ni1d@arrl.net>
2529 Enable LRA register allocator for PDP11.
2530 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
2533 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
2534 * config/pdp11/pdp11.opt (-mlra): New option.
2535 * doc/invoke.texi (PDP-11 Options): Document -mlra.
2537 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
2539 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
2540 (*<absneg:code>extend<mode>xf2): Ditto.
2542 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
2544 PR tree-optimization/87415
2545 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
2548 2018-10-02 Jeff Law <law@redhat.com>
2550 * gimple-fold.c (get_range_strlen): Only set *nonstr when
2551 an unterminated string is discovered. Bubble up range
2552 even for unterminated strings.
2553 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
2554 indicates the string was not terminated via NONSTR.
2556 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
2558 * tree-vrp.c (extract_range_from_unary_expr): Special case all
2559 pointer conversions.
2560 Do not do anything special for anti-ranges.
2562 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
2564 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
2565 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
2567 2018-10-03 Martin Liska <mliska@suse.cz>
2569 PR gcov-profile/86109
2570 * coverage.c (coverage_begin_function): Do not
2571 mark lambdas as artificial.
2572 * tree-core.h (struct GTY): Remove tm_clone_flag
2573 and introduce new lambda_function.
2574 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
2576 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
2579 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
2580 P8_VECTOR and VSX are enabled.
2582 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
2584 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
2585 0x3907 as CPU model number.
2587 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
2589 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
2590 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
2591 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
2592 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
2593 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
2594 * config/s390/s390.md: Likewise. Rename also the cpu attribute
2595 value from arch12 to z14.
2597 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
2599 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
2601 (isinf<mode>2): Ditto.
2603 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
2605 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
2606 before emitting fxam. Perform calculations in XFmode.
2608 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
2610 * match.pd (((X /[ex] A) +- B) * A): New transformation.
2612 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
2615 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
2616 * tree.c (signed_or_unsigned_type_for): Handle complex.
2618 2018-10-02 Jeff Law <law@redhat.com>
2620 * gimple-fold.c (get_range_strlen): Remove dead code.
2622 2018-10-02 Martin Sebor <msebor@redhat.com>
2623 Jeff Law <law@redhat.com>
2625 * builtins.c (unterminated_array): Add new arguments.
2626 If argument is not terminated, bubble up size and exact
2628 (expand_builtin_strnlen): Detect, avoid expanding
2629 and diagnose unterminated arrays.
2630 (c_strlen): Fill in offset of start of unterminated strings.
2631 * builtins.h (unterminated_array): Update prototype.
2633 2018-10-02 Richard Biener <rguenther@suse.de>
2635 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
2636 of haddv4df, first reduce to SSE width and exploit the fact
2637 that we only need element zero with the reduction result.
2638 (reduc_plus_scal_v2df): Likewise.
2640 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
2642 * dojump.h (do_jump): Delete.
2643 (do_jump_1): Likewise.
2644 (split_comparison): Move around.
2645 * dojump.c (do_jump): Make static.
2646 (do_jump_1): Likewise.
2647 (jumpifnot): Move around.
2648 (jumpifnot_1): Likewise.
2650 (jumpif_1): Likewise.
2651 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
2653 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
2655 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
2656 insns in the delay slot and add_insn_after for the jump insn.
2658 2018-10-02 Richard Biener <rguenther@suse.de>
2660 * tree-inline.c (expand_call_inline): Use the location of
2661 the callee declaration for the inline-entry marker.
2662 * final.c (notice_source_line): Remove special-casing of
2663 NOTE_INSN_INLINE_ENTRY.
2665 2018-10-01 Carl Love <cel@us.ibm.com>
2668 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
2669 (__builtin_mtfsb0): New.
2670 (__builtin_mtfsb1): New.
2671 ( __builtin_set_fpscr_rn): New.
2672 (__builtin_set_fpscr_drn): New.
2673 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
2674 (rs6000_expand_set_fpscr_rn_builtin): Add.
2675 (rs6000_expand_set_fpscr_drn_builtin): Add.
2676 (rs6000_expand_builtin): Add case statement entries for
2677 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
2678 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
2679 RS6000_BUILTIN_MFFSL.
2680 (rs6000_init_builtins): Add ftype initialization and def_builtin
2681 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
2682 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
2683 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
2684 rs6000_mffscdrn): Add define_insn.
2685 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
2686 * doc/extend.texi: Add documentation for the builtins.
2688 2018-10-01 Richard Biener <rguenther@suse.de>
2690 PR tree-optimization/87465
2691 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
2692 causing branch miscounts.
2694 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2696 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
2697 aarch64_option_default_param): New.
2698 (params.h): Include.
2699 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
2700 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
2701 stack-clash protection validation code.
2703 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2705 * params.c (validate_param): New.
2706 (add_params): Use it.
2707 (set_param_value): Refactor param validation into validate_param.
2708 (diagnostic.h): Include.
2709 * diagnostic.h (diagnostic_ready_p): New.
2711 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2713 * params.c (set_param_value):
2714 Add index of parameter being validated.
2715 * common/common-target.def (option_validate_param): New.
2716 * common/common-targhooks.h (default_option_validate_param): New.
2717 * common/common-targhooks.c (default_option_validate_param): New.
2718 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
2719 * doc/tm.texi: Regenerate.
2721 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2724 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2725 Add validation for stack-clash parameters and set defaults.
2727 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2730 * configure.ac: Add stack-clash-protection-guard-size.
2731 * doc/install.texi: Document it.
2732 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
2733 * params.def: Update comment for guard-size.
2734 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
2735 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
2736 * configure: Regenerate.
2738 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2741 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
2742 STACK_DYNAMIC_OFFSET): New.
2743 * config/aarch64/aarch64.c (aarch64_layout_frame):
2744 Update outgoing args size.
2745 (aarch64_stack_clash_protection_alloca_probe_range,
2746 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
2748 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2751 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
2753 * target.def (stack_clash_protection_alloca_probe_range): New.
2754 (stack_clash_protection_final_dynamic_probe): Remove.
2755 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
2756 (default_stack_clash_protection_final_dynamic_probe): Remove.
2757 * targhooks.c: Likewise.
2758 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
2759 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
2760 * doc/tm.texi: Regenerate.
2762 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2765 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
2766 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
2767 aarch64_clamp_to_uimm12_shift): New.
2768 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
2769 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
2771 2018-10-01 Tamar Christina <tamar.christina@arm.com>
2774 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
2776 2018-10-01 Jeff Law <law@redhat.com>
2777 Richard Sandiford <richard.sandiford@linaro.org>
2778 Tamar Christina <tamar.christina@arm.com>
2781 * config/aarch64/aarch64.md
2782 (probe_stack_range): Add k (SP) constraint.
2783 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
2784 STACK_CLASH_MAX_UNROLL_PAGES): New.
2785 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
2786 stack probes for stack clash.
2787 (aarch64_allocate_and_probe_stack_space): New.
2788 (aarch64_expand_prologue): Use it.
2789 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
2790 (aarch64_sub_sp): Add emit_move_imm optional param.
2792 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
2794 PR tree-optimization/87261
2795 * match.pd: Remove trailing whitespace.
2796 Add (x & y) | ~(x | y) -> ~(x ^ y),
2797 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
2799 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
2801 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
2803 (ashlsi3_insn): Update instruction constraints.
2804 (ashrsi3_insn): Likewise.
2805 (rotrsi3): Likewise.
2806 (add_shift): Likewise.
2807 * config/arc/constraints.md (Csz): New 32 bit constraint. It
2808 avoids placing in the limm field small constants which, otherwise,
2809 could end into a small instruction.
2811 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
2813 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
2814 destination register is not odd-even.
2815 (umaddsidi4_split): Likewise.
2817 2018-10-01 Richard Biener <rguenther@suse.de>
2819 * tree-inline.c (expand_call_inline): Store origin of fn
2820 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
2821 * tree.c (block_ultimate_origin): Simplify and do some
2824 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
2826 * config/i386/mmx.md (EMMS): New int iterator.
2827 (emms): New int attribute.
2828 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
2829 EMMS int iterator. Explicitly declare clobbers.
2830 (mmx_emms): Remove expander.
2832 * config/i386/predicates.md (emms_operation): Remove predicate.
2833 (vzeroall_pattern): New predicate.
2834 (vzeroupper_pattern): Rename from vzeroupper_operation.
2835 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
2836 vzeroupper_pattern and vzeroall_pattern predicates.
2838 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
2840 PR rtl-optimization/86939
2841 * ira-lives.c (make_hard_regno_born): Rename from this...
2842 (make_hard_regno_live): ... to this. Remove update to conflict
2843 information. Update function comment.
2844 (make_hard_regno_dead): Add conflict information update. Update
2846 (make_object_born): Rename from this...
2847 (make_object_live): ... to this. Remove update to conflict information.
2848 Update function comment.
2849 (make_object_dead): Add conflict information update. Update function
2851 (mark_pseudo_regno_live): Call make_object_live.
2852 (mark_pseudo_regno_subword_live): Likewise.
2853 (mark_hard_reg_dead): Update function comment.
2854 (mark_hard_reg_live): Call make_hard_regno_live.
2855 (process_bb_node_lives): Likewise.
2856 * lra-lives.c (make_hard_regno_born): Rename from this...
2857 (make_hard_regno_live): ... to this. Remove update to conflict
2858 information. Remove now uneeded check_pic_pseudo_p argument.
2859 Update function comment.
2860 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
2861 to conflict information. Update function comment.
2862 (mark_pseudo_live): Remove update to conflict information. Update
2864 (mark_pseudo_dead): Add conflict information update.
2865 (mark_regno_live): Call make_hard_regno_live.
2866 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
2867 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
2869 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
2872 * config/i386/i386.c (construct_container): Use TImode for
2873 BLKmode values in 2 integer registers.
2875 2018-09-29 Jeff Law <law@redhat.com>
2877 * builtins.c (unterminated_array): Pass in c_strlen_data * to
2878 c_strlen rather than just a tree *.
2879 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
2880 Update recursive calls appropriately. If caller did not provide a
2881 suitable data pointer, create a local one. When a non-terminated
2882 string is discovered, bubble up information about the string via the
2883 c_strlen_data object.
2884 * builtins.h (c_strlen): Update prototype.
2885 (c_strlen_data): New structure.
2886 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
2887 For a type 2 call, if c_strlen indicates a non-terminated string
2888 use the length of the non-terminated string.
2889 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
2891 2018-09-29 Jakub Jelinek <jakub@redhat.com>
2894 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
2895 __m512d type for __A argument rather than __m512.
2897 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
2899 * match.pd (simple_comparison): Don't optimize if either operand is
2900 a function pointer when target needs function pointer canonicalization.
2902 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
2904 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
2905 power5 .. power9 to remove indirection.
2906 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
2907 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
2908 ASM_CPU_476_SPEC): Delete.
2909 (ASM_CPU_SPEC): Adjust.
2910 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
2911 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
2913 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
2915 * config.in: Delete HAVE_AS_DCI.
2916 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
2917 * config/rs6000/rs6000.h: Ditto.
2918 * configure.ac: Delete HAVE_AS_DCI.
2919 * configure: Regenerate.
2921 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
2923 * config.in (HAVE_AS_LWSYNC): Delete.
2924 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
2925 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
2927 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
2928 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
2930 * configure.ac: Delete HAVE_AS_LWSYNC.
2931 * configure: Regenerate.
2933 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
2934 Pierre-Marie de Rodat <derodat@adacore.com>
2936 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
2937 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
2938 (cgraph_node::create_thunk): Add indirect_offset parameter.
2939 (thunk_adjust): Likewise.
2940 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
2941 and initialize the corresponding field with it.
2942 (cgraph_node::dump): Dump indirect_offset field.
2943 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
2944 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
2945 (thunk_adjust): Add indirect_offset parameter and deal with it.
2946 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
2947 pass it to thunk_adjust. Do not call the target hook if it's non-zero
2948 or if the thunk is external or local. Fix formatting. Do not chain
2949 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
2950 if any, in the GIMPLE representation.
2951 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
2952 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
2953 (input_node): Read indirect_offset field.
2954 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
2955 call to thunk_adjust.
2956 * tree-nested.c (struct nesting_info): Add thunk_p field.
2957 (create_nesting_tree): Set it.
2958 (convert_all_function_calls): Copy static chain from targets to thunks.
2959 (finalize_nesting_tree_1): Return early for thunks.
2960 (unnest_nesting_tree_1): Do not finalize thunks.
2961 (gimplify_all_functions): Do not gimplify thunks.
2963 2018-09-28 David Malcolm <dmalcolm@redhat.com>
2965 * opt-suggestions.c (option_proposer::build_option_suggestions):
2966 Release "option_values".
2968 2018-09-28 David Malcolm <dmalcolm@redhat.com>
2970 * coverage.c (get_coverage_counts): Convert problem-reporting dump
2971 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
2972 * dumpfile.c (kind_as_string): New function.
2973 (dump_loc): Rather than a hardcoded prefix of "note: ", use
2974 kind_as_string to vary the prefix based on dump_kind.
2975 (selftest::test_capture_of_dump_calls): Update for above.
2977 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
2979 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
2980 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
2982 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
2984 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
2985 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
2986 INVALID_REGNUM instead of FPSR_REG.
2987 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
2988 * config/i386/i386.md: Update comment of FP compares.
2989 (fldenv): Do not clobber FPSR_REG.
2991 2018-09-28 Richard Biener <rguenther@suse.de>
2993 * tree.h (BLOCK_ORIGIN): New.
2994 * omp-expand.c (grid_expand_target_grid_body): Assign
2995 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
2996 * tree-inline.c (remap_block): Likewise.
2997 * auto-profile.c (get_function_decl_from_block): Simplify
2998 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
2999 * langhooks.c (lhd_print_error_function): Likewise.
3000 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
3002 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
3003 * tree.c (block_nonartificial_location): Likewise.
3004 (block_ultimate_origin): Likewise.
3005 * tree-pretty-print.c (percent_K_format): Likewise. Remove
3006 no longer needed LTO case.
3008 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
3009 Jan Hubicka <jh@suse.cz>
3010 Martin Jambor <mjambor@suse.cz>
3012 * simplify-rtx.c (simplify_merge_mask): New function.
3013 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
3014 same masks are used in op1 or op2.
3015 (test_vec_merge): New function.
3016 (test_vector_ops): Call test_vec_merge.
3018 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
3020 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
3021 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
3022 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
3023 (cypress_costs): Set it.
3024 (supersparc_costs): Likewise.
3025 (hypersparc_costs): Likewise.
3026 (leon_cost): Likewise.
3027 (leon3_costs): Likewise.
3028 (sparclet_costs): Likewise.
3029 (ultrasparc_costs): Likewise.
3030 (ultrasparc_costs): Likewise.
3031 (niagara_costs): Likewise.
3032 (niagara2_costs): Likewise.
3033 (niagara3_costs): Likewise.
3034 (niagara4_costs): Likewise.
3035 (niagara7_costs): Likewise.
3036 (m8_costs): Likewise.
3037 (TARGET_CAN_FOLLOW_JUMP): Define.
3038 (pass_work_around_errata::gate): Minor tweak.
3039 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
3040 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
3041 Automaitcally clear MASK_FSMULD mask for V7 processors.
3042 (sparc_can_follow_jump): New static function.
3043 (output_ubranch): Deal with CROSSING_JUMP_P.
3044 (sparc_use_sched_lookahead): Rewrite using switch statement.
3045 (sparc_issue_rate): Reorder.
3046 (sparc_branch_cost): New function.
3048 2018-09-27 Martin Sebor <msebor@redhat.com>
3050 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
3051 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
3052 (int_fits_type_p): Same.
3054 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
3056 * config/i386/i386.md (FPCR_REG): Remove.
3057 (UNSPEC_FLDCW): Remove.
3058 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
3059 (x86_fldcw_1): Remove insn pattern.
3060 (fnstenv): Do not clobber FPCR_REG.
3062 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
3063 (CALL_USED_REGISTERS): Ditto.
3064 (REG_ALLOC_ORDER): Ditto.
3065 (REG_CLASS_CONTENTS): Ditto.
3066 (HI_REGISTER_NAMES): Ditto.
3067 (ADDITIONAL_REGISTER_NAMES): Use defines instead
3068 of numerical constants.
3069 * config/i386/i386.c (regclass_map): Remove fpsr register.
3070 (dbx_register_map): Ditto.
3071 (dbx64_register_map): Ditto.
3072 (svr4_dbx_register_map): Ditto.
3073 (print_reg): Do not handle FPCR_REG.
3075 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
3078 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
3079 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
3080 Delete, always treat as true.
3081 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
3082 Ditto. Simplify remaining code.
3083 * config/powerpcspe/powerpcspe.h: Ditto.
3084 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
3085 Simplify remaining code.
3086 (rs6000_expand_builtin): Ditto.
3087 * config/rs6000/rs6000.h: Ditto.
3088 * configure.ac: Ditto.
3089 * configure: Regenerate.
3091 2018-09-27 Martin Liska <mliska@suse.cz>
3093 * coverage.c (get_coverage_counts): Revert the formatting
3094 of missing profile opt info.
3096 2018-09-27 Richard Biener <rguenther@suse.de>
3100 * dwarf2out.c (set_block_origin_self): Do not mark outermost
3101 block as we do not output that.
3102 (gen_inlined_subroutine_die): Elide the originally outermost
3103 block, matching what we do for concrete instances.
3104 (decls_for_scope): Add parameter specifying whether to recurse
3107 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
3108 Tom de Vries <tom@codesourcery.com>
3112 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
3113 STORE_FLAG_VALUE == 1.
3115 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
3117 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
3118 constant definitions.
3119 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
3120 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
3121 ("speculation_barrier"): New expander definition.
3123 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
3125 PR gcov-profile/86957
3126 * common.opt: New warning option -Wmissing-profile.
3127 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
3128 * doc/invoke.texi: Document -Wmissing-profile.
3130 2018-09-26 Jim Wilson <jimw@sifive.com>
3132 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
3133 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
3135 2018-09-26 Martin Sebor <msebor@redhat.com>
3137 * tree.c (zerop): Change return type to bool.
3138 (integer_zerop, integer_onep, integer_each_onep): Same.
3139 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
3140 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
3141 (real_onep, real_minus_onep, chain_index): Same.
3142 (print_type_hash_statistics, type_list_equal): Same.
3143 * tree.h (zerop): Same.
3144 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
3145 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
3146 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
3147 (real_onep, real_minus_onep, chain_index): Same.
3148 (print_type_hash_statistics, type_list_equal): Same.
3150 2018-09-26 Jim Wilson <jimw@sifive.com>
3152 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
3154 2018-09-26 Jakub Jelinek <jakub@redhat.com>
3157 * config/i386/i386.c: Include debug.h and dwarf2out.h.
3158 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
3161 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
3163 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
3165 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
3167 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
3168 and FP_SECOND_SSE_REGS.
3169 (REG_CLASS_NAMES): Ditto.
3170 (REG_CLASS_CONTENTS): Ditto.
3171 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
3172 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
3173 (ix86_preferred_output_reload_class): Ditto.
3174 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
3175 clobber constraint to "=&f".
3176 (fix_truncdi_i387): Ditto.
3177 (lrintxfdi2): Ditto.
3178 (fistdi2_<rounding>): Ditto.
3179 (fpremxf4_i387): Change "=u" constraint to "=f".
3180 (fprem1xf4_i387): Ditto.
3182 (fptanxf4_i387): Ditto.
3183 (fxtractxf3_i387): Ditto.
3184 (fscalexf4_i387): Ditto.
3185 (atan2xf3): Change "u" constraint to "f".
3186 (fyl2xxf3_i387): Ditto.
3187 (fyl2xp1xf3_i387): Ditto.
3189 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
3192 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
3193 for removed I387_MASK_PM entity.
3196 2018-09-26 Jeff Law <law@redhat.com>
3198 2018-09-26 Alexey Neyman <stilor@att.net>
3200 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
3201 headers are no longer pulled in by <isl/val.h>.
3203 2018-09-26 Richard Biener <rguenther@suse.de>
3206 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
3207 or concrete instance DIE to the tree. Create abstract origin
3208 attributes also for concrete instances.
3210 2018-09-26 Alexey Neyman <stilor@att.net>
3212 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
3213 headers are no longer pulled in by <isl/val.h>.
3215 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
3217 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
3218 Use new helper functions.
3219 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
3220 Use new helper functions.
3221 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
3222 aarch_mm_needs_release): New declarations.
3223 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
3224 aarch_mm_needs_release): New.
3226 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
3228 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
3229 (arm32_output_mi_thunk): Deal with long calls.
3231 2018-09-26 Richard Biener <rguenther@suse.de>
3235 * tree-inline.c (expand_call_inline): When the location
3236 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
3237 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
3238 the inserted BLOCK to make inlined_function_outer_scope_p
3240 * dwarf2out.c (add_call_src_coords_attributes): Do not add
3241 coords for reserved locations.
3243 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3245 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
3246 (*call_indirect_nonlocal_sysv<mode>): Ditto.
3247 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
3248 (*sibcall_nonlocal_sysv<mode>): Ditto.
3249 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
3250 (<bd>_<mode>): Ditto.
3251 (<bd>tf_<mode>): Ditto.
3253 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3255 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
3256 control string as a list of templates instead of as C code.
3257 (*altivec_movti): Ditto.
3258 * config/rs6000/darwin.md (movdf_low_di): Ditto.
3260 2018-09-25 Jim Wilson <jimw@sifive.com>
3262 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
3263 when target symbol is weak.
3265 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
3268 * builtins.c (unterminated_array): Simplify.
3269 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
3270 where pointer arithmetic is safe.
3272 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3275 * config/rs6000/altivec.md (altivec_vspltb): Use
3276 const_0_to_15_operand instead of u5bit_cint_operand.
3277 (*altivec_vspltb_internal): Ditto.
3278 (altivec_vspltb_direct): Ditto.
3279 (altivec_vsplth): Use const_0_to_7_operand instead of
3281 (*altivec_vsplth_internal): Ditto.
3282 (altivec_vsplth_direct): Ditto.
3283 (altivec_vspltw): Use const_0_to_3_operand instead of
3285 (*altivec_vspltw_internal): Ditto.
3286 (altivec_vspltw_direct): Ditto.
3287 (altivec_vspltsf): Ditto.
3288 (*altivec_vspltsf_internal): Ditto.
3289 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
3290 various splats with the proper size immediate. Reorder the various
3291 cases by ascending size of immediate, and put all such together.
3293 2018-09-25 Richard Biener <rguenther@suse.de>
3296 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
3298 (maybe_create_die_with_external_ref): Do not create
3299 DW_TAG_imported_unit here.
3300 (add_abstract_origin_attribute): Handle external BLOCK refs.
3301 (dwarf2out_abstract_function): Simplify LTO case.
3302 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
3303 rather than using maybe_create_die_with_external_ref.
3305 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
3308 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
3309 (frndintxf2_mask_pm_i387): Ditto.
3310 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
3311 Enable for !flag_trapping_math.
3312 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
3313 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
3314 Change operand 1 predicate to nonimmediate_operand.
3315 (attr "i387_cw"): Remove mask_pm.
3316 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
3317 (enum ix86_entity): Remove I387_MASK_PM.
3318 * config/i386/i386.c (ix86_i387_mode_needed): Do not
3319 handle I387_MASK_PM.
3320 (ix86_mode_needed): Ditto.
3321 (ix86_mode_after): Ditto.
3322 (ix86_mode_entry): Ditto.
3323 (ix86_mode_exit): Ditto.
3324 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
3326 2018-09-25 Jakub Jelinek <jakub@redhat.com>
3328 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
3329 to_update_switch_stmts to vNULL instead of calling create on them
3332 2018-09-25 Richard Biener <rguenther@suse.de>
3334 PR tree-optimization/87402
3335 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
3336 (visit_phi): Re-instantiate handling of supposed to be VARYING
3337 but non-VARYING backedge value.
3339 2018-09-25 Richard Biener <rguenther@suse.de>
3342 * dwarf2out.c (struct sym_off_pair): New.
3343 (external_die_map): New global.
3344 (lookup_decl_die): When in LTO create DIEs lazily from the
3346 (lookup_block_die): New function, create DIEs lazily in LTO.
3347 (equate_block_to_die): New function.
3348 (dwarf2out_die_ref_for_decl): During WPA get the association
3349 from the external DIE map.
3350 (dwarf2out_register_external_die): Record mapping into the
3352 (maybe_create_die_with_external_ref): New function split out from
3353 DIE generation part of old dwarf2out_register_external_die.
3354 (add_abstract_origin_attribute): Do not return the DIE. When
3355 in LTO reference externals directly.
3356 (dwarf2out_abstract_function): When in LTO ignore calls for
3357 decls with external DIEs (already present abstract instances).
3358 (gen_call_site_die): Adjust.
3359 (add_high_low_attributes): Likewise.
3360 (gen_lexical_block_die): Likewise.
3361 (gen_inlined_subroutine_die): Likewie.
3362 (gen_block_die): Likewise.
3363 (dwarf2out_inline_entry): Likewise.
3364 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
3367 2018-09-25 Martin Liska <mliska@suse.cz>
3369 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
3370 integers and not by a float value.
3372 2018-09-25 Martin Liska <mliska@suse.cz>
3375 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
3377 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
3378 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
3379 * godump.c (go_format_type): Remove extra parenthesis.
3381 2018-09-25 Martin Liska <mliska@suse.cz>
3383 * alias.c (set_dest_equal_p): Remove unused function.
3384 * config/i386/i386.c (def_builtin_pure2): Likewise.
3385 * diagnostic-show-locus.c (class layout): Remove
3387 (layout::layout): Likewise here.
3388 * dump-context.h (class temp_dump_context): Likewise.
3389 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
3390 (add_AT_loclistsptr): Likewise.
3391 (add_AT_offset): Likewise.
3392 (get_AT_hi_pc): Likewise.
3393 (is_comdat_die): Likewise.
3394 (type_is_enum): Likewise.
3395 (ceiling): Likewise.
3396 (add_AT_vms_delta): Likewise.
3397 (is_class_die): Likewise.
3398 * edit-context.c (class line_event): Remove unused field.
3399 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
3401 * ipa-cp.c (ipa_get_vr_lat): Likewise.
3402 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
3403 (ok_for_base_p_nonstrict): Likewise.
3404 * tree-chrec.c (is_not_constant_evolution): Likewise.
3405 (chrec_fold_poly_cst): Likewise.
3406 * tree-if-conv.c (has_pred_critical_p): Likewise.
3407 * tree-ssa-coalesce.c (print_exprs): Likewise.
3408 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
3409 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
3410 * tree-vrp.c (value_ranges_intersect_p): Likewise.
3411 (value_range_nonnegative_p): Likewise.
3413 2018-09-25 Martin Liska <mliska@suse.cz>
3415 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
3416 Do not handle "GNU Pascal".
3417 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
3419 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
3420 from documentation. Likewise.
3421 * dbxout.c (dbxout_range_type): Likewise.
3422 * doc/cpp.texi: Likewise.
3423 * doc/extend.texi: Likewise.
3424 * doc/frontends.texi: Likewise.
3425 * doc/invoke.texi: Remove Pascal entry.
3426 * tree.def (CLEANUP_POINT_EXPR): Likewise.
3427 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
3429 2018-09-25 Martin Liska <mliska@suse.cz>
3432 * doc/invoke.texi: Document all parameters and remove default
3435 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
3438 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
3439 contains HOST_WIDE_INTs when computing its size.
3441 2018-09-24 Jim Wilson <jimw@sifive.com>
3444 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
3446 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
3448 2018-09-24 Andrew Pinski <apinski@marvell.com>
3450 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
3451 access prev before checking it for NULLness in the
3452 AARCH64_FUSE_CMP_BRANCH case.
3454 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
3457 * config/i386/i386.c (rest_of_insert_endbranch): Set
3458 endbr_queued_at_entrance to true and don't insert ENDBR if
3459 x86_function_profiler will be called.
3460 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
3462 * config/i386/i386.h (machine_function): Add
3463 endbr_queued_at_entrance.
3465 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
3467 * genattrtab.c (mk_attr_alt): Use alternative_mask.
3468 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
3470 (check_attr_test): Use alternative_mask.
3471 (get_attr_value): Likewise.
3472 (compute_alternative_mask): Use alternative_mask and XWINT.
3473 (make_alternative_compare): Use alternative_mask.
3474 (attr_alt_subset_p): Use XWINT.
3475 (attr_alt_subset_of_compl_p): Likewise.
3476 (attr_alt_intersection): Use alternative_mask and XWINT.
3477 (attr_alt_union): Likewise.
3478 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
3479 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
3480 (simplify_test_exp): Use alternative_mask and XWINT.
3481 (write_test_expr): Use alternative_mask and XWINT, adjust bit
3482 number calculation to support 64 bits. Generate code that
3483 checks 64-bit masks.
3484 (main): Use alternative_mask.
3485 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
3487 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
3490 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
3491 RETURN+USE when returning via %r14.
3493 2018-09-24 Martin Liska <mliska@suse.cz>
3495 * gcov.c (output_lines): Print colorization legend
3496 for both flag_use_colors and flag_use_hotness_colors.
3499 2018-09-24 Martin Liska <mliska@suse.cz>
3501 * coverage.c (get_coverage_counts): Use warning_at
3502 with current_function_decl location. Use %qD in warning
3505 2018-09-24 Martin Liska <mliska@suse.cz>
3507 * memory-block.h (memory_block_pool::release): Annotate with
3508 valgrind that the memory is not accessible.
3510 2018-09-24 Martin Liska <mliska@suse.cz>
3513 * asan.c: Make asan_handled_variables extern.
3515 * cfgexpand.c (expand_stack_vars): Make sure
3516 a representative is unpoison if another
3517 variable in the partition is handled by
3518 use-after-scope sanitization.
3520 2018-09-24 Richard Biener <rguenther@suse.de>
3522 PR tree-optimization/63155
3523 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
3524 the worklist when the edge of the respective argument isn't
3527 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3529 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
3530 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
3531 (MASK_CLASS_P): Update for rename.
3532 (MAYBE_MASK_CLASS_P): Ditto.
3533 (REG_CLASS_NAMES): Update.
3534 (REG_CLASS_CONTENT): Update.
3535 * config/i386/i386.c (regclass_map): Update for MASK_REG
3536 and ALL_MASK_REGS rename.
3537 * config/i386/constraints.md (Yk): Update for rename.
3540 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3542 * config/i386/i386.h (enum reg_class): Remove
3543 EVEX_SSE_REGS and MOD4_SSE_REGS.
3544 (REG_CLASS_NAMES): Update.
3545 (REG_CLASS_CONTENT): Update.
3546 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
3547 registers as ALL_SSE_REGS.
3548 (ix86_additional_allocno_class_p): Remove.
3549 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
3550 (ix86_register_priority): Lower priority of EVEX SSE registers.
3551 Use IN_RANGE macro where appropriate.
3552 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
3553 AVX-5124VNNIW checks.
3554 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
3555 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
3556 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
3557 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
3558 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
3559 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
3560 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
3561 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
3562 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
3563 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
3564 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
3565 * config/i386/constraints.md (Yh): Remove.
3567 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3569 * config/i386/i386.c (regclass_map): Declare integer REX registers
3572 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
3574 * doc/service.texi (Service): Switch the fsf.org link to https.
3576 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
3579 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3580 Define to speculation_safe_value_not_needed.
3582 2018-09-21 Florian Weimer <fweimer@redhat.com>
3585 * doc/extend.texi (Common Function Attributes): Mention that
3586 noreturn suppresses tail call optimization.
3588 2018-09-21 Jeff Law <law@redhat.com>
3590 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
3591 vr_values::cleanup_edges_and_switches.
3592 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
3594 (identify_jump_threads): Remove EDGE_IGNORE handling.
3595 (execute_vrp): Move handling of to_remove_edges and
3596 to_update_switch_stmts into vr_values class member functions.
3597 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
3598 (to_update_switch_stmts): Likewise.
3599 * vr-values.c: Include cfghooks.h.
3600 (vr_values::vr_values): Initialize to_remove_edges and
3601 to_update_switch_stmts.
3602 (vr_values::~vr_values): Verify to_remove_edges and
3603 to_update_switch_stmts are empty.
3604 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
3605 (vr_values::cleanup_edges_and_switches): New member function.
3606 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
3607 function. Add new data members.
3609 2018-09-21 David Malcolm <dmalcolm@redhat.com>
3611 PR tree-optimization/87309
3612 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
3613 calls with pflags and alt_flags.
3614 (selftest::test_capture_of_dump_calls): Add test of interaction of
3615 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
3617 2018-09-21 Olivier Hainque <hainque@adacore.com>
3619 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
3621 2018-09-21 Olivier Hainque <hainque@adacore.com>
3623 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
3625 2018-09-21 Olivier Hainque <hainque@adacore.com>
3627 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
3628 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
3629 (VXWORKS_LIBS_RTP): Minor reordering.
3631 2018-09-21 Olivier Hainque <hainque@adacore.com>
3633 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
3634 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
3636 2018-09-21 Olivier Hainque <hainque@adacore.com>
3638 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
3639 (PTRDIFF_TYPE): Likewise.
3641 2018-09-21 Olivier Hainque <hainque@adacore.com>
3643 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
3644 triplet, similar to support for VxWorks7.
3645 * config/vxworks-dummy.h: Provide a default definition
3646 of TARGET_VXWORKS64 to 0.
3648 2018-09-21 Olivier Hainque <hainque@adacore.com>
3650 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
3651 * config/vxworks-dummy.h: here.
3653 2018-09-21 Olivier Hainque <hainque@adacore.com>
3655 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
3657 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
3658 Bo Zhou <zbo.zhou@hisilicon.com>
3660 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
3661 * config/aarch64/aarch64-tune.md: Regenerated.
3662 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
3663 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
3664 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
3666 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
3667 Julian Brown <julian@codesourcery.com>
3669 * builtins.c (get_builtin_sync_mem): Handle address spaces.
3671 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
3673 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
3674 if the call takes a static chain.
3676 2018-09-21 Martin Liska <mliska@suse.cz>
3678 * auto-profile.c (autofdo_source_profile::read): Do not
3680 (read_profile): Do not add working sets.
3681 (read_autofdo_file): Remove sum_all.
3682 (afdo_callsite_hot_enough_for_early_inline): Remove const
3684 * coverage.c (struct counts_entry): Remove gcov_summary.
3685 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
3686 do not support GCOV_TAG_PROGRAM_SUMMARY.
3687 (get_coverage_counts): Remove summary and expected
3689 * coverage.h (get_coverage_counts): Likewise.
3690 * doc/gcov-dump.texi: Remove -w option.
3691 * gcov-dump.c (dump_working_sets): Remove.
3692 (main): Do not support '-w' option.
3693 (print_usage): Likewise.
3694 (tag_summary): Likewise.
3695 * gcov-io.c (gcov_write_summary): Do not dump
3697 (gcov_read_summary): Likewise.
3698 (gcov_histo_index): Remove.
3699 (gcov_histogram_merge): Likewise.
3700 (compute_working_sets): Likewise.
3701 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
3703 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
3704 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
3705 (GCOV_HISTOGRAM_SIZE): Remove.
3706 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
3707 (struct gcov_summary): Simplify rapidly just
3708 to runs and sum_max fields.
3709 (gcov_histo_index): Remove.
3710 (NUM_GCOV_WORKING_SETS): Likewise.
3711 (compute_working_sets): Likewise.
3712 * gcov-tool.c (print_overlap_usage_message): Remove
3713 trailing empty line.
3714 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
3715 (output_lines): Remove program related line.
3716 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
3717 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
3719 (input_profile_summary): Do not read it.
3720 (merge_profile_summaries): And do not merge it.
3721 (input_symtab): Do not call removed function.
3722 * modulo-sched.c (sms_schedule): Do not print sum_max.
3723 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
3724 removed when histogram method was invented.
3725 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
3727 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
3729 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
3730 and dump selected value.
3731 * profile.c (add_working_set): Remove.
3732 (get_working_sets): Likewise.
3733 (find_working_set): Likewise.
3734 (get_exec_counts): Do not work with working sets.
3735 (read_profile_edge_counts): Do not inform as sum_max is removed.
3736 (compute_branch_probabilities): Likewise.
3737 (compute_value_histograms): Remove argument for call of
3738 get_coverage_counts.
3739 * profile.h: Do not make gcov_summary const.
3741 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
3743 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
3745 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
3747 PR tree-optimization/86990
3748 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
3749 Check that the entire merged store group is made of constants only for
3752 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
3754 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
3755 (VTABLE_VERIFICATION_SPEC): Likewise.
3756 (SANITIZER_EARLY_SPEC): Likewise.
3757 (SANITIZER_SPEC): Likewise.
3758 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
3759 * doc/invoke.texi (Link Options): Document -r.
3761 2018-09-20 Richard Biener <rguenther@suse.de>
3764 * gimplify.c (gimplify_expr): Retain alignment of
3765 addressable lvalue in dereference.
3767 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
3770 * configure.ac: Check for .loc is_stmt support.
3771 * configure, config.in: Rebuilt.
3772 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
3775 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3777 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
3778 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
3781 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3783 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
3785 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
3788 2018-09-20 Marek Polacek <polacek@redhat.com>
3790 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
3792 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
3794 PR tree-optimization/87288
3795 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
3796 into account when determining PEELING_FOR_NITERS.
3798 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
3800 PR tree-optimization/86877
3801 * tree-vect-loop.c (vect_analyze_loop_2): Call
3802 vect_verify_datarefs_alignment.
3804 2018-09-19 Marek Polacek <polacek@redhat.com>
3806 * doc/invoke.texi: Document -Wclass-conversion.
3808 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
3810 * config/pa/pa.c (pa_adjust_priority): Delete.
3811 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
3813 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
3814 (atomic_storehi): Likewise.
3815 (atomic_storesi): Likewise.
3816 (atomic_loaddi): Restore compare and swap exchange loop code.
3818 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
3820 PR rtl-optimization/86902
3821 * combine.c (try_combine): When changing the CC mode used, don't change
3822 an unrelated mode in other_insn to that new CC mode.
3824 2018-09-19 David Malcolm <dmalcolm@redhat.com>
3826 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
3827 with %T in place of calls to dump_generic_expr.
3828 (prune_runtime_alias_test_list): Likewise.
3829 (create_runtime_alias_checks): Likewise.
3830 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
3831 (vect_analyze_data_ref_dependence): Likewise.
3832 (vect_slp_analyze_data_ref_dependence): Likewise.
3833 (vect_record_base_alignment): Likewise. Use %G in place of call
3834 to dump_gimple_stmt.
3835 (vect_compute_data_ref_alignment): Likewise.
3836 (verify_data_ref_alignment): Likewise.
3837 (vect_find_same_alignment_drs): Likewise.
3838 (vect_analyze_group_access_1): Likewise.
3839 (vect_analyze_data_ref_accesses): Likewise.
3840 (dependence_distance_ge_vf): Likewise.
3841 (dump_lower_bound): Likewise.
3842 (vect_prune_runtime_alias_test_list): Likewise.
3843 (vect_find_stmt_data_reference): Likewise.
3844 (vect_analyze_data_refs): Likewise.
3845 (vect_create_addr_base_for_vector_ref): Likewise.
3846 (vect_create_data_ref_ptr): Likewise.
3847 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
3848 (vect_can_advance_ivs_p): Likewise.
3849 (vect_update_ivs_after_vectorizer): Likewise.
3850 (vect_gen_prolog_loop_niters): Likewise.
3851 (vect_prepare_for_masked_peels): Likewise.
3852 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
3853 (vect_determine_vectorization_factor): Likewise.
3854 (vect_is_simple_iv_evolution): Likewise.
3855 (vect_analyze_scalar_cycles_1): Likewise.
3856 (vect_analyze_loop_operations): Likewise.
3857 (report_vect_op): Likewise.
3858 (vect_is_slp_reduction): Likewise.
3859 (check_reduction_path): Likewise.
3860 (vect_is_simple_reduction): Likewise.
3861 (vect_create_epilog_for_reduction): Likewise.
3862 (vect_finalize_reduction:): Likewise.
3863 (vectorizable_induction): Likewise.
3864 (vect_transform_loop_stmt): Likewise.
3865 (vect_transform_loop): Likewise.
3866 (optimize_mask_stores): Likewise.
3867 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
3868 (vect_split_statement): Likewise.
3869 (vect_recog_over_widening_pattern): Likewise.
3870 (vect_recog_average_pattern): Likewise.
3871 (vect_determine_min_output_precision_1): Likewise.
3872 (vect_determine_precisions_from_range): Likewise.
3873 (vect_determine_precisions_from_users): Likewise.
3874 (vect_mark_pattern_stmts): Likewise.
3875 (vect_pattern_recog_1): Likewise.
3876 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
3877 (vect_record_max_nunits): Likewise.
3878 (vect_build_slp_tree_1): Likewise.
3879 (vect_build_slp_tree_2): Likewise.
3880 (vect_print_slp_tree): Likewise.
3881 (vect_analyze_slp_instance): Likewise.
3882 (vect_detect_hybrid_slp_stmts): Likewise.
3883 (vect_detect_hybrid_slp_1): Likewise.
3884 (vect_slp_analyze_operations): Likewise.
3885 (vect_slp_analyze_bb_1): Likewise.
3886 (vect_transform_slp_perm_load): Likewise.
3887 (vect_schedule_slp_instance): Likewise.
3888 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
3889 (vect_mark_stmts_to_be_vectorized): Likewise.
3890 (vect_init_vector_1): Likewise.
3891 (vect_get_vec_def_for_operand): Likewise.
3892 (vect_finish_stmt_generation_1): Likewise.
3893 (vect_check_load_store_mask): Likewise.
3894 (vectorizable_call): Likewise.
3895 (vectorizable_conversion): Likewise.
3896 (vectorizable_operation): Likewise.
3897 (vectorizable_load): Likewise.
3898 (vect_analyze_stmt): Likewise.
3899 (vect_is_simple_use): Likewise.
3900 (vect_get_vector_types_for_stmt): Likewise.
3901 (vect_get_mask_type_for_stmt): Likewise.
3902 * tree-vectorizer.c (increase_alignment): Likewise.
3904 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
3906 * doc/rtl.texi: Adjust vec_select description.
3907 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
3908 non-constant selectors.
3910 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
3912 * config/aarch64/aarch64-protos.h
3913 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
3914 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
3915 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
3916 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
3917 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
3918 (AARCH64_FL_PROFILE): Move index so flags are ordered.
3919 (AARCH64_ISA_RCPC8_4): New flag.
3920 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
3921 to aarch64_offset_9bit_signed_unscaled_p.
3922 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
3924 * config/aarch64/constraints.md (Ust): New constraint.
3925 * config/aarch64/predicates.md.
3926 (aarch64_9bit_offset_memory_operand): New predicate.
3927 (aarch64_rcpc_memory_operand): New predicate.
3929 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
3931 PR rtl-optimization/87361
3932 * rtlanal.c (nonzero_bits1): Revert accidental change.
3934 2018-09-19 Richard Biener <rguenther@suse.de>
3936 PR tree-optimization/87349
3937 PR tree-optimization/87342
3938 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
3940 2018-09-18 Marek Polacek <polacek@redhat.com>
3942 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
3943 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
3945 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
3947 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
3949 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
3951 PR rtl-optimization/86882
3952 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
3954 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
3956 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
3957 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
3959 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
3962 * doc/invoke.texi (Link Options): Fix formatting and grammar.
3964 2018-09-18 Richard Biener <rguenther@suse.de>
3967 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
3968 (compute_samebase_partition_bases): Likewise.
3969 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
3970 (gimple_can_coalesce_p): Simplify.
3972 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
3974 Handle a library implementation of ffs calling __builtin_ffs.
3975 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
3976 (mmix_init_libfuncs): New function: make __builtin_ffs expand
3979 2018-09-17 David Malcolm <dmalcolm@redhat.com>
3981 * diagnostic-show-locus.c (class layout_range): Add field
3983 (layout_range::layout_range): Add "original_idx" param and use it
3984 to initialize new field.
3985 (make_range): Use 0 for original_idx.
3986 (layout::layout): Pass in index to calls to
3987 maybe_add_location_range.
3988 (layout::maybe_add_location_range): Add param "original_idx" and
3989 pass it on to layout_range.
3990 (layout::print_any_labels): Pass on range->m_original_idx to
3992 (gcc_rich_location::add_location_if_nearby): Use 0 for
3994 * gcc-rich-location.h (text_range_label::get_text): Update for new
3996 (range_label_for_type_mismatch::get_text): Likewise.
3998 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
4000 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
4002 2018-09-17 David Malcolm <dmalcolm@redhat.com>
4004 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
4005 format_string_diagnostic_t.
4006 (fmtwarn_n): Likewise.
4007 * substring-locations.c
4008 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
4009 (format_warning_n_va): Convert to...
4010 (format_string_diagnostic_t::emit_warning_n_va): ...this.
4011 (format_warning_va): Convert to...
4012 (format_string_diagnostic_t::emit_warning_va): ...this.
4013 (format_warning_at_substring): Convert to...
4014 (format_string_diagnostic_t::emit_warning): ...this.
4015 (format_warning_at_substring_n): Convert to...
4016 (format_string_diagnostic_t::emit_warning_n): ...this.
4017 * substring-locations.h (class format_string_diagnostic_t): New
4019 (format_warning_va): Convert to
4020 format_string_diagnostic_t::emit_warning_va.
4021 (format_warning_n_va): Convert to
4022 format_string_diagnostic_t::emit_warning_n_va.
4023 (format_warning_at_substring): Convert to
4024 format_string_diagnostic_t::emit_warning.
4025 (format_warning_at_substring_n): Convert to
4026 format_string_diagnostic_t::emit_warning_n.
4028 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
4029 Bernd Schmidt <bernds_cb1@t-online.de>
4031 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
4034 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
4036 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
4037 operand 0 predicate to nonimmediate operand.
4038 (rint<mode>2_frndint): Remove insn pattern.
4039 (rint<mode>2): Change operand 1 predicate to general_operand.
4040 Extend operand 1 to XFmode and generate rintxf2 insn.
4041 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
4042 Do not use X87MODEF mode macro.
4043 (frndintxf2_<rounding>_i387): Rename from
4044 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
4045 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
4046 to XFmode and generate significandxf3 insn.
4048 2018-09-17 Richard Biener <rguenther@suse.de>
4050 PR tree-optimization/87328
4051 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
4052 visiting unexecutable backedges when not iterating.
4053 (do_rpo_vn): Mark all edges not executable even when not
4056 2018-09-17 Martin Jambor <mjambor@suse.cz>
4059 * doc/invoke.texi (Warning Options): Likewise.
4061 2018-09-17 Richard Biener <rguenther@suse.de>
4063 PR tree-optimization/87301
4064 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
4065 clean EH info from leftover copy assignments.
4067 2018-09-17 Martin Liska <mliska@suse.cz>
4069 PR gcov-profile/85871
4070 * gcov.c (output_intermediate_file): Fix out of bounds
4073 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
4075 * config/arc/arc.c: Object attributes for core4 not reflected
4077 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
4080 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
4082 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
4084 2018-09-17 Martin Liska <mliska@suse.cz>
4086 * doc/gcov.texi: Document new option --use-hotness-colors.
4087 * gcov.c (struct source_info): Declare new field.
4088 (source_info::source_info): Set default for maximum_count.
4089 (print_usage): Add new -q option.
4090 (process_args): Process it.
4091 (accumulate_line_info): Save src->maximum_count.
4092 (output_line_beginning): Make color line number if
4093 flag_use_hotness_colors is set.
4094 (output_line_details): Pass default argument value.
4095 (output_lines): Pass src->maximum_count.
4097 2018-09-17 Martin Liska <mliska@suse.cz>
4099 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
4100 Use processor_names table.
4101 * config/i386/i386.c (ix86_default_align): Use
4102 processor_cost_table for alignment values.
4103 (ix86_option_override_internal): Use processor_names.
4104 (ix86_function_specific_print): Likewise.
4105 * config/i386/i386.h (struct processor_costs):
4106 Add alignment values.
4107 (struct ptt): Remove and replace with const char *.
4108 * config/i386/x86-tune-costs.h (struct processor_costs):
4109 Declare default alignments for all costs.
4111 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
4113 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
4114 symbolics or VR_VARYING ranges for ABS_EXPR.
4115 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
4116 when range will wrap.
4118 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
4121 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
4122 before and after a JUMP_TABLE_DATA.
4124 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
4127 * dojump.c (do_compare_and_jump): Canonicalize function pointers
4128 when one operand is a function pointer. Use POINTER_TYPE_P and
4129 FUNC_OR_METHOD_TYPE_P.
4130 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
4131 * fold-const.c (build_range_check): Likewise.
4132 * match.pd (simple_comparison): Likewise.
4134 2018-09-14 David Malcolm <dmalcolm@redhat.com>
4137 * spellcheck.c (get_edit_distance_cutoff): New function.
4138 (selftest::test_edit_distance_unit_test_oneway): Rename to...
4139 (selftest::test_get_edit_distance_one_way): ...this.
4140 (selftest::test_get_edit_distance_unit): Rename to...
4141 (selftest::test_get_edit_distance_both_ways): ...this.
4142 (selftest::test_edit_distances): Move tests to this new function,
4143 and test some more pairs of strings. Update for above renaming.
4144 (selftest::get_old_cutoff): New function.
4145 (selftest::test_get_edit_distance_cutoff): New function.
4146 (selftest::assert_suggested_for): New function.
4147 (ASSERT_SUGGESTED_FOR): New macro.
4148 (selftest::assert_not_suggested_for): New function.
4149 (ASSERT_NOT_SUGGESTED_FOR): New macro.
4150 (selftest::test_suggestions): New function.
4151 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
4152 tests to selftest::test_edit_distances and call it. Add calls to
4153 selftest::test_get_edit_distance_cutoff and
4154 selftest::test_suggestions.
4155 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
4156 (best_match::consider): Replace hard-coded cutoff calculation with
4158 (best_match::get_cutoff): New declaration.
4159 (best_match::get_best_meaningful_candidate): Likewise.
4161 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
4163 * builtins.c (fold_builtin_strlen): Remove TODO comment.
4165 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
4168 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
4170 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
4171 terminated string literal.
4173 2018-09-14 Martin Sebor <msebor@redhat.com>
4175 * builtins.c (unterminated_array): Handle ARRAY_REF.
4176 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
4177 * builtins.h (unterminated_array): Declare extern.
4178 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
4180 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
4183 2018-09-14 Martin Sebor <msebor@redhat.com>
4184 Jeff Law <law@redhat.com>
4186 * builtins.c (unterminated_array): New.
4187 (expand_builtin_strcpy): Adjust.
4188 (expand_builtin_strcpy_args): Detect unterminated arrays.
4189 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
4190 unterminated arrays.
4191 * gimple-fold.h (get_maxval_strlen): Add argument.
4192 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
4194 * gimple-fold.c (get_range_strlen): Add argument.
4195 (get_maxval_strlen): Adjust.
4196 * gimple-fold.h (get_range_strlen): Add argument.
4198 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
4200 * config/i386/movdirintrin.h: Fix copyright year.
4202 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
4204 * reg-stack.c: Include regs.h.
4205 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
4206 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
4207 FIRST_STACK_REG, not DFmode.
4208 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
4209 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
4210 (change stack): Default register mode to the reg_raw_mode of
4211 FIRST_STACK_REG, not DFmode.
4212 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
4213 (*swapxf): Rename from swapxf.
4215 2018-09-14 Carl Love <cel@us.ibm.com>
4217 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
4218 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
4220 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
4223 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
4226 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
4229 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
4231 2018-09-14 Jason Merrill <jason@redhat.com>
4233 Fix --enable-gather-detailed-mem-stats.
4234 * hash-table.c (hash_table_usage): Change from variable to function.
4235 * hash-table.h: Adjust.
4236 * Makefile.in: Add missing dependencies on hash-table.h.
4238 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4240 PR tree-optimization/87259
4242 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
4243 execute_cse_reciprocals_1 has tried transforming.
4245 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
4247 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
4248 VR_VARYING for PLUS/MINUS_EXPR.
4250 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
4252 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
4255 2018-09-14 Richard Biener <rguenther@suse.de>
4258 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
4259 bits for the merged partition.
4261 2018-09-13 Martin Sebor <msebor@redhat.com>
4262 Bernd Edlinger <bernd.edlinger@hotmail.de>
4264 * builtins.h (c_srlen): Add argument.
4265 * builtins.c (warn_string_no_nul): New function.
4266 (c_strlen): Add argument and use it. Update recursive calls.
4267 Pass DECL argument to string_constant to get info on non
4268 terminated strings. Update *NONSTR as needed.
4269 (fold_builtin_strlen): Add argument to calls to c_strlen.
4270 Warn for unterminated arrays.
4271 (warn_string_no_null): Add prototype.
4272 * expr.c (string_constant): Update arguments. Update recursive
4273 calls appropriately. Detect missing NUL terminator and outermost
4274 declaration its missing in.
4275 Improve checks for arrays with nonzero lower bound or elements
4276 that are not a single byte. Simplify offset computation.
4277 Simplify checks for non-NUL terminated strings.
4278 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
4279 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
4281 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
4283 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
4285 * fold-const.c (c_getstr): Fix function comment. Remove unused third
4286 argument. Fix range checks.
4287 * fold-const.h (c_getstr): Adjust protoype.
4288 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
4289 string is constant but contains no NUL byte.
4291 * expr.c (string_constant): Adjust function comment.
4292 Remove bogus check for zero termination.
4294 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
4296 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
4297 (get_constant_size): Don't make STRING_CSTs larger than they are.
4298 (check_string_literal): New check function for STRING_CSTs.
4299 (output_constant): Use it.
4301 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
4304 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
4306 2018-09-13 Richard Biener <rguenther@suse.de>
4308 PR tree-optimization/87263
4309 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
4310 (struct unwind_state): Add max_rpo field.
4311 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
4312 Compute max_rpo, the max RPO number a block can be backwards reached
4313 from. Re-write non-iterating mode to a RPO ordered worklist approach,
4314 separating it from the iterating mode.
4316 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
4318 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
4319 (rfs_decision): New scheduling decision.
4321 2018-09-13 Richard Biener <rguenther@suse.de>
4324 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
4325 (vn_nary_op_insert_pieces_predicated): Do not write useless
4326 valid_dominated_by_p entry outside of the allocated storage.
4328 2018-09-13 Omar Sandoval <osandov@osandov.com>
4329 Tom de Vries <tdevries@suse.de>
4332 * dwarf2out.c (is_c): New function.
4333 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
4335 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
4338 * config/aarch64/aarch64.md (*aarch64_bfxil):
4340 * config/aarch64/constraints.md (Ulc): Define.
4341 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
4343 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
4346 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
4348 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
4349 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
4350 aarch64_layout_frame call.
4351 (aarch64_expand_epilogue): Likewise.
4352 (aarch64_initial_elimination_offset): Likewise.
4353 (aarch64_get_separate_components): Likewise.
4354 (aarch64_use_return_insn_p): Likewise.
4355 (aarch64_layout_frame): Remove unneeded check.
4357 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4359 * configure.ac: Only append
4360 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
4361 gcc_config_arguments if it was never reconfigured or last reconfigure
4362 was with different arguments.
4363 * configure: Regenerated.
4365 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4366 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4369 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
4370 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
4372 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4374 PR tree-optimization/87287
4375 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
4376 X % C == 0 to X % (unsigned) C == 0 optimization to ...
4377 * match.pd (X % C == 0): ... here. New optimization.
4379 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4382 * expr.h (maybe_optimize_mod_cmp): Declare.
4383 * expr.c (mod_inv): New function.
4384 (maybe_optimize_mod_cmp): New function.
4385 (do_store_flag): Use it.
4386 * cfgexpand.c (expand_gimple_cond): Likewise.
4388 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
4389 Julian Brown <julian@codesourcery.com>
4392 * gimplify.c (gimplify_scan_omp_clauses): Set
4393 target_firstprivatize_array_bases in OpenACC parallel and kernels
4394 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
4395 OpenACC data regions.
4397 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
4399 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
4400 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
4402 2018-09-12 Richard Biener <rguenther@suse.de>
4404 PR tree-optimization/87280
4405 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
4406 edge but unreachable target.
4407 (do_rpo_vn): For conservatively handling a PHI only mark
4408 the backedge executable but not the block reachable.
4410 2018-09-12 Richard Biener <rguenther@suse.de>
4412 PR tree-optimization/87266
4413 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
4416 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
4418 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
4420 ("trunc<BFP:mode><DFP_ALL:mode>2")
4421 ("trunc<DFP_ALL:mode><BFP:mode>2")
4422 ("extend<BFP:mode><DFP_ALL:mode>2")
4423 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
4424 according to the target operand type.
4426 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4427 Andreas Krebbel <krebbel@linux.ibm.com>
4429 PR tree-optimization/86844
4430 * gimple-ssa-store-merging.c
4431 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
4432 there are any overlapping stores in between them, make sure they are
4433 also coalesced or we give up completely.
4435 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4438 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
4439 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
4441 2018-09-12 Tom de Vries <tdevries@suse.de>
4443 * common.opt (gdescribe-dies): Add option.
4444 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
4445 attribute for artifical and nameless decls.
4446 (dwarf2out_register_external_die): Add description attribute to
4447 external reference die.
4448 (add_desc_attribute): New functions.
4449 (gen_subprogram_die): Add description attribute to
4450 DW_TAG_call_site_parameter.
4451 * tree-pretty-print.c (print_generic_expr_to_str): New function.
4452 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
4453 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
4455 (@item -gdescribe-dies): Add.
4457 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
4459 * tree-vrp.c (vrp_shift_undefined_p): Remove.
4460 (extract_range_from_binary_expr_1: Call
4461 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
4462 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
4465 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
4467 * gimple-ssa-warn-alloca.c
4468 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
4469 field for ALLOCA_BOUND_*_LARGE.
4471 2018-09-11 Nathan Sidwell <nathan@acm.org>
4473 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
4475 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4477 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
4478 for clobbers. Remove obsolete comment.
4480 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4482 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
4483 mpxchk, mpxld and mpxst types.
4484 (define_attr length_immediate): Remove all processing of mpx types.
4485 (define_attr prefix_0f): Ditto.
4486 (define_attr memory): Ditto.
4488 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4490 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
4491 (log<mode>2): Change operand 1 predicate to general_operand.
4492 Extend operand 1 to XFmode and generate logxf3 insn.
4493 (log10<mode>2): Change operand 1 predicate to general_operand.
4494 Extend operand 1 to XFmode and generate log10xf3 insn.
4495 (log2<mode>2): Change operand 1 predicate to general_operand.
4496 Extend operand 1 to XFmode and generate log2xf3 insn.
4497 (fyl2xp1_extend<mode>xf3_i387): Remove.
4498 (log1p<mode>2): Change operand 1 predicate to general_operand.
4499 Extend operand 1 to XFmode and generate log1pxf3 insn.
4500 (fxtract_extend<mode>xf3_i387): Remove.
4501 (logb<mode>2): Change operand 1 predicate to general_operand.
4502 Extend operand 1 to XFmode and generate logbxf3 insn.
4503 (ilogb<mode>2): Change operand 1 predicate to general_operand.
4504 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
4505 (significand<mode>2): Change operand 1 predicate to general_operand.
4506 Extend operand 1 to XFmode and generate significandxf3 insn.
4508 2018-09-11 Nathan Sidwell <nathan@acm.org>
4510 * gcc.c (perror_with_name, pfatal_with_name): Delete.
4511 (load_specs): Use fatal_error.
4512 (DELETE_IF_ORDINARY, process_command): Use error.
4513 (execute, run_attempt): Use fatal_error.
4515 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
4517 * diagnostic-core.h (sorry_at): New prototype.
4518 * diagnostic.c (sorry_at): New function.
4520 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
4522 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
4523 by zero as VR_UNDEFINED.
4525 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
4527 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
4528 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
4529 (<sincos>mode2): New expander.
4530 (sincos_extend<mode>xf3_i387): Remove insn pattern.
4531 (sincos -> sin, cos splitters): Remove splitter patterns.
4532 (sincos<mode>3): Change operand 2 predicate to general_operand.
4533 Extend operand 2 to XFmode and generate sincosxf3 insn.
4534 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
4535 Change operand 3 predicate to const1_operand.
4536 (fptan_extend<mode>xf4_i387): Remove insn pattern.
4537 (tanxf2): Update operands in the call to fptanxf4_i387.
4538 (tan<mode>2): Change operand 1 predicate to general_operand.
4539 Extend operand 1 to XFmode and generate tanxf3 insn.
4540 (atan2xf3): Rename from *fpatanxf3_i387.
4541 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
4542 (atan2xf3): Remove expander.
4543 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
4544 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
4545 (atan<mode>2): Change operand 1 predicate to general_operand.
4546 Extend operand 1 to XFmode and generate atanxf3 insn.
4548 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
4550 * config/i386/i386.md (x87/SSE constant load splitter): Use
4551 memory_operand instead of nonimmediate_operand for input operand
4554 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
4556 * config/i386/i386.md (float partial SSE register stall splitter): Move
4557 splitter near its instruction pattern.
4558 (float_extend partial SSE register stall splitter): Ditto.
4559 (float_truncate partial SSE register stall splitter): Ditto.
4561 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
4564 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
4565 to speculation_safe_value_not_needed.
4568 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
4569 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
4570 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
4571 leaf_function_p, instead use has_hard_reg_initial_val.
4573 2018-09-09 Nathan Sidwell <nathan@acm.org>
4575 * gcc.h (pfatal_with_name): Don't declare here.
4576 * gcc.c (pfatal_with_name): Make static.
4578 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
4580 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
4583 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
4585 PR rtl-optimization/85458
4586 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
4587 priority hook to reduce the priority of EXPR.
4589 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
4591 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
4592 DImode for x87 on 32bit targets. Conditionally disable x87 modes
4593 with X87_ENABLE_FLOAT. Remove preparation code.
4594 (*float<SWI48:mode><MODEF:mode>2): Rename from
4595 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
4596 math using "enabled" attribute.
4597 (*floatdi<MODEF:mode>2_i387): Rename from
4598 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
4599 enable for 32bit targets only.
4600 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
4602 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
4603 as operand 1 predicate. Rewrite as define_insn_and_split.
4604 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
4606 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
4608 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
4609 to fallthru to FLOAT case.
4611 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
4614 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
4615 around folding of vec_sl to handle out of range shift values.
4617 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
4619 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
4620 Update callers to gen_fix_trunc<mode>_i387_fisttp
4621 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
4622 nonimmediate_operand.
4623 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
4624 and corresponding splitters.
4625 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
4626 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
4627 (fix_truncdi_i387_with_temp): Remove insn pattern
4628 and corresponding splitters.
4629 (fix_trunc<mode>_i387): Change operand 0 predicate to
4630 nonimmediate_operand.
4631 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
4632 and corresponding splitters.
4633 (*fistdi2_1): Remove.
4635 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
4636 (lrintxfdi2): Remove expander. Reimplement as define_insn.
4637 (*fist<mode>2_1): Remove.
4638 (fist<mode>2): Ditto.
4639 (fist<mode>2_with_temp): Remove insn pattern and corresponding
4641 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
4642 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
4643 (fistdi2_<rounding>): Change operand 0 predicate to
4644 nonimmediate_operand.
4645 (fistdi2_<rounding>_with_temp): Remove insn pattern
4646 and corresponding splitters.
4647 (fist<mode>2_<rounding>): Change operand 0 predicate to
4648 nonimmediate_operand.
4649 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
4650 and corresponding splitters.
4652 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
4654 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
4656 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
4659 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
4661 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4662 early gimple folding of vec_splat().
4663 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
4664 * gimple-fold.h: Add an extern define for tree_vec_extract().
4666 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
4668 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
4669 wrappers around TREE_TYPE comparisons.
4671 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
4674 * config/s390/predicates.md: Add nonsym_memory_operand.
4675 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
4676 contains a SYMBOL_REF, load it into an intermediate pseudo.
4677 (s390_emit_compare_and_swap): Legitimize operand.
4678 * config/s390/s390.md: Use the new nonsym_memory_operand
4679 with UNSPECV_CAS patterns.
4681 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
4684 * config/s390/s390-passes.def: New file.
4685 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
4687 (make_pass_s390_early_mach): Add declaration.
4688 * config/s390/s390.c (make_pass_s390_early_mach):
4689 (s390_option_override): Remove dynamic registration.
4690 * config/s390/t-s390: Add s390-passes.def.
4692 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
4694 * config/s390/s390.c (s390_decompose_constant_pool_ref):
4695 Remove UNSPEC_LTREL_BASE check.
4696 (annotate_constant_pool_refs): Likewise.
4697 (find_constant_pool_ref): Likewise.
4698 (find_ltrel_base): Removed.
4699 (replace_ltrel_base): Removed.
4700 (s390_mainpool_finish): Remove replace_ltrel_base call.
4701 (s390_chunkify_start): Remove pending LTREL_BASE logic.
4702 (s390_chunkify_finish): Remove replace_ltrel_base call.
4703 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
4705 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
4708 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
4709 to speculation_safe_value_not_needed.
4711 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
4712 Bernd Schmidt <bernds_cb1@t-online.de>
4714 * config/nvptx/nvptx-opts.h: New file.
4715 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
4716 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
4718 (TARGET_SM35): New macro.
4719 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
4721 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
4723 (misa=): New option.
4724 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
4726 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
4728 * config/i386/i386.md (truncdfsf2): Remove expander.
4729 (truncdfsf2_with_temp): Ditto.
4730 (truncxf<mode>2): Ditto.
4731 (*truncdfsf_fast_mixed): Remove insn pattern.
4732 (*truncdfsf_fast_i387): Ditto.
4733 (*truncdfsf_mixed): Ditto.
4734 (*truncdfsf_i387): Ditto.
4735 (*truncdfsf2_i387_1): Ditto.
4736 (*truncxfsf2_mixed): Ditto.
4737 (*truncxfdf2_mixed): Ditto.
4738 (*truncxf<mode>2_i387_noop): Ditto. Update callers
4739 to call gen_truncxf<mode>2 instead.
4740 (*truncxf<mode>2_i387): Remove.
4741 (reg->reg splitters): Remove splitter pattern.
4742 (reg->mem splitters): Ditto.
4744 (truncdfsf2): New insn pattern.
4745 (truncxf<mode>2): Ditto.
4747 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4749 * tree-ssa-math-opts.c (is_mult_by): New function.
4750 (is_square_of): Use the above.
4751 (optimize_recip_sqrt): New function.
4752 (pass_cse_reciprocals::execute): Use the above.
4754 2018-09-05 Richard Biener <rguenther@suse.de>
4757 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
4758 to zero-init the emplaced vec.
4760 2018-09-05 Martin Liska <mliska@suse.cz>
4762 PR tree-optimization/87205
4763 * tree-switch-conversion.c (pass_lower_switch::execute):
4764 Group cases for switch statements.
4766 2018-09-05 Richard Biener <rguenther@suse.de>
4768 PR tree-optimization/87217
4769 * tree-ssa-sccvn.c (vuse_valueize): New.
4770 (vn_reference_lookup_pieces): Use it.
4771 (vn_reference_lookup): Likewise.
4773 2018-09-05 Nathan Sidwell <nathan@acm.org>
4776 * stor-layout.c (place_field): Scan forwards to check last
4777 bitfield when ms_bitfield_placement is in effect.
4779 2018-09-05 Richard Biener <rguenther@suse.de>
4782 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
4785 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
4786 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
4788 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
4789 * config.gcc (extra_objs): Build it.
4790 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
4792 * config/aarch64/aarch64-passes.def
4793 (pass_tag_collision_avoidance): New pass.
4794 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
4795 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
4796 (aarch64_classify_address): Remove static qualifier.
4797 (aarch64_address_info, aarch64_address_type): Move to...
4798 * config/aarch64/aarch64-protos.h: ... here.
4799 (make_pass_tag_collision_avoidance): New function.
4800 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
4803 2018-09-05 Martin Liska <mliska@suse.cz>
4805 * doc/gcov.texi: Update documentation of humar
4807 * gcov.c (format_count): Print one decimal place, it provides
4808 more fine number of situations like '1G' vs. '1.4G'.
4810 2018-09-05 Martin Liska <mliska@suse.cz>
4813 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
4814 * optc-gen.awk: Allow 'Var' for Deprecated options in order
4815 to generate a MASK value.
4817 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
4820 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
4821 if frame pointer isn't used.
4822 (compute_frame_pointer_to_fb_displacement): Likewise.
4824 2018-09-04 Jakub Jelinek <jakub@redhat.com>
4827 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
4828 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
4829 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
4830 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
4831 and OPTION_MASK_ISA_XSAVEC_UNSET.
4833 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
4835 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
4836 XOR operations in NAND case.
4838 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
4840 * wide-int-range.cc (wide_int_range_convert): New.
4841 * wide-int-range.h (wide_int_range_convert): New.
4842 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
4843 code into wide_int_range_convert.
4844 (extract_range_into_wide_ints): Do not munge anti range constants
4845 into the entire domain. Just return the range back.
4847 2018-09-04 Martin Liska <mliska@suse.cz>
4849 * genmatch.c (output_line_directive): Add new argument
4851 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
4853 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
4855 * doc/invoke.texi (Option Summary): Add whitespace.
4857 * doc/invoke.texi (Option Summary): Add -Waligned-new.
4859 2018-09-04 Richard Biener <rguenther@suse.de>
4861 PR tree-optimization/87211
4862 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
4863 backedge value we're supposed to treat as VARYING also number
4864 the PHI to VARYING in case it got a different value-number already.
4866 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
4868 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
4869 (extract_range_from_binary_expr_1): Do not call
4870 vrp_can_optimize_bit_op.
4871 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
4873 (wide_int_range_get_mask_and_bounds): New.
4874 (wide_int_range_optimize_bit_op): New.
4875 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
4876 (wide_int_range_bit_and): Same.
4877 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
4878 (wide_int_range_optimize_bit_op): New.
4879 (wide_int_range_get_mask_and_bounds): New.
4881 2018-09-04 Richard Biener <rguenther@suse.de>
4883 PR tree-optimization/87176
4884 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
4885 variable. When value-numbering a virtual PHI node make sure
4886 to not value-number to the backedge value.
4888 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
4890 * doc/extend.texi (Long Long, Hex Floats): Document support for
4891 long long and hex floats in more recent versions of ISO C++.
4893 2018-09-03 Richard Biener <rguenther@suse.de>
4895 PR tree-optimization/87177
4896 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
4899 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4901 * bb-reorder.c (edge_order): Convert to C-qsort-style
4902 tri-state comparator.
4903 (reorder_basic_blocks_simple): Change std::stable_sort to
4906 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4908 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
4909 tri-state comparator.
4910 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
4912 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4914 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
4915 (mergesort): ... here as maximum count for using netsort.
4916 (gcc_qsort): Set nlim to 3 if stable sort is requested.
4917 (gcc_stablesort): New.
4918 * system.h (gcc_stablesort): Declare.
4920 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
4922 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
4923 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
4924 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
4926 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
4928 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
4929 lxsdx and stxsdx alternatives.
4930 (*mov<mode>_hardfloat64): Ditto.
4931 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
4933 2018-09-03 Richard Biener <rguenther@suse.de>
4935 PR tree-optimization/87200
4936 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
4939 2018-09-03 Martin Liska <mliska@suse.cz>
4941 PR tree-optimization/87201
4942 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
4943 Fix parenthesis in an expression.
4945 2018-09-03 Richard Biener <rguenther@suse.de>
4947 PR tree-optimization/87197
4948 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
4949 visited. CSE the VN_INFO hashtable lookup.
4951 PR tree-optimization/87169
4952 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
4953 iterating make sure there's no extra backedges from irreducible
4954 regions feeding the header. Mark the destination block
4957 2018-09-03 Martin Liska <mliska@suse.cz>
4960 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
4961 * common/common-targhooks.c (default_get_valid_option_values):
4963 * common/common-targhooks.h (default_get_valid_option_values):
4965 * common/config/i386/i386-common.c: Move processor_target_table
4967 (ix86_get_valid_option_values): New function.
4968 (TARGET_GET_VALID_OPTION_VALUES): New macro.
4969 * config/i386/i386.c (struct ptt): Move to i386-common.c.
4970 (PTA_*): Move all defined masks into i386-common.c.
4971 (ix86_function_specific_restore): Use new processor_cost_table.
4972 * config/i386/i386.h (struct ptt): Moved from i386.c.
4973 (struct pta): Likewise.
4974 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
4975 * doc/tm.texi.in: Likewise.
4976 * opt-suggestions.c (option_proposer::suggest_option):
4977 Pass prefix to build_option_suggestions.
4978 (option_proposer::get_completions): Likewise.
4979 (option_proposer::build_option_suggestions): Use the new target
4981 * opts.c (struct option_help_tuple): New struct.
4982 (print_filtered_help): Use the new target hook.
4984 2018-09-03 Martin Liska <mliska@suse.cz>
4987 * predict.c (set_even_probabilities): Add likely_edges
4988 argument and handle cases where we have precisely one
4990 (combine_predictions_for_bb): Catch also likely_edges.
4991 (tree_predict_by_opcode): Handle gswitch statements.
4992 * tree-cfg.h (find_case_label_for_value): New declaration.
4993 (find_taken_edge_switch_expr): Likewise.
4994 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
4995 Find pivot in decision tree based on probabily, not by number of
4998 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
5000 * doc/standards.texi (Standards): Update Objective-C reference.
5002 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
5004 * doc/install.texi (Prerequisites): Update link for MPC.
5006 2018-09-01 Michael Matz <matz@suse.de>
5008 PR tree-optimization/87074
5009 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
5010 PHIs for outer-loop uses.
5012 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
5014 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
5015 * doc/invoke.texi (C Dialect Options): Ditto.
5017 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
5019 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
5021 2018-08-31 Richard Biener <rguenther@suse.de>
5023 PR tree-optimization/87168
5024 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
5025 (rpo_elim::eliminate_avail): When OP was not visited it must
5028 2018-08-31 David Malcolm <dmalcolm@redhat.com>
5030 * tree-vrp.c (copy_value_range): Convert param "from" from
5031 "value_range *" to "const value_range *".
5032 (range_is_null): Likewise for param "vr".
5033 (range_int_cst_p): Likewise.
5034 (range_int_cst_singleton_p): Likewise.
5035 (symbolic_range_p): Likewise.
5036 (value_ranges_intersect_p): Likewise for both params.
5037 (value_range_nonnegative_p): Likewise for param "vr".
5038 (value_range_constant_singleton): Likewise.
5039 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
5040 (extract_range_into_wide_ints): Likewise for param "vr".
5041 (extract_range_from_multiplicative_op): Likewise for params "vr0"
5043 (vrp_can_optimize_bit_op): Likewise.
5044 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
5046 (extract_range_from_unary_expr): Likewise.
5047 (debug_value_range): Likewise for param "vr".
5048 (value_range::dump): Add "const" qualifier.
5049 (vrp_prop::check_array_ref): Convert local "vr" from
5050 "value_range *" to "const value_range *".
5051 (vrp_prop::check_mem_ref): Likewise.
5052 (vrp_prop::visit_stmt): Likewise for local "old_vr".
5053 (vrp_intersect_ranges_1): Likewise for param "vr_1".
5054 (vrp_intersect_ranges): Likewise.
5055 (simplify_stmt_for_jump_threading): Likewise for local "vr".
5056 (vrp_prop::vrp_finalize): Likewise.
5057 * tree-vrp.h (value_range::dump): Add "const" qualifier.
5058 (vrp_intersect_ranges): Add "const" qualifier to params as above.
5059 (extract_range_from_unary_expr): Likewise.
5060 (value_range_constant_singleton): Likewise.
5061 (symbolic_range_p): Likewise.
5062 (copy_value_range): Likewise.
5063 (extract_range_from_binary_expr_1): Likewise.
5064 (range_int_cst_p): Likewise.
5065 (vrp_set_zero_nonzero_bits): Likewise.
5066 (range_int_cst_singleton_p): Likewise.
5068 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
5070 * config/aarch64/arm_neon.h (vabsd_s64): New.
5071 (vnegd_s64): Likewise.
5073 2018-08-31 Martin Jambor <mjambor@suse.cz>
5075 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
5077 2018-08-31 Martin Liska <mliska@suse.cz>
5079 * ipa-icf.c (sem_item::add_type): Use
5080 sem_item::m_type_hash_cache.
5081 * ipa-icf.h: Move the cache from sem_item_optimizer
5084 2018-08-31 Nathan Sidwell <nathan@acm.org>
5086 * doc/extend.texi (Backwards Compatibility): Remove implicit
5087 extern C leeway of () being (...).
5089 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5091 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
5093 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
5097 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
5099 2018-08-31 Jakub Jelinek <jakub@redhat.com>
5102 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
5103 gen_int_mode. Formatting fixes.
5105 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
5107 * target.def (custom_function_descriptors): Improve documentation.
5108 * doc/tm.texi.in (Trampolines): Expand discussion of function
5109 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
5110 beginning of the section.
5111 * doc/tm.texi: Regenerated.
5113 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
5115 * cfg.h (class auto_edge_flag): Spell out the template-id of the
5116 base class in the initializer list. This is a workaround for
5117 building with older GCC.
5118 (class auto_bb_flag): Likewise.
5120 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
5122 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
5123 (altivec_vcmpequ<VI_char>_p): Remove star.
5124 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
5126 (expand_strncmp_vec_sequence): New function.
5127 (emit_final_str_compare_vec): New function.
5128 (expand_strn_compare): Add support for vector strncmp.
5129 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
5130 length specification to bytes.
5131 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
5132 (vcmpnezb_p): New pattern.
5133 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
5134 for option -mstring-compare-inline-limit.
5136 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
5138 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
5139 (PTA_SKYLAKE): Add PTA_AES.
5140 (PTA_GOLDMONT): Likewise.
5142 2018-08-29 Jan Hubicka <jh@suse.cz>
5145 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
5146 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
5148 2018-08-29 Jan Hubicka <jh@suse.cz>
5150 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
5152 (hash_tree): Do not visit TYPE_STUB_DECL.
5153 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
5154 stream TYPE_STUB_DECL.
5155 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
5156 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
5157 after free_lang_data.
5158 (type_in_anonymous_namespace_p): Likewise.
5160 2018-08-29 Jan Hubicka <jh@suse.cz>
5162 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
5163 comment that it has to be even number.
5164 (class sreal): Change m_sig type to int32_t.
5165 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
5166 int64_t for temporary calculations.
5167 (sreal_verify_basics): Drop one bit from minimum and maximum.
5169 2018-08-30 Richard Biener <rguenther@suse.de>
5171 PR tree-optimization/87147
5172 * tree-ssa-sccvn.c (SSA_VISITED): New function.
5173 (visit_phi): When the degenerate result is from the backedge and
5174 we didn't visit its definition yet drop to VARYING.
5175 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
5177 2018-08-29 Jan Hubicka <jh@suse.cz>
5179 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
5181 (hash_tree): Likewise.
5183 2018-08-29 Jan Hubicka <jh@suse.cz>
5185 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
5186 and TYPE_NEXT_REF_TO.
5188 2018-08-29 Jan Hubicka <jh@suse.cz>
5190 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
5191 comment that it has to be even number.
5192 (class sreal): Change m_sig type to int32_t.
5193 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
5194 int64_t for temporary calculations.
5195 (sreal_verify_basics): Drop one bit from minimum and maximum.
5197 2018-08-30 Tamar Christina <tamar.christina@arm.com>
5199 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
5201 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
5204 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
5205 if to_add is negative.
5207 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
5210 * builtins.c (c_strlen): Improve range checks.
5212 2018-08-29 Martin Sebor <msebor@redhat.com>
5213 Jeff Law <law@redhat.com>
5215 PR tree-optimization/86714
5216 PR tree-optimization/86711
5217 * builtins.c (c_strlen): Add arguments to call to string_constant.
5218 * expr.c (string_constant): Add argument. Detect missing nul
5219 terminator and outermost declaration it's missing in.
5220 * expr.h (string_constant): Add argument.
5221 * fold-const.c (read_from_constant_string): Add arguments to call to
5223 (c_getstr): Likewise.
5224 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
5226 * tree-ssa-strlen.c (get_stridx): Likewise.
5228 2018-08-29 Jan Hubicka <jh@suse.cz>
5230 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
5231 Do not stream DECL_VINDEX.
5232 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
5233 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
5234 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
5236 2018-08-29 Richard Biener <rguenther@suse.de>
5238 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
5239 virtual operands that are not default defs to honor region
5241 (rpo_vn_valueize): Remove ineffective code here.
5243 2018-08-29 Richard Biener <rguenther@suse.de>
5245 PR tree-optimization/87132
5246 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
5247 when skipping defs reachable over backedges.
5249 2018-08-29 Richard Biener <rguenther@suse.de>
5251 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
5252 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
5253 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
5254 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
5255 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
5256 (vn_reference_lookup_3): Remove use of const_parms.
5257 (free_rpo_vn): Do not free const_parms.
5258 (do_rpo_vn): Do not call init_const_parms.
5259 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
5260 SSA_NAME_POINTS_TO_READONLY_MEMORY.
5261 (call_may_clobber_ref_p_1): Likewise.
5263 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
5266 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
5268 (-ftree-scev-cprop): Document.
5270 2018-08-29 Jan Hubicka <jh@suse.cz>
5272 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
5274 (sreal constructor): Update.
5275 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
5276 sreal:operator/): Update.
5278 2018-08-29 Martin Liska <mliska@suse.cz>
5280 * tree-switch-conversion.c (switch_conversion::expand):
5281 Strenghten assumption about gswitch statements.
5283 2018-08-29 Richard Biener <rguenther@suse.de>
5285 PR tree-optimization/87117
5286 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
5287 re-value-number released SSA VDEFs.
5289 2018-08-29 Richard Biener <rguenther@suse.de>
5291 PR tree-optimization/87126
5292 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
5294 2018-08-28 Jim Wilson <jimw@sifive.com>
5296 * config/riscv/pic.md: Rewrite.
5297 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
5299 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
5300 (SOFTF, default_load, softload, softstore): New.
5302 2018-08-28 Jeff Law <law@redhat.com>
5304 * fold-const.c (fold_binary_loc): Remove recently added assert.
5306 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
5308 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
5309 to OP parmeter of generated function.
5311 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
5313 PR tree-optimization/87009
5314 * match.pd: Add boolean optimizations.
5316 2018-08-28 Martin Sebor <msebor@redhat.com>
5319 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
5320 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
5321 (pass_walloca::gate): Use it.
5322 (alloca_call_type): Same.
5323 (pass_walloca::execute): Same.
5324 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
5326 2018-08-28 David Malcolm <dmalcolm@redhat.com>
5328 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
5329 GCC_VERSION for usage of "__gcc_dump_printf__" format from
5332 2018-08-28 Richard Biener <rguenther@suse.de>
5334 PR tree-optimization/87124
5335 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
5336 constants before looking up avail.
5338 2018-08-28 Jakub Jelinek <jakub@redhat.com>
5341 * calls.c (maybe_warn_nonstring_arg): Punt early if
5342 warn_stringop_overflow is zero. Don't call get_range_strlen
5343 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
5344 Swap comparison operands to have constants on rhs. Only use
5345 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
5346 increment lenrng[0].
5348 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
5350 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
5351 use of tree_to_shwi. Remove duplicated test for the size being
5352 a whole number of bytes.
5354 2018-08-28 Richard Biener <rguenther@suse.de>
5356 PR tree-optimization/87117
5357 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
5358 Handle removed stmt without LHS (GIMPLE_NOP).
5360 2018-08-28 Richard Biener <rguenther@suse.de>
5362 PR tree-optimization/87117
5363 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
5364 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
5366 2018-08-28 Richard Biener <rguenther@suse.de>
5368 PR tree-optimization/87117
5369 * tree-ssa-pre.c (compute_avail): Do not make expressions
5370 with predicated values available.
5371 (get_expr_value_id): Assert we do not run into predicated value
5374 2018-08-28 Richard Biener <rguenther@suse.de>
5376 PR tree-optimization/87117
5377 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
5378 get virtual operands.
5379 (get_expr_operands): Handle STRING_CST like other decls.
5381 2018-08-28 Martin Liska <mliska@suse.cz>
5383 * tree.h: Update documentation of fndecl_built_in_p
5387 2018-08-27 Jeff Law <law@redhat.com>
5388 PR tree-optimization/87110
5389 * tree-ssa-dse.c (compute_trims): Handle non-constant
5392 2018-08-27 Martin Sebor <msebor@redhat.com>
5394 PR tree-optimization/86914
5395 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
5397 2018-08-27 Martin Sebor <msebor@redhat.com>
5399 PR tree-optimization/87112
5400 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
5401 the type of the bound argument.
5403 2018-08-27 Jeff Law <law@redhat.com>
5405 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
5406 type does not have a TYPE_SIZE_UNIT.
5408 2018-08-27 Steve Ellcey <sellcey@cavium.com>
5410 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
5411 with include of backend.h.
5413 2018-08-27 Richard Biener <rguenther@suse.de>
5415 PR tree-optimization/86927
5416 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
5417 use const cond reduction code.
5419 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
5421 PR tree-optimization/85758
5422 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
5424 2018-08-27 David Malcolm <dmalcolm@redhat.com>
5427 * diagnostic-show-locus.c (class layout_range): Update for
5428 conversion of show_caret_p to a tri-state.
5429 (layout_range::layout_range): Likewise.
5430 (make_range): Likewise.
5431 (layout::maybe_add_location_range): Likewise.
5432 (layout::should_print_annotation_line_p): Don't show annotation
5433 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
5434 (layout::get_state_at_point): Update for conversion of
5435 show_caret_p to a tri-state. Bail out early for
5436 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
5437 underlining or source colorization.
5438 (gcc_rich_location::add_location_if_nearby): Update for conversion
5439 of show_caret_p to a tri-state.
5440 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
5441 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
5443 (selftest::test_one_liner_labels): Likewise.
5444 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
5445 conversion of show_caret_p to a tri-state.
5446 * pretty-print.c (text_info::set_location): Likewise.
5447 * pretty-print.h (text_info::set_location): Likewise.
5448 * substring-locations.c (format_warning_n_va): Likewise.
5449 * tree-diagnostic.c (default_tree_printer): Likewise.
5450 * tree-pretty-print.c (newline_and_indent): Likewise.
5452 2018-08-27 David Malcolm <dmalcolm@redhat.com>
5455 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
5456 line above for line-insertion fix-it hints.
5457 (selftest::test_fixit_insert_containing_newline): Update the
5458 expected results, and add a test with line-numbering enabled.
5460 2018-08-27 Martin Liska <mliska@suse.cz>
5463 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
5464 params with DECL_HAS_VALUE_EXPR_P.
5466 2018-08-27 Martin Liska <mliska@suse.cz>
5468 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
5469 selected expansion strategy.
5471 2018-08-27 Martin Liska <mliska@suse.cz>
5473 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
5474 * builtins.c (is_builtin_fn): Likewise.
5475 * attribs.c (diag_attr_exclusions): Use new function
5476 fndecl_built_in_p and remove check for FUNCTION_DECL if
5478 (builtin_mathfn_code): Likewise.
5479 (fold_builtin_expect): Likewise.
5480 (fold_call_expr): Likewise.
5481 (fold_builtin_call_array): Likewise.
5482 (fold_call_stmt): Likewise.
5483 (set_builtin_user_assembler_name): Likewise.
5484 (is_simple_builtin): Likewise.
5485 * calls.c (gimple_alloca_call_p): Likewise.
5486 (maybe_warn_nonstring_arg): Likewise.
5487 * cfgexpand.c (expand_call_stmt): Likewise.
5488 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
5489 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
5490 (cgraph_node::verify_node): Likewise.
5491 * cgraphclones.c (build_function_decl_skip_args): Likewise.
5492 (cgraph_node::create_clone): Likewise.
5493 * config/arm/arm.c (arm_insert_attributes): Likewise.
5494 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
5495 * dse.c (scan_insn): Likewise.
5496 * expr.c (expand_expr_real_1): Likewise.
5497 * fold-const.c (operand_equal_p): Likewise.
5498 (fold_binary_loc): Likewise.
5499 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
5500 * gimple-low.c (lower_stmt): Likewise.
5501 * gimple-pretty-print.c (dump_gimple_call): Likewise.
5502 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
5504 * gimple.c (gimple_build_call_from_tree): Likewise.
5505 (gimple_call_builtin_p): Likewise.
5506 (gimple_call_combined_fn): Likewise.
5507 * gimplify.c (gimplify_call_expr): Likewise.
5508 (gimple_boolify): Likewise.
5509 (gimplify_modify_expr): Likewise.
5510 (gimplify_addr_expr): Likewise.
5511 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
5512 * ipa-cp.c (determine_versionability): Likewise.
5513 * ipa-fnsummary.c (compute_fn_summary): Likewise.
5514 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
5515 * ipa-split.c (visit_bb): Likewise.
5516 (split_function): Likewise.
5517 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
5518 * lto-cgraph.c (input_node): Likewise.
5519 * lto-streamer-out.c (write_symbol): Likewise.
5520 * omp-low.c (setjmp_or_longjmp_p): Likewise.
5521 (lower_omp_1): Likewise.
5522 * predict.c (strip_predict_hints): Likewise.
5523 * print-tree.c (print_node): Likewise.
5524 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
5525 * trans-mem.c (is_tm_irrevocable): Likewise.
5526 (is_tm_load): Likewise.
5527 (is_tm_simple_load): Likewise.
5528 (is_tm_store): Likewise.
5529 (is_tm_simple_store): Likewise.
5530 (is_tm_abort): Likewise.
5531 (tm_region_init_1): Likewise.
5532 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
5533 * tree-cfg.c (verify_gimple_call): Likewise.
5534 (move_stmt_r): Likewise.
5535 (stmt_can_terminate_bb_p): Likewise.
5536 * tree-eh.c (lower_eh_constructs_2): Likewise.
5537 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
5538 * tree-inline.c (remap_gimple_stmt): Likewise.
5539 (copy_bb): Likewise.
5540 (estimate_num_insns): Likewise.
5541 (fold_marked_statements): Likewise.
5542 * tree-sra.c (scan_function): Likewise.
5543 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
5544 (optimize_stack_restore): Likewise.
5545 (pass_fold_builtins::execute): Likewise.
5546 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5547 (mark_all_reaching_defs_necessary_1): Likewise.
5548 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
5549 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
5550 (pass_forwprop::execute): Likewise.
5551 * tree-ssa-loop-im.c (stmt_cost): Likewise.
5552 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5553 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
5554 * tree-ssa-strlen.c (get_string_length): Likewise.
5555 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
5556 (find_func_aliases_for_call): Likewise.
5557 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
5558 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
5559 * tree-tailcall.c (find_tail_calls): Likewise.
5560 * tree.c (need_assembler_name_p): Likewise.
5561 (free_lang_data_in_decl): Likewise.
5562 (get_call_combined_fn): Likewise.
5563 * ubsan.c (is_ubsan_builtin_p): Likewise.
5564 * varasm.c (incorporeal_function_p): Likewise.
5565 * tree.h (DECL_BUILT_IN): Remove and replace with
5567 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
5568 (fndecl_built_in_p): New.
5570 2018-08-27 Martin Liska <mliska@suse.cz>
5572 PR tree-optimization/86847
5573 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
5574 Dump also subtree probability.
5575 (switch_decision_tree::do_jump_if_equal): New function.
5576 (switch_decision_tree::emit_case_nodes): Handle special
5577 situations in balanced tree that can be emitted much simpler.
5578 Fix calculation of probabilities that happen in tree expansion.
5579 * tree-switch-conversion.h (struct cluster): Add
5581 (struct simple_cluster): Likewise.
5582 (struct case_tree_node): Add new function has_child.
5583 (do_jump_if_equal): New.
5585 2018-08-27 Martin Liska <mliska@suse.cz>
5587 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
5588 Add new argument to bit_test_cluster constructor.
5589 (bit_test_cluster::emit): Set bits really number of values
5591 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
5592 probability argument.
5593 * tree-switch-conversion.h (struct bit_test_cluster):
5594 Add m_handles_entire_switch.
5596 2018-08-27 Martin Liska <mliska@suse.cz>
5598 PR tree-optimization/86702
5599 * tree-switch-conversion.c (jump_table_cluster::emit):
5600 Make probabilities even for values in jump table
5601 according to number of cases handled.
5602 (switch_decision_tree::compute_cases_per_edge): Pass
5603 argument to reset_out_edges_aux function.
5604 (switch_decision_tree::analyze_switch_statement): Likewise.
5605 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
5608 2018-08-27 Martin Liska <mliska@suse.cz>
5610 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
5611 cfun argument explicitly.
5612 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
5613 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
5614 function gimple_switch_default_bb.
5615 (convert_switch_statements):
5617 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
5618 * stmt.c (label_to_block_fn): Use label_to_block and pass
5619 cfun argument explicitly and use gimple_switch_label_bb.
5620 (expand_case): Likewise.
5621 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
5622 cfun argument explicitly. Likewise.
5623 (make_edges_bb): Likewise.
5624 (make_cond_expr_edges): Likewise.
5625 (get_cases_for_edge): Likewise.
5626 (make_gimple_switch_edges): Likewise.
5627 (label_to_block_fn): Likewise.
5628 (label_to_block): Likewise.
5629 (make_goto_expr_edges): Likewise.
5630 (make_gimple_asm_edges): Likewise.
5631 (main_block_label): Likewise.
5632 (group_case_labels_stmt): Likewise.
5633 (find_taken_edge_computed_goto): Likewise.
5634 (find_taken_edge_switch_expr): Likewise.
5635 (gimple_verify_flow_info): Likewise.
5636 (gimple_redirect_edge_and_branch): Likewise.
5637 (gimple_switch_label_bb): New function.
5638 (gimple_switch_default_bb): Likewise.
5639 (gimple_switch_edge): Likewise.
5640 (gimple_switch_default_edge): Likewise.
5641 * tree-cfg.h (label_to_block_fn): Remove and replace ...
5642 (label_to_block): ... with this.
5643 (gimple_switch_label_bb): New.
5644 (gimple_switch_default_bb): Likewise.
5645 (gimple_switch_edge): Likewise.
5646 (gimple_switch_default_edge): Likewise.
5647 * tree-cfgcleanup.c (convert_single_case_switch): Use
5648 new gimple functions and pass new argument to label_to_block.
5649 (cleanup_control_flow_bb):
5650 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
5651 cfun argument explicitly.
5652 (make_eh_edges): Likewise.
5653 (redirect_eh_dispatch_edge): Likewise.
5654 (lower_resx): Likewise.
5655 (lower_eh_dispatch): Likewise.
5656 (maybe_remove_unreachable_handlers): Likewise.
5657 (unsplit_eh): Likewise.
5658 (cleanup_empty_eh): Likewise.
5659 (verify_eh_edges): Likewise.
5660 (verify_eh_dispatch_edge): Likewise.
5661 * tree-ssa-dom.c (record_edge_info): Likewise.
5662 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
5663 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
5664 (thread_through_normal_block): Likewise.
5665 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
5666 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
5667 * tree-switch-conversion.c (switch_conversion::collect): Use new
5669 (switch_conversion::check_final_bb): Likewise.
5670 (switch_conversion::gather_default_values): Pass new argument
5672 (switch_conversion::build_constructors): Likewise.
5673 (switch_decision_tree::compute_cases_per_edge): Use new
5674 gimple_switch_edge function.
5675 (switch_decision_tree::analyze_switch_statement): Pass new argument
5677 (switch_decision_tree::try_switch_expansion): Use
5678 gimple_switch_default_edge.
5679 * tree-vrp.c (find_switch_asserts): Pass new argument
5681 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
5682 (vr_values::simplify_switch_using_ranges): Likewise.
5684 2018-08-27 Richard Biener <rguenther@suse.de>
5686 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
5687 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
5689 * tree-ssa-sccvn.h (struct vn_pval): New structure.
5690 (struct vn_nary_op_s): Add unwind_to member. Add
5691 predicated_values flag and put result into a union together
5692 with a linked list of vn_pval.
5693 (struct vn_ssa_aux): Add name member to make maintaining
5694 a map of SSA name to vn_ssa_aux possible. Remove no longer
5695 needed info, dfsnum, low, visited, on_sccstack, use_processed
5696 and range_info_anti_range_p members.
5697 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
5698 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
5700 (vn_valueize): New global.
5701 (vn_context_bb): Likewise.
5702 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
5703 VN_INFO_PTR_INFO): Remove.
5704 * tree-ssa-sccvn.c: ... (rewrite)
5705 (pass_fre::execute): For -O2+ initialize loops and run
5706 RPO VN in optimistic mode (iterating). For -O1 and -Og
5707 run RPO VN in non-optimistic mode.
5708 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
5709 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
5710 * doc/invoke.texi (sccvn-max-scc-size): Remove.
5711 (rpo-vn-max-loop-depth): Document.
5712 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
5713 when valuezing the VUSE signals we walked out of the region.
5714 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
5715 (phi_translate): Set VN context block to use for availability
5717 (compute_avail): Likewise.
5718 (pre_valueize): New function.
5719 (pass_pre::execute): Adjust to the RPO VN API.
5721 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
5722 (propagate_constants_for_unrolling): Remove.
5723 (tree_unroll_loops_completely): Perform value-numbering
5724 on the unrolled bodies loop parent.
5726 2018-08-27 Richard Biener <rguenther@suse.de>
5728 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
5729 for partial antic compute.
5731 2018-08-27 Jakub Jelinek <jakub@redhat.com>
5733 PR rtl-optimization/87065
5734 * combine.c (simplify_if_then_else): Formatting fix.
5735 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
5737 (known_cond): Don't return const_true_rtx for vector modes. Use
5738 CONST0_RTX instead of const0_rtx. Formatting fixes.
5740 2018-08-27 Martin Liska <mliska@suse.cz>
5742 PR gcov-profile/87069
5743 * gcov.c (process_file): Record files already processed
5744 and warn about a file being processed multiple times.
5746 2018-08-27 Martin Liska <mliska@suse.cz>
5749 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5750 Set default values for x_aarch64_*_string strings.
5751 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
5752 prefix. For -mabi do not print '=ABI' in help and use
5753 <option_value> format for -msve-vector-bits and -moverride
5756 2018-08-26 Jeff Law <law@redhat.com>
5758 * config/mips/frame-header-opt.c: Include "backend.h" rather than
5761 2018-08-26 Marek Polacek <polacek@redhat.com>
5763 PR c++/87029, Implement -Wredundant-move.
5764 * doc/invoke.texi: Document -Wredundant-move.
5766 2018-08-25 Martin Sebor <msebor@redhat.com>
5768 PR tree-optimization/87059
5769 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
5770 to the same type as the other.
5771 * fold-const.c (fold_binary_loc): Assert expectation.
5773 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
5775 * config/darwin.c (machopic_legitimize_pic_address): Clean up
5776 extraneous parentheses, dead code section and formatting.
5778 2018-08-24 David Malcolm <dmalcolm@redhat.com>
5781 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
5782 wide enough for jumps in the line-numbering to be visible.
5783 (layout::print_gap_in_line_numbering): New member function.
5784 (layout::calculate_line_spans): When using line numbering, merge
5785 line spans that are only 1 line apart.
5786 (diagnostic_show_locus): When printing line numbers, show gaps in
5787 line numbering directly, rather than printing headers.
5788 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
5789 line-numbering with multiple line spans.
5790 (selftest::test_fixit_insert_containing_newline_2): Add test of
5791 line-numbering, in which the spans are close enough to be merged.
5793 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
5795 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
5796 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
5797 * tree-vrp.c (range_is_nonnull): Remove.
5798 (range_includes_zero_p): Accept value_range instead of min/max.
5799 (extract_range_from_binary_expr_1): Do not early bail on
5801 Use range_includes_zero_p instead of range_is_nonnull.
5802 (extract_range_from_unary_expr): Use range_includes_zero_p instead
5803 of range_is_nonnull.
5804 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
5805 special case VR_ANTI_RANGE.
5806 (vrp_finalize): Same.
5807 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
5809 (range_is_nonnull): Remove.
5810 * vr-values.c (vrp_stmt_computes_nonzero): Use
5811 range_includes_zero_p instead of range_is_nonnull.
5812 (extract_range_basic): Pass value_range to range_includes_zero_p
5813 instead of range_is_nonnull.
5815 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
5817 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
5818 * emit-rtl.h (rtl_data): Remove return_bnd.
5819 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
5820 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
5821 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
5822 (POINTER_BOUNDS_MODE): Remove definition.
5823 (make_pointer_bounds_mode): Remove.
5824 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
5825 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
5826 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
5827 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
5828 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
5829 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
5830 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
5832 * config/i386/i386-modes.def (BND32, BND64): Remove.
5833 * config/i386/i386.c (dbx_register_map): Remove bound registers.
5834 (dbx64_register_map): Ditto.
5835 (svr4_dbx_register_map): Ditto.
5836 (indirect_thunk_bnd_needed): Remove.
5837 (indirect_thunks_bnd_used): Ditto.
5838 (indirect_return_bnd_needed): Ditto.
5839 (indirect_return_via_cx_bnd): Ditto.
5840 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
5841 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
5842 (output_indirect_thunk): Ditto. Remove need_prefix argument.
5843 (output_indirect_thunk_function): Remove handling of
5844 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
5845 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
5846 (ix86_save_reg): Remove handling of crtl->return_bnd.
5847 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
5848 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
5849 and UNSPEC_BNDLX_ADDR.
5850 (ix86_output_indirect_branch_via_reg): Remove handling of
5851 indirect_thunk_prefix_bnd.
5852 (ix86_output_indirect_branch_via_push): Ditto.
5853 (ix86_output_function_return): Ditto.
5854 (ix86_output_indirect_function_return): Ditto.
5855 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
5856 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
5857 (CALL_USED_REGISTERS): Ditto.
5858 (REG_ALLOC_ORDER): Update for removal of bound registers.
5859 (HI_REGISTER_NAMES): Ditto.
5860 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
5861 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
5862 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
5863 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
5864 (FIRST_PSEUDO_REG): Update.
5865 (BND): Remove mode iterator.
5866 * config/i386/predicates.md (bnd_mem_operator): Remove.
5868 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
5870 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
5873 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
5875 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
5876 the case in which the permute needs only a single element and
5877 repeats for every vector of the result. Extend that case to
5878 handle variable-length vectors.
5879 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
5881 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
5884 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
5885 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
5887 2018-08-24 Richard Biener <rguenther@suse.de>
5889 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
5890 bb_flags_allocated members.
5891 (auto_flag): New RAII class for allocating flags.
5892 (auto_edge_flag): New RAII class for allocating edge flags.
5893 (auto_bb_flag): New RAII class for allocating bb flags.
5894 * cfgloop.c (verify_loop_structure): Allocate temporary edge
5896 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
5897 in favor of temporarily allocated BB flag.
5898 * hsa-brig.c: Re-order includes.
5899 * hsa-dump.c: Likewise.
5900 * hsa-regalloc.c: Likewise.
5901 * print-rtl.c: Likewise.
5902 * profile-count.c: Likewise.
5904 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
5907 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
5910 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
5913 * wide-int-range.cc (wide_int_range_div): Do not ignore result
5914 from wide_int_range_multiplicative_op.
5916 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5918 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
5921 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
5923 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
5926 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
5928 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
5929 full GENERIC expression used for replacement.
5931 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
5933 * tree-vrp.c (abs_extent_range): Remove.
5934 (extract_range_into_wide_ints): Pass wide ints by reference.
5935 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
5936 Pass wide ints by reference in all calls to
5937 extract_range_into_wide_ints.
5938 * wide-int-range.cc (wide_int_range_div): New.
5939 * wide-int-range.h (wide_int_range_div): New.
5940 (wide_int_range_includes_zero_p): New.
5941 (wide_int_range_zero_p): New.
5943 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
5945 * config/aarch64/aarch64.md (arches): New enum.
5946 (arch): New enum attr.
5947 (arch_enabled): New attr.
5948 (enabled): Now uses arch_enabled only.
5949 (simd, sve, fp16): Removed attribute.
5950 (fp): Attr now defined in terms of 'arch'.
5951 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
5952 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
5953 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
5954 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
5955 attributes into 'arch'.
5956 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
5957 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
5958 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
5959 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
5960 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
5961 'simd' attribute into 'arch'.
5962 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
5963 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
5964 Convert use of 'fp' attribute to 'arch'.
5965 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
5966 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
5968 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
5969 (different modes) Merge 'fp' and 'simd' into 'arch'.
5970 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
5973 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
5975 PR rtl-optimization/87026
5976 * expmed.c (canonicalize_comparison): If we can no longer create
5977 pseudoregisters, don't.
5979 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
5982 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
5984 * config/arm/arm.c (speculation_barrier_libfunc): New static
5986 (arm_init_libfuncs): Initialize it.
5987 (arm_emit_speculation_barrier): New function.
5988 * config/arm/arm.md (speculation_barrier): Call
5989 arm_emit_speculation_barrier for architectures that do not have
5991 (speculation_barrier_insn): Only match on Armv7 or later.
5993 2018-08-23 Richard Biener <rguenther@suse.de>
5996 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
5999 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
6001 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
6002 of single-vector TBLs.
6003 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
6006 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
6009 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
6010 aarch64_evpc_tbl guard.
6012 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
6014 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
6017 2018-08-22 Martin Sebor <msebor@redhat.com>
6020 * tree-pretty-print.c (pretty_print_string): Add argument.
6021 (dump_generic_node): Call to pretty_print_string with string size.
6023 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
6025 PR rtl-optimization/86771
6026 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
6027 of two SETs into those two SETs, one to be placed at i2, if that SETs
6028 destination is modified between i2 and i3.
6030 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6032 PR tree-optimization/86725
6033 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
6035 (vect_analyze_scalar_cycles_1): Check it.
6037 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6039 PR tree-optimization/86725
6040 * tree-vect-loop.c (vect_is_simple_reduction): When treating
6041 an outer loop phi as a double reduction, make sure that the
6042 single user of the phi result is an inner loop phi.
6044 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6046 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
6047 grouped stores with gaps to a strided group.
6049 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6051 * tree-vect-stmts.c (get_group_load_store_type)
6052 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
6053 first statement in a group.
6055 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6057 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
6058 the sequence used in gcc/gcc.c.
6060 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6063 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
6066 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6068 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
6069 Darwin10-specific unwinder-shim.
6070 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
6071 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
6072 New to cater for Darwin10 Rosetta.
6074 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6076 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
6077 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
6079 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6084 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
6085 (dwarf2out_switch_text_section): Generate a local label for the
6086 second function sub-section and apply it as the second FDE start
6088 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
6089 second sub-section start.
6091 2018-08-22 Richard Biener <rguenther@suse.de>
6093 PR tree-optimization/86988
6094 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
6096 2018-08-22 Richard Biener <rguenther@suse.de>
6098 PR tree-optimization/86945
6099 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
6101 2018-08-22 Alexandre Oliva <oliva@adacore.com>
6103 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
6104 a comment about how uses of r2 for .sdata2 come about.
6106 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
6108 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
6110 2018-08-21 Marek Polacek <polacek@redhat.com>
6112 PR c++/86981, Implement -Wpessimizing-move.
6113 * doc/invoke.texi: Document -Wpessimizing-move.
6115 2018-08-21 Jan Hubicka <jh@suse.cz>
6117 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
6118 * tree.h (is_redundant_typedef): Remove.
6119 * dwarf2out.c (is_redundant_typedef): Turn into static function.
6121 2018-08-21 Jan Hubicka <jh@suse.cz>
6123 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
6126 2018-08-21 Tamar Christina <tamar.christina@arm.com>
6128 * expmed.c (extract_low_bits): Reject invalid subregs early.
6130 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
6133 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
6136 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6138 * config/vxworks.h: Guard vxworks_asm_out_constructor and
6139 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
6140 * config/vxworks.c: Likewise.
6142 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6144 * config/vxworks.c: Set targetm.have_ctors_dtors
6145 if HAVE_INITFINI_ARRAY_SUPPORT.
6146 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
6147 if HAVE_INITFINI_ARRAY_SUPPORT.
6149 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6151 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
6152 default search path for VxWorks < 7.
6154 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
6156 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
6157 (get_format_string): Refer to c_getstr.
6159 2018-08-21 Tom de Vries <tdevries@suse.de>
6161 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
6162 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
6163 (debuginfo_early_stop): Declare.
6164 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
6165 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
6166 (debuginfo_early_stop): New function.
6167 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
6168 and debuginfo_early_stop.
6169 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
6170 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
6171 (general_init): Call debuginfo_early_init.
6172 (finalize): Call debuginfo_fini.
6173 (do_compile): Call debuginfo_init.
6174 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
6176 (@item -fdump-debug, @item -fdump-earlydebug): Add.
6178 2018-08-21 Tom de Vries <tdevries@suse.de>
6180 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
6181 flag_dump_noaddr and flag_dump_unnumbered.
6183 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
6185 * wide-int-range.cc (wide_int_range_abs): New.
6186 (wide_int_range_order_set): Rename from wide_int_range_min_max.
6187 * wide-int-range.h (wide_int_range_abs): New.
6188 (wide_int_range_min_max): New.
6189 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
6190 case to call wide_int_range_abs.
6191 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
6192 (extract_range_from_abs_expr): Delete.
6194 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
6197 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
6198 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
6201 2018-08-20 Nathan Sidwell <nathan@acm.org>
6202 Jeff Law <law@redhat.com>
6204 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
6205 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
6207 2018-08-20 David Malcolm <dmalcolm@redhat.com>
6210 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
6211 (decl_attributes): Likewise.
6212 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
6214 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
6215 * diagnostic-core.h (class auto_diagnostic_group): New class.
6216 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
6217 (diagnostic_report_diagnostic): Handle the first diagnostics within
6219 (emit_diagnostic): Add auto_diagnostic_group instance.
6221 (inform_n): Likewise.
6222 (warning): Likewise.
6223 (warning_at): Likewise.
6224 (warning_n): Likewise.
6225 (pedwarn): Likewise.
6226 (permerror): Likewise.
6228 (error_n): Likewise.
6229 (error_at): Likewise.
6231 (fatal_error): Likewise.
6232 (internal_error): Likewise.
6233 (internal_error_no_backtrace): Likewise.
6234 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
6235 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
6236 * diagnostic.h (struct diagnostic_context): Add fields
6237 "diagnostic_group_nesting_depth",
6238 "diagnostic_group_emission_count", "begin_group_cb",
6240 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
6241 Add auto_diagnostic_group instance(s).
6242 (find_explicit_erroneous_behavior): Likewise.
6243 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
6244 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
6245 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
6246 (gimplify_va_arg_expr): Likewise.
6247 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
6248 (HSA_SORRY_AT): Likewise.
6249 * ipa-devirt.c (compare_virtual_tables): Likewise.
6250 (warn_odr): Likewise.
6251 * multiple_target.c (expand_target_clones): Likewise.
6252 * opts-common.c (cmdline_handle_error): Likewise.
6253 * reginfo.c (globalize_reg): Likewise.
6254 * substring-locations.c (format_warning_n_va): Likewise.
6255 * tree-inline.c (expand_call_inline): Likewise.
6256 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
6257 * tree-ssa-loop-niter.c
6258 (do_warn_aggressive_loop_optimizations): Likewise.
6259 * tree-ssa-uninit.c (warn_uninit): Likewise.
6260 * tree.c (warn_deprecated_use): Likewise.
6262 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
6265 * config/i386/i386.md (eh_return): Always update EH return
6266 address in word_mode.
6268 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
6270 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
6271 TARGET_SPLIT_COMPLEX_ARG is defined.
6273 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
6275 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
6277 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
6280 * expr.c (expand_assignment): Assert that bitpos is positive.
6281 (store_field): Likewise
6282 (expand_expr_real_1): Make sure that bitpos is positive.
6283 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
6286 2018-08-20 Nathan Sidwell <nathan@acm.org>
6288 * Makefile.in (CPP_ID_DATA_H): Delete.
6289 (CPP_INTERNAL_H): Don't add it.
6290 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
6291 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
6293 2018-08-20 Richard Biener <rguenther@suse.de>
6295 PR tree-optimization/78655
6296 * tree-vrp.c (extract_range_from_binary_expr_1): Make
6297 pointer + offset nonnull if either operand is nonnull work.
6299 2018-08-20 Tom de Vries <tdevries@suse.de>
6301 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
6302 unless the referenced die describes the added property using
6303 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
6304 Otherwise, add a DW_AT_location to the referenced die.
6306 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
6309 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
6310 register_operand when calling ix86_set_reg_reg_cost.
6311 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
6312 Set *total to 0 for operands that satisfy x86_64_immediate_operand
6313 predicate and to 1 otherwise.
6315 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
6317 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
6318 emit a diagnostic that it is not supported and reset the option.
6319 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
6320 supported and consume the option. (ASM_FINAL_SPEC): New.
6322 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
6324 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
6327 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
6329 C-SKY port: Documentation
6331 * doc/extend.texi (C-SKY Function Attributes): New section.
6332 * doc/invoke.texi (Option Summary): Add C-SKY options.
6333 (C-SKY Options): New section.
6334 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
6336 2018-08-17 Jojo <jijie_rong@c-sky.com>
6337 Huibin Wang <huibin_wang@c-sky.com>
6338 Sandra Loosemore <sandra@codesourcery.com>
6339 Chung-Lin Tang <cltang@codesourcery.com>
6341 C-SKY port: Backend implementation
6343 * config/csky/*: New.
6344 * common/config/csky/*: New.
6346 2018-08-17 Jojo <jijie_rong@c-sky.com>
6347 Huibin Wang <huibin_wang@c-sky.com>
6348 Sandra Loosemore <sandra@codesourcery.com>
6349 Chung-Lin Tang <cltang@codesourcery.com>
6350 Andrew Jenner <andrew@codesourcery.com>
6352 C-SKY port: Configury
6354 * config.gcc (csky-*-*): New.
6355 * configure.ac: Add csky to targets for dwarf2 debug_line support.
6356 * configure: Regenerated.
6358 2018-08-17 David Malcolm <dmalcolm@redhat.com>
6360 * dump-context.h: Include "dumpfile.h".
6361 (dump_context::dump_printf_va): Convert final param from va_list
6362 to va_list *. Convert from ATTRIBUTE_PRINTF to
6363 ATTRIBUTE_GCC_DUMP_PRINTF.
6364 (dump_context::dump_printf_loc_va): Likewise.
6365 * dumpfile.c: Include "stringpool.h".
6366 (make_item_for_dump_printf_va): Delete.
6367 (make_item_for_dump_printf): Delete.
6368 (class dump_pretty_printer): New class.
6369 (dump_pretty_printer::dump_pretty_printer): New ctor.
6370 (dump_pretty_printer::emit_items): New member function.
6371 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
6373 (dump_pretty_printer::emit_item): New member function.
6374 (dump_pretty_printer::stash_item): New member function.
6375 (dump_pretty_printer::format_decoder_cb): New member function.
6376 (dump_pretty_printer::decode_format): New member function.
6377 (dump_context::dump_printf_va): Reimplement in terms of
6378 dump_pretty_printer.
6379 (dump_context::dump_printf_loc_va): Convert final param from va_list
6381 (dump_context::begin_scope): Reimplement call to
6382 make_item_for_dump_printf.
6383 (dump_printf): Update for change to dump_printf_va.
6384 (dump_printf_loc): Likewise.
6385 (selftest::test_capture_of_dump_calls): Convert "stmt" from
6386 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
6387 with %T, %E, and %G.
6388 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
6389 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
6390 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
6391 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
6392 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
6393 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
6394 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
6395 within a dump_printf_loc call to "%wu".
6396 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
6397 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
6398 missing space after "=".
6399 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
6400 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
6401 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
6402 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
6403 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
6404 duplicate "vectorized" from message.
6406 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
6408 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
6409 polyNxK_t element's TYPE_STRING_FLAG.
6411 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
6413 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
6414 (they were unnamed before). Fix comments.
6416 2018-08-17 Nathan Sidwell <nathan@acm.org>
6418 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
6420 2018-08-17 Richard Biener <rguenther@suse.de>
6422 PR tree-optimization/86841
6423 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
6425 2018-08-17 Martin Liska <mliska@suse.cz>
6427 * common.opt: Remove Warn, Init and Report for options with
6428 Ignore/Deprecated flag. Warning is done automatically for
6430 * config/i386/i386.opt: Likewise.
6431 * config/ia64/ia64.opt: Likewise.
6432 * config/rs6000/rs6000.opt: Likewise.
6433 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
6434 Remove usage of flag_check_pointer_bounds.
6435 * lto-wrapper.c (merge_and_complain): Do not handle
6436 OPT_fcheck_pointer_bounds.
6437 (append_compiler_options): Likewise.
6438 * opt-functions.awk: Do not handle Deprecated.
6439 * optc-gen.awk: Check that Var, Report and Init are not
6440 used for an option with Ignore/Deprecated flag.
6441 * opts-common.c (decode_cmdline_option): Do not report
6443 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
6445 * opts.h (struct cl_option): Remove cl_deprecated flag.
6446 (CL_ERR_DEPRECATED): Remove error enum value.
6448 2018-08-17 Richard Biener <rguenther@suse.de>
6451 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
6452 across a va-arg-pack using call adjust its return value accordingly.
6454 2018-08-16 Martin Sebor <msebor@redhat.com>
6456 PR tree-optimization/86853
6457 * gimple-ssa-sprintf.c (struct format_result): Rename member.
6458 (struct fmtresult): Add member and initialize it in ctors.
6459 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
6460 (format_string): Handle %S the same as %ls. Set MAYFAIL.
6461 (format_directive): Set POSUNDER4K when MAYFAIL is set.
6462 (parse_directive): Handle %C same as %c.
6463 (sprintf_dom_walker::compute_format_length): Adjust.
6464 (is_call_safe): Adjust.
6466 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
6468 * builtins.c (c_strlen): Add new parameter eltsize. Use it
6469 for determining how to count the elements.
6470 * builtins.h (c_strlen): Adjust prototype.
6471 * expr.c (string_constant): Add new parameter mem_size.
6472 Set *mem_size appropriately.
6473 * expr.h (string_constant): Adjust protoype.
6474 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
6475 * gimple-fold.h (get_range_strlen): Adjust prototype.
6476 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
6477 (format_string): Call get_string_length with eltsize.
6479 2018-08-16 David Malcolm <dmalcolm@redhat.com>
6481 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
6482 to emit the span, rather than setting it as the prefix.
6484 2018-08-16 David Malcolm <dmalcolm@redhat.com>
6486 * diagnostic-show-locus.c (layout::start_annotation_line): Add
6487 "margin_char" parameter, defaulting to space. Use it in place
6488 of pp_space for the initial part of the margin.
6489 (layout::print_leading_fixits): Use '+' when filling the margin
6490 of line-insertion fix-it hints.
6492 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6494 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
6497 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6499 * config/rs6000/altivec.md: Don't set length attribute to the default
6501 * config/rs6000/darwin.md: Ditto.
6502 * config/rs6000/dfp.md: Ditto.
6503 * config/rs6000/htm.md: Ditto.
6504 * config/rs6000/rs6000.md: Ditto.
6505 * config/rs6000/sync.md: Ditto.
6506 * config/rs6000/vsx.md: Ditto.
6508 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6510 * config/rs6000/altivec.md: Don't set length attribute to the default
6511 value, for branch instructions.
6512 * config/rs6000/darwin.md: Ditto.
6513 * config/rs6000/rs6000.md: Ditto.
6515 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6517 * config/rs6000/rs6000.md (length): Always define as const_int 4.
6518 (unnamed conditional branch define_insn): Set length to 4 or 8
6519 depending on offset.
6520 (<bd>_<mode>): Similar, for alternative 0.
6521 (<bd>tf_<mode>): Ditto.
6523 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6525 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
6527 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
6529 * doc/rtl.texi: Replace old RTX class names with new names.
6532 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
6534 * expmed.h (canonicalize_comparison): New declaration.
6535 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
6536 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
6537 * optabs.c (prepare_cmp_insn): Likewise.
6538 * rtl.h (unsigned_condition_p): New function which checks if a
6539 comparison operator is unsigned.
6541 2018-08-16 Nathan Sidwell <nathan@acm.org>
6543 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
6544 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
6546 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6549 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
6550 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
6551 (mov<mov>): ..this and enable unconditionally.
6553 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6555 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
6557 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
6559 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
6560 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
6561 "Common" with "Target".
6563 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
6565 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
6566 * doc/invoke.texi (mmitigate-rop): Remove.
6567 * config/i386/i386.c: Do not include "regrename.h".
6568 (ix86_rop_should_change_byte_p, reg_encoded_number)
6569 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
6571 (ix86_reorg): Remove call to ix86_mitigate_rop.
6572 * config/i386/i386.md (attr "modrm_class"): Remove.
6573 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
6574 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
6575 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
6577 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6579 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
6580 allow folding of mergeh() and mergel() for the float and double types.
6581 (fold_mergehl_helper): Rework to handle building a permute tree
6584 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
6586 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
6589 2018-08-15 David Malcolm <dmalcolm@redhat.com>
6591 * common.opt (fdiagnostics-show-labels): New option.
6592 * diagnostic-show-locus.c (class layout_range): Add field
6594 (class layout): Add field "m_show_labels_p".
6595 (layout_range::layout_range): Add param "label" and use it to
6597 (make_range): Pass in NULL for new "label" param of layout_range's
6599 (layout::layout): Initialize m_show_labels_p.
6600 (layout::maybe_add_location_range): Pass in loc_range->m_label
6601 when constructing layout_range instances.
6602 (struct line_label): New struct.
6603 (layout::print_any_labels): New member function.
6604 (layout::print_line): Call it if label-printing is enabled.
6605 (selftest::test_one_liner_labels): New test.
6606 (selftest::test_diagnostic_show_locus_one_liner): Call it.
6607 * diagnostic.c (diagnostic_initialize): Initialize
6608 context->show_labels_p.
6609 * diagnostic.h (struct diagnostic_context): Add field
6611 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
6612 -fno-diagnostics-show-labels.
6613 * dwarf2out.c (gen_producer_string): Add
6614 OPT_fdiagnostics_show_labels to the ignored options.
6615 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
6617 (gcc_rich_location::maybe_add_expr): Likewise.
6618 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
6619 label" param, defaulting to NULL.
6620 (gcc_rich_location::add_expr): Add "label" param.
6621 (gcc_rich_location::maybe_add_expr): Likewise.
6622 (class text_range_label): New class.
6623 (class range_label_for_type_mismatch): New class.
6624 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
6625 of format_warning_va.
6626 (fmtwarn_n): Likewise for new params of format_warning_n_va.
6627 * lto-wrapper.c (merge_and_complain): Add
6628 OPT_fdiagnostics_show_labels to the "pick one setting" options.
6629 (append_compiler_options): Likewise to the dropped options.
6630 (append_diag_options): Likewise to the passed-on options.
6631 * opts.c (common_handle_option): Handle the new option.
6632 * selftest-diagnostic.c
6633 (test_diagnostic_context::test_diagnostic_context): Enable
6635 * substring-locations.c: Include "gcc-rich-location.h".
6636 (format_warning_n_va): Add "fmt_label" and "param_label" params
6637 and use them as appropriate.
6638 (format_warning_va): Add "fmt_label" and "param_label" params,
6639 passing them on to format_warning_n_va.
6640 (format_warning_at_substring): Likewise.
6641 (format_warning_at_substring_n): Likewise.
6642 * substring-locations.h (format_warning_va): Add "fmt_label" and
6643 "param_label" params.
6644 (format_warning_n_va): Likewise.
6645 (format_warning_at_substring): Likewise.
6646 (format_warning_at_substring_n): Likewise.
6647 * toplev.c (general_init): Initialize global_dc->show_labels_p.
6649 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
6652 * builtins.c (expand_builtin_memcmp): Do not expand the call
6653 when overflow is detected.
6655 2018-08-15 Martin Sebor <msebor@redhat.com>
6657 PR tree-optimization/71625
6658 * config/aarch64/aarch64-builtins.c
6659 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
6661 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
6663 * config/s390/s390.c (s390_reorg): Remove loop.
6665 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
6668 (darwin_function_switched_text_sections): Delete.
6669 * gcc/config/darwin.h
6670 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
6672 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
6675 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
6676 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
6677 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
6679 2018-08-15 Martin Liska <mliska@suse.cz>
6681 PR tree-optimization/86925
6682 * predict.c (expr_expected_value_1): When taking
6683 later predictor, assign also probability.
6684 Use fold_build2_initializer_loc in order to fold
6685 the expression in -frounding-math.
6687 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
6689 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
6691 (expand_vec_perm_1): Try the new method.
6693 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
6696 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
6697 Check whether lra_live_max_point is 0 before dividing.
6699 2018-08-14 Martin Sebor <msebor@redhat.com>
6701 PR tree-optimization/86650
6702 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
6703 (vrp_prop::check_mem_ref): Same.
6705 2018-08-13 Liu Hao <lh_mouse@126.com>
6707 * pretty-print.c (eat_esc_sequence): Swap the foreground and
6708 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
6709 and clear it thereafter, as it only works for DBCS.
6711 2018-08-13 Liu Hao <lh_mouse@126.com>
6713 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
6714 handle returned by _get_osf_handle().
6716 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6718 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
6719 for folding vec_perm.
6721 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6723 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
6724 Add support for gimple-folding of vec_pack() and vec_unpack()
6727 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6729 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
6730 vec_xst variants to the list.
6731 (rs6000_gimple_fold_builtin): Add support for folding unaligned
6732 vector loads and stores.
6734 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
6736 * config.gcc (rs6000-ibm-aix4.x): Delete.
6737 (rs6000-ibm-aix5.1): Delete.
6738 (rs6000-ibm-aix5.2): Delete.
6739 (rs6000-ibm-aix5.3): Delete.
6740 * config/rs6000/aix43.h: Delete.
6741 * config/rs6000/aix51.h: Delete.
6742 * config/rs6000/aix52.h: Delete.
6743 * config/rs6000/t-aix43: Delete.
6745 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
6747 * config/s390/s390.c (s390_decompose_constant_pool_ref):
6749 (s390_decompose_address): Factor out constant pool ref
6752 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6754 * config/nds32/nds32-predicates.c
6755 (nds32_can_use_bclr_p): Change return type as bool.
6756 (nds32_can_use_bset_p): Ditto.
6757 (nds32_can_use_btgl_p): Ditto.
6758 (nds32_can_use_bitci_p): Ditto.
6759 * config/nds32/nds32-protos.h
6760 (nds32_can_use_bclr_p): Change declaration.
6761 (nds32_can_use_bset_p): Ditto.
6762 (nds32_can_use_btgl_p): Ditto.
6763 (nds32_can_use_bitci_p): Ditto.
6765 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6767 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
6768 Support -msched-prolog-epilog option.
6769 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
6771 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6773 * common/config/nds32/nds32-common.c
6774 (nds32_option_optimization_table): Enalbe -malways-align.
6776 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
6778 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
6780 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
6781 OPT_misr_secure_ case.
6782 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
6783 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
6784 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
6786 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
6787 (nds32_isr_info): New field security_level.
6788 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
6789 * config/nds32/nds32.md (return_internal): Consider critical attribute.
6790 * config/nds32/nds32.opt (misr-secure): New option.
6791 * config/nds32/nds32_init.inc: New file.
6792 * config/nds32/nds32_isr.h: New file.
6794 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
6796 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
6797 Update comment for atomic instructions.
6798 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
6799 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
6801 (atomic_loaddi): Revise fence expansion to only emit fence prior to
6802 load for __ATOMIC_SEQ_CST model.
6803 (atomic_loaddi_1): Remove float register target.
6804 (atomic_storedi): Handle CONST_INT values.
6805 (atomic_storedi_1): Remove float register source. Add special case
6807 (memory_barrier): New expander and insn.
6809 2018-08-11 Jakub Jelinek <jakub@redhat.com>
6811 PR tree-optimization/86835
6812 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
6813 new_stmt after def_gsi, make sure to insert new_square_stmt after
6814 that stmt, not 2 stmts before it.
6816 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
6819 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
6822 2018-08-10 Martin Liska <mliska@suse.cz>
6825 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
6827 * builtins.c (expand_builtin_expect_with_probability):
6829 (expand_builtin_expect_with_probability): New function.
6830 (build_builtin_expect_predicate): Add new argumnet probability
6831 for BUILT_IN_EXPECT_WITH_PROBABILITY.
6832 (fold_builtin_expect):
6835 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
6836 * builtins.h (fold_builtin_expect): Set new argument.
6837 * doc/extend.texi: Document __builtin_expect_with_probability.
6838 * doc/invoke.texi: Likewise.
6839 * gimple-fold.c (gimple_fold_call): Pass new argument.
6840 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
6841 also BUILT_IN_EXPECT_WITH_PROBABILITY.
6842 * predict.c (get_predictor_value): New function.
6843 (expr_expected_value): Add new argument probability. Assume
6844 that predictor and probability are always non-null.
6845 (expr_expected_value_1): Likewise. For __builtin_expect and
6846 __builtin_expect_with_probability set probability. Handle
6847 combination in binary expressions.
6848 (tree_predict_by_opcode): Simplify code by simply calling
6849 get_predictor_value.
6850 (pass_strip_predict_hints::execute): Add handling of
6851 BUILT_IN_EXPECT_WITH_PROBABILITY.
6852 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
6854 * tree.h (DECL_BUILT_IN_P): New function.
6856 2018-08-10 Martin Liska <mliska@suse.cz>
6858 PR tree-optimization/85799
6859 * passes.def: Add argument for pass_strip_predict_hints.
6860 * predict.c (class pass_strip_predict_hints): Add new argument
6862 (strip_predictor_early): New function.
6863 (pass_strip_predict_hints::execute): Call the function to
6865 (strip_predict_hints): New function.
6866 * predict.def: Fix comment.
6868 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
6870 * Makefile.in: Clarify which tm.texi to copy over to assert the
6871 right to grant a GFDL license for all.
6873 2018-08-09 Jeff Law <law@redhat.com>
6875 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
6878 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
6880 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
6883 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
6885 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
6886 reductions for variable-length vectors.
6888 2018-08-09 David Malcolm <dmalcolm@redhat.com>
6891 * common.opt (fdiagnostics-show-line-numbers): New option.
6892 * diagnostic-show-locus.c (class layout): Add fields
6893 "m_show_line_numbers_p" and "m_linenum_width";
6894 (num_digits): New function.
6895 (test_num_digits): New function.
6896 (layout::layout): Initialize new fields. Update m_x_offset
6897 logic to handle any left margin.
6898 (layout::print_source_line): Print line number when requested.
6899 (layout::start_annotation_line): New member function.
6900 (layout::print_annotation_line): Call it.
6901 (layout::print_leading_fixits): Likewise.
6902 (layout::print_trailing_fixits): Likewise. Update calls to
6903 move_to_column for new parameter.
6904 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
6905 it to potentially call start_annotation_line.
6906 (layout::show_ruler): Call start_annotation_line.
6907 (selftest::test_line_numbers_multiline_range): New selftest.
6908 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
6909 and selftest::test_line_numbers_multiline_range.
6910 * diagnostic.c (diagnostic_initialize): Initialize
6911 show_line_numbers_p.
6912 * diagnostic.h (struct diagnostic_context): Add field
6913 "show_line_numbers_p".
6914 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
6915 -fno-diagnostics-show-line-numbers.
6916 * dwarf2out.c (gen_producer_string): Add
6917 OPT_fdiagnostics_show_line_numbers to the ignored options.
6918 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
6919 one setting" options.
6920 (append_compiler_options): Likewise to the dropped options.
6921 (append_diag_options): Likewise to the passed-on options.
6922 * opts.c (common_handle_option): Handle the new option.
6923 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
6925 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
6927 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
6928 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
6929 third argument of type "const signed char" to descriptions of
6930 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
6931 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
6932 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
6933 __builtin_bcdsub_ov functions.
6935 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
6937 PR tree-optimization/86858
6938 * tree-vect-loop.c (vect_is_simple_reduction): Restore
6939 flow_bb_inside_loop_p calls.
6941 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
6943 PR tree-optimization/86871
6944 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
6945 instead of gimple_assign_lhs.
6947 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
6950 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
6951 register constraint to operand 0.
6952 (add<mode>3_carryinC): Likewise.
6953 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
6955 2018-08-09 Martin Liska <mliska@suse.cz>
6958 * common.opt: Remove extra line.
6960 2018-08-09 Martin Liska <mliska@suse.cz>
6962 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
6963 at the end of a line, make first letter capital and end up
6964 a sentence with a dot.
6965 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
6966 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
6967 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
6968 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
6969 (PARAM_MAX_ISL_OPERATIONS): Likewise.
6970 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
6971 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
6972 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
6973 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
6974 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
6975 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
6976 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
6977 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
6978 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
6979 (PARAM_TREE_REASSOC_WIDTH): Likewise.
6980 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
6981 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
6982 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
6984 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
6987 * config/s390/s390.c (s390_option_override_internal): Reduce the
6988 stack-clash-protection-probe-interval param if it would be too big
6991 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
6994 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
6995 * config/m68k/m68k.c (handle_move_double): Don't call
6996 m68k_final_prescan_insn.
6997 (m68k_adjust_decorated_operand): Renamed from
6998 m68k_final_prescan_insn, remove first and third operand and
7000 (print_operand): Call it.
7001 (print_operand_address): Call it.
7003 2018-08-08 Nathan Sidwell <nathan@acm.org>
7005 * diagnostic.c (diagnostic_report_current_module): Use
7006 linemap_included_from & linemap_included_from_linemap.
7008 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
7010 * config/aarch64/aarch64-cores.def: Add phecda core.
7011 * config/aarch64/aarch64-tune.md: Regenerate.
7012 * doc/invoke.texi: Add phecda core.
7014 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
7017 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
7019 * config/s390/s390.md ("movti"): Add more alternatives for
7020 constant to GPR copies.
7022 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
7024 * config/s390/s390.c: Fix whitespace damage throughout the file.
7025 * config/s390/s390.h: Likewise.
7026 * config/s390/tpf.h: Likewise.
7028 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
7030 * config/s390/s390.c (s390_loadrelative_operand_p):
7031 Remove TARGET_CPU_ZARCH usages.
7032 (s390_rtx_costs): Likewise.
7033 (s390_legitimate_constant_p): Likewise.
7034 (s390_cannot_force_const_mem): Likewise.
7035 (legitimate_reload_constant_p): Likewise.
7036 (s390_preferred_reload_class): Likewise.
7037 (legitimize_pic_address): Likewise.
7038 (legitimize_tls_address): Likewise.
7039 (s390_split_branches): Removed.
7040 (s390_add_execute): Removed.
7041 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
7042 (s390_mainpool_start): Likewise.
7043 (s390_mainpool_finish): Likewise.
7044 (s390_mainpool_cancel): Removed.
7045 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
7046 (s390_chunkify_cancel): Likewise.
7047 (s390_return_addr_rtx): Likewise.
7048 (s390_register_info): Remove split_branches_pending_p uages.
7049 (s390_optimize_register_info): Likewise.
7050 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
7051 split_branches_pending_p usages.
7052 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
7053 (s390_load_got): Likewise.
7054 (s390_expand_split_stack_prologue): Likewise.
7055 (output_asm_nops): Likewise.
7056 (s390_function_profiler): Likewise.
7057 (s390_emit_call): Likewise.
7058 (s390_conditional_register_usage): Likewise.
7059 (s390_optimize_prologue): Likewise.
7060 (s390_reorg): Remove TARGET_CPU_ZARCH and
7061 split_branches_pending_p usages.
7062 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
7064 (s390_output_indirect_thunk_function): Likewise.
7065 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
7066 (TARGET_CPU_ZARCH_P): Removed.
7067 (struct machine_function): Remove split_branches_pending_p.
7068 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
7070 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
7072 * common/config/s390/s390-common.c (processor_flags_table):
7074 * config.gcc: Remove with_arch/with_tune support.
7075 * config/s390/2064.md: Remove cpu attribute comparisons.
7076 * config/s390/driver-native.c (s390_host_detect_local_cpu):
7078 * config/s390/linux.h (ASM_SPEC):
7079 Remove -march support.
7080 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
7081 Use a table to get an arch level.
7082 * config/s390/s390-opts.h (enum processor_type):
7084 * config/s390/s390.c
7085 (processor_table): Remove entries, add arch_level values.
7086 (s390_issue_rate): Remove cases.
7087 (s390_option_override): Adjust
7088 s390_option_override_internal() call.
7089 (s390_option_override_internal): Remove deprecation warning.
7090 (s390_valid_target_attribute_tree): Adjust
7091 s390_option_override_internal() call.
7092 * config/s390/s390.h (struct s390_processor):
7093 Share with s390-c.c, add arch_level field.
7094 * config/s390/s390.md:
7095 Remove occurrences in cpu attribute.
7096 * config/s390/s390.opt: Remove -march/-mtune support.
7097 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
7098 * doc/invoke.texi: Remove deprecation warning.
7100 2018-08-08 Luis Machado <luis.machado@linaro.org>
7102 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
7104 (qdf24xx_tunings): Set vector cost structure to
7105 qdf24xx_vector_cost.
7107 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
7108 <register_sextend>: Set to 3.
7110 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
7113 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
7114 * config/aarch64/aarch64-simd.md
7115 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
7116 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
7117 (aarch64_frecpx<mode>): ...this new pattern.
7118 * config/aarch64/aarch64-simd-builtins.def: Remove comment
7119 about aarch64_frecp<FRECP:frecp_suffix><mode>.
7121 2018-08-07 Martin Liska <mliska@suse.cz>
7124 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
7125 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
7126 * predict.def (PRED_MALLOC_NONNULL): New predictor.
7127 * doc/extend.texi: Document that malloc attribute adds
7130 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
7133 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7134 Define to speculation_safe_value_not_needed.
7136 2018-08-06 Jeff Law <law@redhat.com>
7138 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
7139 the vr_values instance to cprop_into_stmt.
7140 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
7141 (cprop_operand): Also query EVRP to determine if OP is a constant.
7143 2018-08-06 Nathan Sidwell <nathan@acm.org>
7145 * diagnostic.c (diagnostic_report_current_module): Reroll
7146 included-at loop. Translate text.
7148 2018-08-06 David Malcolm <dmalcolm@redhat.com>
7150 * function-tests.c (selftest::test_expansion_to_rtl): Call
7151 free_after_compilation.
7153 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7155 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
7157 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
7159 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
7160 loops with memory block operations from getting unrolled.
7162 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
7165 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7166 Define to speculation_safe_value_not_needed.
7168 2018-08-06 Jeff Law <law@redhat.com>
7170 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
7173 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
7176 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
7177 with all enabled __intN types.
7179 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
7181 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7183 * alias.c (record_set): Check for clobber high.
7184 * cfgexpand.c (expand_gimple_stmt): Likewise.
7185 * combine-stack-adj.c (single_set_for_csa): Likewise.
7186 * combine.c (find_single_use_1): Likewise.
7187 (set_nonzero_bits_and_sign_copies): Likewise.
7188 (get_combine_src_dest): Likewise.
7189 (is_parallel_of_n_reg_sets): Likewise.
7190 (try_combine): Likewise.
7191 (record_dead_and_set_regs_1): Likewise.
7192 (reg_dead_at_p_1): Likewise.
7193 (reg_dead_at_p): Likewise.
7194 * dce.c (deletable_insn_p): Likewise.
7195 (mark_nonreg_stores_1): Likewise.
7196 (mark_nonreg_stores_2): Likewise.
7197 * df-scan.c (df_find_hard_reg_defs): Likewise.
7198 (df_uses_record): Likewise.
7199 (df_get_call_refs): Likewise.
7200 * dwarf2out.c (mem_loc_descriptor): Likewise.
7201 * haifa-sched.c (haifa_classify_rtx): Likewise.
7202 * ira-build.c (create_insn_allocnos): Likewise.
7203 * ira-costs.c (scan_one_insn): Likewise.
7204 * ira.c (equiv_init_movable_p): Likewise.
7205 (rtx_moveable_p): Likewise.
7206 (interesting_dest_for_shprep): Likewise.
7207 * jump.c (mark_jump_label_1): Likewise.
7208 * postreload-gcse.c (record_opr_changes): Likewise.
7209 * postreload.c (reload_cse_simplify): Likewise.
7210 (struct reg_use): Add source expr.
7211 (reload_combine): Check for clobber high.
7212 (reload_combine_note_use): Likewise.
7213 (reload_cse_move2add): Likewise.
7214 (move2add_note_store): Likewise.
7215 * print-rtl.c (print_pattern): Likewise.
7216 * recog.c (decode_asm_operands): Likewise.
7217 (store_data_bypass_p): Likewise.
7218 (if_test_bypass_p): Likewise.
7219 * regcprop.c (kill_clobbered_value): Likewise.
7220 (kill_set_value): Likewise.
7221 * reginfo.c (reg_scan_mark_refs): Likewise.
7222 * reload1.c (maybe_fix_stack_asms): Likewise.
7223 (eliminate_regs_1): Likewise.
7224 (elimination_effects): Likewise.
7225 (mark_not_eliminable): Likewise.
7226 (scan_paradoxical_subregs): Likewise.
7227 (forget_old_reloads_1): Likewise.
7228 * reorg.c (find_end_label): Likewise.
7229 (try_merge_delay_insns): Likewise.
7230 (redundant_insn): Likewise.
7231 (own_thread_p): Likewise.
7232 (fill_simple_delay_slots): Likewise.
7233 (fill_slots_from_thread): Likewise.
7234 (dbr_schedule): Likewise.
7235 * resource.c (update_live_status): Likewise.
7236 (mark_referenced_resources): Likewise.
7237 (mark_set_resources): Likewise.
7238 * rtl.c (copy_rtx): Likewise.
7239 * rtlanal.c (reg_referenced_p): Likewise.
7240 (single_set_2): Likewise.
7241 (noop_move_p): Likewise.
7242 (note_stores): Likewise.
7243 * sched-deps.c (sched_analyze_reg): Likewise.
7244 (sched_analyze_insn): Likewise.
7246 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7248 * cse.c (invalidate_reg): New function extracted from...
7249 (invalidate): ...here.
7250 (canonicalize_insn): Check for clobber high.
7251 (invalidate_from_clobbers): invalidate clobber highs.
7252 (invalidate_from_sets_and_clobbers): Likewise.
7253 (count_reg_usage): Check for clobber high.
7254 (insn_live_p): Likewise.
7255 * cselib.c (cselib_expand_value_rtx_1):Likewise.
7256 (cselib_invalidate_regno): Check for clobber in setter.
7257 (cselib_invalidate_rtx): Pass through setter.
7258 (cselib_invalidate_rtx_note_stores):
7259 (cselib_process_insn): Check for clobber high.
7260 * cselib.h (cselib_invalidate_rtx): Add operand.
7262 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7264 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
7265 (mark_not_eliminable): Likewise.
7266 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
7267 * lra-lives.c (process_bb_lives): Check for clobber high.
7268 * lra.c (new_insn_reg): Remember clobber highs.
7269 (collect_non_operand_hard_regs): Check for clobber high.
7270 (lra_set_insn_recog_data): Likewise.
7271 (add_regs_to_insn_regno_info): Likewise.
7272 (lra_update_insn_regno_info): Likewise.
7274 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7276 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
7277 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
7279 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7281 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
7282 (copy_insn_1): Likewise.
7283 (gen_hard_reg_clobber_high): New gen function.
7284 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
7285 * genemit.c (gen_exp): Likewise.
7286 (gen_emit_seq): Pass through info.
7287 (gen_insn): Check for CLOBBER_HIGH.
7288 (gen_expand): Pass through info.
7289 (gen_split): Likewise.
7290 (output_add_clobbers): Likewise.
7291 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
7292 (remove_clobbers): Likewise.
7293 * rtl.h (gen_hard_reg_clobber_high): New declaration.
7295 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7297 * doc/rtl.texi (clobber_high): Add.
7298 (parallel): Add in clobber high
7299 * rtl.c (rtl_check_failed_code3): Add function.
7300 * rtl.def (CLOBBER_HIGH): Add expression.
7301 * rtl.h (RTL_CHECKC3): Add macro.
7302 (rtl_check_failed_code3): Add declaration.
7303 (XC3EXP): Add macro.
7305 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
7308 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
7309 cfun->machine->max_used_stack_alignment if needed.
7311 2018-08-04 Martin Sebor <msebor@redhat.com>
7313 PR tree-optimization/86571
7314 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
7317 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
7320 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7323 2018-08-03 Jeff Law <law@redhat.com>
7326 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7327 Define to speculation_safe_value_not_needed.
7329 2018-08-03 David Malcolm <dmalcolm@redhat.com>
7331 * doc/gcov.texi (-x): Remove duplicate "to".
7332 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
7333 (-Wif-not-aligned): Remove duplicate "is".
7334 (-flto): Remove duplicate "the".
7335 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
7336 duplicate "v5.00.b".
7337 (MSP430 Options): Remove duplicate "and" from the description
7338 of "-mgprel-sec=regexp".
7339 (x86 Options): Remove duplicate copies of "vmldLog102" and
7340 vmlsLog104 from description of "-mveclibabi=type".
7342 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
7344 * internal-fn.h (first_commutative_argument): Declare.
7345 * internal-fn.c (first_commutative_argument): New function.
7346 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
7347 restrictions for pattern statements. Use first_commutative_argument
7348 to look for commutative operands in calls to internal functions.
7350 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
7352 * Makefile.in (wide-int-range.o): New.
7353 * tree-vrp.c: Move all the wide_int_* functions to...
7354 * wide-int-range.cc: ...here.
7355 * tree-vrp.h: Move all the wide_int_* prototypes to...
7356 * wide-int-range.h: ...here.
7358 2018-08-03 Tom de Vries <tdevries@suse.de>
7360 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
7362 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
7363 * except.c (output_function_exception_table): Do early exit if
7364 targetm_common.except_unwind_info (&global_options) == UI_NONE.
7366 2018-08-03 Martin Liska <mliska@suse.cz>
7368 * predict.c (dump_prediction): Change to 2 digits
7371 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
7373 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
7375 (falkor_am_1_gtov_gtov): ... a new insn reservation.
7377 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
7379 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
7380 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
7381 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
7382 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
7383 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
7385 2018-08-02 David Malcolm <dmalcolm@redhat.com>
7387 * diagnostic-show-locus.c (diagnostic_show_locus): Use
7388 pp_take_prefix when saving the existing prefix.
7389 * diagnostic.c (diagnostic_append_note): Likewise.
7390 * langhooks.c (lhd_print_error_function): Likewise.
7391 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
7392 param's type. Free the existing prefix.
7393 (pp_take_prefix): New function.
7394 (pretty_printer::pretty_printer): Drop the prefix parameter.
7395 Rename the length parameter to match the comment.
7396 (pretty_printer::~pretty_printer): Free the prefix.
7397 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
7399 (struct pretty_printer): Drop the "const" from "prefix" field's
7400 type and clarify memory management.
7401 (pp_set_prefix): Drop the "const" from the 2nd param.
7402 (pp_take_prefix): New decl.
7404 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
7406 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
7407 for word_mode_ok here instead of passing as argument.
7408 (expand_block_compare): Change select_block_compare_mode() call.
7409 (expand_strncmp_gpr_sequence): New function.
7410 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
7412 2018-08-02 Jeff Law <law@redhat.com>
7415 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7416 Define to speculation_safe_value_not_needed.
7419 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7420 Define to speculation_safe_value_not_needed.
7422 2018-08-02 Tom de Vries <tdevries@suse.de>
7425 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
7426 function. Return UI_TARGET unconditionally.
7427 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
7428 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
7430 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7432 * genemit.c (print_overload_test): Fix typo.
7434 2018-08-02 Richard Biener <rguenther@suse.de>
7436 PR tree-optimization/86816
7437 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
7438 which checks for value availability before querying it.
7439 (gvn_uses_equal): Use it.
7440 (same_succ_hash): Likewise.
7441 (gimple_equal_p): Likewise.
7443 2018-08-02 Nick Clifton <nickc@redhat.com>
7446 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7447 Define to speculation_safe_value_not_needed.
7450 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7451 Define to speculation_safe_value_not_needed.
7454 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7455 Define to speculation_safe_value_not_needed.
7458 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7459 Define to speculation_safe_value_not_needed.
7462 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7463 Define to speculation_safe_value_not_needed.
7466 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7467 Define to speculation_safe_value_not_needed.
7470 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7471 Define to speculation_safe_value_not_needed.
7474 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7475 Define to speculation_safe_value_not_needed.
7478 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
7479 speculation_safe_value_not_needed.
7482 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
7483 to speculation_safe_value_not_needed.
7485 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7487 * doc/md.texi: Expand the documentation of instruction names
7488 to mention port-local uses. Document '@' in pattern names.
7489 * read-md.h (overloaded_instance, overloaded_name): New structs.
7491 (md_reader::handle_overloaded_name): New member function.
7492 (md_reader::get_overloads): Likewise.
7493 (md_reader::m_first_overload): New member variable.
7494 (md_reader::m_next_overload_ptr): Likewise.
7495 (md_reader::m_overloads_htab): Likewise.
7496 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
7497 m_next_overload_ptr and m_overloads_htab.
7498 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
7499 (get_mode_token, get_code_token, get_int_token): New functions.
7500 (map_attr_string): Add an optional argument that passes back
7501 the associated iterator.
7502 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
7503 (md_reader::handle_overloaded_name, add_overload_instance): New
7505 (apply_iterators): Handle '@' names. Report an error if '@'
7506 is used without iterators.
7507 (initialize_iterators): Initialize the new iterator_group fields.
7508 * genopinit.c (handle_overloaded_code_for)
7509 (handle_overloaded_gen): New functions.
7510 (main): Use them to print declarations of maybe_code_for_* and
7511 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
7512 * genemit.c (print_overload_arguments, print_overload_test)
7513 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
7514 (main): Use it to print definitions of maybe_code_for_* and
7515 maybe_gen_* functions.
7516 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
7517 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
7518 instead of explicit mode checks.
7519 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
7520 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
7521 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
7522 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
7523 (aarch64_expand_compare_and_swap): Likewise
7524 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
7525 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
7526 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
7527 (aarch64_constant_pool_reload_icode): Delete.
7528 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
7529 instead of aarch64_constant_pool_reload_icode. Use
7530 code_for_aarch64_reload_mov instead of explicit mode checks.
7531 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
7532 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
7533 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
7534 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
7535 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
7536 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
7537 (aarch64_atomic_load_op_code): Delete.
7538 (aarch64_emit_atomic_load_op): Likewise.
7539 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
7540 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
7541 instead of aarch64_emit_atomic_load_op.
7542 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
7543 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
7544 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
7545 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
7546 character before the pattern name.
7547 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7548 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
7549 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
7550 (aarch64_frecps<mode>): Likewise.
7551 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
7552 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
7553 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
7554 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
7555 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
7557 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7559 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
7560 Allow HFmode constants if TARGET_FP_F16INST.
7562 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
7565 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
7566 No longer check last store for clobber of address register.
7568 2018-08-02 Martin Liska <mliska@suse.cz>
7570 PR gcov-profile/86817
7571 * gcov.c (process_all_functions): New function.
7573 (process_file): Move functions processing to
7574 process_all_functions.
7576 2018-08-02 David Malcolm <dmalcolm@redhat.com>
7578 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
7579 "const" to the "gimple *" and "rtx_insn *" parameters.
7580 * dumpfile.h (dump_user_location_t::dump_user_location_t):
7582 (dump_location_t::dump_location_t): Likewise.
7584 2018-08-01 Martin Sebor <msebor@redhat.com>
7586 PR tree-optimization/86650
7587 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
7588 rather than a "gcall *". Directly pass the data of interest
7589 to percent_K_format, rather than building a temporary CALL_EXPR
7591 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
7592 (gimple_fold_builtin_strncat): Adjust.
7593 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
7594 gcall* argument with gimple*.
7595 * gimple-ssa-warn-restrict.c (check_call): Same.
7596 (wrestrict_dom_walker::before_dom_children): Same.
7597 (builtin_access::builtin_access): Same.
7598 (check_bounds_or_overlap): Same
7599 (maybe_diag_overlap): Same.
7600 (maybe_diag_offset_bounds): Same.
7601 * tree-diagnostic.c (default_tree_printer): Move usage of
7602 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
7604 * tree-pretty-print.c (percent_K_format): Add argument.
7605 * tree-pretty-print.h: Add argument.
7606 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
7607 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
7608 (maybe_diag_stxncpy_trunc): Same.
7609 (handle_builtin_stxncpy): Same.
7610 (handle_builtin_strcat): Same.
7612 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7614 * match.pd: Optimise pointer range checks.
7616 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7618 PR tree-optimization/86758
7619 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
7620 to remove pattern statements.
7622 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7624 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
7625 result of dfs_enumerate_from when constructing stmt_vec_infos,
7626 instead of additionally calling get_loop_body.
7628 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7630 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
7632 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
7633 When creating an iv, assert that the step is not known to be zero.
7634 (vect_setup_realignment): Update call accordingly.
7635 * tree-vect-stmts.c (vectorizable_store): Likewise.
7636 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
7638 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7640 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
7641 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
7642 (vectorizable_reduction): Likewise.
7643 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
7644 (vect_detect_hybrid_slp_stmts): Likewise.
7645 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
7647 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
7649 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
7650 (wide_int_set_zero_nonzero_bits): ...this.
7651 (zero_nonzero_bits_from_vr): Rename to...
7652 (vrp_set_zero_nonzero_bits): ...this.
7653 (extract_range_from_multiplicative_op_1): Abstract wide int
7655 (wide_int_range_multiplicative_op): ...here.
7656 (extract_range_from_binary_expr_1): Extract wide int binary
7657 operations into their own functions.
7658 (wide_int_range_lshift): New.
7659 (wide_int_range_can_optimize_bit_op): New.
7660 (wide_int_range_shift_undefined_p): New.
7661 (wide_int_range_bit_xor): New.
7662 (wide_int_range_bit_ior): New.
7663 (wide_int_range_bit_and): New.
7664 (wide_int_range_trunc_mod): New.
7665 (extract_range_into_wide_ints): New.
7666 (vrp_shift_undefined_p): New.
7667 (extract_range_from_multiplicative_op): New.
7668 (vrp_can_optimize_bit_op): New.
7669 * tree-vrp.h (value_range::dump): New.
7670 (wide_int_range_multiplicative_op): New.
7671 (wide_int_range_lshift):New.
7672 (wide_int_range_shift_undefined_p): New.
7673 (wide_int_range_bit_xor): New.
7674 (wide_int_range_bit_ior): New.
7675 (wide_int_range_bit_and): New.
7676 (wide_int_range_trunc_mod): New.
7677 (zero_nonzero_bits_from_bounds): Rename to...
7678 (wide_int_set_zero_nonzero_bits): ...this.
7679 (zero_nonzero_bits_from_vr): Rename to...
7680 (vrp_set_zero_nonzero_bits): ...this.
7681 (range_easy_mask_min_max): Rename to...
7682 (wide_int_range_can_optimize_bit_op): this.
7683 * vr-values.c (simplify_bit_ops_using_ranges): Rename
7684 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
7686 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7688 * tree-vectorizer.h (vect_orig_stmt): New function.
7689 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
7690 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
7691 (vect_create_epilog_for_reduction): Likewise.
7692 (vectorizable_live_operation): Likewise.
7693 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
7694 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
7695 * tree-vect-stmts.c (vectorizable_call): Likewise.
7696 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
7698 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7700 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
7702 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
7703 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
7704 (vect_transform_loop): Likewise.
7705 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7707 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7709 * tree-vectorizer.h (vect_schedule_slp): Return void.
7710 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7711 (vect_schedule_slp): Likewise.
7713 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7715 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
7717 (vect_transform_loop): Update calls accordingly. Schedule SLP
7718 instances before the main loop, if any exist.
7720 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7722 PR tree-optimization/86749
7723 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
7724 If the lhs is used in a COND_EXPR, check that it is being used
7725 as the "then" or "else" value.
7727 2018-08-01 Tom de Vries <tdevries@suse.de>
7730 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
7731 speculation_safe_value_not_needed.
7733 2018-08-01 Richard Biener <rguenther@suse.de>
7735 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
7738 2018-08-01 Martin Liska <mliska@suse.cz>
7740 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
7741 format how successful transformation is dumped.
7742 (gimple_mod_pow2_value_transform): Likewise.
7743 (gimple_mod_subtract_transform): Likewise.
7744 (gimple_stringops_transform): Likewise.
7746 2018-08-01 Martin Liska <mliska@suse.cz>
7749 * value-prof.c (interesting_stringop_to_profile_p):
7750 Simplify the code and add BUILT_IN_MEMMOVE.
7751 (gimple_stringops_transform): Likewise.
7753 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
7755 * config/aarch64/aarch64-simd.md
7756 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
7757 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
7758 use GPI iterator instead of SI mode.
7760 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
7762 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
7763 rs6000_speculation_barrier.
7764 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
7765 new barrier pattern name.
7767 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
7769 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
7770 (speculation_barrier): New insn.
7772 2018-08-01 Richard Biener <rguenther@suse.de>
7775 * graphite.h: Include isl/id.h and isl/space.h to allow build
7778 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
7781 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
7782 mode for COFF targets.
7783 * defaults.h (TARGET_COFF): Define.
7784 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
7785 TARGET_COFF): Define.
7786 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
7787 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
7788 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
7790 2018-07-31 Alexandre Oliva <oliva@adacore.com>
7792 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
7793 * gimple-streamer-out.c (output_bb): Save it.
7794 * lto-streamer-in.c (input_struct_function_base): Restore
7795 instance discriminator if available. Create map on demand.
7796 * lto-streamer-out.c (output_struct_function_base): Save it if
7798 * final.c (decl_to_instance_map): Document LTO strategy.
7800 2018-07-31 Alexandre Oliva <oliva@adacore.com>
7801 Olivier Hainque <hainque@adacore.com>
7803 * debug.h (decl_to_instance_map_t): New type.
7804 (decl_to_instance_map): Declare.
7805 (maybe_create_decl_to_instance_map): New inline function.
7806 * final.c (bb_discriminator, last_bb_discriminator): New statics,
7807 to track basic block discriminators.
7808 (final_start_function_1): Initialize them.
7809 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
7811 (decl_to_instance_map): New variable.
7812 (map_decl_to_instance, maybe_set_discriminator): New functions.
7813 (notice_source_line): Set discriminator.
7815 2018-07-31 Ian Lance Taylor <iant@golang.org>
7817 * targhooks.c (default_have_speculation_safe_value): Add
7820 2018-07-31 David Malcolm <dmalcolm@redhat.com>
7822 * dump-context.h: Include "pretty-print.h".
7823 (dump_context::refresh_dumps_are_enabled): New decl.
7824 (dump_context::emit_item): New decl.
7825 (class dump_context): Add fields "m_test_pp" and
7827 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
7828 (temp_dump_context::get_dumped_text): New decl.
7829 (class temp_dump_context): Add field "m_pp".
7830 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
7831 (dump_context::refresh_dumps_are_enabled): ...and add a test for
7833 (set_dump_file): Update for above change.
7834 (set_alt_dump_file): Likewise.
7835 (dump_loc): New overload, taking a pretty_printer *.
7836 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
7837 to any test pretty-printer.
7838 (make_item_for_dump_gimple_stmt): New function, adapted from
7839 optinfo::add_gimple_stmt.
7840 (dump_context::dump_gimple_stmt): Call it, and use the result,
7841 eliminating the direct usage of dump_file and alt_dump_file in
7842 favor of indirectly using them via emit_item.
7843 (make_item_for_dump_gimple_expr): New function, adapted from
7844 optinfo::add_gimple_expr.
7845 (dump_context::dump_gimple_expr): Call it, and use the result,
7846 eliminating the direct usage of dump_file and alt_dump_file in
7847 favor of indirectly using them via emit_item.
7848 (make_item_for_dump_generic_expr): New function, adapted from
7850 (dump_context::dump_generic_expr): Call it, and use the result,
7851 eliminating the direct usage of dump_file and alt_dump_file in
7852 favor of indirectly using them via emit_item.
7853 (make_item_for_dump_printf_va): New function, adapted from
7854 optinfo::add_printf_va.
7855 (make_item_for_dump_printf): New function.
7856 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
7857 and use the result, eliminating the direct usage of dump_file and
7858 alt_dump_file in favor of indirectly using them via emit_item.
7859 (make_item_for_dump_dec): New function.
7860 (dump_context::dump_dec): Call it, and use the result,
7861 eliminating the direct usage of dump_file and alt_dump_file in
7862 favor of indirectly using them via emit_item.
7863 (make_item_for_dump_symtab_node): New function, adapted from
7864 optinfo::add_symtab_node.
7865 (dump_context::dump_symtab_node): Call it, and use the result,
7866 eliminating the direct usage of dump_file and alt_dump_file in
7867 favor of indirectly using them via emit_item.
7868 (dump_context::begin_scope): Reimplement, avoiding direct usage
7869 of dump_file and alt_dump_file in favor of indirectly using them
7871 (dump_context::emit_item): New member function.
7872 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
7873 Set up test pretty-printer on the underlying context. Call
7874 refresh_dumps_are_enabled.
7875 (temp_dump_context::~temp_dump_context): Call
7876 refresh_dumps_are_enabled.
7877 (temp_dump_context::get_dumped_text): New member function.
7878 (selftest::verify_dumped_text): New function.
7879 (ASSERT_DUMPED_TEXT_EQ): New macro.
7880 (selftest::test_capture_of_dump_calls): Run all tests twice, with
7881 and then without optinfo enabled. Add uses of
7882 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
7883 * dumpfile.h: Update comment for the dump_* API.
7884 * optinfo-emit-json.cc
7885 (selftest::test_building_json_from_dump_calls): Update for new
7886 param for temp_dump_context ctor.
7887 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
7888 and "m_owned" field.
7889 (optinfo_item::~optinfo_item): Likewise.
7890 (optinfo::add_item): New member function.
7891 (optinfo::emit): Update comment.
7892 (optinfo::add_string): Delete.
7893 (optinfo::add_printf): Delete.
7894 (optinfo::add_printf_va): Delete.
7895 (optinfo::add_gimple_stmt): Delete.
7896 (optinfo::add_gimple_expr): Delete.
7897 (optinfo::add_tree): Delete.
7898 (optinfo::add_symtab_node): Delete.
7899 (optinfo::add_dec): Delete.
7900 * optinfo.h (class dump_context): New forward decl.
7901 (optinfo::add_item): New decl.
7902 (optinfo::add_string): Delete.
7903 (optinfo::add_printf): Delete.
7904 (optinfo::add_printf_va): Delete.
7905 (optinfo::add_gimple_stmt): Delete.
7906 (optinfo::add_gimple_expr): Delete.
7907 (optinfo::add_tree): Delete.
7908 (optinfo::add_symtab_node): Delete.
7909 (optinfo::add_dec): Delete.
7910 (optinfo::add_poly_int): Delete.
7911 (optinfo_item::optinfo_item): Remove "owned" param.
7912 (class optinfo_item): Remove field "m_owned".
7914 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
7917 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
7918 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
7919 requested variable alignment.
7920 (expand_one_ssa_partition): Likewise.
7921 (expand_one_var): Likewise.
7923 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7925 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
7926 to speculation_safe_value_not_needed.
7928 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7930 * targhooks.h (speculation_safe_value_not_needed): New prototype.
7931 * targhooks.c (speculation_safe_value_not_needed): New function.
7932 * target.def (have_speculation_safe_value): Update documentation.
7933 * doc/tm.texi: Regenerated.
7935 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7937 * config/aarch64/iterators.md (ALLI_TI): New iterator.
7938 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
7940 (despeculate_copy<ALLI:mode>_insn): New insn.
7941 (despeculate_copyti_insn): New insn.
7942 (despeculate_simple<ALLI:mode>): New insn
7943 (despeculate_simpleti): New insn.
7944 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
7946 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
7947 aarch64_speculation_safe_value.
7948 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
7950 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7952 * config/aarch64/aarch64-speculation.cc: New file.
7953 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
7954 before pass_reorder_blocks.
7955 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
7957 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
7958 X14 and X15 when tracking speculation.
7959 * config/aarch64/aarch64.md (register name constants): Add
7960 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
7961 (unspec): Add UNSPEC_SPECULATION_TRACKER.
7962 (speculation_barrier): New insn attribute.
7963 (cmp<mode>): Allow SP in comparisons.
7964 (speculation_tracker): New insn.
7965 (speculation_barrier): Add speculation_barrier attribute.
7966 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
7967 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
7968 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
7970 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7972 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
7973 aarch64_track_speculation is true.
7974 (tb<optab><mode>1): Likewise.
7975 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
7976 generate CB[N]Z when tracking speculation.
7977 (aarch64_split_compare_and_swap): Likewise.
7978 (aarch64_split_atomic_op): Likewise.
7980 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7982 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
7984 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7986 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
7987 (speculation_barrier): New insn.
7989 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7991 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
7992 * config/arm/arm.md (speculation_barrier): New expand.
7993 (speculation_barrier_insn): New pattern.
7995 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
7997 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
7998 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
7999 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
8000 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
8002 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
8003 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
8004 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
8005 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
8006 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
8007 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
8008 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
8009 * builtins.c (expand_speculation_safe_value): New function.
8010 (expand_builtin): Call it.
8011 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
8012 * doc/extend.texi: Document __builtin_speculation_safe_value.
8013 * doc/md.texi: Document "speculation_barrier" pattern.
8014 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
8015 TARGET_HAVE_SPECULATION_SAFE_VALUE.
8016 * doc/tm.texi: Regenerated.
8017 * target.def (have_speculation_safe_value, speculation_safe_value): New
8019 * targhooks.c (default_have_speculation_safe_value): New function.
8020 (default_speculation_safe_value): New function.
8021 * targhooks.h (default_have_speculation_safe_value): Add prototype.
8022 (default_speculation_safe_value): Add prototype.
8024 2018-07-31 David Malcolm <dmalcolm@redhat.com>
8026 * dump-context.h (dump_context::dump_loc): New decl.
8027 * dumpfile.c (dump_context::dump_loc): New member function.
8028 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
8029 and dump_gimple_stmt.
8030 (dump_context::dump_gimple_expr_loc): Likewise, using
8032 (dump_context::dump_generic_expr_loc): Likewise, using
8034 (dump_context::dump_printf_loc_va): Likewise, using
8036 (dump_context::begin_scope): Explicitly using the global function
8037 "dump_loc", rather than the member function.
8039 2018-07-31 Martin Sebor <msebor@redhat.com>
8041 PR tree-optimization/86741
8042 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
8044 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
8046 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
8047 depend on whether prefetch instructions will be emitted or not.
8048 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
8049 will be emitted or not.
8050 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
8051 (TARGET_SETMEM_PFD): New macros.
8053 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8055 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
8056 (NULL_STMT_VEC_INFO): Delete.
8057 (stmt_vec_info::operator*): Likewise.
8058 (stmt_vec_info::operator gimple *): Likewise.
8059 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
8060 of NULL_STMT_VEC_INFO.
8061 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8062 (vect_reassociating_reduction_p): Likewise.
8063 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
8064 (vectorizable_store): Likewise.
8065 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
8066 (vec_info::free_stmt_vec_infos): Likewise.
8068 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8070 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
8071 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
8073 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8075 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
8076 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
8077 (vec_info::free_stmt_vec_info): New private member functions.
8078 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
8079 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
8080 * tree-parloops.c (gather_scalar_reductions): Remove calls to
8081 set_stmt_vec_info_vec and free_stmt_vec_infos.
8082 * tree-vect-loop.c (_loop_vec_info): Remove call to
8083 set_stmt_vec_info_vec.
8084 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
8085 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
8086 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
8087 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
8088 (vec_info::free_stmt_vec_info): ...these new functions. Remove
8089 assignments in {vec_info::,}new_stmt_vec_info that are redundant
8090 with the clearing in the xcalloc.
8091 (stmt_vec_info_vec): Delete.
8092 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
8093 (vectorize_loops): Likewise.
8094 (vec_info::~vec_info): Remove argument from call to
8095 free_stmt_vec_infos.
8096 (vec_info::add_stmt): Remove vinfo argument from call to
8099 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8101 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
8102 rather than a gimple stmt.
8103 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
8104 information for pattern statements when passed the original
8105 statement; instead wait to be passed the pattern statement itself.
8106 Don't call set_vinfo_for_stmt here.
8107 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
8108 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
8109 stmt_vec_infos here.
8110 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
8111 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
8112 stmt_vec_infos entry.
8114 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8116 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
8117 * tree-vectorizer.c (vec_info::replace_stmt): New function.
8118 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
8119 * tree-vect-stmts.c (vectorizable_call): Likewise.
8120 (vectorizable_simd_clone_call): Likewise.
8122 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8124 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
8125 * tree-vectorizer.c (vec_info::remove_stmt): New function.
8126 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
8127 * tree-vect-loop.c (vect_transform_loop): Likewise.
8128 * tree-vect-slp.c (vect_schedule_slp): Likewise.
8129 * tree-vect-stmts.c (vect_remove_stores): Likewise.
8131 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8133 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
8134 (vect_dr_stmt): Delete.
8135 * tree-vectorizer.c (vec_info::lookup_dr): New function.
8136 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
8138 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
8139 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
8140 (vect_verify_datarefs_alignment, vect_peeling_supportable)
8141 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
8142 (vect_analyze_data_refs): Likewise.
8143 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
8145 (vect_find_same_alignment_drs): Likewise.
8146 (vect_slp_analyze_node_dependences): Update calls accordingly.
8147 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
8148 instead of DR_VECT_AUX.
8149 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
8150 of a vector data references. Use vec_info::lookup_dr instead of
8152 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
8153 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
8154 instead of DR_VECT_AUX.
8156 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8158 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
8160 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
8162 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8163 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
8164 (vect_gen_prolog_loop_niters): Likewise.
8166 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8168 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
8169 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
8170 (vect_known_alignment_in_bytes, vect_dr_behavior)
8171 (vect_get_scalar_dr_size): Take references as dr_vec_infos
8172 instead of data_references. Update calls to other routines for
8173 which the same change has been made.
8174 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
8175 dr_vec_infos instead of stmt_vec_infos.
8176 (vect_analyze_data_ref_dependence): Update call accordingly.
8177 (vect_slp_analyze_data_ref_dependence)
8178 (vect_record_base_alignments): Use DR_VECT_AUX.
8179 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
8180 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
8181 (vector_alignment_reachable_p, vect_get_data_access_cost)
8182 (vect_peeling_supportable, vect_analyze_group_access_1)
8183 (vect_analyze_group_access, vect_analyze_data_ref_access)
8184 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
8185 (vect_compile_time_alias, vect_small_gap_p)
8186 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
8187 (vect_supportable_dr_alignment): Take references as dr_vec_infos
8188 instead of data_references. Update calls to other routines for
8189 which the same change has been made.
8190 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
8191 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
8192 (vect_slp_analyze_and_verify_node_alignment)
8193 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
8194 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
8195 (vect_setup_realignment): Use dr_vec_infos. Update calls after
8197 (_vect_peel_info::dr): Replace with...
8198 (_vect_peel_info::dr_info): ...this new field.
8199 (vect_peeling_hash_get_most_frequent)
8200 (vect_peeling_hash_choose_best_peeling): Update accordingly.
8201 (vect_peeling_hash_get_lowest_cost):
8202 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
8203 routines for which the same change has been made.
8204 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
8206 * tree-vect-loop-manip.c (get_misalign_in_elems)
8207 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
8209 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8210 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
8211 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
8212 (get_group_load_store_type, get_negative_load_store_type)
8213 (vect_get_data_ptr_increment, vectorizable_store)
8214 (vectorizable_load): Likewise.
8215 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
8216 Update calls to other routines for which the same change has been made.
8218 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8220 * tree-vectorizer.h (vec_info::move_dr): New member function.
8221 (dataref_aux): Rename to...
8222 (dr_vec_info): ...this and add "dr" and "stmt" fields.
8223 (_stmt_vec_info::dr_aux): Update accordingly.
8224 (_stmt_vec_info::data_ref_info): Delete.
8225 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
8226 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
8227 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
8228 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
8230 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
8231 (STMT_VINFO_DR_INFO): New macro.
8232 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
8233 (set_dr_misalignment): Update after rename of dataref_aux.
8234 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
8235 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
8236 initialization of STMT_VINFO_DATA_REF.
8237 * tree-vectorizer.c (vec_info::move_dr): New function.
8238 * tree-vect-patterns.c (vect_recog_bool_pattern)
8239 (vect_recog_mask_conversion_pattern)
8240 (vect_recog_gather_scatter_pattern): Use it.
8241 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
8242 the "dr" and "stmt" fields of dr_vec_info instead of
8243 STMT_VINFO_DATA_REF.
8245 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8247 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
8248 (is_pattern_stmt_p): Use it.
8249 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
8250 on pattern statements.
8252 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8254 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
8255 original stmt as a stmt_vec_info rather than a gimple stmt.
8256 (vect_pattern_recog_1): Take the statement directly as a
8257 stmt_vec_info, rather than via a gimple_stmt_iterator.
8258 Update call to vect_mark_pattern_stmts.
8259 (vect_pattern_recog): Update calls accordingly.
8261 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8263 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
8264 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
8265 a vect_def_type for the first argument.
8266 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
8267 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
8268 operand if it isn't defined by a vectorized statement.
8269 (vect_build_gather_load_calls): Remove the mask_dt argument and
8270 update calls to vect_get_vec_def_for_stmt_copy.
8271 (vectorizable_bswap): Likewise the dt argument.
8272 (vectorizable_call): Update calls to vectorizable_bswap and
8273 vect_get_vec_def_for_stmt_copy.
8274 (vectorizable_simd_clone_call, vectorizable_assignment)
8275 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
8276 (vectorizable_comparison): Update calls to
8277 vect_get_vec_def_for_stmt_copy.
8278 (vectorizable_store): Likewise. Remove now-unnecessary calls to
8280 (vect_get_loop_based_defs): Remove dt argument and update call
8281 to vect_get_vec_def_for_stmt_copy.
8282 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
8283 and vect_get_vec_def_for_stmt_copy.
8284 (vectorizable_load): Update calls to vect_build_gather_load_calls
8285 and vect_get_vec_def_for_stmt_copy.
8286 * tree-vect-loop.c (vect_create_epilog_for_reduction)
8287 (vectorizable_reduction, vectorizable_live_operation): Update calls
8288 to vect_get_vec_def_for_stmt_copy.
8290 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8292 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
8293 and gimple stmt arguments with a stmt_vec_info.
8294 (vect_record_base_alignments): Update calls accordingly.
8295 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
8296 and gimple stmt arguments with a stmt_vec_info.
8297 (vect_build_slp_tree_1): Remove vinfo argument and update call
8298 to vect_record_max_nunits.
8299 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
8300 and vect_record_max_nunits.
8302 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8304 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
8305 file and take a stmt_vec_info instead of a gimple stmt.
8306 (supportable_widening_operation, vect_finish_replace_stmt)
8307 (vect_finish_stmt_generation, vect_get_store_rhs)
8308 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
8309 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
8310 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
8311 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
8312 (vect_create_data_ref_ptr, bump_vector_ptr)
8313 (vect_permute_store_chain, vect_setup_realignment)
8314 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
8315 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
8316 (vectorizable_reduction, vectorizable_induction)
8317 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
8318 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
8319 than gimple stmts as arguments.
8320 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
8321 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
8322 (can_group_stmts_p, vect_check_gather_scatter)
8323 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
8324 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
8325 (vect_permute_load_chain, vect_shift_permute_load_chain)
8326 (vect_transform_grouped_load)
8327 (vect_record_grouped_load_vectors): Likewise.
8328 * tree-vect-loop.c (vect_fixup_reduc_chain)
8329 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
8330 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
8331 (vectorizable_reduction, vectorizable_induction)
8332 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
8333 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
8334 (vect_get_load_store_mask): Likewise.
8335 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
8336 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
8337 * tree-vect-stmts.c (vect_mark_relevant)
8338 (is_simple_and_all_uses_invariant)
8339 (exist_non_indexing_operands_for_use_p, process_use)
8340 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
8341 (vect_get_vec_def_for_operand, vect_get_vec_defs)
8342 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
8343 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
8344 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
8345 (get_negative_load_store_type, get_load_store_type)
8346 (vect_check_load_store_mask, vect_check_store_rhs)
8347 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
8348 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
8349 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
8350 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8351 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
8352 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
8353 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
8354 (supportable_widening_operation): Likewise.
8356 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8358 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
8359 a stmt_vec_info instead of a gcall.
8360 (vect_check_gather_scatter): Update call accordingly.
8361 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
8363 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
8364 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
8365 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
8366 instead of a gimple stmt.
8367 (vect_transform_loop): Update calls accordingly.
8368 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
8369 stmt_vec_infos instead of gimple stmts.
8370 (vect_analyze_slp_instance): Update use accordingly.
8371 * tree-vect-stmts.c (read_vector_array, write_vector_array)
8372 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
8373 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
8374 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
8375 (vect_gen_widened_results_half, vect_get_loop_based_defs)
8376 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
8377 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
8378 down to subroutines.
8380 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8382 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
8383 of the worklist from a vector of gimple stmts to a vector of
8385 * tree-vect-stmts.c (vect_mark_relevant, process_use)
8386 (vect_mark_stmts_to_be_vectorized): Likewise
8388 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8390 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
8391 statement before passing it to vect_analyze_stmt.
8392 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
8393 the chain of phi vector definitions. Track the exit phi via its
8395 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
8396 STMT_VINFO_REDUC_DEF.
8397 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
8398 stmt_vec_infos to handle the statement chains.
8399 (vect_get_slp_defs): Record the first statement in the node
8400 using a stmt_vec_info.
8401 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
8402 statements here and pass their stmt_vec_info down to subroutines.
8403 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
8404 down to vect_finish_stmt_generation.
8405 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
8406 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
8407 stmt_vec_infos to subroutines.
8408 (vect_remove_stores): Use stmt_vec_infos to handle the statement
8411 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8413 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
8414 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
8415 (vect_permute_store_chain, vect_setup_realignment)
8416 (vect_permute_load_chain, vect_shift_permute_load_chain)
8417 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
8418 stmts internally, and when passing values to other vectorizer routines.
8419 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
8420 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
8421 (vect_analyze_loop_operations, get_initial_def_for_reduction)
8422 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
8423 (vectorizable_reduction, vectorizable_induction)
8424 (vectorizable_live_operation, vect_transform_loop_stmt)
8425 (vect_transform_loop): Likewise.
8426 * tree-vect-patterns.c (vect_reassociating_reduction_p)
8427 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
8428 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
8429 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8430 (vect_slp_analyze_node_operations_1): Likewise.
8431 * tree-vect-stmts.c (vect_mark_relevant, process_use)
8432 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
8433 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
8434 (vect_finish_stmt_generation_1, get_group_load_store_type)
8435 (get_load_store_type, vect_build_gather_load_calls)
8436 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
8437 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
8438 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8439 (vectorizable_store, vectorizable_load, vectorizable_condition)
8440 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
8441 (supportable_widening_operation): Likewise.
8442 (vect_get_vector_types_for_stmt): Likewise.
8443 * tree-vectorizer.h (vect_dr_behavior): Likewise.
8445 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8447 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
8448 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
8449 (vect_permute_store_chain, vect_permute_load_chain)
8450 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
8451 repeated stmt_vec_info lookups.
8452 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
8453 (vect_update_ivs_after_vectorizer): Likewise.
8454 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
8455 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
8456 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
8457 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8458 (vect_bb_slp_scalar_cost): Likewise.
8459 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
8461 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8463 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
8464 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
8465 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
8466 to get gassigns and gcalls, rather than operating on generc gimple
8468 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
8469 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
8470 (vectorizable_load, vect_analyze_stmt): Likewise.
8471 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
8473 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8475 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
8476 return stmt_vec_infos rather than gimple stmts. Do not accept
8478 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
8480 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
8481 Update use of get_later_stmt.
8482 (vect_get_constant_vectors): Update call accordingly.
8483 (vect_schedule_slp_instance): Likewise
8484 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
8485 (vect_slp_analyze_instance_dependence): Likewise.
8486 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
8488 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8490 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
8491 (stmt_info_for_cost::stmt_info): ...this new field.
8492 (add_stmt_costs): Update accordingly.
8493 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
8494 (vect_get_known_peeling_cost): Likewise.
8495 (vect_estimate_min_profitable_iters): Likewise.
8496 * tree-vect-stmts.c (record_stmt_cost): Likewise.
8498 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8500 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
8501 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
8502 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
8504 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
8506 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8508 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
8509 a gimple stmt to a stmt_vec_info.
8510 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
8512 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8514 * tree-vectorizer.h (vec_info::grouped_stores): Change from
8515 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
8516 (_loop_vec_info::reduction_chains): Likewise.
8517 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
8519 * tree-vect-slp.c (vect_analyze_slp): Likewise.
8521 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8523 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
8524 a gimple stmt to a stmt_vec_info.
8525 (_stmt_vec_info::next_element): Likewise.
8526 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
8527 (vect_slp_analyze_and_verify_node_alignment)
8528 (vect_analyze_group_access_1, vect_analyze_group_access)
8529 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
8530 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
8531 (vect_supportable_dr_alignment): Update accordingly.
8532 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
8533 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
8534 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
8535 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
8536 * tree-vect-slp.c (vect_build_slp_tree_1)
8537 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
8538 (vect_split_slp_store_group, vect_analyze_slp_instance)
8539 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
8540 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
8541 (get_group_load_store_type, get_load_store_type)
8542 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
8543 (vect_transform_stmt, vect_remove_stores): Likewise.
8545 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8547 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
8549 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
8550 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
8551 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
8552 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
8553 (vector_alignment_reachable_p, vect_get_data_access_cost)
8554 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
8555 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
8556 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
8557 (vect_analyze_group_access_1, vect_analyze_group_access)
8558 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
8559 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
8560 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
8561 result of vect_dr_stmt and use the stmt_vec_info instead of
8562 the associated gimple stmt.
8563 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
8564 (vect_gen_prolog_loop_niters): Likewise.
8565 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8567 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8569 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
8570 to a vec<stmt_vec_info>.
8571 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
8572 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
8574 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
8575 to a vec<stmt_vec_info>.
8576 (bst_traits::value_type, bst_traits::value_type): Likewise.
8577 (bst_traits::hash): Update accordingly.
8578 (vect_get_and_check_slp_defs): Change the stmts parameter from
8579 a vec<gimple *> to a vec<stmt_vec_info>.
8580 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
8581 (vect_build_slp_tree): Likewise.
8582 (vect_build_slp_tree_2): Likewise. Update uses of
8583 SLP_TREE_SCALAR_STMTS.
8584 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
8585 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
8586 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
8587 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
8588 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
8589 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
8590 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
8591 (vect_get_constant_vectors, vect_get_slp_defs)
8592 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
8593 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
8594 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
8595 instead of gimple stmts.
8596 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
8597 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
8598 (vect_slp_analyze_instance_dependence): Update uses of
8599 SLP_TREE_SCALAR_STMTS.
8600 (vect_slp_analyze_and_verify_node_alignment): Likewise.
8601 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
8602 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
8603 (get_initial_defs_for_reduction): Likewise.
8604 (vect_create_epilog_for_reduction): Likewise.
8605 (vectorize_fold_left_reduction): Likewise.
8606 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
8607 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
8608 (can_vectorize_live_stmts): Likewise.
8610 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8612 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
8613 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
8614 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
8616 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
8618 (gather_scalar_reductions): Update after above interface changes.
8619 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
8620 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
8622 (vect_force_simple_reduction): Likewise.
8623 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
8624 LOOP_VINFO_REDUCTIONS.
8625 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8627 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8629 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
8630 a gimple stmt to a stmt_vec_info.
8631 * tree-vect-loop.c (vect_active_double_reduction_p)
8632 (vect_force_simple_reduction, vectorizable_reduction): Update
8635 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8637 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
8638 vec<gimple *> to a vec<stmt_vec_info>.
8639 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
8640 the reduction_phis argument from a vec<gimple *> to a
8642 (vectorizable_reduction): Likewise the phis local variable that
8643 is passed to vect_create_epilog_for_reduction. Update for new type
8644 of SLP_TREE_VEC_STMTS.
8645 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
8646 (vectorizable_live_operation): Likewise.
8647 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
8648 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
8650 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8652 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
8653 a gimple stmt to a stmt_vec_info.
8654 (vectorizable_condition, vectorizable_live_operation)
8655 (vectorizable_reduction, vectorizable_induction): Pass back the
8656 vectorized statement as a stmt_vec_info.
8657 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
8658 use of STMT_VINFO_VEC_STMT.
8659 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
8660 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
8661 as stmt_vec_infos rather than gimple stmts.
8662 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
8664 (vectorizable_live_operation): Likewise.
8665 (vectorizable_reduction, vectorizable_induction): Likewise,
8666 updating use of STMT_VINFO_VEC_STMT.
8667 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
8668 of STMT_VINFO_VEC_STMT.
8669 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
8670 (vectorizable_simd_clone_call, vectorizable_conversion)
8671 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8672 (vectorizable_store, vectorizable_load, vectorizable_condition)
8673 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
8674 from a gimple stmt to a stmt_vec_info.
8675 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
8676 pointer to a stmt_vec_info to the vectorizable_* routines.
8678 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8680 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
8681 a gimple stmt to a stmt_vec_info.
8682 (is_pattern_stmt_p): Update accordingly.
8683 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
8684 (vect_record_grouped_load_vectors): Likewise.
8685 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
8686 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
8687 (vect_model_reduction_cost): Likewise.
8688 (vect_create_epilog_for_reduction): Likewise.
8689 (vectorizable_reduction, vectorizable_induction): Likewise.
8690 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8691 Return the stmt_vec_info for the pattern statement.
8692 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
8693 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
8694 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
8695 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
8696 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
8697 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
8698 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
8699 (free_stmt_vec_info, vect_is_simple_use): Likewise.
8701 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8703 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
8704 (vect_finish_stmt_generation): Likewise.
8705 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
8706 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
8707 (vect_build_gather_load_calls): Use the return value of the above
8708 functions instead of a separate call to vinfo_for_stmt. Use narrow
8709 scopes for the input gimple stmt and wider scopes for the associated
8710 stmt_vec_info. Use vec_info::lookup_def when setting these
8711 stmt_vec_infos from an SSA_NAME definition.
8712 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
8713 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
8714 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8715 (vectorizable_store, vectorizable_load, vectorizable_condition)
8716 (vectorizable_comparison): Likewise.
8717 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
8718 (vectorizable_reduction): Likewise.
8720 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8722 * tree-vectorizer.h (vect_is_simple_use): Add an optional
8723 stmt_vec_info * parameter before the optional gimple **.
8724 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
8725 (process_use, vect_get_vec_def_for_operand_1): Update callers.
8726 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
8727 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8728 (vectorizable_live_operation): Likewise.
8729 * tree-vect-patterns.c (type_conversion_p): Likewise.
8730 (vect_look_through_possible_promotion): Likewise.
8731 (vect_recog_rotate_pattern): Likewise.
8732 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
8734 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8736 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
8737 a typedef to a wrapper class.
8738 (NULL_STMT_VEC_INFO): New macro.
8739 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
8740 (stmt_vec_info::operator*): New function.
8741 (stmt_vec_info::operator gimple *): Likewise.
8742 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
8743 (add_stmt_costs): Likewise.
8744 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
8745 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
8746 (vect_get_known_peeling_cost): Likewise.
8747 (vect_estimate_min_profitable_iters): Likewise.
8748 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8749 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
8750 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
8751 (vectorizable_store, free_stmt_vec_infos): Likewise.
8752 (new_stmt_vec_info): Change return type of xcalloc to
8755 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8757 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
8758 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
8759 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
8760 a single_imm_use-based sequence.
8761 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
8763 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8765 * tree-vectorizer.h (vec_info::lookup_def): Declare.
8766 * tree-vectorizer.c (vec_info::lookup_def): New function.
8767 * tree-vect-patterns.c (vect_get_internal_def): Use it.
8768 (vect_widened_op_tree): Likewise.
8769 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
8770 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
8771 (vectorizable_reduction): Likewise.
8772 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
8774 (vect_is_slp_reduction): Update calls accordingly. Use
8775 vec_info::lookup_def.
8776 (vect_is_simple_reduction): Likewise
8777 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
8779 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8781 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
8782 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
8783 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
8785 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
8786 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
8787 (vect_update_vf_for_slp, vect_analyze_loop_operations)
8788 (vect_is_slp_reduction, vectorizable_induction)
8789 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
8790 * tree-vect-patterns.c (vect_init_pattern_stmt):
8791 (vect_determine_min_output_precision_1, vect_determine_precisions)
8792 (vect_pattern_recog): Likewise.
8793 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
8794 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
8795 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
8796 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
8797 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
8798 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
8799 info field from a loop to a loop_vec_info.
8801 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8803 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
8804 (vec_info::add_stmt): Declare.
8805 * tree-vectorizer.c (vec_info::add_stmt): New function.
8806 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
8807 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
8808 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
8809 (vectorizable_induction): Likewise.
8810 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
8811 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
8812 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
8813 (vectorizable_load): Likewise.
8814 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8815 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
8816 (vect_recog_gather_scatter_pattern): Likewise.
8817 (append_pattern_def_seq): Likewise. Remove a check that is
8818 performed by add_stmt itself.
8820 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8822 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
8823 which make_ssa_name was called with new_stmt before new_stmt
8826 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8828 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
8830 (vect_is_slp_reduction): ...here...
8831 (vect_is_simple_reduction): ...and here. Remove repetition of tests
8832 that are already known to be false.
8834 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8836 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
8837 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
8838 STMT_VINFO_NUM_SLP_USES when it's true.
8839 (vect_free_slp_instance): Add a final_p parameter and pass it to
8841 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
8842 (vect_analyze_slp_instance): Likewise.
8843 (vect_slp_analyze_operations): Likewise.
8844 (vect_slp_analyze_bb_1): Likewise.
8845 * tree-vectorizer.c (vec_info): Likewise.
8846 * tree-vect-loop.c (vect_transform_loop): Likewise.
8848 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8850 * tree-vect-loop.c (vectorizable_reduction): Assert that the
8851 function is not called for second and subsequent members of
8854 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8856 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
8857 cases for nested loops from here to ...
8858 (vect_create_epilog_for_reduction): ...here. Only call
8859 vect_is_simple_use for inner-loop reductions.
8861 2018-07-31 Martin Liska <mliska@suse.cz>
8863 PR gcov-profile/85338
8864 PR gcov-profile/85350
8865 PR gcov-profile/85372
8866 * profile.c (struct location_triplet): New.
8867 (struct location_triplet_hash): Likewise.
8868 (output_location): Do not output a BB that
8869 is already recorded for a line.
8870 (branch_prob): Use streamed_locations.
8872 2018-07-31 Martin Liska <mliska@suse.cz>
8874 PR gcov-profile/85370
8875 * coverage.c (coverage_begin_function): Do not mark target
8876 clones as artificial functions.
8878 2018-07-31 Martin Liska <mliska@suse.cz>
8880 PR gcov-profile/83813
8881 PR gcov-profile/84758
8882 PR gcov-profile/85217
8883 PR gcov-profile/85332
8884 * profile.c (branch_prob): Do not record GOTO expressions
8885 for GIMPLE statements which locations are already streamed.
8887 2018-07-31 Olivier Hainque <hainque@adacore.com>
8889 * gcc.c (handle_spec_function): Accept a soft_matched_part
8890 argument, as do_spec_1. Pass it down to ...
8891 (eval_spec_function): Accept a soft_matched_part argument,
8892 and pass it down to ...
8893 (do_spec_2): Accept a soft_matched_part argument, and pass
8894 it down to do_spec_1.
8895 (do_spec_1): Pass soft_matched_part to handle_spec_function.
8896 (handle_braces): Update call to handle_spec_function.
8897 (driver::set_up_specs): Update calls to do_spec_2.
8898 (compare_debug_dump_opt_spec_function): Likewise.
8899 (compare_debug_self_opt_spec_function): Likewise.
8901 2018-07-31 Olivier Hainque <hainque@adacore.com>
8903 * common.opt (nolibc): New option.
8904 * doc/invoke.texi (Link Options): Document it.
8905 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
8906 * config/alpha/linux.h: Likewise.
8907 * config/arc/elf.h: Likewise.
8908 * config/arm/uclinux-elf.h: Likewise.
8909 * config/arm/unknown-elf.h: Likewise.
8910 * config/avr/avrlibc.h: Likewise.
8911 * config/bfin/bfin.h: Likewise.
8912 * config/bfin/linux.h: Likewise.
8913 * config/bfin/uclinux.h: Likewise.
8914 * config/darwin.h: Likewise.
8915 * config/darwin10.h: Likewise.
8916 * config/darwin12.h: Likewise.
8917 * config/gnu-user.h: Likewise.
8918 * config/lm32/uclinux-elf.h: Likewise.
8919 * config/pa/pa-hpux11.h: Likewise.
8920 * config/pa/pa64-hpux.h: Likewise.
8921 * config/sparc/sparc.h: Likewise.
8923 2018-07-31 Olivier Hainque <hainque@adacore.com>
8925 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
8926 undefined variables.
8928 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
8931 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
8934 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
8936 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
8937 terminated string literal.
8939 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
8941 PR rtl-optimization/85160
8942 * combine.c (is_just_move): New function.
8943 (try_combine): Allow combining two instructions into two if neither of
8944 the original instructions was a move.
8946 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
8949 * doc/extend.texi (Global Register Variables): Discourage use of type
8951 (Local Register Variables): Likewise.
8953 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
8955 PR tree-optimization/86506
8956 * hwint.h (ceil_log2): Resync with hwint.c implementation.
8958 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
8961 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
8962 hard_regno, make sure no insn between `from` and `to` clobbers it.
8964 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
8965 Tom de Vries <tdevries@suse.de>
8967 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
8968 (PTX_DEFAULT_RUNTIME_DIM): ... this.
8969 (nvptx_goacc_validate_dims): Set default worker and gang dims to
8970 PTX_DEFAULT_RUNTIME_DIM.
8971 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
8973 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
8975 * config/pa/pa.c (pa_output_addr_vec): Align address table.
8976 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
8977 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
8979 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
8981 * config/rs6000/constraints.md (wG constraint): Delete, no longer
8983 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
8984 predicate to reflect toc fusion has been deleted.
8985 (toc_fusion_mem_raw): Delete, no longer used.
8986 (toc_fusion_mem_wrapped): Likewise.
8987 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
8989 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
8990 Delete, no longer used.
8991 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
8992 meant to be used for toc fusion.
8993 (rs6000_debug_print_mode): Delete toc fusion debugging.
8994 (rs6000_debug_reg_global): Likewise.
8995 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
8996 fusion and secondary reload support that were never used.
8997 (rs6000_option_override_internal): Delete TOC fusion, that was only
8998 partially defined, and it did not work unless you also used the
9000 (rs6000_legitimate_address_p): Delete TOC fusion support.
9001 (rs6000_opt_masks): Likewise.
9002 (fusion_wrap_memory_address): Delete function, no longer used.
9003 (fusion_split_address); Delete TOC fusion support.
9004 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
9005 longer used with toc fusion being deleted.
9006 (TARGET_TOC_FUSION_FP): Likewise.
9007 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
9009 (toc fusion spliter): Delete TOC fusion support.
9010 (toc_fusionload_<mode>): Likewise.
9011 (toc_fusionload_di): Likewise.
9012 (fusion_gpr_load_<mode>): Delete generator function, this insn no
9013 longer needs to be named. Rename predicate to delete TOC fusion.
9014 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
9015 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
9016 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
9017 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
9018 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
9020 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
9022 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
9023 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
9024 __int128 in built-in function prototypes.
9025 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
9026 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
9028 2018-07-27 Martin Sebor <msebor@redhat.com>
9030 PR tree-optimization/86696
9031 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
9032 types, including enums.
9033 (handle_char_store): Be prepared for the above function to fail.
9035 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
9037 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
9038 when optimization level is lower than 2 or optimize for size.
9040 2018-07-26 Martin Sebor <msebor@redhat.com>
9042 PR tree-optimization/86043
9043 PR tree-optimization/86042
9044 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
9045 (get_string_cst_length): Rename...
9046 (get_min_string_length): ...to this. Add argument.
9047 (handle_char_store): Extend to handle multi-character stores by
9049 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
9050 * tree.h (initializer_zerop): Add argument.
9052 2018-07-26 Jakub Jelinek <jakub@redhat.com>
9055 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
9056 declare target to variables if they have always,{to,from,tofrom} map
9059 2018-07-26 Martin Liska <mliska@suse.cz>
9062 * lto-wrapper.c: Add linker_output as prefix
9063 for ltrans_output_file.
9065 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
9067 PR rtl-optimization/85805
9068 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
9069 value for hard registers if that was written in the same mode.
9071 2018-07-26 Martin Liska <mliska@suse.cz>
9073 PR gcov-profile/86536
9074 * gcov.c (format_gcov): Use printf format %.*f directly
9075 and do not handle special values.
9077 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9079 * common/config/arc/arc-common.c (arc_option_optimization_table):
9080 Update default optimizations for size.
9082 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9084 * config/arc/arc.md (movsf_insn): Add short instruction selection.
9085 * config/arc/constraints.md (CfZ): New constraint.
9086 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
9087 (subsf3_fpu): Likewise.
9088 (cmpsf_fpu): Likewise.
9089 (cmpsf_fpu_uneq): Likewise.
9091 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9093 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
9095 (arc_is_uncached_mem_p): uncached applies to both the variable and
9098 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9100 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
9103 2018-07-25 David Malcolm <dmalcolm@redhat.com>
9105 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
9106 field "m_scopes" from vec to auto_vec.
9108 2018-07-25 Martin Liska <mliska@suse.cz>
9110 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
9113 2018-07-25 Richard Biener <rguenther@suse.de>
9116 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
9117 special wrt context_die late.
9118 (gen_subprogram_die): Re-use DIEs in local scope.
9120 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
9122 PR tree-optimization/86644
9123 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
9125 2018-07-25 Martin Liska <mliska@suse.cz>
9128 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
9129 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
9131 2018-07-25 Martin Liska <mliska@suse.cz>
9134 * params.def: Explain ASan abbreviation and provide
9135 a documentation link.
9137 2018-07-24 Martin Sebor <msebor@redhat.com>
9139 PR tree-optimization/86622
9140 PR tree-optimization/86532
9141 * builtins.h (string_length): Declare.
9142 * builtins.c (c_strlen): Correct handling of non-constant offsets.
9143 (check_access): Be prepared for non-constant length ranges.
9144 (string_length): Make extern.
9145 * expr.c (string_constant): Only handle the minor non-constant
9146 array index. Use string_constant to compute the length of
9147 a generic string constant.
9149 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
9151 PR tree-optimization/86618
9152 * tree-vect-stmts.c (vectorizable_call): Don't take the address
9153 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
9155 2018-07-24 David Malcolm <dmalcolm@redhat.com>
9157 PR tree-optimization/86636
9158 * json.cc (json::object::set): Fix comment. Add assertions.
9159 (json::array::append): Move here from json.h. Add comment and an
9161 (json::string::string): Likewise.
9162 * json.h (json::array::append): Move to json.cc.
9163 (json::string::string): Likewise.
9164 * optinfo-emit-json.cc
9165 (optrecord_json_writer::impl_location_to_json): Assert that we
9166 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
9167 wrapper around it. Expand the location once, rather than three
9169 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
9170 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
9172 (optrecord_json_writer::optinfo_to_json): Likewise, in four
9173 places. Fix some overlong lines.
9175 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
9177 * config/aarch64/aarch64-simd.md
9178 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
9179 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
9180 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
9181 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
9182 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
9183 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
9184 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
9185 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
9186 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
9188 2018-07-24 Jakub Jelinek <jakub@redhat.com>
9191 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
9192 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
9193 and abs_d == d, do the power of two handling if profitable.
9195 2018-07-24 Richard Biener <rguenther@suse.de>
9197 * match.pd: Add BIT_FIELD_REF canonicalizations.
9199 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9202 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
9204 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9206 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
9207 terminated STRING_CST object.
9209 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9211 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
9213 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
9215 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
9216 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
9217 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
9218 the elements into a register.
9219 (rs6000_split_v4si_init_di_reg): Delete.
9220 (rs6000_split_v4si_init): Delete.
9221 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
9222 (vsx_init_v4si): Rewrite as a define_expand.
9224 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
9226 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
9227 zero_extend argument from memory): New.
9229 2018-07-22 Martin Sebor <msebor@redhat.com>
9232 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
9233 diagnosing calls with unknown arguments unless -Walloca-larger-than
9234 is restricted to less than PTRDIFF_MAX bytes.
9236 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
9238 * doc/gcov.texi (Invoking Gcov): Editorial changes.
9240 2018-07-20 David Malcolm <dmalcolm@redhat.com>
9242 * pretty-print.c (text_info::set_location): Remove redundant
9243 "line_table" parameter from call to rich_location::set_range.
9245 2018-07-20 Martin Sebor <msebor@redhat.com>
9248 * builtins.c (expand_builtin_alloca): Adjust.
9249 * calls.c (alloc_max_size): Simplify.
9250 * cgraphunit.c (cgraph_node::expand): Adjust.
9251 * common.opt (larger_than_size, warn_frame_larger_than): Remove
9253 (frame_larger_than_size): Same.
9254 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
9255 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
9257 * doc/invoke.texi (GCC Command Options): Document option arguments.
9258 Explain byte-size arguments and suffixes.
9259 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
9260 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
9261 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
9262 * doc/options.texi (UInteger): Expand.
9263 (Host_Wide_Int, ByteSize): Document new properties.
9264 * final.c (final_start_function_1): Include sizes in an error message.
9265 * function.c (frame_offset_overflow): Same.
9266 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
9267 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
9268 Diagnose unbounded alloca calls only for limits of less than
9270 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
9271 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
9273 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
9274 only for limits of less than PTRDIFF_MAX.
9275 * langhooks-def.h (lhd_handle_option): Change function argument
9277 * langhooks.c (lhd_handle_option): Same.
9278 * langhooks.h (handle_option): Same.
9279 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
9281 (var_type, var_type_struct): Same.
9282 (var_set): Handle ByteSize flag.
9283 * optc-gen.awk: Add comments to output to ease debugging. Make
9284 use of HOST_WIDE_INT where appropriate.
9285 * opts-gen-save.awk: Use %lx to format unsigned long.
9286 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
9287 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
9288 arguments. Parse bytes-size suffixes.
9289 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
9290 (enum_value_to_arg): Same.
9291 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
9292 (handle_option): Adjust.
9293 (generate_option): Change function argument to HOST_WIDE_INT.
9294 (cmdline_handle_error): Adjust.
9295 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
9296 (set_option): Change function argument to HOST_WIDE_INT.
9297 (option_enabled): Handle cl_host_wide_int.
9298 (get_option_state): Handle CLVC_SIZE.
9299 (control_warning_option): Same.
9300 * opts.c (common_handle_option): Change function argument to
9301 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
9302 OPT_Wvla_larger_than_.
9303 * opts.h (enum cl_var_type): Add an enumerator.
9304 * stor-layout.c (layout_decl): Print a more meaningful warning.
9305 * toplev.c (output_stack_usage): Adjust.
9307 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
9309 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
9310 call to inline_expand_builtin_string_cmp.
9311 (expand_builtin_strcmp): Likewise.
9312 (expand_builtin_strncmp): Likewise.
9313 (inline_string_cmp): Delete the last parameter, change char_type_node
9314 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
9316 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
9317 the inlining expansion on target where the type of the call has same or
9318 narrower precision than unsigned char.
9320 2018-07-20 David Malcolm <dmalcolm@redhat.com>
9322 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
9323 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
9324 * common.opt (fsave-optimization-record): New option.
9325 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
9326 * doc/invoke.texi (-fsave-optimization-record): New option.
9327 * dumpfile.c: Include "optinfo-emit-json.h".
9328 (struct kv_pair): Move to coretypes.h.
9329 (optgroup_options): Make non-static.
9330 (dump_context::end_scope): Call
9331 optimization_records_maybe_pop_dump_scope.
9332 * dumpfile.h (optgroup_options): New decl.
9333 * json.cc: New file.
9335 * optinfo-emit-json.cc: New file.
9336 * optinfo-emit-json.h: New file.
9337 * optinfo.cc: Include "optinfo-emit-json.h".
9338 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
9339 (optinfo_enabled_p): Check optimization_records_enabled_p.
9340 (optinfo_wants_inlining_info_p): Likewise.
9341 * optinfo.h: Update comment.
9342 * profile-count.c (profile_quality_as_string): New function.
9343 * profile-count.h (profile_quality_as_string): New decl.
9344 (profile_count::quality): New accessor.
9345 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
9346 and optinfo_emit_json_cc_tests.
9347 * selftest.h (selftest::json_cc_tests): New decl.
9348 (selftest::optinfo_emit_json_cc_tests): New decl.
9349 * toplev.c: Include "optinfo-emit-json.h".
9350 (compile_file): Call optimization_records_finish.
9351 (do_compile): Call optimization_records_start.
9352 * tree-ssa-live.c: Include optinfo.h.
9353 (remove_unused_scope_block_p): Retain inlining information if
9354 optinfo_wants_inlining_info_p returns true.
9356 2018-07-20 Richard Biener <rguenther@suse.de>
9359 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
9360 to cover -flto-partition=none.
9362 2018-07-20 Martin Liska <mliska@suse.cz>
9364 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
9365 (get_decl_source_range): Remove unused function.
9367 2018-07-20 Richard Biener <rguenther@suse.de>
9369 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
9370 (struct vn_phi_s): Likewise.
9371 (struct vn_reference_s): Likewise.
9372 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
9373 for searching the slot of an entry known to be in the hash itself.
9374 (vn_phi_hasher::equal): Likewise.
9375 (vn_reference_hasher::equal): Likewise.
9376 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
9378 (optimistic_info, current_info): Remove, keeping only valid_info.
9379 (vn_reference_lookup_1): Remove fallback lookup.
9380 (vn_reference_lookup_2): Likewise.
9381 (vn_nary_op_lookup_1): Likewise.
9382 (vn_phi_lookup): Likewise.
9383 (vn_nary_build_or_lookup_1): Make sure to not chain the built
9385 (vn_reference_insert): Adjust, chain the inserted hash element
9386 at last_inserted_ref.
9387 (vn_reference_insert_pieces): Likewise.
9388 (visit_reference_op_call): Likewise.
9389 (vn_nary_op_insert_into): Chain the inserted hash element at
9391 (vn_nary_op_insert_pieces): Adjust.
9392 (vn_nary_op_insert): Likewise.
9393 (vn_nary_op_insert_stmt): Likewise.
9394 (vn_phi_insert): Adjust, chain the inserted hash element at
9396 (process_scc): Remove clearing and copying the optimistic
9397 table. Instead remove elements inserted during an optimistic
9398 iteration from the single table we maintain.
9399 (init_scc_vn): Adjust.
9400 (free_scc_vn): Likewise.
9401 (sccvn_dom_walker::record_cond): Likewise.
9402 (sccvn_dom_walker::after_dom_children): Likewise.
9404 2018-07-19 Martin Sebor <msebor@redhat.com>
9406 PR tree-optimization/84047
9407 PR tree-optimization/83776
9408 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
9409 (check_array_bounds): Call it.
9411 2018-07-19 Martin Sebor <msebor@redhat.com>
9413 * align.h (align_flags): Use member initialization.
9415 2018-07-19 David Malcolm <dmalcolm@redhat.com>
9417 * Makefile.in (OBJS): Add optinfo.o.
9418 * coretypes.h (class symtab_node): New forward decl.
9419 (struct cgraph_node): New forward decl.
9420 (class varpool_node): New forward decl.
9421 * dump-context.h: New file.
9422 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
9424 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
9425 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
9426 (set_alt_dump_file): Likewise.
9427 (dump_context::~dump_context): New dtor.
9428 (dump_gimple_stmt): Move implementation to...
9429 (dump_context::dump_gimple_stmt): ...this new member function.
9430 Add the stmt to any pending optinfo, creating one if need be.
9431 (dump_gimple_stmt_loc): Move implementation to...
9432 (dump_context::dump_gimple_stmt_loc): ...this new member function.
9433 Start a new optinfo and add the stmt to it.
9434 (dump_gimple_expr): Move implementation to...
9435 (dump_context::dump_gimple_expr): ...this new member function.
9436 Add the stmt to any pending optinfo, creating one if need be.
9437 (dump_gimple_expr_loc): Move implementation to...
9438 (dump_context::dump_gimple_expr_loc): ...this new member function.
9439 Start a new optinfo and add the stmt to it.
9440 (dump_generic_expr): Move implementation to...
9441 (dump_context::dump_generic_expr): ...this new member function.
9442 Add the tree to any pending optinfo, creating one if need be.
9443 (dump_generic_expr_loc): Move implementation to...
9444 (dump_context::dump_generic_expr_loc): ...this new member
9445 function. Add the tree to any pending optinfo, creating one if
9447 (dump_printf): Move implementation to...
9448 (dump_context::dump_printf_va): ...this new member function. Add
9449 the text to any pending optinfo, creating one if need be.
9450 (dump_printf_loc): Move implementation to...
9451 (dump_context::dump_printf_loc_va): ...this new member function.
9452 Start a new optinfo and add the stmt to it.
9453 (dump_dec): Move implementation to...
9454 (dump_context::dump_dec): ...this new member function. Add the
9455 value to any pending optinfo, creating one if need be.
9456 (dump_context::dump_symtab_node): New member function.
9457 (dump_context::get_scope_depth): New member function.
9458 (dump_context::begin_scope): New member function.
9459 (dump_context::end_scope): New member function.
9460 (dump_context::ensure_pending_optinfo): New member function.
9461 (dump_context::begin_next_optinfo): New member function.
9462 (dump_context::end_any_optinfo): New member function.
9463 (dump_context::s_current): New global.
9464 (dump_context::s_default): New global.
9465 (dump_scope_depth): Delete global.
9466 (dumpfile_ensure_any_optinfo_are_flushed): New function.
9467 (dump_symtab_node): New function.
9468 (get_dump_scope_depth): Reimplement in terms of dump_context.
9469 (dump_begin_scope): Likewise.
9470 (dump_end_scope): Likewise.
9471 (selftest::temp_dump_context::temp_dump_context): New ctor.
9472 (selftest::temp_dump_context::~temp_dump_context): New dtor.
9473 (selftest::verify_item): New function.
9474 (ASSERT_IS_TEXT): New macro.
9475 (ASSERT_IS_TREE): New macro.
9476 (ASSERT_IS_GIMPLE): New macro.
9477 (selftest::test_capture_of_dump_calls): New test.
9478 (selftest::dumpfile_c_tests): Call it.
9479 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
9480 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
9481 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
9482 descriptive comment.
9483 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
9484 (dump_node, dump_bb): Move these unrelated decls.
9485 (class dump_manager): Add leading comment.
9486 * optinfo.cc: New file.
9487 * optinfo.h: New file.
9489 2018-07-19 Michael Collison <michael.collison@arm.com>
9490 Richard Henderson <rth@redhat.com>
9492 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
9493 (subti): Handle op1 zero.
9494 (subvti4, usub4ti4): New.
9495 (*sub<GPI>3_compare1_imm): New.
9496 (sub<GPI>3_carryinCV): New.
9497 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
9498 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
9500 2018-07-19 Michael Collison <michael.collison@arm.com>
9501 Richard Henderson <rth@redhat.com>
9503 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
9504 (addti3): Create simpler code if low part is already known to be 0.
9505 (addvti4, uaddvti4): New.
9506 (*add<GPI>3_compareC_cconly_imm): New.
9507 (*add<GPI>3_compareC_cconly): New.
9508 (*add<GPI>3_compareC_imm): New.
9509 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
9510 handle constants within this pattern..
9511 (*add<GPI>3_compareV_cconly_imm): New.
9512 (*add<GPI>3_compareV_cconly): New.
9513 (*add<GPI>3_compareV_imm): New.
9514 (add<GPI>3_compareV): New.
9515 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
9516 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
9517 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
9518 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
9519 with 'comparison' operator.
9520 (*add<GPI>3_compareV_cconly_imm): Ditto.
9521 (*add<GPI>3_compareV_cconly): Ditto.
9522 (*add<GPI>3_compareV_imm): Ditto.
9523 (add<GPI>3_compareV): Ditto.
9524 (add<mode>3_carryinC): Ditto.
9525 (*add<mode>3_carryinC_zero): Ditto.
9526 (*add<mode>3_carryinC): Ditto.
9527 (add<mode>3_carryinV): Ditto.
9528 (*add<mode>3_carryinV_zero): Ditto.
9529 (*add<mode>3_carryinV): Ditto.
9531 2018-07-19 Michael Collison <michael.collison@arm.com>
9532 Richard Henderson <rth@redhat.com>
9534 * config/aarch64/aarch64-modes.def (CC_V): New.
9535 * config/aarch64/aarch64-protos.h
9536 (aarch64_addti_scratch_regs): Declare
9537 (aarch64_subvti_scratch_regs): Declare.
9538 (aarch64_expand_subvti): Declare.
9539 (aarch64_gen_unlikely_cbranch): Declare
9540 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
9541 for signed overflow using CC_Vmode.
9542 (aarch64_get_condition_code_1): Handle CC_Vmode.
9543 (aarch64_gen_unlikely_cbranch): New function.
9544 (aarch64_addti_scratch_regs): New function.
9545 (aarch64_subvti_scratch_regs): New function.
9546 (aarch64_expand_subvti): New function.
9548 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9550 * config/aarch64/aarch64-option-extensions.def: New entry for profile
9552 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
9553 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
9556 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9559 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
9560 address check not strict.
9562 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9564 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
9566 (store_pair_lanes<mode>): Likewise.
9567 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
9568 enum value 'ADDR_QUERY_LDP_STP_N'.
9569 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
9570 (aarch64_print_address_internal): Add declaration.
9571 (aarch64_print_ldpstp_address): Remove.
9572 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
9573 (aarch64_print_operand): Change printing of 'y'.
9574 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
9575 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
9576 'true' rather than '1'.
9577 * config/aarch64/constraints.md (Uml): Likewise.
9578 (Uml): Rename to Umn.
9581 2018-07-19 Richard Biener <rguenther@suse.de>
9583 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
9585 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
9586 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
9587 (vn_reference_hasher): Likewise.
9588 (struct vn_tables_s): Remove obstack and alloc-pool members.
9589 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
9590 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
9591 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
9592 (vn_reference_insert_pieces): Likewise.
9593 (alloc_vn_nary_op_noinit): Adjust.
9594 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
9595 (vn_phi_eq): Adjust.
9596 (shared_lookup_phiargs): Remove.
9597 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
9598 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
9599 (visit_reference_op_call): Likewise.
9600 (copy_nary, copy_phi, copy_reference): Remove.
9601 (process_scc): Rewind the obstack when iterating. Do not
9602 copy the elements to valid_info but just move them from one
9603 hashtable to the other.
9604 (allocate_vn_table): Adjust.
9605 (free_vn_table): Likewise.
9606 (init_scc_vn): Likewise.
9607 (free_scc_vn): Likewise.
9609 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
9612 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
9613 indirect_return as function type attribute.
9614 (ix86_attribute_table): Change indirect_return to function
9616 * doc/extend.texi: Update indirect_return attribute.
9618 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
9620 * wide-int.h (widest2_int): New.
9621 * gimple-fold.c (arith_overflowed_p): Use it.
9622 * tree.h (widest2_int_cst): New.
9623 * tree-vrp.c (wide_int_binop_overflow): Rename from
9624 vrp_int_const_binop.
9625 Rewrite to work on trees.
9626 (extract_range_from_multiplicative_op_1): Abstract code to...
9627 (wide_int_range_min_max): ...here.
9628 (wide_int_range_cross_product): ...and here.
9629 (extract_range_from_binary_expr_1): Abstract overflow code to...
9630 (wide_int_range_mult_wrapping): ...here.
9631 * tree-vrp.h (wide_int_range_cross_product): New.
9632 (wide_int_range_mult_wrapping): New.
9634 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
9635 Julia Koval <julia.koval@intel.com>
9637 * config/i386/x86-tune-costs.h (skylake_memcpy,
9638 skylake_memset): Replace rep_prefix with unrolling for size 512.
9640 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
9643 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
9644 comparision with EQ_EXPR in last stmt.
9646 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
9648 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
9649 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
9650 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
9651 previously known as "PowerPC AltiVec Built-in Functions". Move
9652 some material to new subsubsections "PowerPC AltiVec Built-in
9653 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
9655 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
9656 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
9657 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
9658 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
9660 2018-07-18 Richard Biener <rguenther@suse.de>
9662 PR tree-optimization/86557
9663 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
9666 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
9668 * config/s390/s390.c (s390_function_profiler): Generate CFI.
9670 2018-07-17 Jeff Law <law@redhat.com>
9672 * config/arm/arm.c (get_label_padding): Update for recent
9673 changes to label_to_alignment.
9675 PR tree-optimization/86010
9676 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
9678 * config/mips/mips.c (vr4130_align_insns): Update for recent
9679 changes to label_to_alignment.
9681 * config/frv/frv.c (frv_label_align): Update for recent changes
9682 to label_to_alignment.
9684 * config/nios2/nios2.c (nios2_label_align): Update for recent
9685 changes which dropped ALIGN_LABELS_LOG.
9687 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
9689 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
9690 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
9692 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
9694 * config/arc/arc.c (arc_label_align): Use align_labels instead of
9695 deprecated align_labels_log.
9697 2018-07-17 Richard Biener <rguenther@suse.de>
9700 * dwarf2out.c (init_sections_and_labels): Always generate
9701 a debug_line_str_section for early LTO debug.
9702 (dwarf2out_finish): Reset debug_line_str_hash output early.
9703 Bump counter for extra dwarf5 .debug_loc labels to not conflict
9705 (dwarf2out_early_finish): Output debug_line_str.
9707 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
9709 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
9710 index register on z196 or later.
9712 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
9714 * config/s390/s390.c (s390_default_align): Set default function
9716 (s390_override_options_after_change): Call s390_default align.
9717 (s390_option_override_internal): Call s390_default_align.
9718 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
9720 2018-07-17 Jakub Jelinek <jakub@redhat.com>
9723 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
9724 to _looptemp_ clauses, other than the first two.
9726 2018-07-17 Martin Liska <mliska@suse.cz>
9728 * opts.c: Do not enable OPT_falign_* for -Os.
9730 2018-07-17 Martin Liska <mliska@suse.cz>
9732 * align.h (MAX_CODE_ALIGN): New.
9733 (MAX_CODE_ALIGN_VALUE): New.
9734 * common/config/i386/i386-common.c (ix86_handle_option):
9735 (MAX_CODE_ALIGN): Moved to align.h.
9736 * final.c (MAX_CODE_ALIGN): Likewise.
9737 * opts.c (parse_and_check_align_values):
9738 (MAX_CODE_ALIGN): Likewise.
9739 (MAX_CODE_ALIGN_VALUE): Likewise.
9741 2018-07-17 Martin Liska <mliska@suse.cz>
9743 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
9744 in order to fulfil coding style.
9745 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
9746 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9747 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
9748 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
9749 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
9750 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
9751 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
9752 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
9753 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9755 2018-07-17 Martin Liska <mliska@suse.cz>
9757 * align.h: New file.
9758 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
9760 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
9761 align_flags of label_to_alignment.
9762 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
9764 * config/m68k/m68k.c: Do not use removed align_labels_value and
9766 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
9767 (LOOP_ALIGN): Likewise.
9768 (LABEL_ALIGN): Likewise.
9769 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
9770 Remove not used macro.
9771 (rs6000_loop_align): Change return type to align_flags.
9772 (rs6000_loop_align_max_skip): Remove.
9773 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
9774 Change return type to align_flags.
9775 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
9776 Remove not used macro.
9777 (rs6000_loop_align): Change return type to align_flags.
9778 (rs6000_loop_align_max_skip): Remove.
9779 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
9780 * config/rx/rx-protos.h (rx_align_for_label): Make it
9782 * config/rx/rx.c (rx_align_for_label): Change return type
9784 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
9786 into align_flags class.
9787 (LABEL_ALIGN): Likewise.
9788 (LOOP_ALIGN): Likewise.
9789 * config/s390/s390.c (s390_label_align): Use align_flags
9791 (s390_asm_output_function_label): Likewise.
9792 * config/sh/sh.c (sh_override_options_after_change):
9793 Use align_flags class directly without macros.
9794 (find_barrier): Likewise.
9795 (barrier_align): Likewise.
9796 (sh_loop_align): Likewise.
9797 * config/spu/spu.c (spu_option_override):
9798 Use align_flags_tuple::get_value instead of removed macros.
9799 (spu_sched_init): Likewise.
9800 * config/spu/spu.h (GTY): Likewise.
9801 * config/visium/visium.c (visium_option_override):
9802 Set "8" as default secondary alignment.
9803 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
9804 in order to guarantee secondary alignment of 8.
9805 * coretypes.h: Include align.h header file.
9806 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
9807 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
9808 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
9809 * doc/tm.texi.in: Likewise.
9810 * final.c (struct label_alignment): Remove not used structure.
9811 (LABEL_ALIGN): Change type to align_flags.
9812 (LOOP_ALIGN): Likewise.
9813 (JUMP_ALIGN): Likewise.
9814 (default_loop_align_max_skip): Remove.
9815 (default_label_align_max_skip): Likewise.
9816 (default_jump_align_max_skip): Likewise.
9817 (default_label_align_after_barrier_max_skip):
9818 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
9819 (LABEL_TO_MAX_SKIP): Remove.
9820 (label_to_alignment): Return align_flags type instead of integer.
9821 (label_to_max_skip): Remove.
9822 (align_fuzz): Use align_flags type.
9823 (compute_alignments): Use align_flags type and use align_flags::max
9824 to combine multiple alignments.
9825 (grow_label_align): Grow vec instead of C array.
9826 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
9827 (shorten_branches): Use align_flags type and use align_flags::max
9828 to combine multiple alignments.
9829 (final_scan_insn_1): Remove usage of secondary alignment that comes
9830 from label alignment, but instead use proper secondary alignment
9831 which is computed in grow_label_align.
9832 * flags.h (struct align_flags_tuple): Move to align.h.
9833 (struct align_flags): Likewise.
9834 (state_align_loops): Rename to align_loops.
9835 (state_align_jumps): Rename to align_jumps.
9836 (state_align_labels): Rename to align_labels.
9837 (state_align_functions): Rename to align_functions.
9838 (align_loops_log): Remove.
9839 (align_jumps_log): Remove.
9840 (align_labels_log): Remove.
9841 (align_functions_log): Remove.
9842 (align_loops_max_skip): Remove.
9843 (align_jumps_max_skip): Remove.
9844 (align_labels_max_skip): Remove.
9845 (align_functions_max_skip): Remove.
9846 (align_loops_value): Remove.
9847 (align_jumps_value): Remove.
9848 (align_labels_value): Remove.
9849 (align_functions_value): Remove.
9850 * output.h (label_to_alignment): Change return type to align_flags.
9851 (label_to_max_skip): Remove.
9852 * target.def: Remove loop_align_max_skip, label_align_max_skip,
9853 jump_align_max_skip macros.
9854 * targhooks.h (default_loop_align_max_skip): Remove.
9855 (default_label_align_max_skip): Likewise.
9856 (default_jump_align_max_skip): Likewise.
9857 (default_label_align_after_barrier_max_skip): Remove.
9858 * toplev.c (read_log_maxskip): Use ::normalize function.
9859 (parse_N_M): Remove not used argument and also call ::normalize.
9860 (parse_alignment_opts): Do not pass unused arguments.
9861 * varasm.c (assemble_start_function): Use directly align_functions
9862 instead of removed macros.
9863 * system.h: Do not poison removed macros.
9865 2018-07-17 Jakub Jelinek <jakub@redhat.com>
9868 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
9869 and cond temporaries don't have reference type if iterator has
9870 pointer type. For init use &for_pre_body instead of pre_p if
9871 for_pre_body is non-empty.
9873 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
9875 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
9876 double-double modes to SFmode directly directly.
9877 (trunc<mode>sf2_fprs): Delete.
9879 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
9881 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
9882 for conversions between IFmode and the decimal floating point modes.
9883 (init_float128_ieee): Use the correct names for conversions between
9884 KFmode and the decimal floating point modes.
9886 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
9888 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
9889 for the conversions between TDmode and IFmode.
9890 (init_float128_ieee): Use more correct names for the conversions
9891 between TDmode and KFmode.
9893 2018-07-16 Jakub Jelinek <jakub@redhat.com>
9895 PR tree-optimization/86526
9896 * builtins.c (expand_builtin_memcmp): Formatting fixes.
9897 (inline_expand_builtin_string_cmp): Likewise.
9898 (inline_string_cmp): Likewise. Use c_readstr instead of
9899 builtin_memcpy_read_str. Add unit_mode temporary.
9901 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
9904 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
9905 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
9907 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
9909 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
9910 Alphabetize prototypes of built-in functions, separating out
9911 built-in functions that are listed in this section but should be
9912 described elsewhere.
9914 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
9917 * expmed.c (emit_store_flag): Do not emit setcc followed by a
9918 conditional move when trapping comparison was split to a
9919 non-trapping one (and vice versa).
9921 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
9923 * config/s390/s390.c (s390_function_profiler): Generate nops
9924 instead of profiler call sequences.
9925 * config/s390/s390.opt: Add the new option.
9927 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
9929 * config/s390/s390.c (s390_function_profiler): Generate
9930 __mcount_loc section.
9931 * config/s390/s390.opt: Add the new option.
9933 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
9935 * common.opt: Add the new warning.
9936 * config/s390/s390.c (s390_function_profiler): Emit "brasl
9937 %r0,__fentry__" when -mfentry is specified.
9938 (s390_option_override_internal): Disallow -mfentry for 31-bit
9940 * config/s390/s390.opt: Add the new option.
9942 2018-07-16 Richard Biener <rguenther@suse.de>
9945 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
9946 for function-local FUNCTION_DECL and RESULT_DECL immediately.
9948 2018-07-16 Martin Liska <mliska@suse.cz>
9951 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
9954 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
9956 * config/arc/arcHS.md: Update ARCHS scheduling rules.
9958 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
9960 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
9962 * config/arc/arc-cpus.def (hs4x): New cpu.
9964 * config/arc/arc-tables.opt: Regenerate.
9965 * config/arc/arc.c (arc_sched_issue_rate): New function.
9966 (TARGET_SCHED_ISSUE_RATE): Define.
9967 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
9968 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
9970 (attr tune): Add ARCHS4x tune values.
9971 (attr tune_dspmpy): Define.
9972 (*tst): Correct instruction type.
9973 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
9974 * config/arc/arcHS4x.md: New file.
9975 * config/arc/fpu.md: Update instruction type attributes.
9976 * config/arc/t-multilib: Regenerate.
9978 2018-07-16 Tom de Vries <tdevries@suse.de>
9981 * var-tracking.c (vt_initialize): Fix pre_dec handling.
9983 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9985 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
9988 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
9990 PR tree-optimization/86514
9991 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
9992 conversion to a boolean type from a type with greater precision.
9994 2018-07-16 Tom de Vries <tdevries@suse.de>
9996 * var-tracking.c (vt_initialize): Print adjusted insn slim if
9997 dump_flags request TDF_SLIM.
9999 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
10001 * fold-const.c (int_const_binop_1): Abstract...
10002 (wide_int_binop): ...wide int code here.
10003 (poly_int_binop): ...poly int code here.
10004 Abstract the rest of int_const_binop_1 into int_const_binop.
10005 * fold-const.h (wide_int_binop): New.
10006 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
10007 Remove useless PLUS/MINUS_EXPR case.
10008 (zero_nonzero_bits_from_vr): Move wide int code...
10009 (zero_nonzero_bits_from_bounds): ...here.
10010 (extract_range_from_binary_expr_1): Move mask optimization code...
10011 (range_easy_mask_min_max): ...here.
10012 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
10013 (range_easy_mask_min_max): New.
10015 2018-07-15 Jeff Law <law@redhat.com>
10018 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
10021 2018-07-14 Jim Wilson <jimw@sifive.com>
10023 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
10025 2018-07-14 Paul Koning <ni1d@arrl.net>
10027 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
10029 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
10031 * lto-streamer-out.c (copy_function_or_variable): Dump info about
10034 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
10035 Steve Munroe <munroesj52@gmail.com>
10037 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
10038 (_mm_andnot_si128): Likewise.
10039 (_mm_or_si128): Likewise.
10040 (_mm_xor_si128): Likewise.
10042 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
10044 PR middle-end/78809
10045 * builtins.c (expand_builtin_memcmp): Inline the calls first
10046 when result_eq is false.
10047 (expand_builtin_strcmp): Inline the calls first.
10048 (expand_builtin_strncmp): Likewise.
10049 (inline_string_cmp): New routine. Expand a string compare
10050 call by using a sequence of char comparison.
10051 (inline_expand_builtin_string_cmp): New routine. Inline expansion
10052 a call to str(n)cmp/memcmp.
10053 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
10055 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
10057 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
10059 * config/arm/driver-arm.c: Include arm-native.h.
10060 (host_detect_local_cpu): Use auto-generated data tables.
10061 (vendors, arm_cpu_table): Delete. Move part information to ...
10062 * config/arm/arm-cpus.in: ... here.
10063 * config/arm/parsecpu.awk (gen_native): New function.
10064 (vendor, part): New CPU fields.
10065 (END): Add support for building the native CPU detection tables.
10066 * config/arm/t-arm (arm-native.h): Add build rule.
10067 (driver-arm.o): Add dependency on arm-native.h.
10069 2018-07-13 Richard Biener <rguenther@suse.de>
10071 PR middle-end/85974
10072 * match.pd (addr1 - addr2): Allow either of the operand to
10075 2018-07-13 Tom de Vries <tdevries@suse.de>
10077 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
10080 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
10082 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
10083 arrays instead of numbered variables.
10085 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
10087 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
10088 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
10090 2018-07-13 Richard Biener <rguenther@suse.de>
10093 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
10094 instead of get_context_die.
10096 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
10097 Richard Biener <rguenther@suse.de>
10099 PR middle-end/86489
10100 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
10101 that the loop latch destination where phi is defined.
10103 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
10105 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
10106 (riscv_expand_epilogue): Add assertion to check interrupt mode.
10107 (riscv_set_current_function): Extract getting interrupt type to new
10109 (riscv_get_interrupt_type): New function.
10110 (riscv_merge_decl_attributes): New function, checking interrupt type is
10112 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
10114 2018-07-12 Paul Koning <ni1d@arrl.net>
10116 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
10119 2018-07-12 Paul Koning <ni1d@arrl.net>
10121 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
10122 zero reference, add doloop_end instead.
10123 * doc/md.texi (decrement_and_branch_until_zero): Remove.
10124 (Looping patterns): Remove decrement_and_branch_until_zero. Add
10125 detail for doloop_end.
10127 2018-07-12 Martin Sebor <msebor@redhat.com>
10130 * attribs.c (decl_attributes): Reject conflicting attributes before
10131 calling attribute handlers.
10133 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
10135 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
10137 (gcc::dump_manager::get_dump_file_name): likewise.
10138 (dump_begin): Likewise.
10139 * dumpfile.h (dump_begin): Update prototype.
10140 (gcc::dump_manager::get_dump_file_name,
10141 gcc::dump_manager::get_dump_file_name): Update prototype.
10143 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10145 * internal-fn.h (vectorizable_internal_fn_p): New function.
10146 * tree-vect-slp.c (compatible_calls_p): Likewise.
10147 (vect_build_slp_tree_1): Remove nops argument. Handle calls
10148 to internal functions.
10149 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
10151 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10153 * fold-const.h (inverse_conditions_p): Declare.
10154 * fold-const.c (inverse_conditions_p): New function.
10155 * match.pd: Use inverse_conditions_p. Add folds of view_converts
10156 that test the inverse condition of a conditional internal function.
10157 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
10158 * internal-fn.c (internal_fn_mask_index): Handle conditional
10159 internal functions.
10160 (vectorized_internal_fn_supported_p): New function.
10161 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
10162 (any_pred_load_store): Replace with...
10163 (need_to_predicate): ...this new variable.
10164 (redundant_ssa_names): New variable.
10165 (ifcvt_can_use_mask_load_store): Move initial checks to...
10166 (ifcvt_can_predicate): ...this new function. Handle tree codes
10167 for which a conditional internal function exists.
10168 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
10169 instead of ifcvt_can_use_mask_load_store. Update after variable
10171 (predicate_load_or_store): New function, split out from
10172 predicate_mem_writes.
10173 (check_redundant_cond_expr): New function.
10174 (value_available_p): Likewise.
10175 (predicate_rhs_code): Likewise.
10176 (predicate_mem_writes): Rename to...
10177 (predicate_statements): ...this. Use predicate_load_or_store
10178 and predicate_rhs_code.
10179 (combine_blocks, tree_if_conversion): Update after above name changes.
10180 (ifcvt_local_dce): Handle redundant_ssa_names.
10181 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
10182 general conditional functions.
10183 * tree-vect-stmts.c (vectorizable_call): Likewise.
10185 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10186 Alan Hayward <alan.hayward@arm.com>
10187 David Sherwood <david.sherwood@arm.com>
10189 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
10190 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
10191 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
10192 plus and minus and convert them into IFN_COND_FMA-based sequences.
10193 (convert_mult_to_fma): Handle conditional plus and minus.
10195 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10197 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
10198 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
10199 (cond_fnms_optab): New optabs.
10200 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
10201 internal functions.
10202 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
10203 * internal-fn.h (get_conditional_internal_fn): Declare.
10204 (get_unconditional_internal_fn): Likewise.
10205 * internal-fn.c (cond_ternary_direct): New macro.
10206 (expand_cond_ternary_optab_fn): Likewise.
10207 (direct_cond_ternary_optab_supported_p): Likewise.
10208 (FOR_EACH_COND_FN_PAIR): Likewise.
10209 (get_conditional_internal_fn): New function.
10210 (get_unconditional_internal_fn): Likewise.
10211 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
10212 (gimple_match_op::gimple_match_op): Add a new overload for 5
10214 (gimple_match_op::set_op): Likewise.
10215 (gimple_resimplify5): Declare.
10216 * genmatch.c (decision_tree::gen): Generate simplifications for
10218 * gimple-match-head.c (gimple_simplify): Define an overload for
10219 5 operands. Handle calls with 5 arguments in the top-level overload.
10220 (convert_conditional_op): Handle conversions from unconditional
10221 internal functions to conditional ones.
10222 (gimple_resimplify5): New function.
10223 (build_call_internal): Pass a fifth operand.
10224 (maybe_push_res_to_seq): Likewise.
10225 (try_conditional_simplification): Try converting conditional
10226 internal functions to unconditional internal functions.
10227 Handle 3-operand unconditional forms.
10228 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
10229 Define ternary equivalents of the current rules for binary conditional
10230 internal functions.
10231 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
10232 ternary operations.
10233 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
10234 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
10235 (optab): Handle them.
10236 (SVE_COND_FP_TERNARY): New int iterator.
10237 (sve_fmla_op, sve_fmad_op): New int attributes.
10238 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
10239 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
10240 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
10242 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10244 * target.def (preferred_else_value): New target hook.
10245 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
10246 * doc/tm.texi: Regenerate.
10247 * targhooks.h (default_preferred_else_value): Declare.
10248 * targhooks.c (default_preferred_else_value): New function.
10249 * internal-fn.h (conditional_internal_fn_code): Declare.
10250 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
10251 (get_conditional_internal_fn): Use it.
10252 (conditional_internal_fn_code): New function.
10253 * gimple-match.h (gimple_match_cond): New struct.
10254 (gimple_match_op): Add a cond member function.
10255 (gimple_match_op::gimple_match_op): Update all forms to take a
10257 * genmatch.c (expr::gen_transform): Use the same condition as res_op
10258 for the suboperation, but don't specify a particular else_value.
10259 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
10260 (visit_nary_op, visit_reference_op_load): Pass
10261 gimple_match_cond::UNCOND to the gimple_match_op constructor.
10262 * gimple-match-head.c: Include tree-eh.h
10263 (convert_conditional_op): New function.
10264 (maybe_resimplify_conditional_op): Likewise.
10265 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
10266 (gimple_resimplify2): Likewise.
10267 (gimple_resimplify3): Likewise.
10268 (gimple_resimplify4): Likewise.
10269 (maybe_push_res_to_seq): Return null for conditional operations.
10270 (try_conditional_simplification): New function.
10271 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
10273 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
10275 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
10277 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
10279 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
10281 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
10283 (hash_tree): Do not hash DECL_FCONTEXT
10284 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
10285 Do not stream DECL_FCONTEXT.
10286 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
10287 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
10289 2018-07-12 Richard Biener <rguenther@suse.de>
10292 * dwarf2out.c (gen_block_die): Only output blocks when they have
10293 at least one !DECL_IGNORED_P variable.
10295 2018-07-12 Richard Biener <rguenther@suse.de>
10298 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
10299 Remove -mieee-fp handling.
10301 2018-07-12 Richard Biener <rguenther@suse.de>
10303 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
10304 left-over from last patch.
10306 2018-07-12 Jakub Jelinek <jakub@redhat.com>
10308 PR tree-optimization/86492
10309 * gimple-ssa-store-merging.c
10310 (imm_store_chain_info::coalesce_immediate_stores): Call
10311 check_no_overlap even for the merge_overlapping case. Formatting fix.
10313 2018-07-12 Richard Biener <rguenther@suse.de>
10315 PR middle-end/86479
10316 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
10317 move possibly trapping operations into the conditional.
10319 2018-07-12 Richard Biener <rguenther@suse.de>
10321 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
10322 (vn_lookup_simplify_result): Remove recursion limit applied
10324 (vn_nary_build_or_lookup_1): Adjust.
10325 (try_to_simplify): Likewise.
10326 * gimple-match-head.c (gimple_resimplify1): Instead apply one
10328 (gimple_resimplify2): Likewise.
10329 (gimple_resimplify3): Likewise.
10330 (gimple_resimplify4): Likewise.
10332 2018-07-11 Jakub Jelinek <jakub@redhat.com>
10334 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
10335 Use __mmask64 type instead of __mmask8 for __M argument.
10336 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
10337 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
10339 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
10340 __mmask16 for __M argument.
10341 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
10342 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
10343 to __mmask16 instead of __mmask8.
10344 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
10345 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
10346 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
10347 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
10348 instead of __mmask16 for __U argument.
10349 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
10350 __mmask16 instead of __mmask8 for __U argument.
10351 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
10353 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
10355 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
10357 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
10359 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
10361 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
10363 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
10365 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
10367 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
10369 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
10371 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
10372 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
10373 return type as well as __M argument type and all casts from __mmask8
10375 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
10376 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
10377 return type as well as __M argument type and all casts from __mmask8
10379 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
10380 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
10381 return type as well as __M argument type and all casts from __mmask8
10383 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
10384 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
10385 return type as well as __M argument type and all casts from __mmask8
10387 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
10388 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
10391 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
10393 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
10394 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
10397 2018-07-11 Paul Koning <ni1d@arrl.net>
10399 * doc/md.texi (define_subst): Document how multiple occurrences of
10400 the same argument in the replacement pattern are handled.
10402 2018-07-11 Paul Koning <ni1d@arrl.net>
10404 * doc/extend.texi (Common Variable Attributes): Move "mode" into
10405 alphabetical order.
10406 (Common Type Attributes): Add "mode" attribute.
10408 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
10410 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
10411 stream DECL_ORIGINAL_TYPE.
10412 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
10413 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
10414 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
10415 Do not walk original type.
10416 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
10418 (write_ts_decl_non_common_tree_pointers): Do not stream
10420 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
10421 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
10423 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
10425 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
10426 thread twice from the same starting edge.
10428 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
10430 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
10431 * gimple.c (gimple_call_nonnull_result_p): ...here...
10432 (gimple_call_nonnull_arg): ...and here.
10433 * gimple.h (gimple_call_nonnull_result_p): New.
10434 (gimple_call_nonnull_arg): New.
10436 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
10438 * config/arm/arm-cpus.in: Move information from fpu field of each
10439 cpu definition to the isa field.
10440 * config/arm/parsecpu.awk (fpu): Delete match rule.
10441 (gen_comm_data): Don't add bits from the CPU's FPU entry.
10443 2018-07-11 Richard Biener <rguenther@suse.de>
10446 * dwarf2out.c (init_sections_and_labels): Use
10447 output_asm_line_debug_info consistently.
10448 (dwarf2out_early_finish): Likewise.
10449 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
10452 2018-07-11 Richard Biener <rguenther@suse.de>
10454 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
10455 Rework father_bb setting in a way to avoid propagating constants
10456 multiple times on a loop body.
10458 2018-07-10 Mark Wielaard <mark@klomp.org>
10461 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
10463 2018-07-10 Richard Biener <rguenther@suse.de>
10465 * hash-map.h (hash_map::iterator::operator*): Return
10466 references to key and value.
10468 2018-07-10 Jakub Jelinek <jakub@redhat.com>
10471 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
10472 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
10473 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
10474 (gimplify_omp_for): For composite loops, move outer
10475 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
10476 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
10477 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
10478 TREE_LIST for both the original class iterator and the "last" helper
10479 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
10480 loop, remember has_decl_expr from outer composite loops for the
10481 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
10483 2018-07-09 Martin Sebor <msebor@redhat.com>
10485 PR middle-end/77357
10486 PR middle-end/86428
10487 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
10488 accessing implicitly initialized array elements.
10489 * expr.c (string_constant): Handle string initializers of
10490 character arrays within aggregates.
10491 * gimple-fold.c (fold_array_ctor_reference): Add argument.
10492 Store element offset. As a special case, handle zero size.
10493 (fold_nonarray_ctor_reference): Same.
10494 (fold_ctor_reference): Add argument. Store subobject offset.
10495 * gimple-fold.h (fold_ctor_reference): Add argument.
10497 2018-07-09 Paul Koning <ni1d@arrl.net>
10499 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
10500 (pdp11_insn_cost): New function.
10501 (pdp11_md_asm_adjust): New function.
10502 (TARGET_INVALID_WITHIN_DOLOOP): Define.
10503 (pdp11_rtx_costs): Update to match machine better.
10504 (output_addr_const_pdp11): Correct format mismatch warnings.
10505 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
10506 * config/pdp11/pdp11.md: General change to add base_cost and/or
10507 length attributes for use by new pdp11_insn_cost function.
10508 (MIN_BRANCH): Correct definition.
10510 (doloop_end): Use standard pattern name for looping pattern.
10511 (doloop_end_nocc): New.
10512 (movsf): Add another constraint alternative.
10513 (zero_extendqihi2): Add constraint alternatives for not in place
10515 (zero_extendhisi2): Remove.
10516 (shift patterns): Add CC handling variants.
10520 (define_peephole2): New peephole to recognize mov that sets CC for
10523 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10525 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
10526 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
10529 2018-07-09 Jakub Jelinek <jakub@redhat.com>
10532 * real.c (real_nextafter): Return true if result is denormal.
10534 2018-07-09 Martin Liska <mliska@suse.cz>
10536 * common.opt: Add back wrongly removed attribute.
10538 2018-07-09 Richard Biener <rguenther@suse.de>
10541 * dwarf2out.c (gen_block_die): For an early generated DIE
10542 always output high/low PC attributes.
10544 2018-07-09 Tom de Vries <tdevries@suse.de>
10546 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
10547 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
10548 onto VAR_DECL with abstract origin.
10550 2018-07-07 Jim Wilson <jimw@sifive.com>
10552 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
10554 2018-07-07 Tom de Vries <tdevries@suse.de>
10556 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
10558 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
10560 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
10562 (combine_bound): Use wide-int overflow calculation instead of
10564 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
10566 * fold-const.c (int_const_binop_2): Same.
10567 (extract_muldiv_1): Same.
10568 (fold_div_compare): Same.
10569 (fold_abs_const): Same.
10571 * poly-int.h (add): Same.
10575 * predict.c (predict_iv_comparison): Same.
10576 * profile-count.c (slow_safe_scale_64bit): Same.
10577 * simplify-rtx.c (simplify_const_binary_operation): Same.
10578 * tree-chrec.c (tree_fold_binomial): Same.
10579 * tree-data-ref.c (split_constant_offset_1): Same.
10580 * tree-if-conv.c (idx_within_array_bound): Same.
10581 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
10582 * tree-ssa-phiopt.c (minmax_replacement): Same.
10583 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
10584 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
10585 * vr-values.c (vr_values::adjust_range_with_scev): Same.
10586 * wide-int.cc (wi::add_large): Same.
10587 (wi::mul_internal): Same.
10588 (wi::sub_large): Same.
10589 (wi::divmod_internal): Same.
10590 * wide-int.h: Change overflow type to overflow_type for neg, add,
10591 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
10592 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
10593 mul_internal, divmod_internal.
10594 (overflow_type): New enum.
10595 (accumulate_overflow): New.
10597 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
10599 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
10600 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
10602 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
10604 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
10605 argument is checked for zero before entering loop, avoid checking again.
10607 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
10609 * gimplify.h (generic_expr_could_trap_p): Set as global function.
10610 * gimplify.h (generic_expr_could_trap_p): Likwise.
10611 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
10613 2018-07-06 Jakub Jelinek <jakub@redhat.com>
10615 PR tree-optimization/86401
10616 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
10617 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
10618 (fold_bit_and_mask): ... here. New helper function for match.pd.
10619 * fold-const.h (fold_bit_and_mask): Declare.
10620 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
10622 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
10625 * target.def (translate_mode_attribute): New hook.
10626 * targhooks.h (default_translate_mode_attribute): Declare.
10627 * targhooks.c (default_translate_mode_attribute): New function.
10628 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
10629 * doc/tm.texi: Regenerate.
10630 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
10631 (rs6000_translate_mode_attribute): New function.
10633 2018-07-06 Paul Koning <ni1d@arrl.net>
10635 * doc/md.texi (define_split): Document DONE and FAIL.
10636 (define_peephole2): Ditto.
10638 2018-07-05 Jeff Law <law@redhat.com>
10640 PR tree-optimization/86010
10641 * tree-ssa-dse.c (compute_trims): More aggressively trim at
10642 both the head and tail of mem* and str* calls.
10644 2018-07-05 Jim Wilson <jimw@sifive.com>
10646 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
10648 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
10650 * config/aarch64/aarch64-simd.md: correct flags text for
10651 MIN_EXPR replacement.
10653 2018-07-05 James Clarke <jrtc27@jrtc27.com>
10655 * configure: Regenerated.
10657 2018-07-05 Carl Love <cel@us.ibm.com>
10659 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
10660 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
10661 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
10662 VSX_BUILTIN_DOUBLEL_V4SF.
10664 2018-07-05 Martin Sebor <msebor@redhat.com>
10667 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
10668 than its domain to compute its the upper bound of a char array.
10670 2018-07-05 Nathan Sidwell <nathan@acm.org>
10672 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
10673 * doc/cpp.texi: Update comment.
10674 * doc/tm.texi: Rebuilt.
10675 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
10676 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
10677 * doc/extend.texi (Backwards Compatibility): Clarify it is system
10678 headers affected by extern "C".
10679 * system.h: Poison NO_IMPLICIT_EXTERN_C.
10680 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
10681 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
10682 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
10683 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
10684 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
10685 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
10686 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
10687 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
10688 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
10689 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
10690 config/sparc/sp64-elf.h, config/spu/spu.h,
10691 config/stormy16/stormy16.h, config/v850/v850.h,
10692 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
10693 define NO_IMPLICIT_EXTERN_C.
10694 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
10696 2018-07-05 Tamar Christina <tamar.christina@arm.com>
10699 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
10700 instead of GET_MODE_SIZE when comparing Units.
10702 2018-07-05 Tamar Christina <tamar.christina@arm.com>
10705 * rtlanal.c (set_noop_p): Constrain on mode change,
10706 include hard-reg-set.h
10708 2018-07-05 Tamar Christina <tamar.christina@arm.com>
10710 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
10712 2018-07-05 Jakub Jelinek <jakub@redhat.com>
10715 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
10718 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
10720 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
10722 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
10725 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
10727 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
10729 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
10731 PR middle-end/86380
10732 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
10734 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
10736 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
10737 neg_*_op* variables.
10739 2018-07-04 Martin Liska <mliska@suse.cz>
10741 * tree-switch-conversion.c: Define
10742 max_ratio_for_speed and max_ratio_for_size constants.
10744 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
10745 Martin Liska <mliska@suse.cz>
10747 PR middle-end/66240
10750 * common.opt: Rename align options with 'str_' prefix.
10751 * common/config/i386/i386-common.c (set_malign_value): New
10753 (ix86_handle_option): Use it to set -falign-* options/
10754 * config/aarch64/aarch64-protos.h (struct tune_params): Change
10755 type from int to string.
10756 * config/aarch64/aarch64.c: Update default values from int
10758 * config/alpha/alpha.c (alpha_override_options_after_change):
10760 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
10761 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10762 max skip conditionally.
10763 * config/i386/freebsd.h (SUBALIGN_LOG): New.
10764 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10765 max skip conditionally.
10766 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10767 max skip conditionally.
10768 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
10769 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10770 max skip conditionally.
10771 * config/i386/i386.c (struct ptt): Change type from int to
10773 (ix86_default_align): Set default values.
10774 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
10775 max skip conditionally.
10776 * config/i386/iamcu.h (SUBALIGN_LOG): New.
10777 (ASM_OUTPUT_MAX_SKIP_ALIGN):
10778 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
10779 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10780 max skip conditionally.
10781 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
10782 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
10783 * config/i386/x86-64.h (SUBALIGN_LOG): New.
10784 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
10785 max skip conditionally.
10786 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
10787 * config/ia64/ia64.c (ia64_option_override): Set default values
10788 for alignment options.
10789 * config/m68k/m68k.c: Handle new str_align_* options.
10790 * config/mips/mips.c (mips_set_compression_mode): Change
10792 (mips_option_override): Set default values for options.
10793 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
10795 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10797 * config/rx/rx.c (rx_option_override): Likewise.
10798 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
10799 (LABEL_ALIGN): Use align_labels_log.
10800 (LOOP_ALIGN): Use align_loops_align.
10801 * config/s390/s390.c (s390_asm_output_function_label): Use new
10803 * config/sh/sh.c (sh_override_options_after_change):
10804 Change type of constants.
10805 * config/spu/spu.c (spu_sched_init): Likewise.
10806 * config/sparc/sparc.c (sparc_option_override): Set default
10807 values for options.
10808 * config/visium/visium.c (visium_option_override): Likewise.
10809 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
10810 emit p2align format with last argument if it's not needed.
10811 * doc/invoke.texi: Document extended format of -falign-*.
10812 * final.c: Use align_labels alignment.
10813 * flags.h (struct target_flag_state): Change type to use
10815 (struct align_flags_tuple): New.
10816 (struct align_flags): Likewise.
10817 (align_loops_log): Redefine macro to use new types.
10818 (align_loops_max_skip): Redefine macro to use new types.
10819 (align_jumps_log): Redefine macro to use new types.
10820 (align_jumps_max_skip): Redefine macro to use new types.
10821 (align_labels_log): Redefine macro to use new types.
10822 (align_labels_max_skip): Redefine macro to use new types.
10823 (align_functions_log): Redefine macro to use new types.
10824 (align_loops): Redefine macro to use new types.
10825 (align_jumps): Redefine macro to use new types.
10826 (align_labels): Redefine macro to use new types.
10827 (align_functions): Redefine macro to use new types.
10828 (align_functions_max_skip): Redefine macro to use new types.
10829 (align_loops_value): New macro.
10830 (align_jumps_value): New macro.
10831 (align_labels_value): New macro.
10832 (align_functions_value): New macro.
10833 * function.c (invoke_set_current_function_hook): Propagate
10834 alignment values from flags to global variables default in
10836 * ipa-icf.c (sem_function::equals_wpa): Use
10837 cl_optimization_option_eq instead of memcmp.
10838 * lto-streamer.h (cl_optimization_stream_out): Support streaming
10840 (cl_optimization_stream_in): Likewise.
10841 * optc-save-gen.awk: Support strings in cl_optimization.
10842 * opth-gen.awk: Likewise.
10843 * opts.c (finish_options): Remove error checking of invalid
10845 (MAX_CODE_ALIGN): Remove.
10846 (MAX_CODE_ALIGN_VALUE): Likewise.
10847 (parse_and_check_align_values): New function.
10848 (check_alignment_argument): Likewise.
10849 (common_handle_option): Use check_alignment_argument.
10850 * opts.h (parse_and_check_align_values): Declare.
10851 * toplev.c (init_alignments): Remove.
10852 (read_log_maxskip): New.
10853 (parse_N_M): Likewise.
10854 (parse_alignment_opts): Likewise.
10855 (backend_init_target): Remove usage of init_alignments.
10856 * toplev.h (parse_alignment_opts): Declare.
10857 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
10859 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
10860 * tree.c (cl_option_hasher::equal): New.
10861 * varasm.c: Use new global macros.
10863 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
10865 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
10866 Use a simpler align directive also if MAXSKIP = ALIGN-1.
10867 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10868 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10869 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10870 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
10871 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
10872 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
10873 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
10874 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10875 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10876 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10877 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10879 2018-07-04 Martin Liska <mliska@suse.cz>
10880 Jonathan Wakely <jwakely@redhat.com>
10882 * coverage.c: Use correct type.
10883 * doc/invoke.texi: Language correction.
10885 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
10888 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
10889 ENDBRANCH for non-tail call which may return via indirect branch.
10890 * doc/extend.texi: Document indirect_return attribute.
10892 2018-07-03 Martin Sebor <msebor@redhat.com>
10894 PR tree-optimization/86274
10895 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
10897 (format_floating): Correct handling of infinities and NaNs.
10899 2018-07-03 Martin Sebor <msebor@redhat.com>
10901 * print-tree.c (print_real_cst): New function.
10902 (print_node_brief): Call it.
10903 (print_node): Ditto.
10905 2018-07-03 Jeff Law <law@redhat.com>
10907 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
10908 into a single pattern.
10910 * config/h8300/h8300.md (ors code_iterator): New.
10911 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
10912 a single pattern and single splitter.
10913 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
10914 (iorqi3_1, xorqi3_1): Likewise.
10915 (iorqi3, xorqi3 expanders): Similarly.
10917 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
10918 (movmd_internal) into a single pattern using the P mode iterator.
10919 (movmd splitters): Similarly.
10920 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
10921 (movsd splitters): Similarly.
10923 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
10924 ADDB, ADDW and ADDL into a single ADD attribute which selects the
10925 right table based on the size of the operand.
10926 * config/h8300/h8300.md (length_table): Corresponding changes. All
10927 references to "addb", "addw" and "addl" changed to "add".
10928 (btst patterns): Merge two variants into a single pattern.
10929 (tstqi, tsthi): Likewise.
10930 (addhi3_incdec, addsi3_incdec): Likewise.
10931 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
10932 (mulhi3, mulsi3): Likewise.
10933 (udivhi3, udivsi3): Likewise.
10934 (divhi3, divsi3): Likewise.
10935 (andorqi3, andorhi3, andorsi3): Likewise.
10937 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
10940 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
10941 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
10943 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10945 PR tree-optimization/85694
10946 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
10947 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
10949 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
10950 (<u>avg<mode>3_ceil): New patterns.
10952 2018-07-03 David Malcolm <dmalcolm@redhat.com>
10954 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
10955 scan-tree-dump directive.
10956 * gcc.dg/vect/slp-perm-2.c: Likewise.
10957 * gcc.dg/vect/slp-perm-3.c: Likewise.
10958 * gcc.dg/vect/slp-perm-5.c: Likewise.
10959 * gcc.dg/vect/slp-perm-6.c: Likewise.
10960 * gcc.dg/vect/slp-perm-7.c: Likewise.
10961 * gcc.dg/vect/slp-perm-8.c: Likewise.
10963 2018-07-03 Marek Polacek <polacek@redhat.com>
10965 PR middle-end/86202
10966 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
10968 2018-07-03 Richard Biener <rguenther@suse.de>
10971 * tree-ssa-structalias.c (find_func_clobbers): Properly
10972 handle indirect calls.
10974 2018-07-03 Jeff Law <law@redhat.com>
10976 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
10977 (shifts): New code iterator.
10978 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
10979 expander. Fix HImode handling on H8/SX.
10980 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
10981 (subqi3, subhi3, subsi3 expanders): Likewise.
10982 (andqi3, andhi3, andsi3 expanders): Likewise.
10983 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
10984 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
10985 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
10986 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
10987 (zero_extendqihi2, zero_extendqisi2): Likewise.
10988 (extendqihi2, extendqisi2): Likewise.
10989 (rotlqi3, rotlhi3, rotlsi3): Likewise.
10990 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
10991 (rotlqi3_1, rotlhi3_1): Likewise.
10992 (logicalhi3_sn, logicalsi3_sn): Likewise.
10993 (logicalhi3, logicalsi3): Likewise.
10995 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
10997 * tree-vect-patterns.c (vect_recog_rotate_pattern)
10998 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
10999 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
11000 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
11001 type to append_pattern_def_seq instead of creating a stmt_vec_info
11003 (build_mask_conversion): Likewise. Remove vinfo argument.
11004 (vect_add_conversion_to_patterm): Likewise, renaming to...
11005 (vect_add_conversion_to_pattern): ...this.
11006 (vect_recog_mask_conversion_pattern): Update call to
11007 build_mask_conversion. Pass the vector type to
11008 append_pattern_def_seq here too.
11009 (vect_recog_gather_scatter_pattern): Update call to
11010 vect_add_conversion_to_pattern.
11012 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11014 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
11015 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
11016 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
11017 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
11018 STMT_VINFO_PATTERN_DEF_SEQ to null here.
11019 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
11020 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
11021 append_pattern_def_seq instead of new_pattern_def_seq.
11022 (vect_recog_divmod_pattern): Do both of the above.
11023 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
11026 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11028 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
11029 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
11030 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
11031 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
11032 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
11033 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
11034 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
11035 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
11036 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
11037 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
11038 parameter with a single stmt_vec_info.
11039 (vect_recog_func_ptr): Likewise.
11040 (vect_recog_gather_scatter_pattern): Likewise, folding in...
11041 (vect_try_gather_scatter_pattern): ...this.
11042 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
11043 the stmt_vec_info of the statement to be matched. Don't clear
11044 STMT_VINFO_RELATED_STMT.
11045 (vect_pattern_recog): Update call accordingly.
11047 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11049 PR tree-optimization/85694
11050 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
11051 (uavgM3_ceil): Document new optabs.
11052 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
11053 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
11055 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
11056 (savg_ceil_optab): New optabs.
11057 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
11058 (vect_vect_recog_func_ptrs): Add it.
11059 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
11060 constant directly from the associated lhs.
11062 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11064 * tree-vect-patterns.c (vect_split_statement): New function.
11065 (vect_convert_input): Use it to try to split an existing cast.
11067 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11069 * poly-int.h (print_hex): New function.
11070 * dumpfile.h (dump_dec, dump_hex): Declare.
11071 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
11072 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
11073 min_input_precision, operation_precision and operation_sign.
11074 * tree-vect-patterns.c (vect_get_range_info): New function.
11075 (vect_same_loop_or_bb_p, vect_single_imm_use)
11076 (vect_operation_fits_smaller_type): Delete.
11077 (vect_look_through_possible_promotion): Add an optional
11078 single_use_p parameter.
11079 (vect_recog_over_widening_pattern): Rewrite to use new
11080 stmt_vec_info infomration. Handle one operation at a time.
11081 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
11082 (vect_truncatable_operation_p, vect_set_operation_type)
11083 (vect_set_min_input_precision): New functions.
11084 (vect_determine_min_output_precision_1): Likewise.
11085 (vect_determine_min_output_precision): Likewise.
11086 (vect_determine_precisions_from_range): Likewise.
11087 (vect_determine_precisions_from_users): Likewise.
11088 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
11089 (vect_vect_recog_func_ptrs): Put over_widening first.
11091 (vect_pattern_recog): Call vect_determine_precisions.
11093 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11095 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
11096 statements that have been replaced by further pattern statements.
11097 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
11099 2018-07-03 Richard Biener <rguenther@suse.de>
11101 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
11102 always set *dt. Dump vectype in vectype overload.
11103 * dumpfile.h (dump_gimple_expr): New function.
11104 (dump_gimple_expr_loc): Likewise.
11105 * dumpfile.c (dump_gimple_expr): New function.
11106 (dump_gimple_expr_loc): Likewise.
11108 2018-07-02 Jeff Law <law@redhat.com>
11110 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
11111 the H8/300, H8/300H and H8/S variants into a single pattern.
11112 (movhi_h8300, movqi_h8300hs): Similarly.
11113 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
11114 (QHI mode iterator): New.
11116 * config/h8300/h8300.md: Remove trailing whitespace.
11118 2018-07-02 Jim Wilson <jimw@sifive.com>
11120 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
11121 instead of emit_insn for interrupt returns.
11122 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
11123 (riscv_sret, riscv_uret): Likewise.
11125 2018-07-02 David Malcolm <dmalcolm@redhat.com>
11127 * pretty-print.c (selftest::test_pp_format): Move save and restore
11128 of quotes to class auto_fix_quotes, and add an instance.
11129 * selftest.c: Include "intl.h".
11130 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
11131 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
11132 * selftest.h (selftest::auto_fix_quotes): New class.
11134 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
11136 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
11137 (aarch64_sve_prepare_conditional_op): Remove.
11138 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
11139 Allow aarch64_simd_reg_or_zero as select operand; remove
11140 the aarch64_sve_prepare_conditional_op call.
11141 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
11142 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
11143 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
11144 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
11145 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
11146 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
11147 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
11148 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
11149 and a splitters to match all of the *_any patterns.
11150 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
11152 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
11153 (SVE_COND_FP_BINARY_REV): Remove.
11154 (sve_int_op_rev, sve_fp_op_rev): New.
11155 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
11156 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
11157 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
11158 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
11159 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
11160 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
11161 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
11162 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
11163 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
11165 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
11166 Remove match_dup 1 from the inner unspec.
11167 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
11169 * config/aarch64/aarch64.md (movprfx): New attr.
11170 (length): Default movprfx to 8.
11171 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
11172 (*madd<SVE_I>, *msub<SVE_I): Likewise.
11173 (*<su>mul<SVE_I>3_highpart): Likewise.
11174 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
11175 (*v<ASHIFT><SVE_I>3): Likewise.
11176 (*<su><MAXMIN><SVE_I>3): Likewise.
11177 (*<su><MAXMIN><SVE_F>3): Likewise.
11178 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
11179 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
11180 (*div<SVE_F>4): Likewise.
11182 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
11184 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
11187 2018-07-02 Richard Biener <rguenther@suse.de>
11189 PR tree-optimization/86363
11190 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
11191 memset argument refers to a non-variable address.
11193 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
11195 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
11196 {PLUS,MINUS}_EXPR code to...
11197 (adjust_symbolic_bound): ...here,
11198 (combine_bound): ...here,
11199 (set_value_range_with_overflow): ...and here.
11201 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
11203 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
11205 (extract_range_from_abs_expr): ...here.
11207 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
11209 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
11210 -fno-omit-frame-pointer when not optimizing.
11212 2018-07-02 Martin Liska <mliska@suse.cz>
11215 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
11216 (propagate_nothrow): Likewise.
11218 2018-07-02 Martin Liska <mliska@suse.cz>
11221 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
11223 2018-07-02 David Malcolm <dmalcolm@redhat.com>
11225 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
11226 function in r262149, changing "loc" param from source_location to
11227 const dump_location_t &.
11228 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
11229 declaration, as above.
11231 2018-07-01 Paul Koning <ni1d@arrl.net>
11233 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
11234 -munit-asm, -mgnu-asm, -mdec-asm.
11235 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
11236 (pdp11_output_labelref): New.
11237 (pdp11_output_def): New.
11238 (pdp11_output_addr_vec_elt): New.
11239 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
11240 %# and %@ format codes.
11241 (pdp11_option_override): New.
11242 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
11243 (pdp11_output_ident): New.
11244 (pdp11_asm_named_section): New.
11245 (pdp11_asm_init_sections): New.
11246 (pdp11_file_start): New.
11247 (pdp11_file_end): New.
11248 (output_ascii): Use .ascii/.asciz for -mdec-asm.
11249 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
11250 %o, like %c but octal.
11251 (pdp11_option_override): New.
11252 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
11254 (DATA_SECTION_ASM_OP): Ditto.
11255 (READONLY_DATA_SECTION_ASM_OP): New.
11256 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
11257 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
11258 (ASM_OUTPUT_LABELREF): Ditto.
11259 (ASM_OUTPUT_DEF): Ditto.
11260 (ASM_OUTPUT_EXTERNAL): New.
11261 (ASM_OUTPUT_SOURCE_FILENAME): New.
11262 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
11263 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
11264 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
11265 %# and %@ format codes.
11266 * config/pdp11/pdp11.opt (mgnu-asm): New.
11267 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
11268 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
11269 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
11271 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
11273 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
11274 dereferencing path[] beyond its length.
11276 (debug_all_paths): New.
11277 (rewire_first_differing_edge): New.
11278 (adjust_paths_after_duplication): New.
11279 (duplicate_thread_path): Call adjust_paths_after_duplication.
11281 (thread_through_all_blocks): Add new argument to
11282 duplicate_thread_path.
11284 2018-06-30 Jim Wilson <jimw@sifive.com>
11286 * config/riscv/predicates.md (p2m1_shift_operand): New.
11287 (high_mask_shift_operand): New.
11288 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
11289 pattern using p2m1_shift_operand.
11290 (lshsi3_zero_extend_3+2): New combiner pattern using
11291 high_mask_shift_operand.
11293 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11295 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
11297 (vect_recog_rotate_pattern): ...here.
11298 (vect_convert_input): Try to insert casts of invariants in the
11300 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
11301 preheader to be empty.
11303 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11305 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
11306 vector type. If given, install it in the new statement's
11307 STMT_VINFO_VECTYPE.
11308 (vect_element_precision): New function.
11309 (vect_unpromoted_value): New struct.
11310 (vect_unpromoted_value::vect_unpromoted_value): New function.
11311 (vect_unpromoted_value::set_op): Likewise.
11312 (vect_look_through_possible_promotion): Likewise.
11313 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
11314 (vect_widened_op_tree, vect_convert_input): Likewise.
11315 (vect_convert_inputs, vect_convert_output): Likewise.
11316 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
11317 to handle the optional cast of the multiplication result and
11318 vect_widened_op_tree to detect the widened multiplication itself.
11319 Do not require the input and output of promotion casts to have
11320 the same sign, but base the signedness of the operation on the
11321 input rather than the result. If the pattern includes two
11322 promotions, check that those promotions have the same sign.
11323 Do not restrict the MULT_EXPR handling to a double-width result;
11324 handle quadruple-width results and wider. Use vect_convert_inputs
11325 to convert the inputs to the common type.
11326 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
11327 to handle the optional cast of the ABS result. Also allow a sign
11328 change or a sign extension between the ABS and MINUS.
11329 Use vect_widened_op_tree to detect the widened subtraction and use
11330 vect_convert_inputs to convert the inputs to the common type.
11331 (vect_handle_widen_op_by_const): Delete.
11332 (vect_recog_widen_op_pattern): New function.
11333 (vect_recog_widen_mult_pattern): Use it.
11334 (vect_recog_widen_shift_pattern): Likewise.
11335 (vect_recog_widen_sum_pattern): Use
11336 vect_look_through_possible_promotion to handle the promoted
11339 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11341 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
11342 the containing gimple_seq *.
11343 * gimple-iterator.h (gsi_for_stmt): Declare it.
11344 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
11345 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
11346 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
11347 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
11348 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
11350 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
11352 (vect_mark_pattern_stmts): ...here. Handle cases in which the
11353 statement being replaced is part of an existing pattern
11354 definition sequence, inserting the new pattern statements before
11356 (vect_pattern_recog_1): Don't return a bool. If the statement
11357 is already part of a pattern, instead apply pattern matching
11358 to the pattern definition statements. Don't clear the
11359 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
11360 (vect_pattern_recog): Don't break after the first match;
11361 continue processing the pattern definition statements instead.
11362 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
11364 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11366 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
11367 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
11368 (vect_recog_widen_sum_pattern): Use it.
11370 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11372 * tree-vect-loop.c (vectorizable_reduction): Assert that the
11373 phi is not a pattern statement and has not been replaced by
11374 a pattern statement.
11375 * tree-vect-patterns.c (type_conversion_p): Don't check
11376 STMT_VINFO_IN_PATTERN_P.
11377 (vect_recog_vector_vector_shift_pattern): Likewise.
11378 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
11379 the pattern statement rather than the original statement; check
11380 directly for a WIDEN_MULT_EXPR here.
11381 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
11382 vect_is_simple_use to return the pattern statement rather
11383 than the original statement; use is_pattern_stmt_p to check
11384 for such a pattern statement.
11385 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
11386 to return the pattern statement rather than the original statement;
11387 don't do the same transformation here.
11388 (vect_is_simple_use): If the defining statement has been replaced
11389 by a pattern statement, return the pattern statement instead.
11390 Remove the corresponding (local) transformation from the vectype
11393 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11395 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
11396 end and default to null.
11397 * tree-vect-loop.c (vect_create_epilog_for_reduction)
11398 (vectorizable_reduction): Update calls accordingly, dropping the
11399 gimple ** argument if the passed-back statement isn't needed.
11400 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
11401 (vect_recog_rotate_pattern): Likewise.
11402 (vect_recog_mask_conversion_pattern): Likewise.
11403 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
11404 (vect_mask_constant_operand_p): Likewise.
11405 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
11406 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
11407 (get_group_load_store_type, get_load_store_type): Likewise.
11408 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
11409 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
11410 (vectorizable_conversion, vectorizable_assignment): Likewise.
11411 (vectorizable_shift, vectorizable_operation): Likewise.
11412 (vectorizable_store, vect_is_simple_cond): Likewise.
11413 (vectorizable_condition, vectorizable_comparison): Likewise.
11414 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
11415 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
11416 and move it to the end. Cope with null def_stmt_outs.
11418 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
11420 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
11422 2018-06-29 Jeff Law <law@redhat.com>
11424 * config/v850/v850.c (v850_legitimate_address_p): Handle large
11425 displacements for TARGET_V850E2V3 and newer.
11426 (TARGET_LRA_P): Remove. Defaults to LRA now.
11427 * config/v850/v850.md (sign23byte_load): Remove.
11428 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
11429 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
11431 2018-06-29 Martin Liska <mliska@suse.cz>
11434 * coverage.c (coverage_init): Mangle full path name.
11435 * doc/invoke.texi: Document the change.
11436 * gcov-io.c (mangle_path): New.
11437 * gcov-io.h (mangle_path): Likewise.
11438 * gcov.c (mangle_name): Use mangle_path for path mangling.
11440 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11442 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
11443 if starting source register is not even.
11445 2018-06-29 Martin Liska <mliska@suse.cz>
11447 PR tree-optimization/86263
11448 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
11449 Make edge redirection.
11451 2018-06-29 David Malcolm <dmalcolm@redhat.com>
11453 * dumpfile.c (dump_loc): Add indentation based on scope depth.
11454 (dump_scope_depth): New variable.
11455 (get_dump_scope_depth): New function.
11456 (dump_begin_scope): New function.
11457 (dump_end_scope): New function.
11458 * dumpfile.h (get_dump_scope_depth): New declaration.
11459 (dump_begin_scope): New declaration.
11460 (dump_end_scope): New declaration.
11461 (class auto_dump_scope): New class.
11462 (AUTO_DUMP_SCOPE): New macro.
11463 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
11466 2018-06-29 Richard Biener <rguenther@suse.de>
11468 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
11469 compute_all_dependences succeeds.
11470 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
11471 exceed --param loop-max-datarefs-for-datadeps.
11473 2018-06-29 Jakub Jelinek <jakub@redhat.com>
11475 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
11477 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
11480 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
11481 alternative 0 in preferred_for_speed attribute.
11483 2018-06-28 Paul Koning <ni1d@arrl.net>
11485 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
11486 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
11487 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
11488 * config/pdp11/pdp11.md: Correct "length" attribute calculation
11489 for shift insn patterns.
11491 2018-06-28 David Malcolm <dmalcolm@redhat.com>
11493 * cgraph.c (cgraph_node::get_body): Replace assignments to
11494 "dump_file" with calls to set_dump_file.
11495 * dumpfile.c (alt_dump_file): Make static, and group with...
11496 (alt_flags): ...this definition.
11497 (dumps_are_enabled): New variable.
11498 (refresh_dumps_are_enabled): New function.
11499 (set_dump_file): New function.
11500 (set_alt_dump_file): New function.
11501 (gcc::dump_manager::dump_start): Replace assignments to
11502 "dump_file" and "alt_dump_file" with calls to set_dump_file and
11504 (gcc::dump_manager::dump_finish): Likewise.
11505 * dumpfile.h (alt_dump_file): Delete decl.
11506 (dumps_are_enabled): New variable decl.
11507 (set_dump_file): New function decl.
11508 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
11510 * tree-nested.c (lower_nested_functions): Replace assignments to
11511 "dump_file" with calls to set_dump_file.
11513 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
11515 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
11516 goto_locus of each outgoing edge of each basic block.
11518 2018-06-28 Richard Biener <rguenther@suse.de>
11520 * dwarf2out.c (decl_scope_table): Remove.
11521 (push_decl_scope): Likewise.
11522 (pop_decl_scope): Likewise.
11523 (gen_type_die_for_member): Do not call push/pop_decl_scope.
11524 (gen_struct_or_union_type_die): Likewise.
11525 (gen_tagged_type_die): Likewise.
11526 (dwarf2out_init): Do not initialize decl_scope_table.
11527 (dwarf2out_c_finalize): Do not free it.
11529 2018-06-28 Richard Biener <rguenther@suse.de>
11531 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
11532 deciding whether to not re-use a DIE.
11534 2018-06-28 Richard Biener <rguenther@suse.de>
11536 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
11537 DW_AT_abstract_origin attribute.
11539 2018-06-28 Martin Liska <mliska@suse.cz>
11541 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
11542 Use newly introduced constants.
11543 * tree-switch-conversion.h (struct jump_table_cluster):
11544 Define max_ratio_for_size and max_ratio_for_speed.
11546 2018-06-28 Martin Liska <mliska@suse.cz>
11548 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
11549 Add new checking assert to catch invalid state.
11550 (jump_table_cluster::can_be_handled): Handle single case
11552 (jump_table_cluster::is_beneficial): Bail out for such case.
11553 (bit_test_cluster::find_bit_tests):
11554 Add new checking assert to catch invalid state.
11555 (bit_test_cluster::can_be_handled): Handle single case
11557 (bit_test_cluster::is_beneficial): Bail out for such case.
11558 (switch_decision_tree::analyze_switch_statement):
11561 2018-06-28 Martin Liska <mliska@suse.cz>
11563 * common.opt: Introduce -completion option.
11564 * gcc.c (driver_handle_option): Handle it.
11565 (driver::main): Print completions if completion
11567 * opt-suggestions.c (option_proposer::get_completions):
11569 (option_proposer::suggest_completion): Likewise.
11570 (option_proposer::find_param_completions): Likewise.
11571 (verify_autocompletions): Likewise.
11572 (test_completion_valid_options): Likewise.
11573 (test_completion_valid_params): Likewise.
11574 (in_completion_p): Likewise.
11575 (empty_completion_p): Likewise.
11576 (test_completion_partial_match): Likewise.
11577 (test_completion_garbage): Likewise.
11578 (opt_proposer_c_tests): Likewise.
11579 * opt-suggestions.h: Declare new functions.
11580 * opts.c (common_handle_option): Handle OPT__completion_.
11581 * selftest-run-tests.c (selftest::run_tests): Add
11582 opt_proposer_c_tests.
11583 * selftest.c (assert_str_startswith): New.
11584 * selftest.h (assert_str_startswith): Likewise.
11585 (opt_proposer_c_tests): New.
11586 (ASSERT_STR_STARTSWITH): Likewise.
11588 2018-06-28 Martin Liska <mliska@suse.cz>
11590 * Makefile.in: Add opt-suggestions.o.
11591 * gcc-main.c: Include opt-suggestions.h.
11592 * gcc.c (driver::driver): Likewise.
11593 (driver::~driver): Remove m_option_suggestions.
11594 (driver::build_option_suggestions): Moved to option_proposer.
11595 (driver::suggest_option): Likewise.
11596 (driver::handle_unrecognized_options): Use option_proposer.
11597 * gcc.h (class driver): Add new memver m_option_proposer.
11598 * opt-suggestions.c: New file.
11599 * opt-suggestions.h: New file.
11601 2018-06-28 Martin Liska <mliska@suse.cz>
11603 * vec.h (class auto_string_vec): New (moved from auto_argvec).
11604 (auto_string_vec::~auto_string_vec): Likewise.
11606 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
11608 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
11609 prevent_decl_creation_for_types fields up and add reset_location field.
11610 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
11611 statement if id->reset_location is true.
11612 (copy_edges_for_bb): Do not set goto_locus on the new edges if
11613 id->reset_location is true.
11614 (copy_phis_for_bb): Force input_location on the arguments if
11615 id->reset_location is true.
11616 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
11617 is set on the function to be inlined.
11619 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
11621 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
11623 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
11625 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
11626 registers for Pmode.
11627 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
11628 hard registers for the clobbered pseudo.
11630 2018-06-27 Paul Koning <ni1d@arrl.net>
11632 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
11633 mutually exclusive options.
11634 * config/pdp11/constraints.md (h): New constraint.
11635 (O): Update definition to match shift code generation.
11636 (D): New constraint.
11637 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
11639 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
11641 (output_jump): Change arguments.
11642 (pdp11_fixed_cc_regs): New function.
11643 (pdp11_cc_mode): Ditto.
11644 (pdp11_expand_shift): Ditto.
11645 (pdp11_assemble_shift): Ditto.
11646 (pdp11_small_shift): Ditto.
11647 (pdp11_branch_cost): Remove.
11648 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
11650 (pdp11_register_move_cost): Update for CC registers.
11651 (pdp11_rtx_costs): Add case for LSHIFTRT.
11652 (pdp11_output_jump): Add CCNZ mode conditional branches.
11653 (notice_update_cc_on_set): Remove.
11654 (pdp11_cc_mode): New function.
11655 (simple_memory_operand): Correct pre/post decrement case.
11656 (no_side_effect_operand): New function.
11657 (pdp11_regno_reg_class): Add CC_REGS class.
11658 (pdp11_fixed_cc_regs): New function.
11659 (pdp11_small_shift): New function.
11660 (pdp11_expand_shift): New function to expand shift insns.
11661 (pdp11_assemble_shift): New function to output shifts.
11662 (pdp11_branch_cost): Remove.
11663 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
11664 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
11665 (WCHAR_TYPE): Ditto.
11666 (PTRDIFF_TYPE): Ditto.
11667 (ADJUST_INSN_LENGTH): New macro.
11668 (FIXED_REGISTERS): Add CC registers.
11669 (CALL_USED_REGISTERS): Ditto.
11670 (reg_class): Ditto.
11671 (REG_CLASS_NAMES): Ditto.
11672 (REG_CLASS_CONTENTS): Ditto.
11673 (SELECT_CC_MODE): Use new function.
11674 (TARGET_FLAGS_REGNUM): New macro.
11675 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
11676 (cc0_reg_rtx): Remove.
11677 (CC_STATUS_MDEP): Remove.
11678 (CC_STATUS_MDEFP_INIT): Remove.
11679 (CC_IN_FPU): Remove.
11680 (NOTICE_UPDATE_CC): Remove.
11681 (REGISTER_NAMES): Add CC registers.
11682 (BRANCH_COST): Change to constant 1.
11683 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
11685 * config/pdp11/pdp11.opt (mbcopy): Remove.
11686 (mbcopy-builtin): Remove.
11687 (mbranch-cheap): Remove.
11688 (mbranch-expensive): Remove.
11689 * config/pdp11/predicates.md (expand_shift_operand): Update to
11690 match shift code generation.
11691 (ccnz_operator): New predicate.
11692 * doc/invoke.texi (PDP-11 Options): Remove deleted options
11693 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
11694 Remove non-existent option -mabshi, -mno-abshi. Document mutually
11696 * doc/md.texi (PDP-11): Document new D and h constraints. Update
11697 description of O constraint.
11699 2018-06-27 Jeff Law <law@redhat.com>
11700 Austin Law <austinklaw@gmail.com>
11702 * config/v850/v850.md (addsi3_set_flags): New pattern.
11703 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
11704 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
11705 (zero_extendhisi2_v850_set_flags): Likewise.
11706 (zero_extendqisi2_v850_set_flags): Likewise.
11707 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
11708 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
11709 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
11711 * config/v850/v850-protos.h (notice_update_cc): Remove.
11712 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
11713 (v850_print_operand): Handle 'D' and "d".
11714 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
11715 Add handling of arithmetic/logical operations compared against zero.
11716 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
11717 Do not look at v850_compare_op, instead get mode from last argument.
11718 (v850_gen_compare): Remove
11719 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
11720 after reload for prologue insns.
11721 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
11723 (construct_save_jarl): Likewise.
11724 (TARGET_FLAGS_REGNUM): Define.
11725 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
11726 (NOTICE_UPDATE_CC): Remove.
11727 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
11728 than cc0. Conditionalize on reload_completed.
11729 (cmpsi_insn, setfcc_insn): Likewise.
11730 (tst1 splitter): Turn into define_and_split which sets the flags
11732 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
11733 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
11734 (cstoresf4, cstoredf4): Clobber the flags.
11735 (cmpsi, cmpsf, cmpdf): Remove expanders.
11736 (setf_insn): Remove pattern.
11737 (addsi3): Turn into define_and_split which clobbers the flags after
11738 reload and a suitable pattern (addsi3_clobber_flags) for use after
11740 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
11741 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
11742 (ashrsi3, ashrsi3_v850e2): Likewise.
11743 (bins): Clobber the flags.
11744 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
11745 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
11746 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
11747 (call_value_internal_short, call_value_internal_long): Likewise.
11748 (callt_save_interrupt, callt_return_interrupt): Likewise.
11749 (save_interrupt, return_interrupt): Likewise.
11750 (callt_save_all_interrupt, save_all_interrupt): Likewise.
11751 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
11752 (restore_all_interrupt, _restore_all_interrupt): Likewise.
11753 (All FP comparisons): Only allow after reload has completed.
11755 (divh, divhu): Tweak output template.
11756 (branch_z_normal, branch_z_invert): Remove
11757 (branch_nz_normal, branch_nz_invert): Likewise.
11758 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
11760 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
11761 * config/v850/v850.c (notice_update_cc): Remove.
11762 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
11763 (CC_NO_CARRY): Likewise.
11764 (NOTICE_UPDATE_CC): Define to nothing.
11765 * config/v850/v850.md: Remove block comment on cc0 handling
11766 Remove "cc" attribute from all patterns. Remove cc_status handling
11767 from all patterns. Minor formatting fixes.
11769 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11771 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
11772 (cortex-a76.cortex-a55): Likewise.
11773 * config/aarch64/aarch64-tune.md: Regenerate.
11774 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
11775 cortex-a76.cortex-a55.
11777 2018-06-27 Jeff Law <law@redhat.com>
11779 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
11780 (MULTILIB_DIRNAMES): Similarly.
11782 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
11784 * gimple.h (gimple_return_retbnd): Delete.
11785 (gimple_return_set_retbnd): Likewise.
11786 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
11787 gimple_return_set_retbnd.
11788 * gimple-pretty-print.c (dump_gimple_return): Remove call to
11789 gimple_return_retbnd and adjust.
11790 * tree-inline.h (struct copy_body_data): Remove retbnd field.
11791 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
11792 Explicitly return NULL in a couple more cases. Move assertion
11793 on debug statements and remove unreachable code.
11794 (reset_debug_binding): Do not test id->retbnd.
11795 (expand_call_inline): Do not set it.
11797 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
11799 * configure.ac: Add --disable-gcov option.
11800 * configure: Regenerate.
11801 * Makefile.in: Honour @enable_gcov@.
11802 * doc/install.texi: Document --disable-gcov.
11804 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11806 * config/arm/arm-cpus.in (cortex-a76): New entry.
11807 (cortex-a76.cortex-a55): Likewise.
11808 * config/arm/arm-tables.opt: Regenerate.
11809 * config/arm/arm-tune.md: Likewise.
11810 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
11811 * doc/invoke.texi (ARM Options): Document cortex-a76 and
11812 cortex-a76.cortex-a55.
11814 2018-06-27 Tamar Christina <tamar.christina@arm.com>
11817 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
11819 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
11821 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
11823 (EPILOGUE_USES): Likewise.
11825 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
11827 * tree-inline.c (remap_location): New function extracted from...
11828 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
11829 (copy_phis_for_bb): ...here. Call remap_location.
11830 (copy_cfg_body): Adjust call to copy_edges_for_bb.
11832 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
11834 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
11835 unaligned vsx for 16B memset.
11837 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
11840 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
11841 ieee128_float_type_node to long_double_type_node unless
11842 TARGET_LONG_DOUBLE_128 is set.
11844 2018-06-26 David Malcolm <dmalcolm@redhat.com>
11846 * cfgloop.c (get_loop_location): Convert return type from
11847 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
11848 by implicit construction from rtx_insn *, and using
11849 dump_user_location_t::from_function_decl for the fallback case.
11850 * cfgloop.h (get_loop_location): Convert return type from
11851 location_t to dump_user_location_t.
11852 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
11853 dump_printf_loc to pass in a dump_location_t rather than a
11854 location_t, via the gimple stmt.
11855 * coverage.c (get_coverage_counts): Update calls to
11856 dump_printf_loc to pass in dump_location_t rather than a
11858 * doc/optinfo.texi (Dump types): Convert example of
11859 dump_printf_loc from taking "locus" to taking "insn". Update
11860 description of the "_loc" calls to cover dump_location_t.
11861 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
11863 (dump_user_location_t::dump_user_location_t): New constructors,
11864 from gimple *stmt and rtx_insn *.
11865 (dump_user_location_t::from_function_decl): New function.
11866 (dump_loc): Make static.
11867 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
11868 const dump_location_t &.
11869 (dump_generic_expr_loc): Delete.
11870 (dump_printf_loc): Convert param "loc" from location_t to
11871 const dump_location_t &.
11872 (selftest::test_impl_location): New function.
11873 (selftest::dumpfile_c_tests): New function.
11874 * dumpfile.h: Include "profile-count.h".
11875 (class dump_user_location_t): New class.
11876 (struct dump_impl_location_t): New struct.
11877 (class dump_location_t): New class.
11878 (dump_printf_loc): Convert 2nd param from source_location to
11879 const dump_location_t &.
11880 (dump_generic_expr_loc): Delete.
11881 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
11882 const dump_location_t &.
11883 * gimple-fold.c (fold_gimple_assign): Update call to
11884 dump_printf_loc to pass in a dump_location_t rather than a
11885 location_t, via the gimple stmt.
11886 (gimple_fold_call): Likewise.
11887 * gimple-loop-interchange.cc
11888 (loop_cand::analyze_iloop_reduction_var): Update for change to
11889 check_reduction_path.
11890 (tree_loop_interchange::interchange): Update for change to
11891 find_loop_location.
11892 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
11893 change in return-type of find_loop_location.
11894 (graphite_regenerate_ast_isl): Likewise.
11895 * graphite-optimize-isl.c (optimize_isl): Likewise.
11896 * graphite.c (graphite_transform_loops): Likewise.
11897 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
11898 pass in a dump_location_t rather than a location_t, via the
11900 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
11901 * ipa.c (walk_polymorphic_call_targets): Likewise.
11902 * loop-unroll.c (report_unroll): Convert "locus" param from
11903 location_t to dump_location_t.
11904 (decide_unrolling): Update for change to get_loop_location's
11906 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
11907 location_t to dump_user_location_t.
11908 (grid_find_single_omp_among_assignments_1): Updates calls to
11909 dump_printf_loc to pass in a dump_location_t rather than a
11910 location_t, via the gimple stmt.
11911 (grid_parallel_clauses_gridifiable): Convert "tloc" from
11912 location_t to dump_location_t. Updates calls to dump_printf_loc
11913 to pass in a dump_location_t rather than a location_t, via the
11915 (grid_inner_loop_gridifiable_p): Likewise.
11916 (grid_dist_follows_simple_pattern): Likewise.
11917 (grid_gfor_follows_tiling_pattern): Likewise.
11918 (grid_target_follows_gridifiable_pattern): Likewise.
11919 (grid_attempt_target_gridification): Convert initialization
11920 of local "grid" from memset to zero-initialization; FIXME: does
11921 this require C++11? Update call to dump_printf_loc to pass in a
11922 optinfo_location rather than a location_t, via the gimple stmt.
11923 * profile.c (read_profile_edge_counts): Updates call to
11924 dump_printf_loc to pass in a dump_location_t rather than a
11926 (compute_branch_probabilities): Likewise.
11927 * selftest-run-tests.c (selftest::run_tests): Call
11929 * selftest.h (dumpfile_c_tests): New decl.
11930 * tree-loop-distribution.c (pass_loop_distribution::execute):
11931 Update for change in return type of find_loop_location.
11932 * tree-parloops.c (parallelize_loops): Likewise.
11933 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
11934 "locus" from location_t to dump_user_location_t.
11935 (canonicalize_loop_induction_variables): Likewise.
11936 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
11937 for change in return type of find_loop_location.
11938 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
11939 to dump_printf_loc to pass in a dump_location_t rather than a
11940 location_t, via the stmt.
11941 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
11943 * tree-vect-loop-manip.c (find_loop_location): Convert return
11944 type from source_location to dump_user_location_t.
11945 (vect_do_peeling): Update for above change.
11946 (vect_loop_versioning): Update for change in type of
11948 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
11949 from location_t to dump_user_location_t.
11950 (vect_estimate_min_profitable_iters): Update for change in type
11952 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
11953 location_t to dump_location_t.
11954 (vect_slp_bb): Update for change in type of vect_location.
11955 * tree-vectorizer.c (vect_location): Convert from source_location
11956 to dump_user_location_t.
11957 (try_vectorize_loop_1): Update for change in vect_location's type.
11958 (vectorize_loops): Likewise.
11959 (increase_alignment): Likewise.
11960 * tree-vectorizer.h (vect_location): Convert from source_location
11961 to dump_user_location_t.
11962 (find_loop_location): Convert return type from source_location to
11963 dump_user_location_t.
11964 (check_reduction_path): Convert 1st param from location_t to
11965 dump_user_location_t.
11966 * value-prof.c (check_counter): Update call to dump_printf_loc to
11967 pass in a dump_user_location_t rather than a location_t; update
11968 call to error_at for change in type of "locus".
11969 (check_ic_target): Update call to dump_printf_loc to
11970 pass in a dump_user_location_t rather than a location_t, via the
11973 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
11975 * config/s390/s390.h (enum processor_flags): Do not use
11976 default tune parameter when -march was specified.
11978 2018-06-26 Jakub Jelinek <jakub@redhat.com>
11981 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
11982 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
11985 2018-06-26 Richard Biener <rguenther@suse.de>
11987 PR tree-optimization/86287
11989 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
11990 (vect_analyze_loop): Initialize n_stmts.
11992 2018-06-26 Richard Biener <rguenther@suse.de>
11994 PR middle-end/86271
11995 * fold-const.c (fold_convertible_p): Pointer extension
11998 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
12001 * dwarf2out.c (loc_list_has_views): Adjust comments.
12002 (dw_loc_list): Split single cross-partition range with
12005 2018-06-25 Jeff Law <law@redhat.com>
12007 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
12008 on -mbig-switch by default.
12010 * config/v850/predicates.md (const_float_1_operand): Fix match_code
12012 (const_float_0_operand): Remove unused predicate.
12013 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
12014 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
12015 (recipsf2): New expander. Original pattern now called
12017 (recipdf2, recipdf2_insn): Similarly.
12018 (rsqrtsf2, rsqrtsf2_insn): Similarly
12019 (rsqrtdf2, rsqrtdf2_insn): Similarly
12021 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
12023 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
12024 Simplify logic for FreeBSD (twice).
12026 2018-06-25 Martin Sebor <msebor@redhat.com>
12028 PR tree-optimization/86204
12029 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
12030 a strnlen result if it's less than the length of the string.
12032 2018-06-25 Martin Sebor <msebor@redhat.com>
12034 PR tree-optimization/85700
12035 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
12036 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
12037 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
12039 2018-06-25 Martin Sebor <msebor@redhat.com>
12041 * doc/extend.texi (Zero-length arrays): Update and clarify.
12043 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
12045 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
12046 added IEEE/IBM long double multilib support on PowerPC little
12047 endian Linux systems.
12048 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
12049 (MULTILIB_DEFAULTS): Likewise.
12050 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12052 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
12053 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
12054 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
12056 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
12058 PR middle-end/86311
12059 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
12060 (REORDER_45): Likewise.
12062 2018-06-25 Jeff Law <law@redhat.com>
12064 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
12065 dividend to 32 bits. Adjust length.
12066 (udivmodhi4): Cleanup output template. Fix length.
12068 2018-06-25 Carl Love <cel@us.ibm.com>
12070 * config/rs6000/vsx.md: Change word selector to prefered location.
12072 2018-06-25 Richard Biener <rguenther@suse.de>
12074 PR tree-optimization/86304
12075 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
12076 epilogue-if-converted loops as well.
12078 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
12080 * lto-section-out.c (lto_begin_section): Do not print section
12081 name for noaddr and unnumbered dumps.
12083 2018-06-25 Richard Biener <rguenther@suse.de>
12085 * tree-vectorizer.h (struct vec_info_shared): New structure
12086 with parts split out from struct vec_info and loop_nest from
12087 struct _loop_vec_info.
12088 (struct vec_info): Adjust accordingly.
12089 (struct _loop_vec_info): Likewise.
12090 (LOOP_VINFO_LOOP_NEST): Adjust.
12091 (LOOP_VINFO_DATAREFS): Likewise.
12092 (LOOP_VINFO_DDRS): Likewise.
12093 (struct _bb_vec_info): Likewise.
12094 (BB_VINFO_DATAREFS): Likewise.
12095 (BB_VINFO_DDRS): Likewise.
12096 (struct _stmt_vec_info): Add dr_aux member.
12097 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
12098 (DR_MISALIGNMENT_UNINITIALIZED): New.
12099 (set_dr_misalignment): Adjust.
12100 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
12101 (vect_analyze_loop): Adjust prototype.
12102 (vect_analyze_loop_form): Likewise.
12103 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
12104 Compute dependences lazily.
12105 (vect_record_base_alignments): Use shared datarefs/ddrs.
12106 (vect_verify_datarefs_alignment): Likewise.
12107 (vect_analyze_data_refs_alignment): Likewise.
12108 (vect_analyze_data_ref_accesses): Likewise.
12109 (vect_analyze_data_refs): Likewise.
12110 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
12111 constructor parameter for shared part.
12112 (vect_analyze_loop_form): Pass in shared part and adjust.
12113 (vect_analyze_loop_2): Pass in storage for the number of
12114 stmts. Move loop nest finding to the caller. Compute
12116 (vect_analyze_loop): Pass in shared part.
12117 (vect_transform_loop): Verify shared datarefs are unchanged.
12118 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
12119 constructor parameter for shared part.
12120 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
12121 (vect_slp_bb): Verify shared datarefs are unchanged before
12123 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
12125 (new_stmt_vec_info): Initialize DR_AUX misalignment to
12126 DR_MISALIGNMENT_UNINITIALIZED.
12127 * tree-vectorizer.c (vec_info::vec_info): Add constructor
12128 parameter for shared part.
12129 (vec_info::~vec_info): Adjust.
12130 (vec_info_shared::vec_info_shared): New.
12131 (vec_info_shared::~vec_info_shared): Likewise.
12132 (vec_info_shared::save_datarefs): Likewise.
12133 (vec_info_shared::check_datarefs): Likewise.
12134 (try_vectorize_loop_1): Construct shared part live for analyses
12135 of a single loop for multiple vector sizes.
12136 * tree-parloops.c (gather_scalar_reductions): Adjust.
12138 2018-06-25 Richard Biener <rguenther@suse.de>
12140 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
12141 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
12142 (vect_analyze_data_refs): Remove similar code from here and
12143 simplify accordingly.
12145 2018-06-25 Richard Biener <rguenther@suse.de>
12147 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
12148 for reverse storage order accesses rather than asserting
12149 they cannot happen here.
12151 2018-06-25 Tom de Vries <tdevries@suse.de>
12154 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
12155 Use data16 instead of .byte for insn prefix.
12157 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
12160 * parser.c (make_char_string_pack): Pass this literal chars
12161 through cpp_interpret_string.
12162 (cp_parser_userdef_numeric_literal): Check the result of
12163 make_char_string_pack.
12165 2018-06-24 Maya Rashish <coypu@sdf.org>
12167 * ginclude/stddef.h: Simplify conditions around avoiding
12168 re-definition of __size_t.
12170 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12172 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
12173 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
12175 2018-06-22 Maya Rashish <coypu@sdf.org>
12177 * doc/invoke.texi (mno-fancy-math-387): Update for changes
12178 made to OpenBSD and NetBSD through the years.
12180 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
12182 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
12183 behavior of vec_pack (vector double, vector double) to match
12184 behavior of vec_float2 (vector double, vector double).
12186 2018-06-22 Olivier Hainque <hainque@adacore.com>
12188 * gimplify.c (gimplify_function_tree): Prevent creation
12189 of a trampoline for the address of the current function
12190 passed to entry/exit instrumentation hooks.
12192 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
12195 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
12198 2018-06-22 Martin Liska <mliska@suse.cz>
12200 PR tree-optimization/86263
12201 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
12202 Bail out if is_enabled is false.
12203 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
12205 (jump_table_cluster::is_enabled): New function.
12207 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12209 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
12210 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
12211 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
12212 (lto_input_ts_binfo_tree_pointers): Likewise.
12213 * tree-streamer-out.c (streamer_write_tree_bitfields,
12214 write_ts_binfo_tree_pointers): Likewise.
12215 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
12217 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12219 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
12221 2018-06-22 Martin Liska <mliska@suse.cz>
12223 * symbol-summary.h (get): Make it pure and inline move
12224 functionality from ::get function.
12225 (get): Remove and inline into ::get and ::get_create.
12226 (get_create): Move code from ::get function.
12228 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12231 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
12232 -x assembler-with-cpp.
12234 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12236 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
12237 _FILE_OFFSET_BITS=64 for C++.
12239 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
12241 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
12242 conversion insn that shows up when pr85657-3.c is compiled using
12243 IEEE 128-bit long double.
12244 (neg<mode>2_internal): Use the correct mode to check whether the
12245 mode is IBM extended.
12246 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
12247 multiply and divide external functions from being created more
12250 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12252 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
12254 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
12255 the edge can be forwarded.
12256 (cfg_layout_merge_blocks): Likewise.
12258 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12260 * except.c (finish_eh_generation): Commit edge insertions only after
12261 the EH edges have been redirected from post-landing to landing pads.
12263 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12265 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
12266 create_tmp_var_for to create the FRAME decl.
12267 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
12269 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12271 * tree-inline.c (copy_edges_for_bb): Minor tweak.
12272 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
12273 debug statement when resetting its value.
12274 (expand_call_inline): Copy the locus of the call onto the assignment
12275 of the return value, if any. Use local variable in more cases.
12277 2018-06-21 Martin Liska <mliska@suse.cz>
12279 * ipa-pure-const.c (propagate_nothrow): Use
12280 funct_state_summaries->get.
12281 (dump_malloc_lattice): Likewise.
12282 (propagate_malloc): Likewise.
12284 2018-06-21 Richard Biener <rguenther@suse.de>
12286 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
12287 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
12288 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
12290 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
12291 BLOCK_ABSTRACT_ORIGIN unconditionally.
12293 2018-06-21 David Malcolm <dmalcolm@redhat.com>
12295 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
12297 * ipa-reference.c (ipa_reference_c_finalize): Delete
12298 ipa_ref_opt_sum_summaries and set it to NULL.
12300 2018-06-21 Tom de Vries <tdevries@suse.de>
12302 PR tree-optimization/85859
12303 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
12304 test with comment from bb_no_side_effects_p.
12306 2018-06-21 Richard Biener <rguenther@suse.de>
12308 PR tree-optimization/86232
12309 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
12310 max for constant niter.
12312 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
12314 * config/aarch64/aarch64-simd.md
12315 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
12317 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
12319 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
12320 Make opernads of the unspec commutative.
12322 2018-06-21 Richard Biener <rguenther@suse.de>
12324 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
12325 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
12326 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
12327 (vect_analyze_data_ref_dependence): Re-order checks to deal with
12329 (vect_record_base_alignments): Do not record base alignment
12330 for gathers or scatters.
12331 (vect_compute_data_ref_alignment): Drop return value that is always
12332 true. Bail out early for gathers or scatters.
12333 (vect_enhance_data_refs_alignment): Bail out early for gathers
12335 (vect_find_same_alignment_drs): Likewise.
12336 (vect_analyze_data_refs_alignment): Remove dead code.
12337 (vect_slp_analyze_and_verify_node_alignment): Likewise.
12338 (vect_analyze_data_refs): For possible gathers or scatters do
12339 not create an alternate DR, just check their possible validity
12340 and mark them. Adjust DECL_NONALIASED handling to not rely
12341 on DR_BASE_ADDRESS.
12342 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
12343 update inits of gathers or scatters.
12344 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
12345 Also copy gather/scatter flag to pattern vinfo.
12347 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
12349 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
12350 behavior of vec_packsu (vector unsigned long long, vector unsigned
12351 long long) to match behavior of vec_packs with same signature.
12353 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
12354 Thomas Schwinge <thomas@codesourcery.com>
12355 Cesar Philippidis <cesar@codesourcery.com>
12357 * gimplify.c (gimplify_scan_omp_clauses): Add support for
12358 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
12359 (gimplify_adjust_omp_clauses): Likewise.
12360 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
12361 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
12362 (gimplify_omp_target_update): Update handling of acc update and
12364 * omp-low.c (install_var_field): Remove unused parameter
12365 base_pointers_restrict.
12366 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
12367 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
12369 (omp_target_base_pointers_restrict_p): Delete.
12370 (scan_omp_target): Update call to scan_sharing_clauses.
12371 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
12373 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
12374 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
12375 (convert_local_omp_clauses): Likewise.
12376 * tree-pretty-print.c (dump_omp_clause): Likewise.
12377 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
12379 (omp_clause_code_name): Likewise.
12381 2018-06-20 Jakub Jelinek <jakub@redhat.com>
12384 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
12387 PR tree-optimization/86231
12388 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
12389 anti-range don't overwrite *vr0min before using it to compute *vr0max.
12391 2018-06-20 Tom de Vries <tdevries@suse.de>
12393 PR tree-optimization/86097
12394 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
12395 iv type if signedness of iv type is not the same as that of *nit.
12397 2018-06-20 Jakub Jelinek <jakub@redhat.com>
12399 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
12400 EDGE_EH edges, verify they are all EDGE_EH.
12402 2018-06-20 Maya Rashish <coypu@sdf.org>
12404 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
12406 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12408 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
12409 * config/aarch64/aarch64.c (xgene1_tunings): Add
12410 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
12411 (aarch64_mode_valid_for_sched_fusion_p):
12412 Allow 16-byte modes.
12413 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
12414 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
12416 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
12418 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
12419 * config/aarch64/iterators.md (VQ2): New mode iterator.
12421 2018-06-20 Martin Liska <mliska@suse.cz>
12423 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
12424 Change default ratio from 10 to 8.
12426 2018-06-20 Martin Liska <mliska@suse.cz>
12428 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
12430 (bit_test_cluster::find_bit_tests): Likewise.
12431 (switch_decision_tree::analyze_switch_statement): Find clusters.
12432 * tree-switch-conversion.h (struct jump_table_cluster): Document
12435 2018-06-20 Martin Liska <mliska@suse.cz>
12437 * tree-switch-conversion.c (switch_conversion::collect):
12438 Record m_uniq property.
12439 (switch_conversion::expand): Bail out for special conditions.
12440 (group_cluster::~group_cluster): New.
12441 (group_cluster::group_cluster): Likewise.
12442 (group_cluster::dump): Likewise.
12443 (jump_table_cluster::emit): New.
12444 (switch_decision_tree::fix_phi_operands_for_edges): New.
12445 (struct case_node): Remove struct.
12446 (jump_table_cluster::can_be_handled): New.
12447 (case_values_threshold): Moved to header.
12448 (reset_out_edges_aux): Likewise.
12449 (jump_table_cluster::is_beneficial): New.
12450 (bit_test_cluster::can_be_handled): Likewise.
12451 (add_case_node): Remove.
12452 (bit_test_cluster::is_beneficial): New.
12453 (case_bit_test::cmp): New.
12454 (bit_test_cluster::emit): New.
12455 (expand_switch_as_decision_tree_p): Remove.
12456 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
12457 (fix_phi_operands_for_edge): Likewise.
12458 (switch_decision_tree::analyze_switch_statement): New.
12459 (compute_cases_per_edge): Move ...
12460 (switch_decision_tree::compute_cases_per_edge): ... here.
12461 (try_switch_expansion): Likewise.
12462 (switch_decision_tree::try_switch_expansion): Likewise.
12463 (record_phi_operand_mapping): Likewise.
12464 (switch_decision_tree::record_phi_operand_mapping): Likewise.
12465 (emit_case_decision_tree): Likewise.
12466 (switch_decision_tree::emit): Likewise.
12467 (balance_case_nodes): Likewise.
12468 (switch_decision_tree::balance_case_nodes): Likewise.
12469 (dump_case_nodes): Likewise.
12470 (switch_decision_tree::dump_case_nodes): Likewise.
12471 (emit_jump): Likewise.
12472 (switch_decision_tree::emit_jump): Likewise.
12473 (emit_cmp_and_jump_insns): Likewise.
12474 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
12475 (emit_case_nodes): Likewise.
12476 (switch_decision_tree::emit_case_nodes): Likewise.
12477 (conditional_probability): Remove.
12478 * tree-switch-conversion.h (enum cluster_type): New.
12480 (struct cluster): Likewise.
12481 (cluster::cluster): Likewise.
12482 (struct simple_cluster): Likewise.
12483 (simple_cluster::simple_cluster): Likewise.
12484 (struct group_cluster): Likewise.
12485 (struct jump_table_cluster): Likewise.
12486 (struct bit_test_cluster): Likewise.
12487 (struct min_cluster_item): Likewise.
12488 (struct case_tree_node): Likewise.
12489 (case_tree_node::case_tree_node): Likewise.
12490 (jump_table_cluster::case_values_threshold): Likewise.
12491 (struct case_bit_test): Likewise.
12492 (struct switch_decision_tree): Likewise.
12493 (struct switch_conversion): Likewise.
12494 (switch_decision_tree::reset_out_edges_aux): Likewise.
12496 2018-06-20 Martin Liska <mliska@suse.cz>
12498 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
12499 (hoist_edge_and_branch_if_true): Likewise.
12500 (expand_switch_using_bit_tests_p): Likewise.
12501 (struct case_bit_test): Likewise.
12502 (case_bit_test_cmp): Likewise.
12503 (emit_case_bit_tests): Likewise.
12504 (switch_conversion::switch_conversion): New class.
12505 (struct switch_conv_info): Remove old struct.
12506 (collect_switch_conv_info): More to ...
12507 (switch_conversion::collect): ... this.
12508 (check_range): Likewise.
12509 (switch_conversion::check_range): Likewise.
12510 (check_all_empty_except_final): Likewise.
12511 (switch_conversion::check_all_empty_except_final): Likewise.
12512 (check_final_bb): Likewise.
12513 (switch_conversion::check_final_bb): Likewise.
12514 (create_temp_arrays): Likewise.
12515 (switch_conversion::create_temp_arrays): Likewise.
12516 (free_temp_arrays): Likewise.
12517 (gather_default_values): Likewise.
12518 (switch_conversion::gather_default_values): Likewise.
12519 (build_constructors): Likewise.
12520 (switch_conversion::build_constructors): Likewise.
12521 (constructor_contains_same_values_p): Likewise.
12522 (switch_conversion::contains_same_values_p): Likewise.
12523 (array_value_type): Likewise.
12524 (switch_conversion::array_value_type): Likewise.
12525 (build_one_array): Likewise.
12526 (switch_conversion::build_one_array): Likewise.
12527 (build_arrays): Likewise.
12528 (switch_conversion::build_arrays): Likewise.
12529 (gen_def_assigns): Likewise.
12530 (switch_conversion::gen_def_assigns): Likewise.
12531 (prune_bbs): Likewise.
12532 (switch_conversion::prune_bbs): Likewise.
12533 (fix_phi_nodes): Likewise.
12534 (switch_conversion::fix_phi_nodes): Likewise.
12535 (gen_inbound_check): Likewise.
12536 (switch_conversion::gen_inbound_check): Likewise.
12537 (process_switch): Use the newly created class.
12538 (switch_conversion::expand): New.
12539 (switch_conversion::~switch_conversion): New.
12540 * tree-switch-conversion.h: New file.
12542 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12544 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
12545 tree-vect-patterns.c.
12546 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
12547 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
12548 (vect_recog_sad_pattern): Likewise.
12549 (vect_recog_widen_sum_pattern): Likewise.
12550 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
12551 (vect_recog_widen_shift_pattern): Remove the type_in argument.
12552 (vect_recog_rotate_pattern): Likewise.
12553 (vect_recog_mult_pattern): Likewise.
12554 (vect_recog_vector_vector_shift_pattern): Likewise.
12555 (vect_recog_divmod_pattern): Likewise.
12556 (vect_recog_mixed_size_cond_pattern): Likewise.
12557 (vect_recog_bool_pattern): Likewise.
12558 (vect_recog_mask_conversion_pattern): Likewise.
12559 (vect_try_gather_scatter_pattern): Likewise.
12560 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
12561 (vect_recog_over_widening_pattern): Likewise.
12562 (vect_recog_gather_scatter_pattern): Likewise.
12563 (vect_recog_func_ptr): Move from tree-vectorizer.h
12564 (vect_vect_recog_func_ptrs): Move further down the file.
12565 (vect_recog_func): Likewise. Remove the third argument.
12566 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
12567 (vect_pattern_recog_1): Expect the pattern function to do any
12568 necessary target tests. Also expect it to provide a vector type.
12569 Remove the type_in handling.
12571 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12573 * tree-vect-patterns.c (vect_pattern_detected): New function.
12574 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
12575 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
12576 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
12577 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
12578 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
12579 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
12580 (vect_recog_mask_conversion_pattern)
12581 (vect_try_gather_scatter_pattern): Likewise.
12583 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12585 * tree-vect-patterns.c (vect_get_internal_def): New function.
12586 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
12587 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
12588 (search_type_for_mask_1): Use it.
12590 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12592 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
12593 redundant WIDEN_SUM_EXPR handling.
12594 (vect_recog_sad_pattern): Likewise.
12596 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12598 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
12599 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
12600 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
12601 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
12602 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
12604 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12606 * tree-vect-stmts.c (vectorizable_call): Make sure that we
12607 use the stmt_vec_info of the original bb statement for the
12608 new zero assignment, even if the call is part of a pattern.
12610 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12612 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
12613 that the sequence is attached to the original statement rather
12614 than the pattern statement.
12615 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
12616 PATTERN_DEF_SEQ from the original statement rather than
12617 the main pattern statement.
12618 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
12619 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
12620 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
12622 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12624 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
12625 definition statements before the early exit for statements that aren't
12627 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
12629 (vect_transform_loop): ...here. Process pattern definition
12630 statements without first checking whether the main pattern
12631 statement is live or relevant.
12633 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
12635 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
12636 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
12638 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
12640 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
12641 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
12642 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
12643 (expand_block_compare): Change select_block_compare_mode call.
12644 (expand_strncmp_align_check): Use new functions, fix comment.
12645 (emit_final_str_compare_gpr): New function.
12646 (expand_strn_compare): Refactor and clean up code.
12647 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
12649 2018-06-19 Tony Reix <tony.reix@atos.com>
12650 Damien Bergamini <damien.bergamini@atos.com>
12651 David Edelsohn <dje.gcc@gmail.com>
12653 * collect2.c (static_obj): New variable.
12654 (static_libs): New variable.
12655 (is_in_list): Uncomment declaration.
12656 (main): Track AIX libraries linked statically.
12657 (is_in_list): Uncomment definition.
12658 (scan_prog_file): Don't add AIX shared libraries initializer
12659 to constructor list if linking statically.
12661 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
12663 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
12665 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
12667 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
12669 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
12672 2018-06-19 Martin Liska <mliska@suse.cz>
12674 * config/i386/i386.c (ix86_can_inline_p): Do not use
12675 ipa_fn_summaries::get_create.
12676 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
12678 (devirtualization_time_bonus): Likewise.
12679 (ipcp_propagate_stage): Likewise.
12680 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
12681 (edge_set_predicate): Likewise.
12682 (evaluate_conditions_for_known_args): Likewise.
12683 (evaluate_properties_for_edge): Likewise.
12684 (ipa_call_summary::reset): Tranform to ...
12685 (ipa_call_summary::~ipa_call_summary): ... this.
12686 (ipa_fn_summary::reset): Transform to ...
12687 (ipa_fn_summary::~ipa_fn_summary): ... this.
12688 (ipa_fn_summary_t::remove): Rename to ...
12689 (ipa_fn_summary_t::remove_callees): ... this.
12690 (ipa_fn_summary_t::duplicate): Use placement new
12691 instead of memory copy.
12692 (ipa_call_summary_t::duplicate): Likewise.
12693 (ipa_call_summary_t::remove): Remove.
12694 (dump_ipa_call_summary): Change get_create to get.
12695 (ipa_dump_fn_summary): Dump only when summary exists.
12696 (analyze_function_body): Use symbol_summary::get instead
12698 (compute_fn_summary): Likewise.
12699 (estimate_edge_devirt_benefit): Likewise.
12700 (estimate_edge_size_and_time): Likewise.
12701 (inline_update_callee_summaries): Likewise.
12702 (remap_edge_change_prob): Likewise.
12703 (remap_edge_summaries): Likewise.
12704 (ipa_merge_fn_summary_after_inlining): Likewise.
12705 (write_ipa_call_summary): Likewise.
12706 (ipa_fn_summary_write): Likewise.
12707 (ipa_free_fn_summary): Likewise.
12708 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
12709 (struct ipa_call_summary): Likewise.
12710 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
12712 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
12713 (estimate_size_after_inlining): Likewise.
12714 (estimate_growth): Likewise.
12715 (growth_likely_positive): Likewise.
12716 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
12717 (inline_call): Likewise.
12718 * ipa-inline.c (caller_growth_limits): Likewise.
12719 (can_inline_edge_p): Likewise.
12720 (can_inline_edge_by_limits_p): Likewise.
12721 (compute_uninlined_call_time): Likewise.
12722 (compute_inlined_call_time): Likewise.
12723 (want_inline_small_function_p): Likewise.
12724 (edge_badness): Likewise.
12725 (update_caller_keys): Likewise.
12726 (update_callee_keys): Likewise.
12727 (inline_small_functions): Likewise.
12728 (inline_to_all_callers_1): Likewise.
12729 (dump_overall_stats): Likewise.
12730 (early_inline_small_functions): Likewise.
12731 (early_inliner): Likewise.
12732 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12733 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
12734 * ipa-pure-const.c (malloc_candidate_p): Likewise.
12735 * ipa-split.c (execute_split_functions): Likewise.
12736 * symbol-summary.h: Likewise.
12737 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
12739 2018-06-19 Richard Biener <rguenther@suse.de>
12741 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
12742 (vectorize_loops): ... here. Fix dbgcnt handling.
12743 (try_vectorize_loop): Wrap try_vectorize_loop_1.
12745 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
12748 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
12749 ieee128 argument takes up only one (vector) register, not two (floating
12752 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
12754 * gimplify.c (gimplify_init_constructor): Really never clear for an
12755 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
12757 2018-06-19 Richard Biener <rguenther@suse.de>
12759 PR tree-optimization/86179
12760 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
12761 after failed recognition.
12763 2018-06-18 Martin Sebor <msebor@redhat.com>
12765 PR middle-end/85602
12766 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
12767 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
12768 Handle integer subtraction.
12769 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
12770 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
12772 2018-06-18 David Malcolm <dmalcolm@redhat.com>
12774 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
12775 param from rtx to rtx_insn *.
12776 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
12778 (frv_ifcvt_modify_insn): Likwise.
12779 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
12780 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
12781 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
12782 as_a <rtx_insn *> cast to local "unprotected_region" once
12783 it's been established that it's not NULL or pc_rtx.
12784 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
12785 param "sethi" from rtx to rtx_insn *.
12786 (nds32_group_float_insns): Likewise for param "insn".
12787 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
12789 (vax_output_int_subtract): Likewise.
12790 * config/vax/vax.c (vax_output_int_add): Likewise for param
12792 (vax_output_int_subtract): Likewise.
12793 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
12794 (emit_pattern_after): Likewise for param "after".
12795 (emit_insn_after): Likewise.
12796 (emit_jump_insn_after): Likewise.
12797 (emit_call_insn_after): Likewise.
12798 (emit_debug_insn_after): Likewise.
12799 (emit_pattern_before): Likewise for param "before".
12800 (emit_insn_before): Likewise.
12801 (emit_jump_insn_before): Likewise.
12802 * final.c (get_insn_template): Likewise for param "insn", removing
12804 * output.h (get_insn_template): Likewise for 2nd param.
12805 * rtl.h (emit_insn_before): Likewise.
12806 (emit_jump_insn_before): Likewise.
12807 (emit_debug_insn_before_noloc): Likewise.
12808 (emit_insn_after): Likewise.
12809 (emit_jump_insn_after): Likewise.
12810 (emit_call_insn_after): Likewise.
12811 (emit_debug_insn_after): Likewise.
12812 (set_insn_deleted): Likewise for param.
12814 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
12817 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
12818 floating point modes, so that IFmode is numerically greater than
12819 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
12820 to declare the ordering. This prevents IFmode from being
12821 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
12822 machine. Include rs6000-modes.h to share the fractional values
12823 between genmodes* and the rest of the compiler.
12824 (IFmode): Likewise.
12825 (KFmode): Likewise.
12826 (TFmode): Likewise.
12827 * config/rs6000/rs6000-modes.h: New file.
12828 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
12829 meaning of rs6000_long_double_size so that 126..128 selects an
12830 appropriate 128-bit floating point type.
12831 (rs6000_option_override_internal): Likewise.
12832 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
12833 (TARGET_LONG_DOUBLE_128): Change the meaning of
12834 rs6000_long_double_size so that 126..128 selects an appropriate
12835 128-bit floating point type.
12836 (LONG_DOUBLE_TYPE_SIZE): Update comment.
12837 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
12838 source and destination to match the standard usage.
12839 (truncifkf2): Likewise.
12840 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
12841 ISA 2.07 to use an explicit clobber, instead of passing in a
12843 (copysign<mode>3_soft): Likewise.
12845 2018-06-18 David Malcolm <dmalcolm@redhat.com>
12847 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
12848 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
12849 (vect_slp_analyze_instance_dependence): Likewise.
12850 (vect_enhance_data_refs_alignment): Likewise.
12851 (vect_analyze_data_refs_alignment): Likewise.
12852 (vect_slp_analyze_and_verify_instance_alignment
12853 (vect_analyze_data_ref_accesses): Likewise.
12854 (vect_prune_runtime_alias_test_list): Likewise.
12855 (vect_analyze_data_refs): Likewise.
12856 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
12857 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
12858 (vect_analyze_scalar_cycles_1): Likewise.
12859 (vect_get_loop_niters): Likewise.
12860 (vect_analyze_loop_form_1): Likewise.
12861 (vect_update_vf_for_slp): Likewise.
12862 (vect_analyze_loop_operations): Likewise.
12863 (vect_analyze_loop): Likewise.
12864 (vectorizable_induction): Likewise.
12865 (vect_transform_loop): Likewise.
12866 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
12867 * tree-vect-slp.c (vect_analyze_slp): Likewise.
12868 (vect_make_slp_decision): Likewise.
12869 (vect_detect_hybrid_slp): Likewise.
12870 (vect_slp_analyze_operations): Likewise.
12871 (vect_slp_bb): Likewise.
12872 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
12873 (vectorizable_bswap): Likewise.
12874 (vectorizable_call): Likewise.
12875 (vectorizable_simd_clone_call): Likewise.
12876 (vectorizable_conversion): Likewise.
12877 (vectorizable_assignment): Likewise.
12878 (vectorizable_shift): Likewise.
12879 (vectorizable_operation): Likewise.
12880 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
12882 2018-06-18 Martin Sebor <msebor@redhat.com>
12884 PR tree-optimization/81384
12885 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
12886 * builtins.c (expand_builtin_strnlen): New function.
12887 (expand_builtin): Call it.
12888 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
12889 * builtins.def (BUILT_IN_STRNLEN): New.
12890 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
12891 Warn for bounds in excess of maximum object size.
12892 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
12893 single-value ranges. Handle strnlen.
12894 (handle_builtin_strlen): Handle strnlen.
12895 (strlen_check_and_optimize_stmt): Same.
12896 * doc/extend.texi (Other Builtins): Document strnlen.
12898 2018-06-18 Maya Rashish <coypu@sdf.org>
12900 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
12901 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
12902 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
12904 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
12906 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
12908 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12910 * tree.c (escaped_string::escape): Replace cast to char * by
12911 const_cast<char *> (unescaped).
12913 2018-06-18 Nick Clifton <nickc@redhat.com>
12916 * tree.c (escaped_string): New class. Converts an unescaped
12917 string into its escaped equivalent.
12918 (warn_deprecated_use): Use the new class to convert the
12919 deprecation message, if present.
12920 (test_escaped_strings): New self test.
12921 (test_c_tests): Add test_escaped_strings.
12922 * doc/extend.texi (deprecated): Add a note that the
12923 deprecation message is affected by the -fmessage-length
12924 option, and that control characters will be escaped.
12925 (#pragma GCC error): Document this pragma.
12926 (#pragma GCC warning): Likewise.
12927 * doc/invoke.texi (-fmessage-length): Document this option's
12928 effect on the #warning and #error preprocessor directives and
12929 the deprecated attribute.
12931 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
12933 * tree.c (decl_value_expr_lookup): Revert latest change.
12934 (decl_value_expr_insert): Likewise.
12936 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
12938 * gimplify.c (nonlocal_vlas): Delete.
12939 (nonlocal_vla_vars): Likewise.
12940 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
12942 (gimplify_body): Do not create and destroy nonlocal_vlas.
12943 * tree-nested.c: Include diagnostic.h.
12944 (use_pointer_in_frame): Tweak.
12945 (lookup_field_for_decl): Add assertion and declare the transformation.
12946 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
12947 internal error when the reference is in a wrong context. Do not
12948 create a debug decl by default.
12949 (note_nonlocal_block_vlas): Delete.
12950 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
12951 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
12952 create a debug decl by default.
12953 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
12954 call is in a wrong context.
12955 (fixup_vla_decls): New function.
12956 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
12957 debug variables were created.
12958 * tree.c (decl_value_expr_lookup): Add checking assertion.
12959 (decl_value_expr_insert): Likewise.
12961 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
12963 PR middle-end/82479
12964 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
12965 * tree-scalar-evolution.c (interpret_expr): Likewise.
12966 (expression_expensive_p): Likewise.
12967 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
12968 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
12969 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
12970 (ssa_defined_by_minus_one_stmt_p): New.
12972 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
12974 PR middle-end/64946
12975 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
12976 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
12977 * dojump.c (do_jump): Likewise.
12978 * expr.c (expand_expr_real_2): Check operand type's sign.
12979 * fold-const.c (const_unop): Handle ABSU_EXPR.
12980 (fold_abs_const): Likewise.
12981 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
12982 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
12983 (strip_sign_op_1): Likesise.
12984 * match.pd: Add new pattern to generate ABSU_EXPR.
12985 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
12986 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
12987 * tree-eh.c (operation_could_trap_helper_p): Likewise.
12988 * tree-inline.c (estimate_operator_cost): Likewise.
12989 * tree-pretty-print.c (dump_generic_node): Likewise.
12990 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
12991 * tree.def (ABSU_EXPR): New.
12993 2018-06-16 Jakub Jelinek <jakub@redhat.com>
12995 PR middle-end/86095
12996 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
12997 documented as preserved for backward compatibility only.
12998 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
13000 PR rtl-optimization/86108
13001 * bb-reorder.c (create_forwarder_block): Renamed to ...
13002 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
13003 jump from new landing pad to the second part.
13004 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
13007 2018-06-15 Jakub Jelinek <jakub@redhat.com>
13009 PR middle-end/85878
13010 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
13011 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
13012 Only call store_expr for halves if the mode is the same.
13014 PR middle-end/86123
13015 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
13016 Fix up comment formatting.
13018 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
13020 * typed-splay-tree.h (typed_splay_tree::remove): New function.
13021 (typed_splay_tree::closure,
13022 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
13023 (typed_splay_tree::typed_splay_tree,
13024 typed_splay_tree::operator =): Declared private.
13025 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
13026 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
13027 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
13028 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
13029 typed_splay_tree::splay_tree_splay,
13030 typed_splay_tree::splay_tree_foreach_helper,
13031 typed_splay_tree::splay_tree_insert,
13032 typed_splay_tree::splay_tree_remove,
13033 typed_splay_tree::splay_tree_lookup,
13034 typed_splay_tree::splay_tree_predecessor,
13035 typed_splay_tree::splay_tree_successor,
13036 typed_splay_tree::splay_tree_min,
13037 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
13038 (typed_splay_tree::root, typed_splay_tree::comp,
13039 typed_splay_tree::delete_key,
13040 typed_splay_tree::delete_value): New data members.
13041 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
13042 typed_splay_tree::remove.
13044 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
13046 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
13047 -mginv and -mno-ginv to the assembler.
13048 * config/mips/mips.opt (-mcrc): New option.
13049 (-mginv): Likewise.
13050 * doc/invoke.text (-mcrc): Document.
13051 (-mginv): Likewise.
13053 2018-06-15 Nick Clifton <nickc@redhat.com>
13056 * tree.c (escaped_string): New class. Converts an unescaped
13057 string into its escaped equivalent.
13058 (warn_deprecated_use): Use the new class to convert the
13059 deprecation message, if present.
13060 (test_escaped_strings): New self test.
13061 (test_c_tests): Add test_escaped_strings.
13062 * doc/extend.texi (deprecated): Add a note that the
13063 deprecation message is affected by the -fmessage-length
13064 option, and that control characters will be escaped.
13065 (#pragma GCC error): Document this pragma.
13066 (#pragma GCC warning): Likewise.
13067 * doc/invoke.texi (-fmessage-length): Document this option's
13068 effect on the #warning and #error preprocessor directives and
13069 the deprecated attribute.
13071 2018-06-15 Richard Biener <rguenther@suse.de>
13073 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
13074 here, also noting vector size used.
13075 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
13076 size used in MSG_OPTIMIZED_LOCATIONS dump.
13077 (pass_slp_vectorize::execute): Adjust.
13079 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
13082 * config/arc/arc.c (arc_return_address_register): Fix
13085 2018-06-15 Richard Biener <rguenther@suse.de>
13087 PR middle-end/86159
13088 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
13089 leave useless conversion stripping to force_gimple_operand_gsi.
13090 (gimplify_build2): Likewise.
13091 (gimplify_build1): Likewise.
13093 2018-06-15 Richard Biener <rguenther@suse.de>
13095 PR middle-end/86076
13096 * tree-cfg.c (move_stmt_op): unshare invariant addresses
13097 before adjusting their block.
13099 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
13101 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
13102 multilibs for *-*-rtems*.
13103 * config/riscv/t-rtems: New file.
13105 2018-06-14 Jakub Jelinek <jakub@redhat.com>
13107 PR middle-end/86122
13108 * match.pd ((A +- CST1) +- CST2): Punt if last resort
13109 unsigned_type_for returns NULL.
13112 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
13113 subregs of multi-word pseudos unless the float mode has word size.
13115 2018-06-14 Richard Biener <rguenther@suse.de>
13117 PR middle-end/86139
13118 * tree-vect-generic.c (build_word_mode_vector_type): Remove
13119 duplicate and harmful type_hash_canon.
13120 * tree.c (type_hash_canon): Assert we didn't find ourselves.
13122 2018-06-14 Richard Biener <rguenther@suse.de>
13125 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
13128 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
13130 * config/rtems.h (STDINT_LONG32): Define.
13132 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
13133 Prachi Godbole <prachi.godbole@imgtec.com>
13135 * config/mips/mips-cpus.def: Define P6600.
13136 * config/mips/mips-tables.opt: Regenerate.
13137 * config/mips/mips.c (mips_ucbranch_type): New enum.
13138 (mips_rtx_cost_data): Add support for P6600.
13139 (mips_issue_rate): Likewise.
13140 (mips_multipass_dfa_lookahead): Likewise.
13141 (mips_avoid_hazard): Likewise.
13142 (mips_reorg_process_insns): Likewise.
13143 (mips_classify_branch_p6600): New function.
13144 * config/mips/mips.h (TUNE_P6600): New define.
13145 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
13146 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
13147 * config/mips/mips.md: Include p6600.md.
13148 (processor): Add p6600.
13149 * config/mips/p6600.md: New file.
13150 * doc/invoke.texi: Add p6600 to supported architectures.
13152 2018-06-13 Martin Sebor <msebor@redhat.com>
13154 PR tree-optimization/86114
13155 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
13157 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
13159 2018-06-13 Richard Biener <rguenther@suse.de>
13161 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
13162 Properly set vector type of the intermediate stmt.
13163 * tree-vect-stmts.c (vectorizable_operation): The destination
13164 var always has vectype_out type.
13166 2018-06-13 Jeff Law <law@redhat.com>
13168 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
13169 integer 0 for argument to print_rtl_with_bb.
13170 (rl78_reorg): Likewise.
13172 2018-06-13 David Malcolm <dmalcolm@redhat.com>
13174 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
13175 from rtx to rtx_insn *.
13176 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
13178 (add_sched_insns_for_speculation): Likewise for local "target",
13179 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
13180 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
13181 from rtx_insn ** to rtx_code_label **.
13182 (reorg_emit_nops): Likewise.
13183 (c6x_reorg): Likewise for local "call_labels".
13184 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
13186 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
13187 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
13188 the loops over LABEL_REFS.
13189 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
13191 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
13192 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
13193 (split_branches): Strengthen local "olabel" from rtx to
13194 rtx_insn *, adding a safe_as_a cast.
13195 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
13197 (add_insn_after): Likewise for first two params.
13198 (add_insn_before): Likewise.
13199 (remove_insn): Likewise for param.
13200 (emit_pattern_before_noloc): Likewise for second and third params.
13201 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
13202 (emit_call_insn_before_noloc): Likewise.
13203 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
13205 (emit_barrier_before): Likewise.
13206 (emit_label_before): Strengthen "label" param from "rtx" to
13207 "rtx_code_label *". Strengthen "before" param from "rtx" to
13209 (emit_insn_after_1): Strengthen "after" param from "rtx" to
13211 (emit_pattern_after_noloc): Likewise.
13212 (emit_insn_after_noloc): Likewise.
13213 (emit_jump_insn_after_noloc): Likewise.
13214 (emit_call_insn_after_noloc): Likewise.
13215 (emit_debug_insn_after_noloc): Likewise.
13216 (emit_barrier_after): Likewise.
13217 (emit_label_after): Likewise for both params.
13218 (emit_pattern_after_setloc): Likewise for "after" param. Convert
13219 "loc" param from "int" to "location_t".
13220 (emit_insn_after_setloc): Likewise.
13221 (emit_jump_insn_after_setloc): Likewise.
13222 (emit_call_insn_after_setloc): Likewise.
13223 (emit_debug_insn_after_setloc): Likewise.
13224 (emit_pattern_before_setloc): Likewise for "before" param. Convert
13225 "loc" param from "int" to "location_t".
13226 (emit_pattern_before): Convert NULL_RTX to NULL.
13227 (emit_insn_before_setloc): Convert "loc" param from "int" to
13229 (emit_jump_insn_before_setloc): Likewise.
13230 (emit_call_insn_before_setloc): Likewise.
13231 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
13232 rtx_insn *. Convert "loc" param from "int" to "location_t".
13233 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
13234 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
13235 Convert 3rd param from "int" to "location_t".
13236 (emit_barrier_before, emit_barrier_after, next_real_insn):
13237 Strengthen param from rtx to rtx_insn *.
13238 (emit_label_before): Strengthen 1st param from "rtx" to
13239 "rtx_code_label *". Strengthen 2nd param from "rtx" to
13241 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
13242 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
13243 Strengthen 2nd param from "rtx" to "rtx_insn *".
13244 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
13245 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
13246 Likewise. Convert 3rd param from "int" to "location_t".
13247 (emit_label_after): Strengthen 1st param from "rtx" to
13248 "rtx_code_label *".
13249 (next_real_insn, remove_insn): Strengthen param from "rtx" to
13251 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
13252 from "rtx" to "rtx_insn *".
13254 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
13256 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
13257 bodies streamed in with -Q.
13258 * dumpfile.c (dump_files): Add lto-stream-out dump file.
13259 * dumpfile.h (tree_dump_index): Add lto_stream_out.
13260 * gimple-streamer-out.c: Include gimple-pretty-print.h
13261 (output_bb): Dump stmts streamed.
13262 * lto-section-out.c: Include print-tree.h
13263 (lto_begin_section): Dump sections created.
13264 (lto_output_decl_index): Dump decl encoded.
13265 * lto-streamer-out.c: Include print-tree.h
13266 (create_output_block): Dump output block created.
13267 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
13268 (output_function): Dump function output.
13269 (output_constructor): Dump constructor streamed.
13270 (write_global_stream): Output indexes encoded.
13271 (produce_asm_for_decls): Dump streams encoded.
13272 * lto-streamer.c (streamer_dump_file): New global var.
13273 * lto-streamer.h (streamer_dump_file): Declare.
13274 * passes.c (ipa_write_summaries): Initialize streamer dump.
13275 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
13278 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
13281 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
13282 offsets for register save directives. Emit a second batch of save
13283 directives, if need be, when the function accesses prior frames.
13285 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13287 * config/arc/fpu.md (fmasf4): Force operand to register.
13288 (fnmasf4): Likewise.
13290 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13292 * config/arc/arc-protos.h (arc_pad_return): Remove.
13293 * config/arc/arc.c (machine_function): Remove force_short_suffix
13295 (arc_print_operand): Adjust printing of '&'.
13296 (arc_verify_short): Remove conditional printing of short suffix.
13297 (arc_final_prescan_insn): Remove reference to size_reason.
13298 (pad_return): New function.
13299 (arc_reorg): Call pad_return.
13300 (arc_pad_return): Remove.
13301 (arc_init_machine_status): Remove reference to force_short_suffix.
13302 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
13303 (attr length): When attribute iscompact is true force to 2
13304 regardless; in the case of maybe check if we want to force the
13305 instruction to have 4 bytes length.
13306 (nopv): Change it to generate 4 byte long nop as well.
13307 (blockage): New pattern.
13308 (simple_return): Remove call to arc_pad_return.
13309 (p_return_i): Likewise.
13311 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13313 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
13315 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13317 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
13320 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13322 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
13323 for ARC700 and ARCv2.
13325 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
13328 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
13329 operands[2] instead of operands[1].
13332 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13334 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
13335 case, check whether the outer register overlaps an unallocatable
13336 register, not just whether it fits the required class.
13338 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13340 * poly-int.h (can_div_trunc_p): Add new overload in which all values
13342 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
13343 (memrefs_conflict_p): Likewise.
13344 (init_alias_analysis): Likewise.
13345 * cfgexpand.c (expand_debug_expr): Likewise.
13346 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
13347 * cse.c (fold_rtx): Likewise.
13348 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
13349 * expr.c (emit_block_move_hints): Likewise.
13350 (clear_storage_hints, push_block, emit_push_insn): Likewise.
13351 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
13352 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
13353 (emit_group_store): Likewise.
13354 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
13355 to read the PRE/POST_MODIFY increment.
13356 * calls.c (store_one_arg): Use strip_offset.
13357 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
13359 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
13361 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
13362 (simplify_binary_operation_1): Extend CONST_INT handling to
13364 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
13365 than a HOST_WIDE_INT.
13366 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
13368 (adjust_mems, add_stores): Update accodingly.
13369 (vt_canonicalize_addr): Track polynomial offsets.
13370 (emit_note_insn_var_location): Likewise.
13371 (vt_add_function_parameter): Likewise.
13372 (vt_initialize): Likewise.
13374 2018-06-12 Jeff Law <law@redhat.com>
13376 * config.gcc (alpha*-*-freebsd*): Remove.
13377 * config/alpha/freebsd.h: Remove.
13379 2018-06-12 David Malcolm <dmalcolm@redhat.com>
13382 * spellcheck-tree.c (levenshtein_distance): Rename to...
13383 (get_edit_distance): ...this, and update for underlying renaming.
13384 * spellcheck-tree.h (levenshtein_distance): Rename to...
13385 (get_edit_distance): ...this.
13386 * spellcheck.c (levenshtein_distance): Rename to...
13387 (get_edit_distance): ...this. Convert from Levenshtein distance
13388 to Damerau-Levenshtein distance by supporting transpositions of
13389 adjacent characters. Rename "v1" to "v_next" and "v0" to
13391 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
13392 (selftest::test_edit_distance_unit_test_oneway): ...this, and
13393 update for underlying renaming.
13394 (selftest::levenshtein_distance_unit_test): Rename to...
13395 (selftest::test_get_edit_distance_unit): ...this, and update for
13396 underlying renaming.
13397 (selftest::test_find_closest_string): Add example from PR 69968
13398 where transposition helps
13399 (selftest::test_metric_conditions): Update for renaming.
13400 (selftest::test_metric_conditions): Likewise.
13401 (selftest::spellcheck_c_tests): Likewise.
13402 * spellcheck.h (levenshtein_distance): Rename both overloads to...
13403 (get_edit_distance): ...this.
13404 (best_match::consider): Update for renaming.
13406 2018-06-12 Martin Sebor <msebor@redhat.com>
13408 PR tree-optimization/85259
13409 * builtins.c (compute_objsize): Handle constant offsets.
13410 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
13411 true iff a warning has been issued.
13412 * gimple.h (gimple_nonartificial_location): New function.
13413 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
13414 gimple_nonartificial_location and handle -Wno-system-headers.
13415 (handle_builtin_stxncpy): Same.
13417 2018-06-12 Martin Sebor <msebor@redhat.com>
13420 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
13422 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13424 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13425 BUILTIN_VEC_XST entries for pointer to double and long long.
13427 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
13430 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
13432 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
13435 2018-06-12 Martin Liska <mliska@suse.cz>
13437 * doc/options.texi: Document IntegerRange.
13439 2018-06-12 Martin Liska <mliska@suse.cz>
13441 * config/i386/i386.opt: Make MPX-related options as Deprecated.
13442 * opt-functions.awk: Handle Deprecated flag.
13443 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
13445 (read_cmdline_option): Report warning for a deprecated option.
13446 * opts.h (struct cl_option): Add new field cl_deprecated.
13447 (CL_ERR_DEPRECATED): New.
13449 2018-06-12 Martin Liska <mliska@suse.cz>
13451 * doc/options.texi: Document Deprecated option flag.
13453 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13455 * config/arc/arc-arch.h (arc_extras): New enum.
13456 (arc_cpu_t):Add field extra.
13457 (arc_cpu_types): Consider the extras.
13458 * config/arc/arc-cpus.def: Add extras info.
13459 * config/arc/arc-opts.h (processor_type): Consider extra field.
13460 * config/arc/arc.c (arc_override_options): Handle extra field.
13462 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13464 * config/arc/arc-arch.h: Update ARC_OPTX macro.
13465 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
13467 * config/arc/arc.c (arc_init): Update pic warning.
13468 (irq_range): Update irq range parsing warnings.
13469 (arc_override_options): Update various warning messages.
13470 (arc_handle_aux_attribute): Likewise.
13472 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
13474 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
13476 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
13478 * doc/sourcebuild.texi: Document usage of line number 0 in verify
13479 compiler messages directives.
13481 2018-06-12 Matthew Fortune <mfortune@gmail.com>
13483 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
13484 * config/mips/mips-tables.opt: Regenerate.
13485 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
13487 * doc/invoke.texi: Document -march=i6500.
13489 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
13491 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
13492 (i6400_gpmul): Add cpu_unit.
13493 (i6400_gpdiv): Likewise.
13494 (i6400_msa_add_d): Update reservations.
13495 (i6400_msa_int_add) Likewise.
13496 (i6400_msa_short_logic3) Likewise.
13497 (i6400_msa_short_logic2) Likewise.
13498 (i6400_msa_short_logic) Likewise.
13499 (i6400_msa_move) Likewise.
13500 (i6400_msa_cmp) Likewise.
13501 (i6400_msa_short_float2) Likewise.
13502 (i6400_msa_div_d) Likewise.
13503 (i6400_msa_long_logic1) Likewise.
13504 (i6400_msa_long_logic2) Likewise.
13505 (i6400_msa_mult) Likewise.
13506 (i6400_msa_long_float2) Likewise.
13507 (i6400_msa_long_float4) Likewise.
13508 (i6400_msa_long_float5) Likewise.
13509 (i6400_msa_long_float8) Likewise.
13510 (i6400_fpu_fadd): Include frint type.
13511 (i6400_fpu_store): New define_insn_reservation.
13512 (i6400_fpu_load): Likewise.
13513 (i6400_fpu_move): Likewise.
13514 (i6400_fpu_fcmp): Likewise.
13515 (i6400_fpu_fmadd): Likewise.
13516 (i6400_int_mult): Include imul3nc type and update reservation.
13517 (i6400_int_div): Include idiv3 type and update reservation.
13518 (i6400_int_load): Update to check type not move_type.
13519 (i6400_int_store): Likewise.
13520 (i6400_int_prefetch): Set zero latency.
13522 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
13524 * gcc.c: Document new %@{...} sequence.
13525 (LINK_COMMAND_SPEC): Use it for the -L switches.
13526 (cpp_unique_options): Use it for the -I switches.
13527 (at_file_argbuf): New global variable.
13528 (in_at_file): Likewise.
13529 (alloc_args): Create at_file_argbuf.
13530 (clear_args): Truncate at_file_argbuf.
13531 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
13532 (open_at_file): New function.
13533 (close_at_file): Likewise.
13534 (create_at_file): Delete.
13535 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
13538 (validate_switches_from_spec): Deal with %@{...} sequence.
13539 (validate_switches): Likewise.
13540 (driver::finalize): Call clear_args.
13542 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
13544 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
13546 2018-06-11 Martin Sebor <msebor@redhat.com>
13548 * doc/invoke.texi (-Wall): List -Wc++17-compat.
13549 (Wno-class-memaccess): Add @opindex.
13550 (Wno-templates, Wno-multiple-inheritance): Same.
13551 (Wno-virtual-inheritance, Wno-namespaces): Same.
13552 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
13553 (Wno-format-overflow, Wno-format-truncation): Same.
13554 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
13555 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
13556 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
13557 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
13558 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
13559 (Wno-misspelled-isr): Same.
13561 2018-06-11 Martin Sebor <msebor@redhat.com>
13563 * PR tree-optimization/86083
13564 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
13566 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
13568 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
13570 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
13573 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
13574 on the correct operand.
13575 (*movdi_internal64): Ditto.
13577 2018-06-11 Martin Liska <mliska@suse.cz>
13579 PR tree-optimization/86089
13580 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
13582 2018-06-11 Julia Koval <julia.koval@intel.com>
13584 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
13585 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
13586 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
13588 2018-06-11 Olivier Hainque <hainque@adacore.com>
13590 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
13591 for Ada with strict dwarf2.
13593 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
13596 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
13599 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
13601 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
13603 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
13605 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
13608 2018-06-08 Martin Liska <mliska@suse.cz>
13610 * tree-cfg.h (debug_function): Fix argument type to match
13613 2018-06-08 Martin Liska <mliska@suse.cz>
13615 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
13616 Remove usage of MPX-related (and removed) fields.
13617 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
13619 2018-06-08 David Malcolm <dmalcolm@redhat.com>
13621 * cfg.c (debug): Use TDF_NONE rather than 0.
13622 * cfghooks.c (debug): Likewise.
13623 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
13624 (struct dump_option_value_info): Convert to...
13625 (struct kv_pair): ...this template type.
13626 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
13628 (optinfo_verbosity_options): Likewise.
13629 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
13631 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
13632 than int for "optgroup_flags" param.
13633 (dump_generic_expr_loc): Use dump_flags_t rather than int for
13635 (dump_dec): Likewise.
13636 (dump_finish): Use TDF_NONE rather than 0.
13637 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
13638 rather than int for "optgroup_flags" param. Use TDF_NONE rather
13639 than 0. Update for change to option_ptr.
13640 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
13641 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
13642 0. Update for changes to optinfo_verbosity_options and
13644 (opt_info_switch_p): Convert optgroup_flags from int to
13646 (dump_basic_block): Use dump_flags_t rather than int
13647 for "dump_kind" param.
13648 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
13649 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
13650 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
13651 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
13652 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
13653 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
13654 TDF_NONE): Convert from macros to...
13655 (enum dump_flag): ...this new enum.
13656 (dump_flags_t): Update to use enum.
13657 (operator|, operator&, operator~, operator|=, operator&=):
13658 Implement for dump_flags_t.
13659 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
13660 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
13661 Convert from macros to...
13662 (enum optgroup_flag): ...this new enum.
13663 (optgroup_flags_t): New typedef.
13664 (operator|, operator|=): Implement for optgroup_flags_t.
13665 (struct dump_file_info): Convert field "alt_flags" to
13666 dump_flags_t. Convert field "optgroup_flags" to
13668 (dump_basic_block): Use dump_flags_t rather than int for param.
13669 (dump_generic_expr_loc): Likewise.
13670 (dump_dec): Likewise.
13671 (dump_register): Convert param "optgroup_flags" to
13673 (opt_info_enable_passes): Likewise.
13674 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
13676 * gimple-pretty-print.c (debug): Likewise.
13677 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
13678 (merged_store_group::apply_stores): Likewise.
13679 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
13680 * gimple.c (verify_gimple_pp): Likewise.
13681 * graphite-poly.c (print_pbb_body): Likewise.
13682 * passes.c (pass_manager::register_one_dump_file): Convert
13683 local "optgroup_flags" to optgroup_flags_t.
13684 * print-tree.c (print_node): Use TDF_NONE rather than 0.
13686 (debug_body): Likewise.
13687 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
13688 to optgroup_flags_t.
13689 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
13691 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
13692 (convert_mult_to_fma): Likewise.
13693 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
13694 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
13695 * tree-vect-data-refs.c (dump_lower_bound): Convert param
13696 "dump_kind" to dump_flags_t.
13698 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
13700 * config/rs6000/rs6000.c (min, max): Delete.
13702 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
13704 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
13705 -mabi=spe and -mabi=no-spe.
13707 2018-06-08 Martin Liska <mliska@suse.cz>
13709 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
13710 where we expect an existing summary.
13712 2018-06-08 Martin Liska <mliska@suse.cz>
13714 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
13715 * ipa-inline.h (estimate_edge_growth): Likewise.
13717 2018-06-08 Martin Liska <mliska@suse.cz>
13719 * cgraph.c (function_version_hasher::hash): Use
13720 cgraph_node::get_uid ().
13721 (function_version_hasher::equal):
13722 * cgraph.h (cgraph_node::get_uid): New method.
13723 * ipa-inline.c (update_caller_keys): Use
13724 cgraph_node::get_uid ().
13725 (update_callee_keys): Likewise.
13726 * ipa-utils.c (searchc): Likewise.
13727 (ipa_reduced_postorder): Likewise.
13728 * lto-cgraph.c (input_node): Likewise.
13729 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
13730 * symbol-summary.h (symtab_insertion): Likewise.
13731 (symtab_removal): Likewise.
13732 (symtab_duplication): Likewise.
13733 * tree-pretty-print.c (dump_function_header): Likewise.
13734 * tree-sra.c (convert_callers_for_node): Likewise.
13736 2018-06-08 Martin Liska <mliska@suse.cz>
13738 * cgraph.c (symbol_table::create_edge): Always assign a new
13740 (symbol_table::free_edge): Do not recycle numbers.
13741 * cgraph.h (cgraph_edge::get): New method.
13742 * symbol-summary.h (symtab_removal): Use it.
13743 (symtab_duplication): Likewise.
13744 (call_summary::hashable_uid): Remove.
13746 2018-06-08 Martin Liska <mliska@suse.cz>
13748 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
13749 (initialize_growth_caches): Remove.
13750 (free_growth_caches): Likewise.
13751 (do_estimate_edge_time): Use edge_growth_cache.
13752 (do_estimate_edge_size): Likewise.
13753 (do_estimate_edge_hints): Likewise.
13754 * ipa-inline.c (reset_edge_caches): Likewise.
13755 (recursive_inlining): Likewise.
13756 (inline_small_functions): Likewise.
13757 * ipa-inline.h (initialize_growth_caches): Remove.
13758 (estimate_edge_size): Likewise.
13759 (estimate_edge_time): Likewise.
13760 (estimate_edge_hints): Likewise.
13761 (reset_edge_growth_cache): Likewise.
13762 * symbol-summary.h (call_summary::remove): New method.
13764 2018-06-08 Martin Liska <mliska@suse.cz>
13766 * ipa-cp.c (class edge_clone_summary): New summary.
13767 (grow_edge_clone_vectors): Remove.
13768 (ipcp_edge_duplication_hook): Remove.
13769 (class edge_clone_summary_t): New call_summary class.
13770 (ipcp_edge_removal_hook): Remove.
13771 (edge_clone_summary_t::duplicate): New function.
13772 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
13773 (create_specialized_node): Likewise.
13774 (ipcp_driver): Initialize edge_clone_summaries and do not
13777 2018-06-08 Martin Liska <mliska@suse.cz>
13779 * symbol-summary.h (get): New function.
13780 (call_summary::m_initialize_when_cloning): New class member.
13782 2018-06-08 Martin Liska <mliska@suse.cz>
13784 * cgraph.c (cgraph_node::remove): Do not recycle uid.
13785 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
13786 (symbol_table::allocate_cgraph_symbol): Do not set uid.
13787 * passes.c (uid_hash_t): Record removed_nodes by their uids.
13788 (remove_cgraph_node_from_order): Use the removed_nodes set.
13789 (do_per_function_toporder): Likwise.
13790 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
13791 instead of summary_uid.
13792 (symtab_removal): Likewise.
13793 (symtab_duplication): Likewise.
13795 2018-06-08 Martin Liska <mliska@suse.cz>
13797 * ipa-cp.c (ipcp_store_bits_results): Use
13798 ipcp_transformation_sum.
13799 (ipcp_store_vr_results): Likewise.
13800 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
13802 (ipcp_transformation_initialize): ... this.
13803 (ipa_set_node_agg_value_chain):
13804 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
13805 (write_ipcp_transformation_info): Likewise.
13806 (read_ipcp_transformation_info): Likewise.
13807 (ipcp_update_bits): Likewise.
13808 (ipcp_update_vr): Likewise.
13809 (ipcp_transform_function): Likewise.
13810 * ipa-prop.h: Rename ipcp_transformation_summary to
13811 ipcp_transformation.
13812 (class ipcp_transformation_t): New function summary.
13813 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
13814 (ipa_get_agg_replacements_for_node): Likewise.
13816 2018-06-08 Martin Liska <mliska@suse.cz>
13818 * ipa-pure-const.c (struct funct_state_d): Do it class instead
13820 (class funct_state_summary_t): New function_summary class.
13821 (has_function_state): Remove.
13822 (get_function_state): Likewise.
13823 (set_function_state): Likewise.
13824 (add_new_function): Likewise.
13825 (funct_state_summary_t::insert): New function.
13826 (duplicate_node_data): Remove.
13827 (remove_node_data): Remove.
13828 (funct_state_summary_t::duplicate): New function.
13829 (register_hooks): Create new funct_state_summaries.
13830 (pure_const_generate_summary): Use it.
13831 (pure_const_write_summary): Likewise.
13832 (pure_const_read_summary): Likewise.
13833 (propagate_pure_const): Likewise.
13834 (propagate_nothrow): Likewise.
13835 (dump_malloc_lattice): Likewise.
13836 (propagate_malloc): Likewise.
13837 (execute): Do not register hooks, just remove summary
13839 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
13842 2018-06-08 Martin Liska <mliska@suse.cz>
13844 * ipa-reference.c (remove_node_data): Remove.
13845 (duplicate_node_data): Likewise.
13846 (class ipa_ref_var_info_summary_t): New class.
13847 (class ipa_ref_opt_summary_t): Likewise.
13848 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
13849 (get_reference_optimization_summary): Use
13850 ipa_ref_opt_sum_summaries.
13851 (set_reference_vars_info): Remove.
13852 (set_reference_optimization_summary): Likewise.
13853 (ipa_init): Create summaries.
13854 (init_function_info): Use function summary.
13855 (ipa_ref_opt_summary_t::duplicate): New function.
13856 (ipa_ref_opt_summary_t::remove): New function.
13857 (get_read_write_all_from_node): Fix GNU coding style.
13858 (propagate): Use function summary.
13859 (write_node_summary_p): Fix GNU coding style.
13860 (stream_out_bitmap): Likewise.
13861 (ipa_reference_read_optimization_summary): Use function summary.
13862 (ipa_reference_c_finalize): Do not release hooks.
13864 2018-06-08 Martin Liska <mliska@suse.cz>
13866 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
13867 (analyze_function_body): Extract multiple calls of get_create.
13868 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
13869 * ipa-inline.c (recursive_inlining): Use ::get method.
13870 * ipa-inline.h (estimate_edge_growth): Likewise.
13872 2018-06-08 Martin Liska <mliska@suse.cz>
13874 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
13876 (hsa_function_summary::hsa_function_summary): Use the new enum
13878 (hsa_gpu_implementation_p): Use hsa_summaries::get.
13879 * hsa-gen.c (hsa_get_host_function): Likewise.
13880 (get_brig_function_name): Likewise.
13881 * ipa-hsa.c (process_hsa_functions): Likewise.
13882 (ipa_hsa_write_summary): Likewise.
13883 * symbol-summary.h (symtab_duplication): Use ::get function/
13884 (get): New function.
13886 2018-06-08 Martin Liska <mliska@suse.cz>
13888 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
13890 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
13891 (hsa_register_kernel): Likewise.
13892 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
13893 * hsa-gen.c (hsa_get_host_function): Likewise.
13894 (get_brig_function_name): Likewise.
13895 (generate_hsa): Likewise.
13896 (pass_gen_hsail::execute): Likewise.
13897 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
13898 (devirtualization_time_bonus): Likewise.
13899 (ipcp_propagate_stage): Likewise.
13900 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
13901 (edge_set_predicate): Likewise.
13902 (evaluate_conditions_for_known_args): Likewise.
13903 (evaluate_properties_for_edge): Likewise.
13904 (ipa_fn_summary::reset): Likewise.
13905 (ipa_fn_summary_t::duplicate): Likewise.
13906 (dump_ipa_call_summary): Likewise.
13907 (ipa_dump_fn_summary): Likewise.
13908 (analyze_function_body): Likewise.
13909 (compute_fn_summary): Likewise.
13910 (estimate_edge_devirt_benefit): Likewise.
13911 (estimate_edge_size_and_time): Likewise.
13912 (estimate_calls_size_and_time): Likewise.
13913 (estimate_node_size_and_time): Likewise.
13914 (inline_update_callee_summaries): Likewise.
13915 (remap_edge_change_prob): Likewise.
13916 (remap_edge_summaries): Likewise.
13917 (ipa_merge_fn_summary_after_inlining): Likewise.
13918 (ipa_update_overall_fn_summary): Likewise.
13919 (read_ipa_call_summary): Likewise.
13920 (inline_read_section): Likewise.
13921 (write_ipa_call_summary): Likewise.
13922 (ipa_fn_summary_write): Likewise.
13923 (ipa_free_fn_summary): Likewise.
13924 * ipa-hsa.c (process_hsa_functions): Likewise.
13925 (ipa_hsa_write_summary): Likewise.
13926 (ipa_hsa_read_section): Likewise.
13927 * ipa-icf.c (sem_function::merge): Likewise.
13928 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
13929 (do_estimate_edge_time): Likewise.
13930 (estimate_size_after_inlining): Likewise.
13931 (estimate_growth): Likewise.
13932 (growth_likely_positive): Likewise.
13933 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
13934 (inline_call): Likewise.
13935 * ipa-inline.c (caller_growth_limits): Likewise.
13936 (can_inline_edge_p): Likewise.
13937 (can_inline_edge_by_limits_p): Likewise.
13938 (compute_uninlined_call_time): Likewise.
13939 (compute_inlined_call_time): Likewise.
13940 (want_inline_small_function_p): Likewise.
13941 (edge_badness): Likewise.
13942 (update_caller_keys): Likewise.
13943 (update_callee_keys): Likewise.
13944 (recursive_inlining): Likewise.
13945 (inline_small_functions): Likewise.
13946 (inline_to_all_callers_1): Likewise.
13947 (dump_overall_stats): Likewise.
13948 (early_inline_small_functions): Likewise.
13949 (early_inliner): Likewise.
13950 * ipa-inline.h (estimate_edge_growth): Likewise.
13951 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13952 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
13953 * ipa-prop.h (IPA_NODE_REF): Likewise.
13954 (IPA_EDGE_REF): Likewise.
13955 * ipa-pure-const.c (malloc_candidate_p): Likewise.
13956 (propagate_malloc): Likewise.
13957 * ipa-split.c (execute_split_functions): Likewise.
13958 * symbol-summary.h: Rename get to get_create.
13960 (get_create): Likewise.
13961 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
13963 2018-06-08 Martin Liska <mliska@suse.cz>
13965 * symbol-summary.h (release): Move definition out of class
13967 (symtab_removal): Likewise.
13968 (symtab_duplication): Likewise.
13970 2018-06-08 Martin Liska <mliska@suse.cz>
13972 * symbol-summary.h (function_summary): Move constructor
13973 implementation out of class declaration.
13974 (release): Likewise.
13975 (symtab_insertion): Likewise.
13976 (symtab_removal): Likewise.
13977 (symtab_duplication): Likewise.
13980 2018-06-08 Martin Liska <mliska@suse.cz>
13982 * Makefile.in: Remove support for MPX (macros, related functions,
13983 fields in cgraph_node, ...).
13984 * builtin-types.def (BT_BND): Likewise.
13985 (BT_FN_BND_CONST_PTR): Likewise.
13986 (BT_FN_CONST_PTR_BND): Likewise.
13987 (BT_FN_VOID_PTR_BND): Likewise.
13988 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
13989 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
13990 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
13991 (expand_builtin_mempcpy_with_bounds): Likewise.
13992 (expand_builtin_memset_with_bounds): Likewise.
13993 (expand_builtin_memset_args): Likewise.
13994 (std_expand_builtin_va_start): Likewise.
13995 (expand_builtin): Likewise.
13996 (expand_builtin_with_bounds): Likewise.
13997 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
13998 (DEF_LIB_BUILTIN_CHKP): Likewise.
13999 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
14000 (DEF_CHKP_BUILTIN): Likewise.
14001 (BUILT_IN_MEMCPY): Likewise.
14002 (BUILT_IN_MEMMOVE): Likewise.
14003 (BUILT_IN_MEMPCPY): Likewise.
14004 (BUILT_IN_MEMSET): Likewise.
14005 (BUILT_IN_STPCPY): Likewise.
14006 (BUILT_IN_STRCAT): Likewise.
14007 (BUILT_IN_STRCHR): Likewise.
14008 (BUILT_IN_STRCPY): Likewise.
14009 (BUILT_IN_STRLEN): Likewise.
14010 (BUILT_IN_MEMCPY_CHK): Likewise.
14011 (BUILT_IN_MEMMOVE_CHK): Likewise.
14012 (BUILT_IN_MEMPCPY_CHK): Likewise.
14013 (BUILT_IN_MEMSET_CHK): Likewise.
14014 (BUILT_IN_STPCPY_CHK): Likewise.
14015 (BUILT_IN_STRCAT_CHK): Likewise.
14016 (BUILT_IN_STRCPY_CHK): Likewise.
14017 * calls.c (store_bounds): Likewise.
14018 (emit_call_1): Likewise.
14019 (special_function_p): Likewise.
14020 (maybe_warn_nonstring_arg): Likewise.
14021 (initialize_argument_information): Likewise.
14022 (finalize_must_preallocate): Likewise.
14023 (compute_argument_addresses): Likewise.
14024 (expand_call): Likewise.
14025 * cfgexpand.c (expand_call_stmt): Likewise.
14026 (expand_return): Likewise.
14027 (expand_gimple_stmt_1): Likewise.
14028 (pass_expand::execute): Likewise.
14029 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
14030 (cgraph_node::remove): Likewise.
14031 (cgraph_node::dump): Likewise.
14032 (cgraph_node::verify_node): Likewise.
14033 * cgraph.h (chkp_function_instrumented_p): Likewise.
14034 (symtab_node::get_alias_target): Likewise.
14035 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
14036 (cgraph_local_p): Likewise.
14037 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
14038 (cgraph_edge::rebuild_references): Likewise.
14039 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
14040 (walk_polymorphic_call_targets): Likewise.
14041 (cgraph_node::expand_thunk): Likewise.
14042 (symbol_table::output_weakrefs): Likewise.
14043 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
14044 (ix86_handle_option): Likewise.
14045 * config/i386/constraints.md: Likewise.
14046 * config/i386/i386-builtin-types.def (BND): Likewise.
14050 * config/i386/i386-builtin.def (BDESC_END): Likewise.
14051 (BDESC_FIRST): Likewise.
14053 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
14054 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
14055 * config/i386/i386.c (enum reg_class): Likewise.
14056 (ix86_target_string): Likewise.
14057 (ix86_option_override_internal): Likewise.
14058 (ix86_conditional_register_usage): Likewise.
14059 (ix86_valid_target_attribute_inner_p): Likewise.
14060 (ix86_set_indirect_branch_type): Likewise.
14061 (ix86_set_current_function): Likewise.
14062 (ix86_function_arg_regno_p): Likewise.
14063 (init_cumulative_args): Likewise.
14064 (ix86_function_arg_advance): Likewise.
14065 (ix86_function_arg): Likewise.
14066 (ix86_pass_by_reference): Likewise.
14067 (ix86_function_value_regno_p): Likewise.
14068 (ix86_function_value_1): Likewise.
14069 (ix86_function_value_bounds): Likewise.
14070 (ix86_return_in_memory): Likewise.
14071 (ix86_setup_incoming_vararg_bounds): Likewise.
14072 (ix86_va_start): Likewise.
14073 (indirect_thunk_need_prefix): Likewise.
14074 (print_reg): Likewise.
14075 (ix86_print_operand): Likewise.
14076 (ix86_expand_call): Likewise.
14077 (ix86_output_function_return): Likewise.
14078 (reg_encoded_number): Likewise.
14079 (BDESC_VERIFYS): Likewise.
14080 (ix86_init_mpx_builtins): Likewise.
14081 (ix86_init_builtins): Likewise.
14082 (ix86_emit_cmove): Likewise.
14083 (ix86_emit_move_max): Likewise.
14084 (ix86_expand_builtin): Likewise.
14085 (ix86_builtin_mpx_function): Likewise.
14086 (ix86_get_arg_address_for_bt): Likewise.
14087 (ix86_load_bounds): Likewise.
14088 (ix86_store_bounds): Likewise.
14089 (ix86_load_returned_bounds): Likewise.
14090 (ix86_store_returned_bounds): Likewise.
14091 (ix86_class_likely_spilled_p): Likewise.
14092 (ix86_hard_regno_mode_ok): Likewise.
14093 (x86_order_regs_for_local_alloc): Likewise.
14094 (ix86_mitigate_rop): Likewise.
14095 (ix86_bnd_prefixed_insn_p): Likewise.
14096 (ix86_mpx_bound_mode): Likewise.
14097 (ix86_make_bounds_constant): Likewise.
14098 (ix86_initialize_bounds): Likewise.
14099 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
14100 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
14101 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
14102 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
14103 (TARGET_CHKP_BOUND_MODE): Likewise.
14104 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
14105 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
14106 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
14107 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
14108 * config/i386/i386.h (TARGET_MPX): Likewise.
14109 (TARGET_MPX_P): Likewise.
14110 (VALID_BND_REG_MODE): Likewise.
14111 (FIRST_BND_REG): Likewise.
14112 (LAST_BND_REG): Likewise.
14113 (enum reg_class): Likewise.
14114 (BND_REG_P): Likewise.
14115 (BND_REGNO_P): Likewise.
14116 (BNDmode): Likewise.
14117 (ADJUST_INSN_LENGTH): Likewise.
14118 * config/i386/i386.md: Likewise.
14119 * config/i386/i386.opt: Likewise.
14120 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
14121 (defined): Likewise.
14122 (LINK_MPX): Likewise.
14123 (MPX_SPEC): Likewise.
14124 (LIBMPX_SPEC): Likewise.
14125 (LIBMPXWRAPPERS_SPEC): Likewise.
14126 (CHKP_SPEC): Likewise.
14127 * config/i386/predicates.md: Likewise.
14128 * dbxout.c (dbxout_type): Likewise.
14129 * doc/extend.texi: Likewise.
14130 * doc/invoke.texi: Likewise.
14131 * doc/md.texi: Likewise.
14132 * doc/tm.texi: Likewise.
14133 * doc/tm.texi.in: Likewise.
14134 * dwarf2out.c (is_base_type): Likewise.
14135 (gen_formal_types_die): Likewise.
14136 (gen_subprogram_die): Likewise.
14137 (gen_type_die_with_usage): Likewise.
14138 (gen_decl_die): Likewise.
14139 (dwarf2out_late_global_decl): Likewise.
14140 * expr.c (expand_assignment): Likewise.
14141 (emit_storent_insn): Likewise.
14142 (store_expr_with_bounds): Likewise.
14143 (store_expr): Likewise.
14144 (expand_expr_real_1): Likewise.
14145 * expr.h (store_expr_with_bounds): Likewise.
14146 * function.c (use_register_for_decl): Likewise.
14147 (struct bounds_parm_data): Likewise.
14148 (assign_parms_augmented_arg_list): Likewise.
14149 (assign_parm_find_entry_rtl): Likewise.
14150 (assign_parm_is_stack_parm): Likewise.
14151 (assign_parm_load_bounds): Likewise.
14152 (assign_bounds): Likewise.
14153 (assign_parms): Likewise.
14154 (expand_function_start): Likewise.
14155 * gcc.c (CHKP_SPEC): Likewise.
14156 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
14157 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
14158 (wrestrict_dom_walker::check_call): Likewise.
14159 * gimple.c (gimple_build_call_from_tree): Likewise.
14160 * gimple.h (enum gf_mask): Likewise.
14161 (gimple_call_with_bounds_p): Likewise.
14162 (gimple_call_set_with_bounds): Likewise.
14163 * gimplify.c (gimplify_init_constructor): Likewise.
14164 * ipa-cp.c (initialize_node_lattices): Likewise.
14165 (propagate_constants_across_call): Likewise.
14166 (find_more_scalar_values_for_callers_subset): Likewise.
14167 * ipa-hsa.c (process_hsa_functions): Likewise.
14168 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
14169 * ipa-icf.c (sem_function::merge): Likewise.
14170 * ipa-inline.c (early_inliner): Likewise.
14171 * ipa-pure-const.c (warn_function_noreturn): Likewise.
14172 (warn_function_cold): Likewise.
14173 (propagate_pure_const): Likewise.
14174 * ipa-ref.h (enum GTY): Likewise.
14175 * ipa-split.c (find_retbnd): Likewise.
14176 (consider_split): Likewise.
14177 (split_function): Likewise.
14178 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
14179 * ipa.c (walk_polymorphic_call_targets): Likewise.
14180 (symbol_table::remove_unreachable_nodes): Likewise.
14181 (process_references): Likewise.
14182 (cgraph_build_static_cdtor_1): Likewise.
14183 * lto-cgraph.c (lto_output_node): Likewise.
14184 (output_refs): Likewise.
14185 (compute_ltrans_boundary): Likewise.
14186 (input_overwrite_node): Likewise.
14187 (input_node): Likewise.
14188 (input_cgraph_1): Likewise.
14189 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
14190 * passes.c (pass_manager::execute_early_local_passes): Likewise.
14191 (class pass_chkp_instrumentation_passes): Likewise.
14192 (make_pass_chkp_instrumentation_passes): Likewise.
14193 * passes.def: Likewise.
14194 * rtl.h (struct GTY): Likewise.
14195 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
14196 * stor-layout.c (layout_type): Likewise.
14197 * symtab.c: Likewise.
14198 * target.def: Likewise.
14199 * targhooks.c (default_chkp_bound_type): Likewise.
14200 (default_chkp_bound_mode): Likewise.
14201 (default_builtin_chkp_function): Likewise.
14202 (default_chkp_function_value_bounds): Likewise.
14203 (default_chkp_make_bounds_constant): Likewise.
14204 (default_chkp_initialize_bounds): Likewise.
14205 * targhooks.h (default_chkp_bound_type): Likewise.
14206 (default_chkp_bound_mode): Likewise.
14207 (default_builtin_chkp_function): Likewise.
14208 (default_chkp_function_value_bounds): Likewise.
14209 (default_chkp_make_bounds_constant): Likewise.
14210 (default_chkp_initialize_bounds): Likewise.
14211 * toplev.c (compile_file): Likewise.
14212 (process_options): Likewise.
14213 * tree-core.h (DEF_BUILTIN): Likewise.
14214 (DEF_BUILTIN_CHKP): Likewise.
14215 * tree-inline.c (declare_return_variable): Likewise.
14216 (remap_gimple_stmt): Likewise.
14217 (copy_bb): Likewise.
14218 (initialize_inlined_parameters): Likewise.
14219 (expand_call_inline): Likewise.
14220 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
14221 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
14222 (make_pass_ipa_chkp_produce_thunks): Likewise.
14223 (make_pass_chkp): Likewise.
14224 (make_pass_chkp_opt): Likewise.
14225 (make_pass_chkp_instrumentation_passes): Likewise.
14226 * tree-pretty-print.c (dump_generic_node): Likewise.
14227 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
14228 * tree-ssa-dce.c (propagate_necessity): Likewise.
14229 (eliminate_unnecessary_stmts): Likewise.
14230 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
14231 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
14232 * tree-ssa-sccvn.h: Likewise.
14233 * tree-ssa-strlen.c (get_string_length): Likewise.
14234 (valid_builtin_call): Likewise.
14235 (adjust_last_stmt): Likewise.
14236 (handle_builtin_strchr): Likewise.
14237 (handle_builtin_strcpy): Likewise.
14238 (handle_builtin_stxncpy): Likewise.
14239 (handle_builtin_memcpy): Likewise.
14240 (handle_builtin_strcat): Likewise.
14241 (strlen_check_and_optimize_stmt): Likewise.
14242 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
14243 * tree-streamer-in.c: Likewise.
14244 * tree-streamer.c (record_common_node): Likewise.
14245 * tree.c (tree_code_size): Likewise.
14246 (wide_int_to_tree_1): Likewise.
14247 (type_contains_placeholder_1): Likewise.
14248 (build_common_tree_nodes): Likewise.
14249 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
14250 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
14251 (POINTER_BOUNDS_P): Likewise.
14252 (BOUNDED_TYPE_P): Likewise.
14253 (BOUNDED_P): Likewise.
14254 (CALL_WITH_BOUNDS_P): Likewise.
14255 (pointer_bounds_type_node): Likewise.
14256 * value-prof.c (gimple_ic): Likewise.
14257 * var-tracking.c (vt_add_function_parameters): Likewise.
14258 * varasm.c (make_decl_rtl): Likewise.
14259 (assemble_start_function): Likewise.
14260 (output_constant): Likewise.
14261 (maybe_assemble_visibility): Likewise.
14262 * varpool.c (ctor_for_folding): Likewise.
14263 * chkp-builtins.def: Remove.
14264 * ipa-chkp.c: Remove.
14265 * ipa-chkp.h: Remove.
14266 * rtl-chkp.c: Remove.
14267 * rtl-chkp.h: Remove.
14268 * tree-chkp-opt.c: Remove.
14269 * tree-chkp.c: Remove.
14270 * tree-chkp.h: Remove.
14272 2018-06-07 Carl Love <cel@us.ibm.com>
14274 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14275 vextract_fp_from_shortl): Add BE support.
14277 2018-06-07 Paul Koning <ni1d@arrl.net>
14279 * compare-elim.c (try_merge_compare): Don't merge compare if
14280 address contains a side effect.
14281 (try_eliminate_compare): Likewise.
14283 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
14285 * config.gcc: Support "tremont".
14286 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
14287 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14289 * config/i386/i386.c (m_TREMONT): Define.
14290 (processor_target_table): Add "tremont".
14291 (PTA_TREMONT): Define.
14292 (ix86_lea_outperforms): Add TARGET_TREMONT.
14293 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
14294 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
14295 and M_INTEL_GOLDMONT_PLUS.
14296 (fold_builtin_cpu): Add "tremont".
14297 (ix86_add_stmt_cost): Add TARGET_TREMONT.
14298 (ix86_option_override_internal): Add "tremont".
14299 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
14300 (processor_type): Add PROCESSOR_TREMONT.
14301 * config/i386/x86-tune.def: Add m_TREMONT.
14302 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
14304 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14306 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
14307 symbol defined for msp430i* devices to be lower case.
14309 2018-06-07 Richard Biener <rguenther@suse.de>
14311 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
14312 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
14313 Properly wrap signed arithmetic if overflow wraps.
14315 2018-06-07 Jakub Jelinek <jakub@redhat.com>
14317 PR tree-optimization/69615
14318 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
14319 of a cast from a same precision integral SSA_NAME in a bb dominated
14320 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
14321 cast to utype if rhs2 has already a compatible type.
14323 2018-06-07 Richard Biener <rguenther@suse.de>
14325 PR tree-optimization/85935
14326 * graphite-scop-detection.c (find_params_in_bb): Analyze
14327 condition operands with respect to the correct loop. Assert
14328 the analysis doesn't fail.
14330 2018-06-04 Carl Love <cel@us.ibm.com>
14332 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
14333 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
14334 as it is slightly cheaper.
14335 (first_match_or_eos_index_<mode>):
14336 Calculate index using natural element order.
14337 (first_match_index_<mode>):
14338 Calculate index using natural element order.
14339 (first_match_or_eos_index_<mode>):
14340 Calculate index using natural order.
14341 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
14342 for BE and LE modes.
14343 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
14344 P9V_BUILTIN_VCLZLSBB_V16QI.
14345 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
14348 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
14350 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
14351 indentation and line wrap for many prototypes. Add missing
14352 @smallexample directives around block of prototypes for vec_xl and
14355 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
14357 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
14358 track if we pass or return IEEE 128-bit floating point.
14359 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
14360 C++ mangling that is compatible with GCC 8.1.
14361 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
14362 (init_cumulative_args): Note if we pass or return IEEE 128-bit
14363 floating point types.
14364 (rs6000_function_arg_advance_1): Likewise.
14365 (rs6000_mangle_type): Optionally generate mangled names that match
14366 what GCC 8.1 generated for IEEE 128-bit floating point types.
14367 (rs6000_globalize_decl_name): If we have an external function that
14368 passes or returns IEEE 128-bit types, generate a weak reference
14369 from the mangled name used in GCC 8.1 to the current mangled
14371 (rs6000_init_builtins): Make __ibm128 use the long double type if
14372 long double is IBM extended double. Make __float128 use the long
14373 double type if long double is IEEE 128-bit.
14376 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
14377 macro for __ibm128 built-in functions.
14378 (PACK_IF): Add __ibm128 pack/unpack functions.
14379 (UNPACK_IF): Likewise.
14380 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
14381 enable long double built-in functions if long double is IEEE
14382 128-bit floating point.
14383 (rs6000_invalid_builtin): Update long double built-in function
14385 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
14386 functions, adjust the built-in function to use the long double
14387 built-in function if __ibm128 and long double are the same type.
14388 * doc/extend.texi (PowerPC builtins): Update documention for
14389 __builtin_{,un}pack_longdouble. Add documentation for
14390 __builtin_{,un}pack_ibm128.
14392 2018-06-06 Jim Wilson <jimw@sifive.com>
14394 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
14395 (struct machine_function): New field interrupt_mode.
14396 (riscv_handle_type_attribute): New function. Add forward declaration.
14397 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
14398 (riscv_expand_epilogue): Check interrupt_mode field.
14399 (riscv_set_current_function): Check interrupt attribute args and
14400 set interrupt_mode field.
14401 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
14402 (riscv_sret, riscv_uret): New.
14403 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
14404 new arguments to interrupt attribute.
14406 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
14409 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
14410 Don't handle -mcpu=power8 if -mpower9-vector is also used.
14412 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
14414 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
14415 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
14416 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
14417 several redundant entries.
14419 2018-06-06 David Malcolm <dmalcolm@redhat.com>
14421 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
14422 type from "rtx" to "rtx_insn *".
14423 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
14424 for local "call_insn", removing cast.
14425 (ix86_expand_call): Likewise, introducing a "call_insn" local.
14427 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
14429 PR tree-optimization/86066
14430 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
14431 for BIT_INSERT_EXPR stores.
14433 2018-06-06 Richard Biener <rguenther@suse.de>
14435 PR tree-optimization/86062
14436 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
14437 component refs ontop
14438 of to be offsetted base.
14440 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14442 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
14443 to be static and remove check on interrupt attribute name.
14445 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
14447 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
14448 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
14450 2018-06-05 Steve Ellcey <sellcey@cavium.com>
14453 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
14455 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
14456 Remove second argument, change how error is called.
14457 (aarch64_layout_arg): Remove second argument from
14458 aarch64_err_no_fpadvsimd call.
14459 (aarch64_init_cumulative_args): Ditto.
14460 (aarch64_gimplify_va_arg_expr): Ditto.
14461 * config/aarch64/aarch64.md (mov<mode>): Ditto.
14463 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
14465 * config/i386/i386.md (simple_return_indirect_internal): New expander.
14466 (*simple_return_indirect_internal<mode>): Rename from
14467 simple_return_indirect_internal. Use W mode iterator.
14468 (rstorssp): New expander.
14469 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
14470 (clrssbsy): New expander.
14471 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
14473 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
14475 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
14477 (cmse_check_pointed_object): Likewise.
14479 2018-06-05 Martin Liska <mliska@suse.cz>
14481 PR gcov-profile/47618
14482 * doc/invoke.texi: Document how -fprofile-dir format
14485 2018-06-05 Richard Biener <rguenther@suse.de>
14487 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
14488 removal pretend DOM info isn't available so we do not update
14489 it and only remove edges, not dominated blocks. Actually free
14490 DOM info in case we removed something. Remove unreachable blocks.
14491 (mfb_keep_latches): Work with either DOM info or marked backedges.
14492 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
14493 first. Mark backedges if DOM info isn't available.
14494 (Re-)compute DOM info after cleanup_control_flow_pre.
14496 2018-06-05 Richard Biener <rguenther@suse.de>
14498 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
14499 (locus_discrim_hasher::hash): Adjust.
14500 (locus_discrim_hasher::equal): Likewise.
14501 (next_discriminator_for_locus): Work on line directly.
14502 (same_line_p): Pass in expanded locus1 as well.
14503 (assign_discriminators): Avoid redundant location expansions.
14505 2018-06-05 Richard Biener <rguenther@suse.de>
14507 PR tree-optimization/86046
14508 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
14509 if required after clearing TREE_ADDRESSABLE.
14511 2018-06-05 Richard Biener <rguenther@suse.de>
14513 PR tree-optimization/86047
14514 * tree-ssa-loop.c (for_each_index): Glob handling of all
14515 decls and constants and really handle all of them.
14517 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14520 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
14521 qualifier_void_pointer and qualifier_const_void_pointer.
14522 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
14523 (arm_init_builtins): Handle the above.
14524 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
14525 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
14526 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
14529 2018-06-05 Martin Liska <mliska@suse.cz>
14531 * auto-profile.c (read_autofdo_file): Do not use
14532 gcov_ctr_summary struct.
14533 (afdo_callsite_hot_enough_for_early_inline): Likewise.
14534 * coverage.c (struct counts_entry): Likewise.
14535 (read_counts_file): Read just single summary entry.
14536 (get_coverage_counts): Use gcov_summary struct.
14537 * coverage.h (get_coverage_counts): Likewise.
14538 * gcov-dump.c (dump_working_sets): Likewise.
14539 (tag_summary): Dump just single summary.
14540 * gcov-io.c (gcov_write_summary): Write just histogram
14542 (gcov_read_summary): Read just single summary.
14543 (compute_working_sets): Use gcov_summary struct.
14544 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
14545 of GCOV_COUNTERS_SUMMABLE.
14546 (GCOV_COUNTERS_SUMMABLE): Remove.
14547 (GCOV_FIRST_VALUE_COUNTER): Replace with
14548 GCOV_COUNTER_V_INTERVAL.
14549 (struct gcov_ctr_summary): Remove.
14550 (struct gcov_summary): Directly use fields of former
14552 (compute_working_sets): Use gcov_summary struct.
14553 * gcov.c (read_count_file): Do not use ctrs fields.
14554 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
14556 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
14558 * profile.c: Likewise.
14559 * profile.h: Likewise.
14561 2018-06-05 Martin Liska <mliska@suse.cz>
14563 PR gcov-profile/84846
14564 * gcov.c (output_lines): Print working directory only
14565 in intermediate format.
14567 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
14569 * config/s390/s390-builtin-types.def: Add void function type.
14570 * config/s390/s390-builtins.def: Use the function type for the
14573 2018-06-04 Jim Wilson <jimw@sifive.com>
14575 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
14577 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
14578 and maybe_eh_return. Change regno to unsigned int. Use new args to
14579 handle EH_RETURN_DATA_REGNO registers properly.
14580 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
14581 (riscv_expand_epilogue): Update comment. Change argument name and
14582 type. Update code to use new name and type. Pass new args to
14583 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
14585 * config/riscv/riscv.md (NORMAL_RETURN): New.
14586 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
14587 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
14588 (eh_return): Call gen_eh_return_internal and emit barrier.
14589 (eh_return_internal): Call riscv_expand_epilogue.
14591 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
14593 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
14594 bit_insertion field and declare can_be_merged_into method.
14595 (merged_store_group::can_be_merged_into): New method.
14596 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
14597 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
14598 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
14600 2018-06-04 Richard Biener <rguenther@suse.de>
14602 PR tree-optimization/85955
14603 * builtins.c (fold_builtin_sincos): Convert pointers to
14604 destination to appropriate type before dereferencing.
14606 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
14608 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
14610 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
14612 * expr.c (expand_expr_real_1): Force the operand into memory if
14613 its TYPE_MODE is BLKmode and if there is no integer mode for
14614 the number of bits being extracted.
14616 2018-06-04 Jakub Jelinek <jakub@redhat.com>
14620 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
14621 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
14623 2018-06-04 Richard Biener <rguenther@suse.de>
14625 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
14626 (cleanup_tree_cfg_noloop): ... single caller. Do
14627 start_recording_case_labels later.
14629 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
14631 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
14632 to _IMMINTRIN_H_INCLUDED.
14633 * config/i386/pconfigintrin.h: Ditto.
14634 * config/i386/waitpkgintrin.h: Ditto.
14635 * config/i386/immintrin.h: Add includes for sgxintrin.h,
14636 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
14637 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
14638 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
14639 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
14640 waitpkgintrin.h and cldemoteintrin.h.
14642 2018-06-04 Richard Biener <rguenther@suse.de>
14644 PR tree-optimization/86038
14645 * tracer.c (find_best_successor): Check probability for
14646 being initialized, bail out if not.
14648 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
14651 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
14652 of bits to ignore when comparing architectures.
14654 2018-06-04 Jakub Jelinek <jakub@redhat.com>
14656 PR tree-optimization/69615
14657 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
14658 maximum or minimum of the type, try to merge it also as if
14659 range1 is + [-, x - 1] or + [x + 1, -].
14662 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
14664 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
14666 PR tree-optimization/86034
14667 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
14668 the unsigned bitfield type in a bit insertion sequence if it does not
14669 have a larger precision than the bitfield size.
14670 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
14672 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
14674 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
14676 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
14678 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
14679 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
14680 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
14681 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
14683 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
14685 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
14686 Disable -fdelete-null-pointer-checks for ELF toolchain.
14688 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
14689 Kito Cheng <kito.cheng@gmail.com>
14691 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
14692 (nds32le-*-*, nds32be-*-*): Integrate checking process.
14693 (nds32*-*-*): Add glibc and uclibc conditions.
14694 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
14695 (TARGET_EXCEPT_UNWIND_INFO): Define.
14696 * config/nds32/elf.h: New file.
14697 * config/nds32/linux.h: New file.
14698 * config/nds32/nds32-elf.opt: New file.
14699 * config/nds32/nds32-linux.opt: New file.
14700 * config/nds32/nds32-fp-as-gp.c
14701 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
14702 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
14704 (nds32_asm_file_end): Ditto.
14705 (nds32_print_operand): Ditto.
14706 (nds32_insert_attributes): Ditto.
14707 (nds32_init_libfuncs): New function.
14708 (TARGET_HAVE_TLS): Define.
14709 (TARGET_INIT_LIBFUNCS): Define.
14710 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
14712 (TARGET_ELF): Apply different mcmodel setting.
14713 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
14714 been migrated into elf.h and linux.h files.
14715 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
14716 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
14717 (mcmodel): The content has been migrated into nds32-elf.opt and
14718 nds32-linux.opt files.
14719 * config/nds32/t-elf: New file.
14720 * config/nds32/t-linux: New file.
14722 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
14723 Shiva Chen <shiva0217@gmail.com>
14725 * config/nds32/constants.md (unspec_volatile_element): Add
14726 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
14727 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
14729 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
14730 (make_pass_nds32_fp_as_gp): Declare.
14731 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
14733 (nds32_asm_function_end_prologue): Remove unused asm output.
14734 (nds32_asm_function_begin_epilogue): Remove unused asm output.
14735 (nds32_asm_file_start): Output necessary fp_as_gp information.
14736 (nds32_option_override): Adjust register usage.
14737 (nds32_expand_prologue): Consider fp_as_gp situation.
14738 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
14739 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
14742 (simple_return): Ditto.
14743 (omit_fp_begin): Output special directive for fp_as_gp.
14744 (omit_fp_end): Output special directive for fp_as_gp.
14745 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
14746 mforbid-fp-as-gp): New options.
14748 2018-06-01 Mark Wielaard <mark@klomp.org>
14750 * dwarf2out.c (dwarf2out_finish): Remove generation of
14751 DW_AT_loclists_base.
14753 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
14755 * gimple-ssa-store-merging.c: Include gimple-fold.h.
14756 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
14757 (struct merged_store_group): Add bit_insertion field.
14758 (dump_char_array): Use standard hexadecimal format.
14759 (merged_store_group::merged_store_group): Set bit_insertion to false.
14760 (merged_store_group::apply_stores): Use optimal buffer size. Deal
14761 with BIT_INSERT_EXPR stores. Move up code updating the mask and
14762 also print the mask in the dump file.
14763 (pass_store_merging::gate): Minor tweak.
14764 (imm_store_chain_info::coalesce_immediate): Fix wrong association
14765 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
14766 stores with INTEGER_CST stores.
14767 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
14768 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
14769 and use it throughout. Generate bit insertion sequences if need be.
14770 (pass_store_merging::process_store): Remove redundant condition.
14771 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
14773 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
14775 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
14776 the 128-bit floating point types. Fix function comment.
14778 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14780 * config/aarch64/aarch64-simd.md
14781 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
14783 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
14786 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
14788 PR tree-optimization/85989
14789 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
14791 (backprop::intersect_uses): Check it when deciding whether this
14792 is a backedge reference.
14793 (backprop::process_block): Add each phi to m_visited_phis
14794 after visiting it, then clear it at the end.
14796 2018-06-01 Richard Biener <rguenther@suse.de>
14798 * tree-vectorizer.h (vect_dr_stmt): New function.
14799 (vect_get_load_cost): Adjust.
14800 (vect_get_store_cost): Likewise.
14801 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
14802 Use vect_dr_stmt instead of DR_SMTT.
14803 (vect_record_base_alignments): Likewise.
14804 (vect_calculate_target_alignment): Likewise.
14805 (vect_compute_data_ref_alignment): Likewise and make static.
14806 (vect_update_misalignment_for_peel): Likewise.
14807 (vect_verify_datarefs_alignment): Likewise.
14808 (vector_alignment_reachable_p): Likewise.
14809 (vect_get_data_access_cost): Likewise. Pass down
14810 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
14811 (vect_get_peeling_costs_all_drs): Likewise.
14812 (vect_peeling_hash_get_lowest_cost): Likewise.
14813 (vect_enhance_data_refs_alignment): Likewise.
14814 (vect_find_same_alignment_drs): Likewise.
14815 (vect_analyze_data_refs_alignment): Likewise.
14816 (vect_analyze_group_access_1): Likewise.
14817 (vect_analyze_group_access): Likewise.
14818 (vect_analyze_data_ref_access): Likewise.
14819 (vect_analyze_data_ref_accesses): Likewise.
14820 (vect_vfa_segment_size): Likewise.
14821 (vect_small_gap_p): Likewise.
14822 (vectorizable_with_step_bound_p): Likewise.
14823 (vect_prune_runtime_alias_test_list): Likewise.
14824 (vect_analyze_data_refs): Likewise.
14825 (vect_supportable_dr_alignment): Likewise.
14826 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
14827 (vect_gen_prolog_loop_niters): Likewise.
14828 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
14829 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
14831 (vect_recog_mask_conversion_pattern): Likewise.
14832 (vect_try_gather_scatter_pattern): Likewise.
14833 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
14834 to vect_get_store_cost.
14835 (vect_get_store_cost): Get stmt_info instead of DR.
14836 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
14837 (vect_get_load_cost): Get stmt_info instead of DR.
14839 2018-06-01 Richard Biener <rguenther@suse.de>
14841 PR middle-end/86017
14842 * gimple-fold.c (var_decl_component_p): Also allow offsetted
14843 vars wrapped in MEM_REFs.
14845 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
14847 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
14848 Fix subreg tests so that we only return a choice between
14849 GENERAL_REGS and FP_REGS if the original classes included both.
14851 2018-06-01 Richard Biener <rguenther@suse.de>
14854 * tree-ssa-structalias.c (get_function_part_constraint):
14855 Handle NULL fi->decl.
14856 (find_func_aliases_for_call): Properly handle indirect
14857 fi from direct call.
14858 (find_func_clobbers): Likewise.
14859 (ipa_pta_execute): Likewise.
14860 (create_variable_info_for): For functions that are ifunc_resolver
14861 resolve to a varinfo that contains the result of the resolver call.
14862 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
14865 2018-05-31 Michael Collison <michael.collison@arm.com>
14867 * config/aarch64/aarch64.md:
14868 (*fix_to_zero_extenddfdi2): New pattern.
14869 * gcc.target/aarch64/fix_extend1.c: New testcase.
14871 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
14873 PR middle-end/78809
14874 PR middle-end/83026
14875 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
14876 and BUILT_IN_STRNCMP_EQ.
14877 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
14878 BUILT_IN_STRNCMP_EQ.
14879 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
14880 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
14881 (gimple_fold_builtin): Likewise.
14882 * tree-ssa-strlen.c (compute_string_length): New function.
14883 (determine_min_obsize): New function.
14884 (handle_builtin_string_cmp): New function to handle calls to
14885 string compare functions.
14886 (strlen_optimize_stmt): Add handling to builtin string compare
14888 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
14889 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
14890 * tree.c (build_common_builtin_nodes): Add new defines of
14891 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
14893 2018-05-31 Jakub Jelinek <jakub@redhat.com>
14896 * bb-reorder.c (pass_partition_blocks::gate): Return false for
14897 functions with naked attribute.
14899 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
14901 * config/i386/sse.md (avx_vec_concat<mode>):
14902 Substitute concat_tg_mode mode attribute with xtg_mode.
14903 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
14904 (concat_tg_mode): Remove mode attribute.
14906 2018-05-31 Martin Sebor <msebor@redhat.com>
14909 * calls.c (alloc_max_size): Correct a logic error/typo.
14910 Treat excessive arguments as infinite. Warn for invalid arguments.
14911 * doc/invoke.texi (-Walloc-size-larger-than): Update.
14913 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
14916 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
14917 and movx for Haswell.
14919 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
14920 Cesar Philippidis <cesar@codesourcery.com>
14922 PR middle-end/85879
14923 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
14924 when emitting error on private/firstprivate reductions.
14925 * omp-low.c (lower_omp_target): Avoid reference-type processing
14926 on pointers for firstprivate clause.
14928 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
14930 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
14933 * config/aarch64/aarch64-simd.md
14934 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
14935 (aarch64_ld1_x3_<mode>): Likewise
14936 (aarch64_st1x2<VALLDIF:mode>): Likewise
14937 (aarch64_st1_x2_<mode>): Likewise
14938 (aarch64_st1x3<VALLDIF:mode>): Likewise
14939 (aarch64_st1_x3_<mode>): Likewise
14940 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
14941 (vld1_s8_x3): Likewise.
14942 (vld1_u16_x3): Likewise.
14943 (vld1_s16_x3): Likewise.
14944 (vld1_u32_x3): Likewise.
14945 (vld1_s32_x3): Likewise.
14946 (vld1_u64_x3): Likewise.
14947 (vld1_s64_x3): Likewise.
14948 (vld1_f16_x3): Likewise.
14949 (vld1_f32_x3): Likewise.
14950 (vld1_f64_x3): Likewise.
14951 (vld1_p8_x3): Likewise.
14952 (vld1_p16_x3): Likewise.
14953 (vld1_p64_x3): Likewise.
14954 (vld1q_u8_x3): Likewise.
14955 (vld1q_s8_x3): Likewise.
14956 (vld1q_u16_x3): Likewise.
14957 (vld1q_s16_x3): Likewise.
14958 (vld1q_u32_x3): Likewise.
14959 (vld1q_s32_x3): Likewise.
14960 (vld1q_u64_x3): Likewise.
14961 (vld1q_s64_x3): Likewise.
14962 (vld1q_f16_x3): Likewise.
14963 (vld1q_f32_x3): Likewise.
14964 (vld1q_f64_x3): Likewise.
14965 (vld1q_p8_x3): Likewise.
14966 (vld1q_p16_x3): Likewise.
14967 (vld1q_p64_x3): Likewise.
14968 (vst1_s64_x2): Likewise.
14969 (vst1_u64_x2): Likewise.
14970 (vst1_f64_x2): Likewise.
14971 (vst1_s8_x2): Likewise.
14972 (vst1_p8_x2): Likewise.
14973 (vst1_s16_x2): Likewise.
14974 (vst1_p16_x2): Likewise.
14975 (vst1_s32_x2): Likewise.
14976 (vst1_u8_x2): Likewise.
14977 (vst1_u16_x2): Likewise.
14978 (vst1_u32_x2): Likewise.
14979 (vst1_f16_x2): Likewise.
14980 (vst1_f32_x2): Likewise.
14981 (vst1_p64_x2): Likewise.
14982 (vst1q_s8_x2): Likewise.
14983 (vst1q_p8_x2): Likewise.
14984 (vst1q_s16_x2): Likewise.
14985 (vst1q_p16_x2): Likewise.
14986 (vst1q_s32_x2): Likewise.
14987 (vst1q_s64_x2): Likewise.
14988 (vst1q_u8_x2): Likewise.
14989 (vst1q_u16_x2): Likewise.
14990 (vst1q_u32_x2): Likewise.
14991 (vst1q_u64_x2): Likewise.
14992 (vst1q_f16_x2): Likewise.
14993 (vst1q_f32_x2): Likewise.
14994 (vst1q_f64_x2): Likewise.
14995 (vst1q_p64_x2): Likewise.
14996 (vst1_s64_x3): Likewise.
14997 (vst1_u64_x3): Likewise.
14998 (vst1_f64_x3): Likewise.
14999 (vst1_s8_x3): Likewise.
15000 (vst1_p8_x3): Likewise.
15001 (vst1_s16_x3): Likewise.
15002 (vst1_p16_x3): Likewise.
15003 (vst1_s32_x3): Likewise.
15004 (vst1_u8_x3): Likewise.
15005 (vst1_u16_x3): Likewise.
15006 (vst1_u32_x3): Likewise.
15007 (vst1_f16_x3): Likewise.
15008 (vst1_f32_x3): Likewise.
15009 (vst1_p64_x3): Likewise.
15010 (vst1q_s8_x3): Likewise.
15011 (vst1q_p8_x3): Likewise.
15012 (vst1q_s16_x3): Likewise.
15013 (vst1q_p16_x3): Likewise.
15014 (vst1q_s32_x3): Likewise.
15015 (vst1q_s64_x3): Likewise.
15016 (vst1q_u8_x3): Likewise.
15017 (vst1q_u16_x3): Likewise.
15018 (vst1q_u32_x3): Likewise.
15019 (vst1q_u64_x3): Likewise.
15020 (vst1q_f16_x3): Likewise.
15021 (vst1q_f32_x3): Likewise.
15022 (vst1q_f64_x3): Likewise.
15023 (vst1q_p64_x3): Likewise.
15025 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
15027 * config/msp430/msp430.c (msp430_output_labelref): Prepend
15028 user_label_prefix to name.
15030 * tree-core.h: Update comment about the format of NAME string
15031 passed to handler in attribute_spec.
15033 * config/msp430/msp430.md: Remove erroneous subreg expression from
15034 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
15035 zero_extend{q,h}isi2.
15037 2018-05-30 Borislav Petkov <bp@suse.de>
15039 * doc/extend.texi: Document some architecture specific
15040 constraints and sort entries.
15042 2018-05-30 Martin Sebor <msebor@redhat.com>
15044 PR middle-end/85369
15045 * builtins.c (expand_builtin_stpcpy_1): New function.
15046 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
15047 only if the former succeeds.
15049 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
15051 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
15054 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15056 * doc/invoke.texi (-flinker-output): Document
15058 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15060 * passes.c (ipa_write_summaries): Only modify statements if body
15062 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
15063 incrementally linking.
15064 (ipa_passes): Likewise.
15065 * lto-cgraph.c (lto_output_node): When incrementally linking do not
15066 pass down resolution info.
15067 * common.opt (flag_incremental_link): Update info.
15068 * gcc.c (plugin specs): Turn flinker-output=* to
15069 -plugin-opt=-linker-output-known
15070 * toplev.c (compile_file): Also cut compilation when doing incremental
15072 * flag-types. (enum lto_partition_model): Add
15073 LTO_LINKER_OUTPUT_NOLTOREL.
15074 (invoke.texi): Add -flinker-output docs.
15075 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
15076 link same way as WPA; do not stream in dead initializers.
15078 * dwarf2out.c (dwarf2out_die_ref_for_decl,
15079 darf2out_register_external_decl): Support incremental link.
15081 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15083 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
15085 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15087 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
15088 it down to simple_object_copy_lto_debug_sections.
15089 (run_gcc): Determine incremental LTO link time and configure
15090 lto1 into non-wpa mode, disable renaming of debug sections.
15092 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
15094 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
15095 descriptions of various incorrectly documented functions.
15097 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
15100 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
15101 address check not strict.
15103 2018-05-30 Richard Biener <rguenther@suse.de>
15105 PR tree-optimization/85964
15106 * tracer.c (better_p): Drop initialized count check, we only
15107 call the function with initialized counts now.
15108 (find_best_successor): Do find a best edge if one
15109 has uninitialized count.
15110 (find_best_predecessor): Likewise. Do BB frequency check only
15111 if count is initialized.
15113 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
15115 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
15116 (aarch64_ldrstr_offset_compare): New.
15117 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
15118 load/store orderings.
15119 (aarch64_gen_adjusted_ldpstp): Likewise.
15121 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
15123 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
15124 Check for subset of GENERAL_REGS and FP_REGS.
15125 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
15128 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
15130 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
15131 and wi::to_poly_offset. Add the current offset and then check
15132 whether the sum fits, rather than using an unchecked addition of
15133 a checked term. Check for a shwi rather than a uhwi.
15134 * expr.c (get_bit_range): Use tree_to_poly_uint64.
15135 (store_constructor): Use poly_int_tree_p.
15136 (expand_expr_real_1): Likewise.
15137 * function.c (assign_temp): Likewise.
15138 * fold-const.c (const_binop): Use poly_int_tree_p and
15139 wi::to_poly_offset.
15140 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
15142 * ipa-icf-gimple.c (func_checker::compare_operand): Use
15143 to_poly_offset for MEM offsets.
15144 * ipa-icf.c (sem_variable::equals): Likewise.
15145 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
15146 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
15147 wi::to_poly_offset for BIT_FIELD_REF offsets.
15148 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
15149 wi::to_poly_offset.
15150 * var-tracking.c (emit_note_insn_var_location): Use
15151 tree_to_poly_uint64.
15153 2018-05-29 Jim Wilson <jimw@sifive.com>
15155 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
15157 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
15160 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
15161 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
15163 (sse4_1_round<mode>2): Use nonimmediate_operand
15164 for operand 1 predicate.
15166 2018-05-29 Martin Sebor <msebor@redhat.com>
15167 Richard Biener <rguenther@suse.de>
15170 * calls.c (get_size_range): Call determine_value_range instead
15171 of get_value_range..
15172 * tree-vrp.h (determine_value_range): Declared new function.
15173 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
15175 2018-05-29 Richard Biener <rguenther@suse.de>
15177 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
15178 sure to use non-pattern stmts for get_earlier_stmt arguments.
15179 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
15180 called on pattern stmts.
15181 (get_later_stmt): Likewise.
15183 2018-05-29 Martin Liska <mliska@suse.cz>
15185 PR gcov-profile/85759
15186 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
15189 2018-05-29 Jakub Jelinek <jakub@redhat.com>
15191 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
15193 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
15197 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
15198 VEC_PACK_FLOAT_EXPR): New tree codes.
15199 * tree-pretty-print.c (op_code_prio): Handle
15200 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
15201 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
15202 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
15203 * tree-inline.c (estimate_operator_cost): Likewise.
15204 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
15205 * fold-const.c (const_binop): Likewise.
15206 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
15207 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
15208 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
15209 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
15210 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
15211 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
15212 * expr.c (expand_expr_real_2): Likewise.
15213 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
15214 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
15215 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
15217 * optabs.c (expand_widen_pattern_expr): For
15218 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
15219 sign from result type rather than operand's type.
15220 (expand_binop_directly): For vec_packu_float_optab and
15221 vec_packs_float_optab allow result type to be different from operand's
15223 * optabs-tree.c (optab_for_tree_code): Handle
15224 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
15225 VEC_PACK_FLOAT_EXPR. Formatting fixes.
15226 * tree-vect-generic.c (expand_vector_operations_1): Handle
15227 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
15228 VEC_PACK_FLOAT_EXPR.
15229 * tree-vect-stmts.c (supportable_widening_operation): Handle
15231 (supportable_narrowing_operation): Handle FLOAT_EXPR.
15232 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
15233 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
15234 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
15235 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
15237 (vec_pack<floatprefix>_float_<mode>): New expander.
15238 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
15240 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
15241 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
15242 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
15243 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
15244 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
15246 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
15247 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
15248 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
15249 VEC_PACK_FLOAT_EXPR): Document.
15251 2018-05-29 Richard Biener <rguenther@suse.de>
15253 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
15255 (stmt_vec_info_vec): Make pointer.
15256 (init_stmt_vec_info_vec): Remove.
15257 (free_stmt_vec_info_vec): Likewise.
15258 (set_stmt_vec_info_vec): New function.
15259 (free_stmt_vec_infos): Likewise.
15260 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
15261 (set_vinfo_for_stmt): Likewise.
15262 (get_earlier_stmt): Likewise.
15263 (get_later_stmt): Likewise.
15264 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
15265 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
15266 (vec_info::~vec_info): Free stmt_vec_infos.
15267 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
15268 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
15269 (pass_slp_vectorize::execute): Likewise.
15270 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
15271 (free_stmt_vec_info_vec): Likewise.
15272 (set_stmt_vec_info_vec): New function.
15273 (free_stmt_vec_infos): Likewise.
15274 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
15275 the global stmt_vec_info_vec.
15276 * tree-parloops.c (gather_scalar_reductions): Use
15277 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
15280 2018-05-29 Richard Biener <rguenther@suse.de>
15282 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
15284 2018-05-29 Martin Liska <mliska@suse.cz>
15285 David Malcolm <dmalcolm@redhat.com>
15287 * vec.c (test_reverse): New.
15288 (vec_c_tests): Add new test.
15289 * vec.h (vl_ptr>::reverse): New function.
15291 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
15293 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
15295 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
15298 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
15300 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
15302 2018-05-28 Richard Biener <rguenther@suse.de>
15304 PR tree-optimization/85933
15305 * tree-vect-data-refs.c (vect_record_base_alignments): Only
15306 look at stmts marked as vectorizable.
15308 2018-05-28 Richard Biener <rguenther@suse.de>
15310 PR tree-optimization/85934
15311 * tree-vect-generic.c (expand_vector_operations_1): Hoist
15312 vector boolean check before scalar optimization.
15314 2018-05-28 Jakub Jelinek <jakub@redhat.com>
15316 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
15319 2018-05-28 Mark Wielaard <mark@klomp.org>
15321 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
15322 if it is an expression containing a minus sign.
15324 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
15326 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
15328 2018-05-27 Paul Koning <ni1d@arrl.net>
15330 * config/pdp11/pdp11.md (truncsihi2): Remove.
15332 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
15333 Chung-Ju Wu <jasonwucj@gmail.com>
15335 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
15337 (unaligned_store_dw): Ditto.
15338 * config/nds32/nds32-memory-manipulation.c
15339 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
15340 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
15341 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
15342 (emit_setmem_word_loop): Rename to ...
15343 (emit_setmem_doubleword_loop): ... this.
15344 (nds32_gen_dup_4_byte_to_word_value): New function.
15345 (nds32_gen_dup_8_byte_to_double_word_value): New function.
15346 (nds32_expand_setmem_loop): Refine implementation.
15347 (nds32_expand_setmem_loop_v3m): Ditto.
15348 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
15351 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
15353 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
15355 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
15357 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
15358 (nds32_init_machine_status): Initialize machine->attr_naked_p and
15359 machine->attr_no_prologue_p.
15360 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
15361 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
15362 (nds32_expand_epilogue): Consider attr_naked_p.
15363 (nds32_expand_epilogue_v3pop): Likewise.
15364 (nds32_can_use_return_insn): Likewise.
15365 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
15366 attr_no_prologue_p fields.
15367 * config/nds32/nds32.opt (mret-in-naked-func): New option.
15369 2018-05-27 Jakub Jelinek <jakub@redhat.com>
15372 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
15374 * config/i386/sse.md
15375 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
15377 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
15379 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
15381 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
15383 (*<floatsuffix>floatv2div2sf2): Rename to ...
15384 (*float<floatunssuffix>v2div2sf2): ... this.
15385 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
15386 (float<floatunssuffix>v2div2sf2_mask): ... this.
15387 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
15388 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
15389 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
15391 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
15393 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
15395 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
15397 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
15399 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
15401 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
15402 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
15403 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
15404 gen_ufix_truncv8dfv8si2.
15405 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
15406 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
15407 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
15408 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
15409 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
15410 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
15411 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
15412 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
15414 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
15418 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
15420 2018-05-25 Jim Wilson <jimw@sifive.com>
15422 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
15423 * config/riscv/riscv.c (struct machine_function): Add
15424 interrupt_handler_p and attribute_checked_p fields.
15425 (riscv_attribute_table): Add interrupt.
15426 (riscv_interrupt_type_p): New.
15427 (riscv_save_reg_p): Save extra regs for interrupt handler.
15428 (riscv_use_save_libcall): Return false for interrupt handler.
15429 (riscv_first_stack_step): Add forward declaration.
15430 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
15431 for interrupt handler with large frame. Use it for saved reg list.
15432 (riscv_expand_prologue): Move flag_stack_usage_info support to
15433 eliminate duplication.
15434 (riscv_expand_epilogue): Generate mret for interrupt handler.
15435 (riscv_epilogue_uses): New.
15436 (riscv_can_use_return_insn): Return false for interrupt handler.
15437 (riscv_function_ok_for_sibcall): Likewise.
15438 (riscv_set_current_function): Add interrupt handler support.
15439 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
15440 * config/riscv/riscv.md (UNSPECV_MRET): New.
15442 (riscv_frflags, riscv_fsflags): Use tab after opcode.
15444 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
15446 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
15448 PR tree-optimization/85712
15449 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
15450 this candidate has already been replaced in-situ by a copy.
15452 2018-05-25 Jason Merrill <jason@redhat.com>
15454 PR c++/80485 - inline function non-zero address.
15455 * symtab.c (nonzero_address): Check DECL_COMDAT.
15457 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
15460 * config/alpha/alpha.md (ashlsi3): New insn pattern.
15461 (*ashlsi_se): Rename from *ashldi_se. Define as sign
15462 extension of SImode operation. Use const123_operand predicate.
15463 (*saddsi_1): Remove.
15464 (*saddl_se_1): Ditto.
15465 (*ssubsi_1): Ditto.
15466 (*ssubl_se_1): Ditto.
15467 * config/alpha/predicates.md (const123_operand): New predicate.
15468 * config/alpha/constraints.md (P): Use IN_RANGE.
15470 2018-05-25 Richard Biener <rguenther@suse.de>
15472 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
15474 (ref_maybe_used_by_stmt_p): Likewise.
15475 (stmt_may_clobber_ref_p): Likewise.
15476 (stmt_may_clobber_ref_p_1): Likewise.
15477 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
15479 (ref_maybe_used_by_stmt_p): Likewise.
15480 (stmt_may_clobber_ref_p): Likewise.
15481 (stmt_may_clobber_ref_p_1): Likewise.
15482 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
15483 the alias oracle to disambiguate DRs with stmts DR analysis
15485 (vect_analyze_data_refs): Do not give up on not analyzable
15486 DRs for BB vectorization. Remove code truncating the dataref
15489 2018-05-25 Jakub Jelinek <jakub@redhat.com>
15492 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
15493 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
15494 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
15496 2018-05-25 Richard Biener <rguenther@suse.de>
15498 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
15499 function, combining stmt data ref gathering and fatal analysis
15501 (vect_analyze_data_refs): Remove now redudnant code and simplify.
15502 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
15503 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
15504 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
15505 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
15507 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15509 PR tree-optimization/85720
15510 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
15511 SCC if all partitions are builtins.
15512 (version_loop_by_alias_check): New parameter. Generate cancelable
15513 runtime alias check if all partitions are builtins.
15514 (distribute_loop): Update call to above function.
15516 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15518 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
15519 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
15520 (parm_default_def_partition_arg): Ditto.
15521 (set_parm_default_def_partition): Ditto.
15522 (get_parm_default_def_partitions): Ditto and make it static.
15523 (get_undefined_value_partitions): Ditto and make it static.
15524 (remove_ssa_form): Refactor call to init_var_map here.
15525 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
15526 computation for loop region.
15527 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
15528 (register_default_def): Delete.
15529 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
15530 (parm_default_def_partition_arg): Ditto.
15531 (set_parm_default_def_partition): Ditto.
15532 (get_parm_default_def_partitions): Ditto and make it static.
15533 (get_undefined_value_partitions): Ditto and make it static.
15534 (coalesce_with_default, coalesce_with_default): Update comment.
15535 (create_coalesce_list_for_region): New func factored out from
15536 create_outofssa_var_map.
15537 (populate_coalesce_list_for_outofssa): New func factored out from
15538 create_outofssa_var_map and coalesce_ssa_name.
15539 (create_outofssa_var_map): Delete.
15540 (coalesce_ssa_name): Refactor to support live range computation.
15541 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
15542 (get_parm_default_def_partitions): Delete.
15543 (get_undefined_value_partitions): Ditto.
15544 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
15545 computation for loop region.
15546 (new_tree_live_info, loe_visit_block): Ditto.
15547 (live_worklist, set_var_live_on_entry): Ditto.
15548 (calculate_live_on_exit, verify_live_on_entry): Ditto.
15549 * tree-ssa-live.h (struct _var_map): New fields.
15550 (init_var_map): Change decl.
15551 (region_contains_p): New.
15553 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15555 * tree-ssa-live.h (live_merge_and_clear): Delete.
15557 2018-05-25 Richard Biener <rguenther@suse.de>
15560 * tree-dump.c (dequeue_and_dump): Remove access to removed
15561 operand 2 of a SWITCH_EXPR.
15563 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15565 * doc/sourcebuild.texi (vect_double_cond_arith): Include
15566 multiplication and division.
15567 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
15568 (cond_udiv@var{m}, cond_umod@var{m}): Document.
15569 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
15570 (cond_udiv_optab, cond_umod_optab): New optabs.
15571 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
15572 (IFN_COND_RDIV): New internal functions.
15573 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
15574 TRUNC_MOD_EXPR and RDIV_EXPR.
15575 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
15576 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
15578 (SVE_INT_BINARY): Include mult.
15579 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
15580 (optab, sve_int_op): Handle mult.
15581 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
15583 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
15584 for SVE_INT_BINARY_SD.
15586 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15588 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
15589 (optab, sve_int_op): Handle div and udiv.
15590 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
15591 for SVE_INT_BINARY_SD.
15592 (*<optab><mode>3): New insn for the same.
15594 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15596 * tree-vect-patterns.c: Include predict.h.
15597 (vect_recog_divmod_pattern): Restrict check for division support
15598 to when optimizing for size.
15600 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15602 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
15603 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
15604 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
15605 (gimple_match_op::set_op): Likewise.
15606 (gimple_resimplify4): Declare.
15607 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
15608 (expr::gen_transform): Likewise.
15609 (decision_tree::gen): Generate a simplification routine for 4 operands.
15610 * gimple-match-head.c (gimple_simplify): Add an overload for
15611 4 operands. In the top-level function, handle up to 4 call
15612 arguments and call gimple_resimplify4.
15613 (gimple_resimplify4): New function.
15614 (build_call_internal): Pass a fourth operand.
15615 (maybe_push_to_seq): Likewise.
15616 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
15617 Fold VEC_COND_EXPRs of an operation and a default value into
15618 an IFN_COND_* function if possible.
15619 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
15621 (SVE_COND_FP_BINARY): Include them.
15622 (optab, sve_fp_op): Handle them.
15623 (SVE_INT_BINARY_REV): New code iterator.
15624 (SVE_COND_FP_BINARY_REV): New int iterator.
15625 (commutative): New int attribute.
15626 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
15628 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
15630 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
15631 (*cond_<optab><mode>): New patterns for reversed operands.
15633 2018-05-25 Richard Biener <rguenther@suse.de>
15635 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
15636 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
15637 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
15638 (STMT_VINFO_GROUPED_ACCESS): Adjust.
15639 * tree-vect-data-refs.c (everywhere): Adjust users.
15640 * tree-vect-loop.c (everywhere): Likewise.
15641 * tree-vect-slp.c (everywhere): Likewise.
15642 * tree-vect-stmts.c (everywhere): Likewise.
15643 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
15645 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15647 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
15649 (gcc_cv_as_section_exclude): ... this.
15650 Try Solaris as #exclude syntax.
15651 * configure: Regenerate.
15652 * config.in: Regenerate.
15653 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
15655 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
15656 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
15658 * varasm.c (default_elf_asm_named_section): Don't check if
15659 HAVE_GAS_SECTION_EXCLUDE is defined.
15661 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15663 * doc/md.texi: Update the documentation of the cond_* optabs
15664 to mention the new final operand. Fix GET_MODE_NUNITS call.
15665 Describe the scalar case too.
15666 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
15667 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
15669 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
15670 (get_conditional_internal_fn): Update comment.
15671 * tree-vect-loop.c (vectorizable_reduction): Pass the original
15672 accumulator value as a final argument to conditional functions.
15673 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
15674 a define_expand and add an "else" operand. Assert for now that
15675 the else operand is equal to operand 2. Use SVE_INT_BINARY and
15676 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
15677 (*cond_<optab><mode>): New patterns.
15678 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
15679 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
15680 (UNSPEC_COND_EOR): Delete.
15681 (optab): Remove associated mappings.
15682 (SVE_INT_BINARY): New code iterator.
15683 (sve_int_op): Remove int attribute and add "minus" to the code
15685 (SVE_COND_INT_OP): Delete.
15686 (SVE_COND_FP_OP): Rename to...
15687 (SVE_COND_FP_BINARY): ...this.
15689 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15691 * optabs.c (can_reuse_operands_p): New function.
15692 (maybe_legitimize_operands): Try to reuse the results for
15695 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
15697 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
15698 Add {q} suffix to insn mnemonic.
15700 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
15702 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
15703 (msp430_warn_func_return): New.
15705 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
15707 * fold-const.c (tree_nonzero_bits): New function.
15708 * fold-const.h (tree_nonzero_bits): Likewise.
15709 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
15710 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
15712 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
15716 * varasm.c (assemble_alias): Check ifunc_resolver only on
15719 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
15722 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
15723 when memory input operand is handled.
15725 2018-05-24 Luis Machado <luis.machado@linaro.org>
15727 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
15729 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
15731 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
15733 * match.pd: Delay FMA folds until after vectorization.
15735 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
15738 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
15739 address check not strict.
15741 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
15743 * gimple-match.h (gimple_match_op): New class.
15744 (mprts_hook): Replace parameters with a gimple_match_op *.
15745 (maybe_build_generic_op): Likewise.
15746 (gimple_simplified_result_is_gimple_val): Replace parameters with
15747 a const gimple_match_op *.
15748 (gimple_simplify): Replace code_helper * and tree * parameters with
15749 a gimple_match_op * parameter.
15750 (gimple_resimplify1): Replace code_helper *, tree and tree *
15751 parameters with a gimple_match_op * parameter.
15752 (gimple_resimplify2): Likewise.
15753 (gimple_resimplify3): Likewise.
15754 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
15755 parameters with a gimple_match_op * parameter.
15756 * gimple-match-head.c (gimple_simplify): Change prototypes of
15757 auto-generated functions to take a gimple_match_op * instead of
15758 separate code_helper * and tree * parameters. Make the same
15759 change in the top-level overload and update calls to the
15760 gimple_resimplify routines. Update calls to the auto-generated
15761 functions and to maybe_push_res_to_seq in the publicly-facing
15762 operation-specific gimple_simplify overloads.
15763 (gimple_match_op::MAX_NUM_OPS): Define.
15764 (gimple_resimplify1): Replace rcode and ops with a single res_op
15765 parameter. Update call to gimple_simplify.
15766 (gimple_resimplify2): Likewise.
15767 (gimple_resimplify3): Likewise.
15768 (mprts_hook): Replace parameters with a gimple_match_op *.
15769 (maybe_build_generic_op): Likewise.
15770 (build_call_internal): Replace type, nargs and ops with
15771 a gimple_match_op *.
15772 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
15773 with a single gimple_match_op *. Update calls to mprts_hook,
15774 build_call_internal and gimple_simplified_result_is_gimple_val.
15775 Factor out code that is common to the tree_code and combined_fn cases.
15776 * genmatch.c (expr::gen_transform): Replace tem_code and
15777 tem_ops with a gimple_match_op called tem_op. Update calls
15778 to the gimple_resimplify functions and maybe_push_res_to_seq.
15779 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
15780 res_ops. Update call to the gimple_resimplify functions.
15781 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
15782 (decision_tree::gen): Make the functions take a gimple_match_op *
15783 called res_op instead of separate res_code and res_ops parameters.
15784 Update call accordingly.
15785 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
15786 and ops with a single res_op parameter. Update calls to
15787 maybe_build_generic_op and maybe_push_res_to_seq.
15788 (fold_stmt_1): Update calls to gimple_simplify and
15789 replace_stmt_with_simplification.
15790 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
15791 and gimple_simplified_result_is_gimple_val.
15792 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
15794 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
15795 with a gimple_match_op *.
15796 (vn_nary_build_or_lookup): Likewise. Update call to
15797 vn_nary_build_or_lookup_1.
15798 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
15799 gimple_match_op *. Update calls to the gimple_resimplify routines
15800 and to gimple_simplified_result_is_gimple_val.
15801 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
15802 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
15803 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
15804 (visit_nary_op): Likewise.
15805 (visit_reference_op_load): Likewise.
15807 2018-05-23 Luis Machado <luis.machado@linaro.org>
15809 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
15810 modifier for printing the step amount.
15812 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15815 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
15818 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
15820 * doc/sourcebuild.texi (Endianness): New subsubsection.
15822 2018-05-23 Luis Machado <luis.machado@linaro.org>
15824 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
15825 <prefetch_dynamic_strides>: New const bool field.
15826 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
15827 prefetch_dynamic_strides.
15828 (exynosm1_prefetch_tune): Likewise.
15829 (thunderxt88_prefetch_tune): Likewise.
15830 (thunderx_prefetch_tune): Likewise.
15831 (thunderx2t99_prefetch_tune): Likewise.
15832 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
15834 (aarch64_override_options_internal): Update to set
15835 PARAM_PREFETCH_DYNAMIC_STRIDES.
15836 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
15837 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
15838 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
15839 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
15840 prefetch-dynamic-strides setting.
15842 2018-05-23 Luis Machado <luis.machado@linaro.org>
15844 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
15845 <minimum_stride>: New const int field.
15846 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
15847 minimum_stride field defaulting to -1.
15848 (exynosm1_prefetch_tune): Likewise.
15849 (thunderxt88_prefetch_tune): Likewise.
15850 (thunderx_prefetch_tune): Likewise.
15851 (thunderx2t99_prefetch_tune): Likewise.
15852 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
15853 <default_opt_level>: Set to 3.
15854 (aarch64_override_options_internal): Update to set
15855 PARAM_PREFETCH_MINIMUM_STRIDE.
15856 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
15857 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
15858 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
15859 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
15860 stride is constant and is below the minimum stride threshold.
15862 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15864 * config/arm/arm-cpus.in (mode26): Delete.
15865 (armv4): Delete mode26 reference.
15866 * config/arm/arm.c (arm_configure_build_target): Delete use of
15869 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
15871 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
15873 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
15874 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
15876 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
15877 (floatunsdidf2): Ditto.
15879 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
15881 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
15882 (fixuns_trunc<mode>si2_avx512f): Ditto.
15883 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
15884 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
15885 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
15887 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
15889 PR rtl-optimization/79985
15890 * df-scan.c (df_insn_refs_collect): Remove special case for
15891 global registers and asm statements.
15893 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
15895 * extend.texi (Global Register Variables): Rewrite the bullet list.
15896 Note that the register is available for allocation. Note that access
15897 via inline asm must use constraints. Add note about async-signal
15898 handlers. Remove paragraph about automagic register selection.
15900 2018-05-23 Richard Biener <rguenther@suse.de>
15902 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
15903 of fixed offset from memset VN.
15905 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
15907 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
15908 first_interp field.
15909 (alloc_cand_and_find_basis): Initialize first_interp field.
15910 (slsr_process_mul): Modify first_interp field.
15911 (slsr_process_add): Likewise.
15912 (slsr_process_cast): Modify first_interp field for each new
15914 (slsr_process_copy): Likewise.
15915 (dump_candidate): Dump first_interp field.
15916 (replace_mult_candidate): Process all interpretations, not just
15918 (replace_rhs_if_not_dup): Likewise.
15919 (replace_one_candidate): Likewise.
15921 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
15923 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
15925 (aarch64_needs_frame_chain): New function.
15926 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
15928 2018-05-23 Sudakshina Das <sudi.das@arm.com>
15931 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
15932 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
15933 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
15934 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
15935 as true for strict-align.
15936 (aarch64_can_inline_p): Perform checks even when callee has no
15937 attributes to check for strict alignment.
15938 * doc/extend.texi (AArch64 Function Attributes): Document
15940 * doc/invoke.texi: (AArch64 Options): Likewise.
15942 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
15944 PR tree-optimization/85853
15945 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
15946 the handling of the root of the node to...
15947 (vect_slp_analyze_node_operations_1): ...this new function,
15948 and run the whole thing with the child nodes' def types
15949 set according to their SLP node's def type.
15951 2018-05-23 Richard Biener <rguenther@suse.de>
15953 PR middle-end/85874
15954 * tree-data-ref.c (create_runtime_alias_checks): Defer
15955 and ignore overflow warnings.
15957 2018-05-23 Yury Gribov <tetra2005@gmail.com>
15959 PR tree-optimization/85822
15960 * tree-vrp.c (is_masked_range_test): Fix handling of negative
15963 2018-05-23 Richard Biener <rguenther@suse.de>
15965 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
15966 memset constants via native_interpret_expr.
15968 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
15971 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
15973 (cgraph_node::create_alias): Likewise.
15974 (cgraph_node::get_availability): Check ifunc_resolver instead
15975 of looking up ifunc attribute.
15976 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
15977 * varasm.c (do_assemble_alias): Likewise.
15978 (assemble_alias): Likewise.
15979 (default_binds_local_p_3): Likewise.
15980 * cgraph.h (cgraph_node): Add ifunc_resolver.
15981 (cgraph_node::only_called_directly_or_aliased_p): Return false
15982 for IFUNC resolver.
15983 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
15985 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
15986 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
15987 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
15988 instead of looking up ifunc attribute.
15990 2018-05-22 Luis Machado <luis.machado@linaro.org>
15992 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
15994 2018-05-22 Martin Sebor <msebor@redhat.com>
15996 PR middle-end/85359
15997 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
15998 only when expasion succeeds.
15999 (expand_builtin_strcmp): Same.
16000 (expand_builtin_strncmp): Same.
16002 2018-05-22 Martin Sebor <msebor@redhat.com>
16004 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
16006 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
16007 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16009 * config/aarch64/aarch64-ldpstp.md: Replace uses of
16010 aarch64_mem_pair_operand with memory_operand and delete operand swapping
16012 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
16013 Add check for legitimate_address.
16014 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
16015 (aarch64_swap_ldrstr_operands): New.
16016 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
16019 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
16020 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16022 * config/aarch64/aarch64.md: New patterns to generate stp
16024 (store_pair_sw, store_pair_dw): New patterns to generate stp for
16025 single words and double words.
16026 (load_pair_sw, load_pair_dw): Likewise.
16027 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
16029 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
16031 * config/aarch64/aarch64-ldpstp.md: Modify peephole
16032 for different mode ldpstp and add peephole for merged zero stores.
16033 Likewise for loads.
16034 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
16036 (aarch64_gen_store_pair): Rename calls to match new patterns.
16037 (aarch64_gen_load_pair): Rename calls to match new patterns.
16038 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
16039 (load_pair<DREG:mode><DREG2:mode>): ... This.
16040 (store_pair<mode>): Rename to...
16041 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
16042 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
16043 New mode iterators.
16044 (V_INT_EQUIV): Handle SImode.
16045 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
16048 2018-05-22 Martin Sebor <msebor@redhat.com>
16051 * calls.c (maybe_warn_nonstring_arg): Use string length to set
16052 or ajust the presumed bound on an operation to avoid unnecessary
16055 2018-05-22 Martin Sebor <msebor@redhat.com>
16057 PR tree-optimization/85826
16058 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
16059 assuming that a DECL necesarily has a constant size.
16061 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
16063 PR middle-end/85862
16064 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
16066 2018-05-22 Richard Biener <rguenther@suse.de>
16068 PR tree-optimization/85834
16069 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
16070 non-constant and non-zero memset arguments.
16072 2018-05-22 Martin Liska <mliska@suse.cz>
16075 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
16077 2018-05-22 Richard Biener <rguenther@suse.de>
16079 PR tree-optimization/85863
16080 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
16081 comparisons when vectype is specified.
16082 (vectorizable_condition): Do not specify vectype for
16083 vect_is_simple_cond when SLP vectorizing.
16085 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
16088 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
16089 define __ibm128 as long double.
16090 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
16091 as a distinct type when IEEE 128-bit support is enabled.
16092 (init_float128_ieee): Fix up conversions between IFmode and IEEE
16093 128-bit types to use the correct functions.
16094 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
16095 convert between 128-bit floating point types that have different
16096 modes but the same representation, instead of using gen_lowpart to
16098 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
16100 (IFKF_reg): New attributes to give the register constraints for
16102 (extend<mode>tf2_internal): New insns to mark an explicit
16103 conversion between 128-bit floating point types that have a
16104 different mode but share the same representation.
16106 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
16108 PR tree-optimization/85814
16109 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
16110 a null return from get_strinfo when unsharing the next
16111 strinfo in the chain.
16113 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
16116 * varasm.c (weak_finish): Clean up weak_decls.
16118 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16120 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
16121 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
16122 UNSPEC_UADALP values.
16123 * config/aarch64/iterators.md (ABAL): New int iterator.
16126 (sur): Add mappings for the above.
16127 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
16129 (aarch64_<sur>abal<mode>_4): Likewise.
16130 (aarch64_<sur>adalp<mode>_3): Likewise.
16131 (<sur>sadv16qi): New define_expand.
16133 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
16135 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
16136 (*movdf_internal): Ditto.
16137 (*rcpsf2_sse): Ditto.
16138 (*rsqrtsf2_sse): Ditto.
16139 (*sqrt<mode>2_sse): Ditto.
16141 2018-05-21 Tamar Christina <tamar.christina@arm.com>
16143 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
16145 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
16146 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
16147 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
16148 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
16150 * config/aarch64/arm_neon.h: Likewise.
16151 * config/aarch64/iterators.md (VQ_I): New.
16153 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
16155 * config.gcc: Add arc/t-multilib-linux to tmake_file for
16157 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
16160 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
16162 * config/nds32/constraints.md (S): New constraint.
16163 * config/nds32/nds32.md (call_internal): Use constraint S.
16164 (call_value_internal): Likewise.
16165 (sibcall_internal): Likewise.
16166 (sibcall_value_internal): Likewise.
16168 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
16169 Chung-Ju Wu <jasonwucj@gmail.com>
16171 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
16172 into consideration.
16174 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
16175 Chung-Ju Wu <jasonwucj@gmail.com>
16177 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
16178 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
16179 (nds32_rtx_costs_impl): Simplify.
16180 (nds32_address_cost_impl): Simplify.
16181 (nds32_init_rtx_costs): New function.
16182 (nds32_rtx_costs_speed_prefer): Likewise.
16183 (nds32_rtx_costs_size_prefer): Likewise.
16184 (nds32_address_cost_speed_prefer): Likewise.
16185 (nds32_address_cost_speed_fwprop): Likewise.
16186 (nds32_address_cost_size_prefer): Likewise.
16187 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
16188 * config/nds32/nds32.c (nds32_option_override): Use
16189 nds32_init_rtx_costs function.
16191 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
16193 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
16194 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
16195 (TARGET_PIPELINE_N8): Likewise.
16196 (TARGET_PIPELINE_N10): Likewise.
16197 (TARGET_PIPELINE_N13): Likewise.
16198 (TARGET_PIPELINE_GRAYWOLF): Likewise.
16200 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
16202 * config/nds32/nds32-fpu.md: Update copyright year.
16204 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16206 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
16208 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16210 * config/nds32/nds32.c
16211 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
16212 * config/nds32/nds32.opt (minline-asm-r15): New option.
16214 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16216 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
16218 * config/nds32/nds32.md (abssi2): New pattern.
16220 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
16222 * config/i386/i386.md (rex64namesuffix): New mode attribute.
16223 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
16224 Merge insn pattern from sse_cvtsi2ss<round_name> and
16225 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
16226 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
16227 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
16228 using SWI48 mode iterator.
16229 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
16230 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
16231 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
16232 pattern from sse_cvttss2si<round_saeonly_name>
16233 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
16234 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
16235 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
16236 using SWI48 mode iterator.
16237 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
16238 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
16239 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
16240 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
16241 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
16242 using SWI48 mode iterator.
16243 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
16244 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
16245 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
16246 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
16247 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
16248 SWI48 mode iterator.
16249 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
16250 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
16251 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
16252 pattern from sse_cvttsd2si<round_saeonly_name>
16253 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
16255 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16257 * config/nds32/nds32-md-auxiliary.c
16258 (nds32_valid_smw_lwm_base_p): Refine.
16259 (nds32_output_smw_single_word): Refine.
16260 (nds32_output_smw_double_word): New.
16261 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
16263 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16265 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
16266 (nds32_output_stack_pop): Refine.
16267 (nds32_expand_unaligned_load): Refine.
16268 (nds32_expand_unaligned_store): Refine.
16270 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16271 Chung-Ju Wu <jasonwucj@gmail.com>
16273 * config/nds32/constants.md: Add TP_REGNUM constant.
16274 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
16275 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
16277 * config/nds32/nds32-doubleword.md: Consider flag_pic.
16278 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
16279 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
16280 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
16281 and PIC code generation.
16282 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
16284 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
16286 * config/nds32/nds32.md: Support TLS and PIC.
16287 * config/nds32/nds32.c: Support TLS and PIC.
16288 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
16289 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
16292 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16294 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
16295 mode with E_ prefix.
16297 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16298 Chung-Ju Wu <jasonwucj@gmail.com>
16300 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
16301 * config/nds32/nds32-md-auxiliary.c
16302 (symbolic_reference_mentioned_p): New.
16303 (nds32_legitimize_ict_address): New.
16304 (nds32_expand_ict_move): New.
16305 (nds32_indirect_call_referenced_p): New.
16306 (nds32_symbol_binds_local_p): Delete.
16307 (nds32_long_call_p): Modify.
16308 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
16309 * config/nds32/nds32-protos.h
16310 (symbolic_reference_mentioned_p): Declare.
16311 (nds32_legitimize_ict_address): Declare.
16312 (nds32_expand_ict_move): Declare.
16313 (nds32_indirect_call_referenced_p): Declare.
16314 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
16315 (nds32_relax_group): Use nds32_ict_const_p as condition.
16316 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
16317 (nds32_asm_file_start): Output ict_model directive in asm code.
16318 (nds32_legitimate_address_p): Consider indirect call.
16319 (nds32_print_operand): Consider indirect call.
16320 (nds32_print_operand_address): Consider indirect call.
16321 (nds32_insert_attributes): Handle "indirect_call" attribute.
16322 (TARGET_LEGITIMATE_ADDRESS_P): Define.
16323 (TARGET_LEGITIMATE_CONSTANT_P): Define.
16324 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
16325 (TARGET_DELEGITIMIZE_ADDRESS): Define.
16326 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
16327 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
16328 (TARGET_ICT_MODEL_SMALL): Define.
16329 (TARGET_ICT_MODEL_LARGE): Define.
16330 * config/nds32/nds32.md (movsi): Consider ict model.
16331 (call, call_value): Consider ict model.
16332 (sibcall, sibcall_value): Consider ict model.
16333 * config/nds32/nds32.opt (mict-model): New option.
16334 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
16337 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
16338 Monk Chiang <sh.chiang04@gmail.com>
16339 Jim Wilson <jimw@sifive.com>
16341 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
16342 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
16343 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
16344 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
16345 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
16346 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
16347 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
16348 compute save_libcall_adjustment properly.
16349 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
16350 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
16351 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
16352 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
16353 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
16354 (ABI_SPEC): Handle mabi=ilp32e.
16355 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
16356 (RVE): Add RVE mask.
16357 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
16358 <-march>: Add rv32e as an example.
16360 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
16363 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
16364 (intra_create_variable_infos): Handle C++ constructors.
16366 2018-05-18 Martin Liska <mliska@suse.cz>
16368 * passes.def: Remove a redundant pass.
16370 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
16373 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
16375 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16377 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
16379 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
16381 (armv2, armv2a, armv3, armv3m): Delete architectures.
16382 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
16383 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
16384 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
16386 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
16387 (*mulsidi3adddi): Likewise.
16388 (mulsidi3): Likewise.
16389 (*mulsidi3_nov6): Likewise.
16390 (umulsidi3): Likewise.
16391 (umulsidi3_nov6): Likewise.
16392 (umaddsidi4): Likewise.
16393 (*umulsidi3adddi): Likewise.
16394 (smulsi3_highpart): Likewise.
16395 (*smulsi3_highpart_nov6): Likewise.
16396 (umulsi3_highpart): Likewise.
16397 (*umulsi3_highpart_nov6): Likewise.
16398 * config/arm/arm.h (arm_arch3m): Delete.
16399 * config/arm/arm.c (arm_arch3m): Delete.
16400 (arm_option_override_internal): Update armv3-related comment.
16401 (arm_configure_build_target): Delete use of isa_bit_mode32.
16402 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
16403 (arm_rtx_costs_internal): Delete check of arm_arch3m.
16404 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
16405 (mulsa3): Likewise.
16406 (mulusa3): Likewise.
16407 * config/arm/arm-protos.h (arm_arch3m): Delete.
16408 * config/arm/arm-tables.opt: Regenerate.
16409 * config/arm/arm-tune.md: Likewise.
16410 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
16411 deleted architectures.
16413 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16415 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
16416 (armv5t, armv5te): New features.
16417 (ARMv5, ARMv5e): Delete fgroups.
16418 (ARMv5t, ARMv5te): Adjust for above changes.
16419 (ARMv6m): Likewise.
16420 (armv5, armv5e): Delete arches.
16421 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
16423 (*call_reg_arm): Likewise.
16424 (*call_value_reg_armv5): Likewise.
16425 (*call_value_reg_arm): Likewise.
16426 (*call_symbol): Likewise.
16427 (*call_value_symbol): Likewise.
16428 (*sibcall_insn): Likewise.
16429 (*sibcall_value_insn): Likewise.
16430 (clzsi2): Likewise.
16431 (prefetch): Likewise.
16432 (define_split and define_peephole2 dependent on arm_arch5):
16434 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
16436 (TARGET_ARM_QBIT): Likewise.
16437 (TARGET_DSP_MULTIPLY): Likewise.
16438 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
16439 (arm_arch5, arm_arch5e): Delete.
16440 (arm_arch5t, arm_arch5te): Declare.
16441 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
16442 (arm_arch5t): Declare.
16443 (arm_option_reconfigure_globals): Update for the above.
16444 (arm_options_perform_arch_sanity_checks): Update comment, replace
16445 use of arm_arch5 with arm_arch5t.
16446 (use_return_insn): Likewise.
16447 (arm_emit_call_insn): Likewise.
16448 (output_return_instruction): Likewise.
16449 (arm_final_prescan_insn): Likewise.
16450 (arm_coproc_builtin_available): Likewise.
16451 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
16452 arm_arch5e with arm_arch5t and arm_arch5te.
16453 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
16454 (arm_arch5t, arm_arch5te): Declare.
16455 * config/arm/arm-tables.opt: Regenerate.
16456 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
16457 * config/arm/t-multilib: Likewise.
16458 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
16459 instead of arm_arch5.
16460 (*call_reg_thumb1): Likewise.
16461 (*call_value_reg_thumb1_v5): Likewise.
16462 (*call_value_reg_thumb1): Likewise.
16463 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
16465 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
16467 2018-05-18 Martin Liska <mliska@suse.cz>
16469 PR gcov-profile/84846
16470 * doc/gcov.texi: Document -t option of gcov tool.
16472 2018-05-18 Martin Liska <mliska@suse.cz>
16474 PR gcov-profile/84846
16475 * gcov.c (print_usage): Add new -t option.
16476 (process_args): Handle the option.
16477 (generate_results): Use stdout as output when requested by
16480 2018-05-18 Martin Liska <mliska@suse.cz>
16482 PR gcov-profile/84846
16483 * coverage.c (coverage_init): Write PWD to .gcno file.
16484 * doc/gcov.texi: Document how working directory is printed.
16485 * gcov-dump.c (dump_gcov_file): Print PWD.
16486 * gcov.c (output_intermediate_file): Likewise.
16487 (read_graph_file): Read PWD string.
16488 (output_lines): Print PWD.
16490 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16492 PR middle-end/85817
16493 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
16494 for retval and return false if all args to phi are zero.
16496 2018-05-18 Richard Biener <rguenther@suse.de>
16498 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
16500 (evrp_dom_walker::before_dom_children): Call it.
16502 2018-05-18 Richard Biener <rguenther@suse.de>
16504 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
16505 results when processing array refs with variable index.
16507 2018-05-18 Toon Moene <toon@moene.org>
16509 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
16510 directly after that of -floop-interchange. Indicate that both
16511 options are enabled by default when specifying -O3.
16513 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16515 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
16516 iterator. Delete separate integer-mode vec_set<mode> expander.
16517 (aarch64_simd_vec_setv2di): Delete.
16518 (vec_setv2di): Delete.
16519 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
16520 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
16521 the "w, r" alternative.
16523 2018-05-18 Martin Liska <mliska@suse.cz>
16525 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
16526 * tree-pass.h (make_pass_lower_switch_O0): New function.
16527 * tree-switch-conversion.c (node_has_low_bound): Remove.
16528 (node_has_high_bound): Likewise.
16529 (node_is_bounded): Likewise.
16530 (class pass_lower_switch): Make it a template type and create
16532 (pass_lower_switch::execute): Add template argument.
16533 (make_pass_lower_switch): New function.
16534 (make_pass_lower_switch_O0): New function.
16535 (do_jump_if_equal): Remove.
16536 (emit_case_nodes): Simplify to just handle all 3 cases and leave
16537 all the hard work to tree optimization passes.
16539 2018-05-18 Martin Liska <mliska@suse.cz>
16541 * dbgcnt.c (limit_low): Renamed from limit.
16542 (limit_high): New variable.
16543 (dbg_cnt_is_enabled): Check for upper limit.
16544 (dbg_cnt): Adjust dumping.
16545 (dbg_cnt_set_limit_by_index): Add new argument for high
16547 (dbg_cnt_set_limit_by_name): Likewise.
16548 (dbg_cnt_process_single_pair): Parse new format.
16549 (dbg_cnt_process_opt): Use strtok.
16550 (dbg_cnt_list_all_counters): Remove 'value' and add
16552 * doc/invoke.texi: Document changes.
16554 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
16556 * doc/sourcebuild.texi (scalar_all_fma): Document.
16557 * tree.def (FMA_EXPR): Delete.
16558 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
16559 * internal-fn.c (ternary_direct): New macro.
16560 (expand_ternary_optab_fn): Likewise.
16561 (direct_ternary_optab_supported_p): Likewise.
16562 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
16563 * builtins.c (fold_builtin_fma): Delete.
16564 (fold_builtin_3): Don't call it.
16565 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
16566 * expr.c (expand_expr_real_2): Likewise.
16567 * fold-const.c (operand_equal_p): Likewise.
16568 (fold_ternary_loc): Likewise.
16569 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
16570 * gimple.c (DEFTREECODE): Likewise.
16571 * gimplify.c (gimplify_expr): Likewise.
16572 * optabs-tree.c (optab_for_tree_code): Likewise.
16573 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16574 * tree-eh.c (operation_could_trap_p): Likewise.
16575 (stmt_could_throw_1_p): Likewise.
16576 * tree-inline.c (estimate_operator_cost): Likewise.
16577 * tree-pretty-print.c (dump_generic_node): Likewise.
16578 (op_code_prio): Likewise.
16579 * tree-ssa-loop-im.c (stmt_cost): Likewise.
16580 * tree-ssa-operands.c (get_expr_operands): Likewise.
16581 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
16582 * fold-const-call.h (fold_fma): Delete.
16583 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
16584 CFN_FNMA and CFN_FNMS.
16585 (fold_fma): Delete.
16586 * genmatch.c (combined_fn): New enum.
16587 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
16588 (commutative_op): New function.
16589 (commutate): Use it. Handle more than 2 operands.
16590 (dt_operand::gen_gimple_expr): Use commutative_op.
16591 (parser::parse_expr): Allow :c to be used with non-binary
16592 operators if the commutative operand is known.
16593 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
16594 CFN_FMS, CFN_FNMA and CFN_FNMS.
16595 (backprop::process_assign_use): Remove FMA_EXPR handling.
16596 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
16597 (gen_hsa_fma): New function.
16598 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
16599 IFN_FNMA and IFN_FNMS.
16600 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
16601 * gimple-fold.h (follow_all_ssa_edges): Declare.
16602 * gimple-fold.c (follow_all_ssa_edges): New function.
16603 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
16604 gimple_build interface and use follow_all_ssa_edges to fold the result.
16605 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
16606 instead of checking for optabs directly.
16607 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
16608 rather than FMA_EXPRs.
16609 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
16610 call to IFN_FMA instead of an FMA_EXPR.
16612 2018-05-17 Jim Wilson <jimw@sifive.com>
16614 * expr.c (do_tablejump): When converting index to Pmode, if we have a
16615 sign extended promoted subreg, and the range does not have the sign bit
16616 set, then do a sign extend.
16618 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
16619 test, check for sign extended subreg and/or constant operands, and
16620 do a sign extend in that case.
16622 2018-05-17 Steve Ellcey <sellcey@cavium.com>
16624 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
16625 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
16627 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
16628 Change logics_shift_reg to logics_shift_imm.
16629 (thunderx2t99_fp_loadpair_basic): Delete.
16630 (thunderx2t99_fp_storepair_basic): Delete.
16631 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
16632 (thunderx2t99_asimd_polynomial): Delete.
16633 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
16634 and neon_fp_mul_d_scalar_q.
16635 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
16636 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
16637 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
16638 (thunderx2t99_asimd_lut): Add missing tbl types.
16639 (thunderx2t99_asimd_ext): Delete.
16640 (thunderx2t99_asimd_load1_1_mult): Delete.
16641 (thunderx2t99_asimd_load1_2_mult): Delete.
16642 (thunderx2t99_asimd_load1_ldp): New.
16643 (thunderx2t99_asimd_load1): New.
16644 (thunderx2t99_asimd_load2): Add missing *load2* types.
16645 (thunderx2t99_asimd_load3): New.
16646 (thunderx2t99_asimd_load4): New.
16647 (thunderx2t99_asimd_store1_1_mult): Delete.
16648 (thunderx2t99_asimd_store1_2_mult): Delete.
16649 (thunderx2t99_asimd_store2_mult): Delete.
16650 (thunderx2t99_asimd_store2_onelane): Delete.
16651 (thunderx2t99_asimd_store_stp): New.
16652 (thunderx2t99_asimd_store1): New.
16653 (thunderx2t99_asimd_store2): New.
16654 (thunderx2t99_asimd_store3): New.
16655 (thunderx2t99_asimd_store4): New.
16657 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
16659 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
16660 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
16662 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
16663 Segher Boessenkool <segher@kernel.crashing.org>
16666 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
16669 2018-05-17 Richard Biener <rguenther@suse.de>
16671 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
16672 for pruning loop and prune defs feeding only already visited PHIs.
16674 2018-05-17 Richard Biener <rguenther@suse.de>
16676 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
16678 2018-05-17 Bin Cheng <bin.cheng@arm.com>
16679 Richard Biener <rguenther@suse.de>
16681 PR tree-optimization/85793
16682 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
16683 for VMAT_ELEMENTWISE.
16685 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
16687 * internal-fn.h (lookup_internal_fn): Declare
16688 * internal-fn.c (lookup_internal_fn): New function.
16689 * gimple.c (gimple_build_call_from_tree): Handle calls to
16690 internal functions.
16691 * gimple-pretty-print.c (dump_gimple_call): Print "." before
16692 internal function names.
16693 * tree-pretty-print.c (dump_generic_node): Likewise.
16694 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
16696 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
16698 * gimple-fold.h (gimple_build): Make the function forms take
16699 combined_fn rather than built_in_function.
16700 (gimple_simplify): Likewise.
16701 * gimple-match-head.c (gimple_simplify): Likewise.
16702 * gimple-fold.c (gimple_build): Likewise.
16703 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
16704 rather than gimple_build_call_internal.
16705 (get_initial_defs_for_reduction): Likewise.
16706 (vect_create_epilog_for_reduction): Likewise.
16707 (vectorizable_live_operation): Likewise.
16709 2018-05-17 Martin Liska <mliska@suse.cz>
16711 * gimple-ssa-sprintf.c (format_directive): Do not use
16712 space in between 'G_' and '('.
16714 2018-05-17 Jakub Jelinek <jakub@redhat.com>
16717 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
16718 even if the mask is not all ones.
16721 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
16723 (ix86_gimple_fold_builtin): Likewise.
16726 * config/i386/i386.c: Include tree-vector-builder.h.
16727 (ix86_vector_shift_count): New function.
16728 (ix86_fold_builtin): Fold shift builtins by scalar count.
16729 (ix86_gimple_fold_builtin): Likewise.
16731 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
16732 _mm512_setzero): New intrinsics.
16734 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
16735 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16737 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
16738 code generation for cases where splatting a value is not useful.
16739 * simplify-rtx.c (simplify_ternary_operation): Simplify
16740 vec_merge across a vec_duplicate and a paradoxical subreg forming
16741 a vector mode to a vec_concat.
16743 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
16745 * config.gcc: Support "goldmont-plus".
16746 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
16748 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
16749 PROCESSOR_GOLDMONT_PLUS.
16750 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
16751 (processor_target_table): Add "goldmont-plus".
16752 (PTA_GOLDMONT_PLUS): Define.
16753 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
16754 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
16755 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
16756 (fold_builtin_cpu): Add "goldmont-plus".
16757 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
16758 (ix86_option_override_internal): Add "goldmont-plus".
16759 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
16760 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
16761 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
16762 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
16764 2018-05-17 Richard Biener <rguenther@suse.de>
16766 PR tree-optimization/85757
16767 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
16768 remove defs that only feed that PHI from further processing.
16770 2018-05-16 Jim Wilson <jimw@sifive.com>
16772 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
16774 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
16776 2018-05-16 Mark Wielaard <mark@klomp.org>
16778 * dwarf2out.c (count_index_strings): New function.
16779 (output_indirect_strings): Call count_index_strings and generate
16780 header for dwarf_version >= 5.
16782 2018-05-16 Mark Wielaard <mark@klomp.org>
16784 * dwarf2out.c (dwarf_FORM): New function.
16785 (set_indirect_string): Use dwarf_FORM.
16786 (reset_indirect_string): Likewise.
16787 (size_of_die): Likewise.
16788 (value_format): Likewise.
16789 (output_die): Likewise.
16790 (add_skeleton_AT_string): Likewise.
16791 (output_macinfo_op): Likewise.
16792 (index_string): Likewise.
16793 (output_index_string_offset): Likewise.
16794 (output_index_string): Likewise.
16795 (count_index_strings): Likewise.
16797 2018-05-16 Carl Love <cel@us.ibm.com>
16799 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
16800 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
16802 2018-05-16 Martin Jambor <mjambor@suse.cz>
16804 * ipa-prop.c (ipa_free_all_edge_args): Remove.
16805 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
16807 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
16809 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
16810 (fnma<mode>4): Likewise.
16811 (fms<mode>4): Likewise.
16812 (fnms<mode>4): Likewise.
16813 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
16814 (aarch64_fnma<mode>4): Likewise.
16815 (aarch64_fms<mode>4): Likewise.
16816 (aarch64_fnms<mode>4): Likewise.
16817 (aarch64_fnmadd<mode>4): Likewise.
16819 2018-05-16 Jason Merrill <jason@redhat.com>
16821 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
16823 2018-05-16 Richard Biener <rguenther@suse.de>
16825 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
16826 (dump_stmt_cost): Declare.
16827 (add_stmt_cost): Dump cost we add.
16828 (add_stmt_costs): New function.
16829 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
16830 No longer exported.
16831 (vect_analyze_stmt): Adjust prototype.
16832 (vectorizable_condition): Likewise.
16833 (vectorizable_live_operation): Likewise.
16834 (vectorizable_reduction): Likewise.
16835 (vectorizable_induction): Likewise.
16836 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
16837 cost vector to pass to vectorizable_ and record afterwards.
16838 (vect_model_reduction_cost): Take cost vector argument and adjust.
16839 (vect_model_induction_cost): Likewise.
16840 (vectorizable_reduction): Likewise.
16841 (vectorizable_induction): Likewise.
16842 (vectorizable_live_operation): Likewise.
16843 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
16844 SLP_TREE_NUMBER_OF_VEC_STMTS.
16845 (vect_analyze_slp_cost_1): Remove.
16846 (vect_analyze_slp_cost): Likewise.
16847 (vect_slp_analyze_node_operations): Take visited args and
16848 a target cost vector. Avoid processing already visited stmt sets.
16849 (vect_slp_analyze_operations): Use a local cost vector to gather
16850 costs and register those of non-discarded instances.
16851 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
16852 (vect_schedule_slp_instance): Remove copying of
16853 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
16855 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
16856 adding cost. Record cost entry location.
16857 (vect_prologue_cost_for_slp_op): Function to compute cost of
16858 a constant or invariant generated for SLP vect in the prologue,
16859 split out from vect_analyze_slp_cost_1.
16860 (vect_model_simple_cost): Make static. Adjust for SLP costing.
16861 (vect_model_promotion_demotion_cost): Likewise.
16862 (vect_model_store_cost): Likewise, make static.
16863 (vect_model_load_cost): Likewise.
16864 (vectorizable_bswap): Add cost vector arg and adjust.
16865 (vectorizable_call): Likewise.
16866 (vectorizable_simd_clone_call): Likewise.
16867 (vectorizable_conversion): Likewise.
16868 (vectorizable_assignment): Likewise.
16869 (vectorizable_shift): Likewise.
16870 (vectorizable_operation): Likewise.
16871 (vectorizable_store): Likewise.
16872 (vectorizable_load): Likewise.
16873 (vectorizable_condition): Likewise.
16874 (vectorizable_comparison): Likewise.
16875 (can_vectorize_live_stmts): Likewise.
16876 (vect_analyze_stmt): Likewise.
16877 (vect_transform_stmt): Adjust calls to vectorizable_*.
16878 * tree-vectorizer.c: Include gimple-pretty-print.h.
16879 (dump_stmt_cost): New function.
16881 2018-05-16 Richard Biener <rguenther@suse.de>
16883 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
16884 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
16885 * tree-ssa-dse.c: Include tree-ssa-loop.h.
16886 (check_name): New callback.
16887 (dse_classify_store): Track cycles via a visited bitmap of PHI
16888 defs and simplify handling of in-loop and across loop dead stores
16889 and properly fail for loop-variant refs. Handle byte-tracking with
16890 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
16893 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
16895 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
16896 (vect_get_mask_type_for_stmt): Likewise.
16897 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
16899 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
16900 to determine the statement's vector type and the vector type that
16901 should be used for calculating nunits. Deal with cases in which
16902 the type has to be deferred.
16903 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
16904 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
16905 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
16906 (vect_determine_vf_for_stmt): New functions, split out from...
16907 (vect_determine_vectorization_factor): ...here.
16908 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
16909 (vect_get_mask_type_for_stmt): New functions, split out from
16910 vect_determine_vectorization_factor.
16912 2018-05-16 Richard Biener <rguenther@suse.de>
16914 * tree-cfg.c (verify_gimple_assign_ternary): Properly
16915 verify the [VEC_]COND_EXPR embedded comparison.
16917 2018-05-15 Martin Sebor <msebor@redhat.com>
16919 PR tree-optimization/85753
16920 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
16921 RECORD_TYPE in addition to ARRAY_TYPE.
16923 2018-05-15 Martin Sebor <msebor@redhat.com>
16925 PR middle-end/85643
16926 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
16928 2018-05-15 Richard Biener <rguenther@suse.de>
16930 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
16931 add by_clobber_p one. Change algorithm to collect all defs
16932 representing uses we need to walk and try reducing them to
16933 a single one before failing.
16934 (dse_dom_walker::dse_optimize_stmt): Adjust.
16936 2018-05-13 Mark Wielaard <mark@klomp.org>
16938 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
16939 (size_of_loc_descr): Likewise.
16940 (output_loc_operands): Likewise.
16941 (output_loc_operands_raw): Likewise.
16942 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
16943 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
16944 (hash_loc_operands): Likewise.
16945 (compare_loc_operands): Likewise.
16947 2018-05-14 Mark Wielaard <mark@klomp.org>
16949 * dwarf2out.c (count_index_addrs): New function.
16950 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
16952 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16954 PR tree-optimization/83648
16955 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
16956 return value as malloc candidate.
16958 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16961 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
16962 param as true in call to suggest_attribute.
16964 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
16966 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
16967 -mreadonly-in-sdata.
16969 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16971 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
16973 (aarch64_crypto_aesd_fused): Likewise.
16975 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
16977 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
16978 (movsi_aarch64): Likewise.
16979 (load_pairsi): Likewise.
16980 (load_pairdi): Likewise.
16981 (store_pairsi): Likewise.
16982 (store_pairdi): Likewise.
16983 (load_pairsf): Likewise.
16984 (load_pairdf): Likewise.
16985 (store_pairsf): Likewise.
16986 (store_pairdf): Likewise.
16987 (zero_extend): Likewise.
16988 (trunc): Swap alternatives.
16989 (fcvt_target): Add '?' to prefer w over r.
16991 2018-05-14 Jakub Jelinek <jakub@redhat.com>
16994 * config/i386/i386.md: Disallow non-commutative arithmetics in
16995 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
16996 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
16997 in the peephole2 before it.
16999 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
17001 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
17002 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
17003 (ix86_handle_option): Handle -mcldemote.
17004 * config.gcc: New header.
17005 * config/i386/cldemoteintrin.h: New file.
17006 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
17007 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
17009 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17010 OPTION_MASK_ISA_CLDEMOTE.
17011 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
17012 (ix86_valid_target_attribute_inner_p): Ditto.
17013 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
17014 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
17015 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
17016 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
17017 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
17019 * config/i386/i386.opt: Add -mcldemote.
17020 * config/i386/x86intrin.h: New header.
17021 * doc/invoke.texi: Add -mcldemote.
17023 2018-05-14 Richard Biener <rguenther@suse.de>
17025 * doc/match-and-simplify.texi: Adjust :s documentation.
17027 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
17029 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
17030 intended memcpy size.
17031 (REORDER_45): Likewise.
17033 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
17035 * sort.cc: New file.
17036 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
17037 * vec.c (qsort_chk): Use gcc_qsort.
17038 * Makefile.in (OBJS-libcommon): Add sort.o.
17039 (build/sort.o): New target. Use it...
17040 (BUILD_RTL): ... here, and...
17041 (build/gencfn-macros): ... here, and...
17042 (build/genmatch): ... here.
17044 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
17045 Chung-Ju Wu <jasonwucj@gmail.com>
17047 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
17048 * config/nds32/nds32-graywolf.md: New file.
17049 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
17050 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
17052 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
17053 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
17054 * config/nds32/nds32.md (pipeline_model): Add graywolf.
17055 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
17056 * config/nds32/pipelines.md: Include n15 settings.
17058 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
17059 Chung-Ju Wu <jasonwucj@gmail.com>
17061 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
17062 * config/nds32/nds32-n13.md: New file.
17063 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
17064 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
17066 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
17067 * config/nds32/nds32.md (pipeline_model): Add n13.
17068 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
17069 * config/nds32/pipelines.md: Include n13 settings.
17071 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
17072 Chung-Ju Wu <jasonwucj@gmail.com>
17074 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
17075 * config/nds32/nds32-n10.md: New file.
17076 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
17077 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
17079 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
17080 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
17081 * config/nds32/nds32.md (pipeline_model): Add n10.
17082 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
17083 * config/nds32/pipelines.md: Include n10 settings.
17085 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
17086 Kito Cheng <kito.cheng@gmail.com>
17087 Chung-Ju Wu <jasonwucj@gmail.com>
17089 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
17090 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
17091 Add enum values for DSP extension instructions.
17092 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
17094 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
17095 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
17096 New code iterators.
17097 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
17098 * config/nds32/nds32-dspext.md: New file for DSP implementation.
17099 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
17100 * config/nds32/nds32-intrinsic.md: Likewise.
17101 * config/nds32/nds32_intrinsic.h: Likewise.
17102 * config/nds32/nds32-md-auxiliary.c: Likewise.
17103 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
17104 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
17105 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
17106 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
17107 * config/nds32/nds32-protos.h: New declarations for DSP extension.
17108 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
17109 TYPE_DMAC in switch statement.
17110 * config/nds32/nds32.c: New checking and implementation for DSP
17111 extension instructions.
17112 * config/nds32/nds32.h: Likewise.
17113 * config/nds32/nds32.md: Likewise.
17114 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
17115 * config/nds32/predicates.md: Implement new predicates for DSP
17118 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17120 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
17121 Reformat alternatives and attributes so it is easier to identify
17122 which constraints/attributes go with which instruction.
17123 (mov<mode>_hardfloat32, FMOVE64): Likewise.
17124 (mov<mode>_softfloat32, FMOVE64): Likewise.
17125 (mov<mode>_hardfloat64, FMOVE64): Likewise.
17126 (mov<mode>_softfloat64, FMOVE64): Likewise.
17128 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
17130 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17132 (Basic PowerPC Built-in Functions): The new name of the
17133 subsection previously known as "PowerPC Built-in Functions".
17134 (Basic PowerPC Built-in Functions Available on all Configurations):
17136 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
17137 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
17138 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
17139 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
17141 2018-05-11 Martin Jambor <mjambor@suse.cz>
17144 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
17147 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
17150 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
17152 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
17154 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
17155 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
17156 (ix86_handle_option): Handle -mwaitpkg.
17157 * config.gcc: New header.
17158 * config/i386/cpuid.h (bit_WAITPKG): New bit.
17159 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
17160 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
17162 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17163 OPTION_MASK_ISA_WAITPKG.
17164 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
17165 (ix86_option_override_internal): Add PTA_WAITPKG.
17166 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
17167 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
17168 IX86_BUILTIN_TPAUSE.
17169 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
17170 __builtin_ia32_umwait and __builtin_ia32_tpause.
17171 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
17172 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
17173 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
17174 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
17175 UNSPECV_TPAUSE): New.
17176 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
17177 * config/i386/i386.opt: Add -mwaitpkg.
17178 * config/i386/waitpkgintrin.h: New file.
17179 * config/i386/x86intrin.h: New header.
17180 * doc/invoke.texi: Add -mwaitpkg.
17182 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
17185 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
17187 (cortex-m0): Use armv6s-m isa.
17188 (cortex-m0plus): Likewise.
17189 (cortex-m1): Likewise.
17190 (cortex-m0.small-multiply): Likewise.
17191 (cortex-m0plus.small-multiply): Likewise.
17192 (cortex-m1.small-multiply): Likewise.
17194 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
17195 Jakub Jelinek <jakub@redhat.com>
17197 PR tree-optimization/85692
17198 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
17199 source permute as well.
17201 2018-05-11 Martin Liska <mliska@suse.cz>
17204 * doc/extend.texi: Document LLVM style format for no_sanitize
17207 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
17209 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
17210 mode_supports_vsx_dform_quad to mode_supports_dq_form.
17211 (mode_supports_vsx_dform_quad): Likewise.
17212 (mode_supports_vmx_dform): Move these functions to be next to the
17213 other mode_supports functions.
17214 (mode_supports_dq_form): Likewise.
17215 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
17216 mode_supports_dq_form.
17217 (reg_offset_addressing_ok_p): Likewise.
17218 (offsettable_ok_by_alignment): Likewise.
17219 (rs6000_legitimate_offset_address_p): Likewise.
17220 (legitimate_lo_sum_address_p): Likewise.
17221 (rs6000_legitimize_address): Likewise.
17222 (rs6000_legitimize_reload_address): Likewise.
17223 (rs6000_secondary_reload_inner): Likewise.
17224 (rs6000_preferred_reload_class): Likewise.
17225 (rs6000_output_move_128bit): Likewise.
17227 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
17229 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
17230 Generate SImode target register for null target.
17231 <case IX86_BUILTIN_XGETBV>: Ditto.
17232 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
17233 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
17235 2018-05-10 Carl Love <cel@us.ibm.com>
17237 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
17238 dcbtt and dcbtstt if operands[2] is 0.
17240 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
17243 * config/i386/sse.md (usadv64qi): New expander.
17245 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
17247 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
17248 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
17249 -maltivec=be support.
17250 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
17251 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
17252 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17253 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
17254 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
17255 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
17256 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
17257 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
17258 altivec_vsumsws): Adjust.
17259 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
17260 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
17261 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
17263 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
17264 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
17265 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
17266 (altivec_lve<VI_char>x): Delete expand.
17267 (*altivec_lve<VI_char>x_internal): Rename to...
17268 (altivec_lve<VI_char>x): ... this.
17269 (altivec_lvxl_<mode>): Delete expand.
17270 (*altivec_lvxl_<mode>_internal): Rename to ...
17271 (altivec_lvxl_<mode>): ... this.
17272 (altivec_stvxl_<mode>): Delete expand.
17273 (*altivec_stvxl_<mode>_internal): Rename to ...
17274 (altivec_stvxl_<mode>): ... this.
17275 (altivec_stve<VI_char>x): Delete expand.
17276 (*altivec_stve<VI_char>x_internal): Rename to ...
17277 (altivec_stve<VI_char>x): ... this.
17278 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
17279 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
17280 reduc_plus_scal_<mode>): Adjust.
17281 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
17283 (rs6000_cpu_cpp_builtins): Adjust.
17284 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
17285 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
17286 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
17287 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
17288 -maltivec=be support.
17289 (rs6000_split_vec_extract_var): Adjust.
17290 (rs6000_split_v4si_init): Adjust.
17291 (swap_selector_for_mode): Delete.
17292 (altivec_expand_lvx_be, altivec_expand_stvx_be,
17293 altivec_expand_stvex_be): Delete.
17294 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
17295 -maltivec=be support.
17296 (rs6000_gimple_fold_builtin): Ditto.
17297 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
17299 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
17300 (TARGET_DIRECT_MOVE_64BIT): Adjust.
17301 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
17302 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
17303 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
17304 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
17305 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
17306 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
17307 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
17308 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
17309 anonymous split): Adjust.
17310 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
17311 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
17313 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
17315 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
17316 when --with-gxx-include-dir is also specified.
17317 * configure: Regenerate.
17319 2018-05-09 Jim Wilson <jimw@sifive.com>
17322 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
17323 * config/riscv/t-withmultilib: New.
17324 * config/riscv/withmultilib.h: New.
17325 * doc/install.texi: Document RISC-V --with-multilib-list support.
17327 2018-05-09 Richard Biener <rguenther@suse.de>
17329 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
17331 (vect_bb_vectorization_profitable_p): Adjust. Compute
17332 actual scalar cost using the cost vector and the add_stmt_cost
17335 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17337 PR rtl-optimization/85645
17338 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
17339 in the REG_CFA_REGISTER note for LR, don't leave it empty.
17341 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17343 PR rtl-optimization/85645
17344 * shrink-wrap.c (spread_components): Return a boolean saying if
17345 anything was changed.
17346 (try_shrink_wrapping_separate): Iterate spread_components until
17347 nothing changes anymore.
17349 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17351 PR rtl-optimization/85645
17352 * regrename.c (build_def_use): Also kill the chains that include the
17353 destination of a REG_CFA_REGISTER note.
17355 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17357 PR rtl-optimization/85645
17358 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
17359 insn that has a REG_CFA_REGISTER note.
17361 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
17363 * cfgexpand.c (expand_clobber): New function.
17364 (expand_gimple_stmt_1): Use it.
17365 * tree-vect-stmts.c (vect_clobber_variable): New function,
17367 (vectorizable_simd_clone_call): ...here.
17368 (vectorizable_store): Emit a clobber either side of an
17369 IFN_STORE_LANES sequence.
17370 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
17372 2018-05-09 Tom de Vries <tom@codesourcery.com>
17375 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
17376 (define_insn "trap_if_false"): Add exit after trap.
17378 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
17380 PR rtl-optimization/85638
17381 * bb-reorder.c: Include common/common-target.h.
17382 (create_forwarder_block): New function extracted from...
17383 (fix_up_crossing_landing_pad): ...here. Rename into...
17384 (dw2_fix_up_crossing_landing_pad): ...this.
17385 (sjlj_fix_up_crossing_landing_pad): New function.
17386 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
17387 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
17388 from both partitions and exit the loop after one iteration.
17390 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17393 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17395 (Basic PowerPC Built-in Functions): The new name of the
17396 subsection previously known as "PowerPC Built-in Functions".
17397 (Basic PowerPC Built-in Functions Available on all Configurations):
17399 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
17401 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
17403 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
17405 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
17408 2018-05-08 Jim Wilson <jimw@sifive.com>
17410 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
17411 (LD_EMUL_SUFFIX): New.
17412 (LINK_SPEC): Use it.
17414 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17416 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17418 (Basic PowerPC Built-in Functions): The new name of the
17419 subsection previously known as "PowerPC Built-in Functions".
17420 (Basic PowerPC Built-in Functions Available on all Configurations):
17422 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
17424 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
17426 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
17428 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
17431 2018-05-08 Jakub Jelinek <jakub@redhat.com>
17434 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
17435 after cmpelim optimization.
17437 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
17439 * config.gcc: Support "goldmont".
17440 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
17441 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17442 PROCESSOR_GOLDMONT.
17443 * config/i386/i386.c (m_GOLDMONT): Define.
17444 (processor_target_table): Add "goldmont".
17445 (PTA_GOLDMONT): Define.
17446 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
17447 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
17448 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
17449 (fold_builtin_cpu): Add "goldmont".
17450 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
17451 (ix86_option_override_internal): Add "goldmont".
17452 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
17453 (processor_type): Add PROCESSOR_GOLDMONT.
17454 * config/i386/i386.md: Add CPU "glm".
17455 * config/i386/glm.md: New file.
17456 * config/i386/x86-tune.def: Add m_GOLDMONT.
17457 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
17459 2018-05-08 Jakub Jelinek <jakub@redhat.com>
17462 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
17464 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
17465 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
17466 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
17469 * config/i386/i386.c (ix86_fold_builtin): Handle
17470 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
17473 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
17474 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
17476 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
17479 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
17480 (check_arch): Likewise.
17481 (check_fpu): Return the result rather than printing it.
17482 (end arch): Fix operator precedence.
17483 (end cpu): Likewise.
17484 (END): Print the result from check_fpu.
17486 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
17487 Alan Hayward <alan.hayward@arm.com>
17488 David Sherwood <david.sherwood@arm.com>
17490 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
17491 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
17492 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
17493 (*fcmuo<mode>_and): New patterns.
17495 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
17497 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
17498 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
17499 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
17500 (cmp_op, sve_imm_con): New code attributes.
17501 (SVE_COND_INT_CMP, imm_con): Delete.
17502 (cmp_op): Remove above unspecs from int attribute.
17503 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
17505 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
17506 comparison-specific unspecs.
17507 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
17508 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
17509 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
17510 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
17511 (*vec_fcm<cmp_op><mode>): Rename to...
17512 (*fcm<cmp_op><mode>): ...this and adjust likewise.
17513 (*vec_fcmuo<mode>): Rename to...
17514 (*fcmuo<mode>): ...this and adjust likewise.
17515 (*pred_fcm<cmp_op><mode>): New pattern.
17516 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
17517 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
17519 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
17521 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
17522 (aarch64_emit_sve_predicated_cond): New function.
17523 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
17524 (aarch64_emit_unspec_cond_or): Replace with...
17525 (aarch64_emit_sve_or_conds): ...this new function. Use
17526 aarch64_emit_sve_ptrue_op for the individual comparisons and
17527 aarch64_emit_binop to OR them together.
17528 (aarch64_emit_inverted_unspec_cond): Replace with...
17529 (aarch64_emit_sve_inverted_cond): ...this new function. Use
17530 aarch64_emit_sve_ptrue_op for the comparison and
17531 aarch64_emit_unop to invert the result.
17532 (aarch64_expand_sve_vec_cmp_float): Update after the above
17533 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
17535 2018-05-07 Nathan Sidwell <nathan@acm.org>
17537 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
17538 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
17539 (Backwards Compatibility): Likewise.
17541 2018-05-07 Luis Machado <luis.machado@linaro.org>
17545 2018-05-07 Luis Machado <luis.machado@linaro.org>
17547 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17548 <prefetch_dynamic_strides>: New const bool field.
17549 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17550 prefetch_dynamic_strides.
17551 (exynosm1_prefetch_tune): Likewise.
17552 (thunderxt88_prefetch_tune): Likewise.
17553 (thunderx_prefetch_tune): Likewise.
17554 (thunderx2t99_prefetch_tune): Likewise.
17555 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
17557 (aarch64_override_options_internal): Update to set
17558 PARAM_PREFETCH_DYNAMIC_STRIDES.
17559 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
17560 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
17561 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
17562 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
17563 prefetch-dynamic-strides setting.
17565 2018-05-07 Luis Machado <luis.machado@linaro.org>
17567 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17568 <minimum_stride>: New const int field.
17569 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17570 minimum_stride field.
17571 (exynosm1_prefetch_tune): Likewise.
17572 (thunderxt88_prefetch_tune): Likewise.
17573 (thunderx_prefetch_tune): Likewise.
17574 (thunderx2t99_prefetch_tune): Likewise.
17575 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
17576 (aarch64_override_options_internal): Update to set
17577 PARAM_PREFETCH_MINIMUM_STRIDE.
17578 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
17579 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
17580 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
17581 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
17582 stride is constant and is below the minimum stride threshold.
17584 2018-05-07 Luis Machado <luis.machado@linaro.org>
17586 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
17589 2018-05-07 Luis Machado <luis.machado@linaro.org>
17591 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17592 <prefetch_dynamic_strides>: New const bool field.
17593 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17594 prefetch_dynamic_strides.
17595 (exynosm1_prefetch_tune): Likewise.
17596 (thunderxt88_prefetch_tune): Likewise.
17597 (thunderx_prefetch_tune): Likewise.
17598 (thunderx2t99_prefetch_tune): Likewise.
17599 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
17601 (aarch64_override_options_internal): Update to set
17602 PARAM_PREFETCH_DYNAMIC_STRIDES.
17603 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
17604 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
17605 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
17606 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
17607 prefetch-dynamic-strides setting.
17609 2018-05-07 Luis Machado <luis.machado@linaro.org>
17611 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17612 <minimum_stride>: New const int field.
17613 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17614 minimum_stride field.
17615 (exynosm1_prefetch_tune): Likewise.
17616 (thunderxt88_prefetch_tune): Likewise.
17617 (thunderx_prefetch_tune): Likewise.
17618 (thunderx2t99_prefetch_tune): Likewise.
17619 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
17620 (aarch64_override_options_internal): Update to set
17621 PARAM_PREFETCH_MINIMUM_STRIDE.
17622 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
17623 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
17624 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
17625 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
17626 stride is constant and is below the minimum stride threshold.
17628 2018-05-06 Jakub Jelinek <jakub@redhat.com>
17631 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
17632 the type is addressable. Don't force op into register if it has
17635 2018-05-05 Roland McGrath <mcgrathr@google.com>
17638 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
17639 any section for which we don't know a specific type it should have,
17640 regardless of name. Previously this was done only for the exact
17641 names ".init_array", ".fini_array", and ".preinit_array".
17642 (default_elf_asm_named_section): Add comment about
17643 relationship with default_section_type_flags and SECTION_NOTYPE.
17644 (get_section): Don't consider it a type conflict if one side has
17645 SECTION_NOTYPE and the other doesn't, as long as neither has the
17646 SECTION_BSS et al used in the default_section_type_flags logic.
17648 2018-05-05 Tom de Vries <tom@codesourcery.com>
17651 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
17652 (workaround_barsyncs): New function.
17653 (nvptx_reorg): Use workaround_barsyncs.
17654 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
17655 (define_expand "nvptx_membar_cta"): New define_expand.
17656 (define_insn "*nvptx_membar_cta"): New insn.
17658 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17660 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
17661 To improve optimization opportunities.
17662 * builtin-types.def: The new needed builtin types for the above.
17664 2018-05-04 Richard Biener <rguenther@suse.de>
17666 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
17667 * gimple-ssa-store-merging.c
17668 (imm_store_chain_info::output_merged_store): Remove redundant create,
17669 release split_store vector contents on failure.
17670 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
17671 scalar stmt vector on cache hit.
17673 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
17675 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
17677 * config.gcc (powerpc-xilinx-eabi*): Remove.
17678 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
17680 (fusion_addis_mem_combo_load): Ditto.
17681 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
17683 (rs6000_cpu_cpp_builtins): Ditto.
17684 * config/rs6000/rs6000-linux.c
17685 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
17686 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
17687 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
17689 (rs6000_setup_reg_addr_masks): Ditto.
17690 (rs6000_init_hard_regno_mode_ok): Ditto.
17691 (rs6000_option_override_internal): Ditto.
17692 (legitimate_lo_sum_address_p): Ditto.
17693 (rs6000_legitimize_address): Ditto.
17694 (rs6000_legitimize_reload_address): Ditto.
17695 (rs6000_legitimate_address_p): Ditto.
17696 (abi_v4_pass_in_fpr): Ditto.
17697 (setup_incoming_varargs): Ditto.
17698 (rs6000_gimplify_va_arg): Ditto.
17699 (rs6000_split_multireg_move): Ditto.
17700 (rs6000_savres_strategy): Ditto.
17701 (rs6000_emit_prologue_components): Ditto.
17702 (rs6000_emit_epilogue_components): Ditto.
17703 (rs6000_emit_prologue): Ditto.
17704 (rs6000_emit_epilogue): Ditto.
17705 (rs6000_elf_file_end): Ditto.
17706 (rs6000_function_value): Ditto.
17707 (rs6000_libcall_value): Ditto.
17708 * config/rs6000/rs6000.h: Ditto.
17709 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
17710 (TARGET_MINMAX): ... this. New.
17711 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
17712 * config/rs6000/rs6000.md: Remove Xilinx FP support.
17713 (*movsi_internal1_single): Delete.
17714 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
17715 mfpu=, mxilinx-fpu): Delete.
17716 * config/rs6000/singlefp.h: Delete.
17717 * config/rs6000/sysv4.h: Remove Xilinx FP support.
17718 * config/rs6000/t-rs6000: Ditto.
17719 * config/rs6000/t-xilinx: Delete.
17720 * config/rs6000/titan.md: Adjust for fp_type removal.
17721 * config/rs6000/vsx.md: Remove Xilinx FP support.
17722 (VStype_simple): Delete.
17723 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
17724 * config/rs6000/xfpu.h: Delete.
17725 * config/rs6000/xfpu.md: Delete.
17726 * config/rs6000/xilinx.h: Delete.
17727 * config/rs6000/xilinx.opt: Delete.
17728 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
17729 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
17731 2018-05-04 Tom de Vries <tom@codesourcery.com>
17734 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
17737 2018-05-04 Richard Biener <rguenther@suse.de>
17739 PR middle-end/85627
17740 * tree-complex.c (update_complex_assignment): We are always in SSA form.
17741 (expand_complex_div_wide): Likewise.
17742 (expand_complex_operations_1): Likewise.
17743 (expand_complex_libcall): Preserve EH info of the original stmt.
17744 (tree_lower_complex): Handle removed blocks.
17745 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
17746 on complex multiplication and division libcall builtins.
17748 2018-05-04 Richard Biener <rguenther@suse.de>
17750 PR middle-end/85574
17751 * fold-const.c (negate_expr_p): Restrict negation of operand
17752 zero of a division to when we know that can happen without
17754 (fold_negate_expr_1): Likewise.
17756 2018-05-04 Jakub Jelinek <jakub@redhat.com>
17759 * real.h (real_nextafter): Declare.
17760 * real.c (real_nextafter): New function.
17761 * fold-const-call.c (fold_const_nextafter): New function.
17762 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
17763 CASE_CFN_NEXTTOWARD.
17764 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
17765 even when arg1_mode is different from arg0_mode.
17767 2018-05-03 Nathan Sidwell <nathan@acm.org>
17769 * doc/extend.texi (Deprecated Features): Remove
17770 -ffriend-injection.
17771 (Backwards Compatibility): Likewise.
17772 * doc/invoke.texi (C++ Language Options): Likewise.
17773 (C++ Dialect Options): Likewise.
17775 2018-05-03 Jakub Jelinek <jakub@redhat.com>
17778 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
17779 _mm512_mask_mullox_epi64): New intrinsics.
17781 2018-05-03 Tom de Vries <tom@codesourcery.com>
17784 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
17785 dump files): Add offload-tree.
17787 2018-05-03 Richard Biener <rguenther@suse.de>
17789 PR tree-optimization/85615
17790 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
17791 to loops not nested in BBs loop father to avoid creating multi-entry
17794 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17796 PR tree-optimization/70291
17797 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
17798 arguments. Change return type to tree. Emit libcall as a new
17799 statement rather than replacing existing one when inplace_p is true.
17800 (expand_complex_multiplication_components): New function.
17801 (expand_complex_multiplication): Expand floating-point complex
17802 multiplication using the above.
17803 (expand_complex_division): Rename inner_type parameter to type.
17804 Update expand_complex_libcall call-site.
17805 (expand_complex_operations_1): Update expand_complex_multiplication
17806 and expand_complex_division call-sites.
17808 2018-05-02 Jakub Jelinek <jakub@redhat.com>
17811 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
17812 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
17813 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
17814 the highest significant bit of the shift count mask is clear. In
17815 check whether and[sq]i3 is needed verify that all significant bits
17816 of the shift count other than the highest are set.
17818 2018-05-02 Tom de Vries <tom@codesourcery.com>
17821 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
17822 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
17823 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
17824 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
17825 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
17826 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
17827 __builtin_goacc_parlevel_size.
17829 2018-05-02 Richard Biener <rguenther@suse.de>
17831 PR tree-optimization/85597
17832 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
17833 do not use split vect_get_vec_defs call but call vect_get_slp_defs
17836 2018-05-02 Tom de Vries <tom@codesourcery.com>
17839 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
17840 dump files): Add ltrans-tree.
17842 2018-05-02 Tom de Vries <tom@codesourcery.com>
17845 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
17846 dump files): Add wpa-ipa.
17848 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
17850 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
17851 powerpc*-*-linux*paired* target.
17852 * config/rs6000/750cl.h: Delete.
17853 * config/rs6000/paired.h: Delete.
17854 * config/rs6000/paired.md: Delete.
17855 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
17857 * config/rs6000/rs6000-builtin.def: Remove paired float support.
17858 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
17859 comment. Remove paired float support.
17860 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
17861 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
17863 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
17864 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
17866 * config/rs6000/rs6000.c: Remove paired float support.
17867 (paired_expand_vector_init, paired_expand_vector_move,
17868 paired_emit_vector_compare, paired_emit_vector_cond_expr,
17869 (paired_expand_lv_builtin, paired_expand_stv_builtin,
17870 paired_expand_builtin, paired_expand_predicate_builtin,
17871 paired_init_builtins): Delete.
17872 * config/rs6000/rs6000.h: Remove paired float support.
17873 * config/rs6000/rs6000.md: Remove paired float support.
17874 (move_from_CR_ov_bit): Delete.
17875 * config/rs6000/rs6000.opt (mpaired): Delete.
17876 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
17877 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
17879 2018-05-02 Richard Biener <rguenther@suse.de>
17881 PR middle-end/85567
17882 * gimplify.c (gimplify_save_expr): When in SSA form allow
17883 SAVE_EXPRs to compute to SSA vars.
17885 2018-05-02 Jakub Jelinek <jakub@redhat.com>
17888 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
17889 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
17890 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
17891 clobber operands[2], instead use a new pseudo. Formatting fixes.
17893 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
17895 PR tree-optimization/85586
17896 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
17897 exit early for statements in the same group if the accesses are
17900 2018-05-02 Tom de Vries <tom@codesourcery.com>
17903 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
17906 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
17908 PR tree-optimization/85143
17909 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
17911 2018-05-01 Tom de Vries <tom@codesourcery.com>
17914 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
17915 not found" error message.
17917 2018-05-01 Tom de Vries <tom@codesourcery.com>
17920 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
17921 * vec.c (test_ordered_remove_if): New function.
17922 (vec_c_tests): Call test_ordered_remove_if.
17923 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
17924 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
17925 * tree-vect-patterns.c (vect_pattern_recog_1): Use
17926 VEC_ORDERED_REMOVE_IF.
17928 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17930 PR tree-optimization/82665
17931 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
17932 pointer subtraction where arguments come from a memchr call.
17934 2018-05-01 Jakub Jelinek <jakub@redhat.com>
17936 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
17937 --push-state --as-needed and --pop-state instead of --as-needed and
17938 --no-as-needed if ld supports it.
17939 * configure: Regenerated.
17942 * doc/install.texi2html: Replace _002d with - and _002a with * in
17943 generated html files using sed.
17945 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17948 * gcc-rich-location.c (blank_line_before_p): New function.
17949 (use_new_line): New function.
17950 (gcc_rich_location::add_fixit_insert_formatted): New function.
17951 * gcc-rich-location.h
17952 (gcc_rich_location::add_fixit_insert_formatted): New function.
17954 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17956 * selftest.c (assert_streq): Rename "expected" and "actual" to
17957 "val1" and "val2". Extend NULL-handling to cover both inputs
17958 symmetrically, while still requiring both to be non-NULL for a pass.
17959 * selftest.h (assert_streq): Rename "expected" and "actual" to
17961 (ASSERT_EQ): Likewise.
17962 (ASSERT_EQ_AT): Likewise.
17963 (ASSERT_KNOWN_EQ): Likewise.
17964 (ASSERT_KNOWN_EQ_AT): Likewise.
17965 (ASSERT_NE): Likewise.
17966 (ASSERT_MAYBE_NE): Likewise.
17967 (ASSERT_MAYBE_NE_AT): Likewise.
17968 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
17969 the assertion to pass.
17970 (ASSERT_STREQ_AT): Likewise.
17972 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
17974 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
17975 interaction with -pie.
17977 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17979 * selftest.h: Fix alphabetization of per-source-file selftest
17982 2018-04-30 Jason Merrill <jason@redhat.com>
17984 PR c++/61982 - dead stores to destroyed objects.
17985 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
17988 2018-04-30 Jason Merrill <jason@redhat.com>
17990 * tree.c (build_clobber): New.
17991 * tree.h: Declare it.
17992 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
17994 2018-04-30 David Malcolm <dmalcolm@redhat.com>
17996 * diagnostic-show-locus.c (layout::layout): Update for
17997 location_get_source_line returning a char_span.
17998 (struct char_span): Move to input.h.
17999 (struct correction): Update for fields in char_span becoming
18001 (struct source_line): Update for location_get_source_line
18002 returning a char_span.
18003 (layout::print_line): Likewise.
18004 * edit-context.c (edited_file::print_content): Likewise.
18005 (edited_file::print_diff_hunk): Likewise.
18006 (edited_file::print_run_of_changed_lines): Likewise.
18007 (edited_file::get_num_lines): Likewise.
18008 (edited_line::edited_line): Likewise.
18009 * final.c (asm_show_source): Likewise.
18010 * input.c (location_get_source_line): Convert return type
18011 from const char * to char_span, losing the final "line_len"
18013 (dump_location_info): Update for the above.
18014 (get_substring_ranges_for_loc): Likewise. Use a char_span
18015 when handling the literal within the line.
18016 (test_reading_source_line): Update for location_get_source_line
18017 returning a char_span.
18018 * input.h (class char_span): Move here from
18019 diagnostic-show-locus.c, converting from a struct to a class.
18020 Make data members private.
18021 (char_span::operator bool): New.
18022 (char_span::length): New.
18023 (char_span::get_buffer): New.
18024 (char_span::operator[]): New.
18025 (char_span::subspan): Make const.
18026 (char_span::xstrdup): New.
18027 (location_get_source_line): Convert return type from const char *
18028 to char_span, losing the final "line_size" param.
18030 2018-04-30 Jan Hubicka <jh@suse.cz>
18032 * lto-wrapper.c (ltrans_priorities): New static var.
18033 (cmp_priority): New.
18034 (run_gcc): Read priorities and if doing parallel build order
18035 the Makefile by them.
18037 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18039 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
18041 2018-04-30 Richard Biener <rguenther@suse.de>
18043 * tree-cfg.c (verify_address): Remove base argument, add
18044 flag whether to check TREE_ADDRESSABLE and do that.
18045 (verify_expr): Remove.
18046 (verify_types_in_gimple_reference): Add pieces from verify_expr.
18047 (verify_gimple_assign_single): Likewise.
18048 (verify_gimple_switch): Likewise.
18049 (verify_expr_location_1): Dereference tp once. Add (disabled)
18050 piece from verify_expr.
18051 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
18053 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
18055 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
18057 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
18059 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
18060 (small_data_pattern): Likewise.
18061 (arc_rewrite_small_data): Likewise.
18062 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
18063 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
18064 (get_symbol_alignment): New function.
18065 (legitimate_small_data_address_p): Likewise.
18066 (legitimate_scaled_address): Update, call
18067 legitimate_small_data_address_p.
18068 (output_sdata): New static variable.
18069 (arc_print_operand): Update how we handle small data operands.
18070 (arc_print_operand_address): Likewise.
18071 (arc_legitimate_address_p): Update, use
18072 legitimate_small_data_address_p.
18073 (arc_rewrite_small_data_p): Remove.
18074 (arc_rewrite_small_data_1): Likewise.
18075 (arc_rewrite_small_data): Likewise.
18076 (small_data_pattern): Likewise.
18077 (compact_sda_memory_operand): Update to use
18078 legitimate_small_data_address_p and get_symbol_alignment.
18079 (prepare_move_operands): Don't rewite sdata pattern.
18080 (prepare_extend_operands): Remove.
18081 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
18083 (zero_extendqisi2): Likewise.
18084 (zero_extendhisi2): Likewise.
18085 (extendqihi2): Likewise.
18086 (extendqisi2): Likewise.
18087 (extendhisi2): Likewise.
18088 (addsi3): Likewise.
18089 (subsi3): Likewise.
18090 (andsi3): Likewise.
18091 * config/arc/constraints.md (Usd): Change it to memory constraint.
18093 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
18095 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
18096 as source of std instructions.
18097 * config/arc/arc.md (movsi_insn): Update pattern predicate to
18098 allow 6-bit constants as source for store instructions.
18099 (movdi_insn): Update instruction pattern to allow 6-bit constants
18100 as source for store instructions.
18102 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
18104 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
18106 2018-04-30 Nathan Sidwell <nathan@acm.org>
18107 Sandra Loosemore <sandra@codesourcery.com>
18109 * dumpfile.c (dump_open): Allow '-' for stdout.
18110 * doc/invoke.texi (Developer Options): Document dump filename
18111 determination early. Document stdin/stdout selection.
18113 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
18115 Microblaze Target: PIC data text relative
18117 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
18118 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
18120 * config/microblaze/microblaze.h (microblaze_constant_address_p):
18121 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
18122 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
18123 New addressing mode for data-text relative position indepenedent code.
18124 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
18125 'ADDRESS_SYMBOLIC_TXT_REL'.
18126 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
18127 (microblaze_legitimate_pic_operand): Exclude function calls from
18128 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
18129 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
18131 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
18132 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
18133 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
18134 for 'address + offset'.
18135 (microblaze_expand_prologue): Add new function prologue call for
18137 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
18138 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
18139 table in case of TARGET_PIC_DATA_TEXT_REL.
18140 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
18141 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
18142 Add new macros 'UNSPEC_TEXT',
18143 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
18144 + exclude function calls from 'UNSPEC_PLT' in case of data text
18146 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
18147 new target hook for generating address diff vector tables in case of
18149 * doc/tm.texi : Regenerate.
18150 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
18151 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
18152 of addr diff vector generation.
18153 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
18154 target hook definition.
18155 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
18156 Add default function for generate_pic_addr_diff_vec -> flag_pic.
18157 * doc/invoke.texi (Add new pic option): Add new microblaze pic
18158 option for data text relative.
18160 2018-04-30 Richard Biener <rguenther@suse.de>
18162 * tree-chrec.h (evolution_function_is_constant_p): Remove
18164 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
18166 2018-04-30 Richard Biener <rguenther@suse.de>
18169 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
18171 2018-04-30 Richard Biener <rguenther@suse.de>
18173 PR tree-optimization/28364
18174 PR tree-optimization/85275
18175 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
18176 copying first exit test.
18178 2018-04-28 Mark Wielaard <mark@klomp.org>
18180 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
18181 dwarf_version >= 5.
18182 (dwarf_AT): Handle DW_AT_addr_base.
18183 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
18185 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
18188 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
18189 (*ashl<dwi>3_doubleword_mask_1): Ditto.
18190 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
18191 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
18193 2018-04-28 Richard Biener <rguenther@suse.de>
18195 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
18196 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
18197 to reflect use. Only add interesting stmts.
18199 2018-04-27 Martin Jambor <mjambor@suse.cz>
18202 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
18203 the jump function allows for passing through aggregate values.
18205 2018-04-27 David Malcolm <dmalcolm@redhat.com>
18207 * input.h (in_system_header_at): Convert from macro to inline
18209 (from_macro_expansion_at): Likewise.
18210 (from_macro_definition_at): Likewise.
18212 2018-04-27 Jeff Law <law@redhat.com>
18214 * config.gcc: Mark tile* targets as deprecated/obsolete.
18216 2018-04-27 Richard Biener <rguenther@suse.de>
18218 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
18221 2018-04-27 Richard Biener <rguenther@suse.de>
18223 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
18225 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
18227 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
18228 with Yd constraint. Set "preferred_for_speed" attribute from
18229 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
18230 with Yd constraint.
18231 (*movdi_internal): Ditto.
18232 (movti_interunit splitters): Remove
18233 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
18234 (movdi_interunit splitters): Ditto.
18235 * config/i386/constraints.md (Ye): Remove.
18236 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
18238 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18241 * config/aarch64/constraints.md (Usg): Limit to 31.
18242 (Usj): Limit to 63.
18244 2018-04-27 Jakub Jelinek <jakub@redhat.com>
18246 PR tree-optimization/85529
18247 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
18248 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
18249 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
18250 zero extension or masking of the MSB bit.
18251 (optimize_range_tests): Add FIRST_BB argument, pass it through
18252 to optimize_range_tests_var_bound.
18253 (maybe_optimize_range_tests, reassociate_bb): Adjust
18254 optimize_range_tests callers.
18256 2018-04-26 Richard Biener <rguenther@suse.de>
18257 Jakub Jelinek <jakub@redhat.com>
18259 * cgraph.h (symbol_table): Just declare debug method here.
18260 * symtab.c (symbol_table::debug): Define.
18262 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
18264 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
18266 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
18268 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
18269 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
18270 (*movdi_internal): Substitute Yi and Yj constraint with x
18271 and Ym and Yn constraint with y constraint. Update "isa"
18272 attribute and set "preferred_for_speed" attribute from
18273 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
18274 (*movsi_internal): Ditto.
18275 (*movdf_internal): Ditto.
18276 (*movsf_internal): Ditto.
18277 (*zero_extendsidi2): Ditto.
18278 * config/i386/sse.md (vec_set<mode>_0): Ditto.
18279 (sse2_loadld): Ditto.
18280 (*vec_extract<ssevecmodelower>_0): Ditto.
18281 (*vec_extractv4si_0_zext_sse4): Ditto.
18282 (vec_concatv2di): Ditto.
18283 (*vec_dup<mode>): Ditto.
18284 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
18285 * config/i386/constraints.md (Yi): Remove.
18290 2018-04-26 Nathan Sidwell <nathan@acm.org>
18292 * dumpfile.c (dump_open): New.
18293 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
18294 (dump_finish): Detect stdio/stderr by value not name.
18296 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
18298 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
18300 2018-04-26 Tom de Vries <tom@codesourcery.com>
18303 * config/nvptx/nvptx.c (verify_neutering_jumps)
18304 (verify_neutering_labels): New function
18305 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
18307 2018-04-26 Tom de Vries <tom@codesourcery.com>
18310 * config/nvptx/nvptx.c (needs_neutering_p): New function.
18311 (nvptx_single): Use needs_neutering_p to skip over insns that do not
18314 2018-04-26 Richard Biener <rguenther@suse.de>
18315 Tom de Vries <tom@codesourcery.com>
18318 * lto-streamer-out.c (output_function): Fixup loops if required to match
18319 discovery done in the reader.
18321 2018-04-26 Richard Biener <rguenther@suse.de>
18323 PR tree-optimization/85116
18324 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
18325 have a loop exit from the single latch predecessor. Remove
18326 case of header with just condition.
18327 (ch_base::copy_headers): Exclude infinite loops from any
18329 (pass_ch::execute): Record exits.
18331 2018-04-26 Richard Biener <rguenther@suse.de>
18333 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
18334 prologue cost vector and pass it to vect_get_load_cost.
18335 (vect_get_peeling_costs_all_drs): Likewise.
18336 (vect_peeling_hash_get_lowest_cost): Likewise.
18337 (vect_enhance_data_refs_alignment): Likewise.
18339 2018-04-26 Richard Biener <rguenther@suse.de>
18341 PR middle-end/85450
18342 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
18343 checking of integer<->pointer conversions.
18344 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
18345 sign-/zero-extending pointer types.
18346 (expand_omp_for_static_chunk): Likewise.
18348 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
18349 Jean Lee <xiaoyur347@gmail.com>
18351 * config/mips/mips.c (mips_asan_shadow_offset): New function.
18352 (TARGET_ASAN_SHADOW_OFFSET): Define.
18353 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
18354 true for -fsanitize=address.
18356 2018-04-25 Mark Wielaard <mark@klomp.org>
18358 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
18361 2018-04-25 Jakub Jelinek <jakub@redhat.com>
18363 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
18364 than "alu", remove explicit "memory" and "imm_disp" attributes.
18365 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
18367 PR middle-end/85414
18368 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
18369 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
18370 gen_lowpart_no_emit.
18372 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
18375 * config/i386/i386.c (ix86_expand_builtin): Change memory
18376 operand to XI, extend p0 to Pmode.
18377 * config/i386/i386.md: Change unspec volatile and operand
18378 1 mode to XI, change operand 0 mode to P.
18380 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18382 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
18383 GET_MODE_MASK before any checking.
18384 (nds32_can_use_bset_p): Likewise.
18385 (nds32_can_use_btgl_p): Likewise.
18387 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18389 * config/nds32/nds32-doubleword.md: New define_split pattern for
18390 illegal register number.
18392 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18394 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
18396 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18398 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
18400 2018-04-25 Richard Biener <rguenther@suse.de>
18402 * lto-streamer.h (LTO_major_version): Bump to 8.
18404 2018-04-25 Jakub Jelinek <jakub@redhat.com>
18406 * BASE-VER: Set to 9.0.0.
18408 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
18410 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
18411 in __abskf2 and __powikf2.
18413 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18416 * config/aarch64/constraints.md (Usg, Usj): New constraints.
18417 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
18418 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
18419 Use the above on operand 2. Reindent.
18420 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
18422 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
18425 * common/config/i386/i386-common.c (ix86_handle_option): Don't
18427 * config/i386/i386.opt (mcet): Removed.
18428 * doc/install.texi: Remove -mcet documentation.
18429 * doc/invoke.texi: Likewise.
18431 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
18434 * doc/install.texi: Remove -mcet from bootstrap-cet.
18436 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18439 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
18440 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
18444 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
18445 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
18446 containing a CONST_VECTOR.
18448 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
18450 * doc/install.texi: Update newlib dependency for nvptx.
18452 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18455 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
18456 instead of INTVAL when shifting x left.
18458 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
18460 PR tree-optimization/85478
18461 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
18462 vect_grouped_store_supported for single element vectors.
18464 2018-04-24 Richard Biener <rguenther@suse.de>
18467 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
18468 load cost increase to the case of non-constant step.
18470 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18473 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
18474 destination if any_malformed_asm.
18476 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
18478 PR middle-end/85496
18479 * expr.c (store_field): In the bitfield case, if the value comes from
18480 a function call and is returned in registers by means of a PARALLEL,
18481 do not change the mode of the temporary unless BLKmode and VOIDmode.
18483 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
18485 PR rtl-optimization/85423
18486 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
18487 dependencies to debug insns when the previous insn is non-debug.
18489 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
18491 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
18492 enums into a single definition.
18493 (fls): Fix predicates and printing.
18496 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
18498 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
18499 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
18500 and short u6 immediate.
18501 (check_if_valid_sleep_operand): Remove.
18502 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
18504 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18506 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
18507 flag_always_save_lp condition.
18508 * config/nds32/nds32.opt (malways-save-lp): New option.
18510 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18512 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
18513 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
18514 * config/nds32/nds32.h
18515 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
18516 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
18518 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18520 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
18521 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
18523 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18524 Chung-Ju Wu <jasonwucj@gmail.com>
18526 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
18528 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
18529 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
18531 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18533 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
18535 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18537 * config/nds32/nds32-protos.h (nds32_data_alignment,
18538 nds32_local_alignment): Declare.
18539 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
18540 nds32_local_alignment): New functions.
18541 (TARGET_CONSTANT_ALIGNMENT): Define.
18542 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
18544 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18546 * config/nds32/nds32.c
18547 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
18548 (TARGET_MODES_TIEABLE_P): Likewise.
18550 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18552 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
18553 level Ofast and Og.
18555 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
18556 Chung-Ju Wu <jasonwucj@gmail.com>
18558 * config/nds32/constants.md (unspec_volatile_element): Add enum values
18559 for unaligned access.
18560 * config/nds32/nds32-intrinsic.c: Implementation of expanding
18562 * config/nds32/nds32-intrinsic.md: Likewise.
18563 * config/nds32/nds32_intrinsic.h: Likewise.
18564 * config/nds32/nds32.h (nds32_builtins): Likewise.
18565 * config/nds32/nds32.opt (munaligned-access): New option.
18566 * config/nds32/nds32.c (nds32_asm_file_start): Display
18567 flag_unaligned_access status.
18569 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
18571 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
18572 -mno-relax is present.
18573 * config/riscv/linux.h (LINK_SPEC): Ditto.
18575 2018-04-20 Martin Sebor <msebor@redhat.com>
18578 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
18580 (gimple_fold_builtin_stxcpy_chk): Same.
18581 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
18583 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
18586 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
18587 __powikf2 when long double is IEEE 128-bit.
18589 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
18591 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
18592 step to make sure stack always aligned.
18594 2018-04-20 Carl Love <cel@us.ibm.com>
18597 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
18598 size check for arg0.
18600 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
18601 Tom de Vries <tom@codesourcery.com>
18604 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
18605 Emit insns for calls too.
18606 (nvptx_find_par): Always look for worker-level predecessor insn.
18607 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
18609 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
18610 (nvptx_process_pars): Propagate frames for calls.
18612 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
18615 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
18617 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
18618 (ix86_handle_option): Don't handle OPT_mibt.
18619 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
18621 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
18623 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
18624 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
18625 (ix86_target_macros): Define __CET__ with flag_cf_protection
18626 for -fcf-protection.
18627 * config/i386/i386.c (isa2_opts): Remove -mibt.
18628 * config/i386/i386.h (TARGET_IBT): Removed.
18629 (TARGET_IBT_P): Likewise.
18630 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
18631 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
18632 * config/i386/i386.opt (mcet): Update help message.
18633 (mshstk): Likewise.
18635 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
18636 -mcet as an alias for -mshstk.
18638 2018-04-20 Richard Biener <rguenther@suse.de>
18640 PR middle-end/85475
18641 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
18642 complexity by forcing a single use of the multiply operand.
18644 2018-04-20 Martin Jambor <mjambor@suse.cz>
18647 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
18648 recursion dependency to only apply to non-clones.
18650 2018-04-20 Martin Jambor <mjambor@suse.cz>
18653 * ipa-cp.c (create_specialized_node): Check that clones of
18654 self-recursive edges exist during IPA-CP.
18656 2018-04-19 Toon Moene <toon@moene.org>
18658 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
18661 2018-04-19 Jakub Jelinek <jakub@redhat.com>
18663 PR tree-optimization/85467
18664 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
18665 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
18666 VECTOR_CST element to type.
18668 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
18671 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
18672 * config/i386/i386.md (builtin_setjmp_setup): Removed.
18673 (builtin_longjmp): Likewise.
18674 (save_stack_nonlocal): New pattern.
18675 (restore_stack_nonlocal): Likewise.
18677 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
18680 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
18681 Replace ASM_OUTPUT_LABEL with fprintf.
18683 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
18686 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
18687 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
18688 * config/i386/i386-c.c (ix86_target_macros_internal): Also
18689 define __IBT__ and __SHSTK__ for -fcf-protection.
18690 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
18692 (ix86_trampoline_init): Likewise.
18693 (x86_output_mi_thunk): Likewise.
18694 (ix86_notrack_prefixed_insn_p): Likewise.
18695 (ix86_option_override_internal): Don't disallow -fcf-protection.
18696 * config/i386/i386.md (rdssp<mode>): Also enable for
18698 (incssp<mode>): Likewise.
18699 (nop_endbr): Likewise.
18700 * config/i386/i386.opt (mcet): Change help message to built-in
18703 (mshstk): Likewise.
18704 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
18705 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
18706 enable CET built-in functions.
18708 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
18710 * common/config/i386/i386-common.c
18711 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
18712 OPTION_MASK_ISA_MOVDIRI_UNSET,
18713 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
18714 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
18715 * config.gcc (movdirintrin.h): New header.
18716 * config/i386/cpuid.h (bit_MOVDIRI,
18717 bit_MOVDIR64B): New bits.
18718 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
18720 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
18721 (VOID, PVOID, PCVOID)): New function types.
18722 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
18723 __builtin_ia32_directstoreu_u64,
18724 __builtin_ia32_movdir64b): New builtins.
18725 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
18726 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
18728 (ix86_valid_target_attribute_inner_p): Ditto.
18729 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
18730 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
18731 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
18732 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
18733 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
18734 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
18735 (movdiri<mode>, movdir64b_<mode>): New.
18736 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
18737 * config/i386/immintrin.h: Include movdirintrin.h.
18738 * config/i386/movdirintrin.h: New file.
18739 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
18741 2018-04-19 Richard Biener <rguenther@suse.de>
18743 PR middle-end/85455
18744 * cfg.c (clear_bb_flags): When loop state says we have
18745 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
18747 2018-04-19 Richard Biener <rguenther@suse.de>
18749 PR tree-optimization/84737
18750 * tree-vect-data-refs.c (vect_copy_ref_info): New function
18751 copying restrict info.
18752 (vect_setup_realignment): Use it.
18753 * tree-vectorizer.h (vect_copy_ref_info): Declare.
18754 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
18755 the first DR to all generated stores.
18756 (vectorizable_load): Likewise for loads.
18758 2018-04-19 Jakub Jelinek <jakub@redhat.com>
18760 PR tree-optimization/85446
18761 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
18762 the integral and pointer types to have the same precision.
18764 * doc/install.texi: Document --disable-cet being the default and
18767 2018-04-18 Martin Liska <mliska@suse.cz>
18769 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
18772 2018-04-18 Martin Liska <mliska@suse.cz>
18775 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
18778 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
18779 arguments if they are comparable.
18781 2018-04-18 Martin Liska <mliska@suse.cz>
18784 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
18787 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
18790 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
18793 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
18794 ENDBR after calling __morestack.
18796 2018-04-18 David Malcolm <dmalcolm@redhat.com>
18799 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
18800 by using gcc_base_ver to generate a gcc_driver_version, and use
18801 it when generating GCC_DRIVER_NAME.
18802 * configure: Regenerate.
18804 2018-04-18 Jakub Jelinek <jakub@redhat.com>
18807 * config.gcc: Obsolete powerpc*-*-*spe*.
18809 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18812 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
18813 (stabstr_D): Change type of unum from unsigned int to
18814 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
18817 2018-04-17 Jim Wilson <jimw@sifive.com>
18820 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
18821 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
18822 Set arg_pointer_offset after using pretend_args_size.
18824 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18826 PR rtl-optimization/85431
18827 * dse.c (record_store): Ignore zero width stores.
18830 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
18831 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
18832 __builtin_stack_restore rather than after it.
18833 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
18834 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
18835 argument instead of virtual_dynamic_stack_rtx.
18837 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
18839 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
18841 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18842 Add note to error message to explain internal mapping of overloaded
18843 built-in function name to non-overloaded built-in function name.
18844 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
18847 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
18850 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
18851 where the inputs overlap with the output.
18853 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18855 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
18856 (=v, v) alternative and explicit "memory" attribute.
18857 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
18858 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
18860 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
18861 "sselog1" type instead of "sselog".
18862 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
18863 "sselog". Remove explicit "memory" attribute.
18864 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
18865 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
18867 (vec_extract_hi_v32hi): Merge all alternatives into one, use
18868 "sselog1" type instead of "sselog". Remove explicit "memory"
18870 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
18871 use "sselog1" type instead of "sselog". Remove explicit "memory"
18873 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
18874 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
18876 (vec_extract_hi_v64qi): Merge all alternatives into one, use
18877 "sselog1" type instead of "sselog". Remove explicit "memory"
18879 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
18880 use "sselog1" type instead of "sselog". Remove explicit "memory"
18884 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
18886 PR middle-end/85414
18887 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
18890 2018-04-17 Martin Jambor <mjambor@suse.cz>
18893 * ipa-cp.c (create_specialized_node): Call
18894 expand_all_artificial_thunks if necessary.
18896 2018-04-17 Martin Liska <mliska@suse.cz>
18899 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
18900 in message, remote space in between '_G' and '('.
18902 2018-04-17 Jakub Jelinek <jakub@redhat.com>
18905 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
18906 avx512f_vmcmp<mode>3<round_saeonly_name>,
18907 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
18908 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
18909 avx512f_rndscale<mode><round_saeonly_name>,
18910 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
18911 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
18912 Use %<iptr>2 instead of %2 for -masm=intel.
18913 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
18914 avx512f_vcvttss2usi<round_saeonly_name>,
18915 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
18917 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
18918 avx512f_vcvttsd2usi<round_saeonly_name>,
18919 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
18920 Use %q1 instead of %1 for -masm=intel.
18921 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
18922 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
18923 of %3 for -masm=intel.
18924 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
18926 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
18928 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
18930 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
18931 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
18933 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
18935 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
18936 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
18937 %g1 and one with %0 and %1.
18938 (avx512er_vmrcp28<mode><round_saeonly_name>,
18939 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
18940 %1 for -masm=intel.
18941 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
18942 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
18943 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
18944 of %0 and %{%4%} for -masm=intel.
18945 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
18946 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
18947 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
18948 order of %0 and %{%5%}%{z%} for -masm=intel.
18950 2018-04-17 Jan Hubicka <jh@suse.cz>
18953 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
18955 2018-04-17 Martin Liska <mliska@suse.cz>
18958 * multiple_target.c (create_dispatcher_calls): Set apostrophes
18959 for target_clone error message. Make default implementation
18960 clone to be a local declaration.
18961 (separate_attrs): Add new argument and check for an empty
18963 (expand_target_clones): Handle it.
18964 (ipa_target_clone): Make redirection just for target_clones
18967 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
18968 Tom de Vries <tom@codesourcery.com>
18970 PR middle-end/84955
18971 * omp-expand.c (expand_oacc_for): Add dummy false branch for
18972 tiled basic blocks without omp continue statements.
18974 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
18977 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
18978 vec_extract expression as having side effects to make sure it gets
18981 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
18984 * config/i386/i386.c (get_builtin_code_for_version): Check
18987 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
18990 * config.gcc: Support "skylake".
18991 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
18993 * config/i386/i386.c (m_SKYLAKE): Define.
18994 (processor_target_table): Add "skylake".
18995 (ix86_option_override_internal): Add "skylake".
18996 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
18997 PROCESSOR_CANNONLAKE.
18998 (get_builtin_code_for_version): Fix priority for
18999 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
19000 PROCESSOR_SKYLAKE-AVX512.
19001 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
19002 (processor_type): Add PROCESSOR_SKYLAKE.
19004 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
19005 Jason Merrill <jason@redhat.com>
19008 * convert.c (convert_to_integer_1): Use direct recursion for
19009 enumeral types and types with a precision less than the number
19010 of bits in their mode.
19012 2018-04-16 Julia Koval <julia.koval@intel.com>
19015 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
19016 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
19018 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
19021 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
19022 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
19023 and -mno-direct-move.
19025 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
19028 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
19029 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
19030 Ensure negative shifts result in {0}.
19032 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
19034 PR rtl-optimization/79916
19035 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
19036 regs (if any) to define how to gnerate SD moves when LRA is in
19039 2018-04-13 Jakub Jelinek <jakub@redhat.com>
19041 PR rtl-optimization/85393
19042 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
19043 * except.c (expand_dw2_landing_pad_for_region): Make static.
19044 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
19045 a label and unconditional jump to old_bb, rather than
19046 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
19049 PR rtl-optimization/85376
19050 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
19051 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
19052 instead of a specific value.
19054 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
19055 Bin Cheng <bin.cheng@arm.com>
19057 PR tree-optimization/82965
19058 PR tree-optimization/83991
19059 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
19060 by_profile_only parameter.
19061 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
19062 information if the loop was predicted to iterate too many times.
19063 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
19065 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
19068 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
19071 2018-04-13 Martin Liska <mliska@suse.cz>
19072 Jakub Jelinek <jakub@redhat.com>
19074 PR middle-end/81657
19075 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
19076 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
19077 * builtins.c (expand_builtin_memory_copy_args): Use
19078 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
19079 handle dest_addr == pc_rtx.
19081 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
19084 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
19085 asked to not generate direct moves.
19086 (fix_trunc<mode>si2_stfiwx): Similar.
19087 (fix_trunc<mode>si2_internal): Similar.
19089 2018-04-12 Jakub Jelinek <jakub@redhat.com>
19092 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
19093 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
19094 lookup if dest in some wider mode is known to be const0_rtx and
19095 if so, record permanent equivalence for it to be ZERO_EXTEND of
19096 the narrower mode destination.
19098 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
19100 * lto-streamer-out.c (output_function): Revert 259346.
19101 * omp-expand.c (expand_oacc_for): Likewise.
19103 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
19105 PR rtl-optimization/85354
19106 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
19107 * sel-sched.c (sel_global_init): ... here.
19109 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
19112 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
19113 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
19114 mode for PE-COFF targets.
19115 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
19116 (i386_pe_asm_lto_end): Likewise.
19117 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
19118 (TARGET_ASM_LTO_END): Likewise.
19119 * config/i386/winnt.c (saved_debug_info_level): New static variable.
19120 (i386_pe_asm_lto_start): New function.
19121 (i386_pe_asm_lto_end): Likewise.
19123 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
19124 Richard Biener <rguenther@suse.de>
19126 PR middle-end/84955
19127 * lto-streamer-out.c (output_function): Fix CFG loop state before
19129 * omp-expand.c (expand_oacc_for): Handle calls to internal
19130 functions like regular functions.
19132 2018-04-12 Richard Biener <rguenther@suse.de>
19135 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
19136 for the early LTO debug to properly generate references to it
19137 during DIE emission. Do not re-use that for the skeleton for
19139 (dwarf2out_early_finish): Likewise.
19141 2018-04-12 Jakub Jelinek <jakub@redhat.com>
19144 * config/i386/sse.md
19145 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
19146 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
19147 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
19148 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
19149 and output is a reg, avoid creating invalid lowpart subreg, but
19150 instead split into a 512-bit move. Don't split if not AVX512VL,
19151 input is xmm16+ reg and output is a mem.
19152 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
19153 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
19154 xmm16+ reg and output is a mem.
19156 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19158 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
19159 also for flag_dwarf2_cfi_asm.
19161 2018-04-12 Jakub Jelinek <jakub@redhat.com>
19163 PR rtl-optimization/85342
19164 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
19165 a bool scalar var inside of the loop instead. Don't try to update
19166 recog_data.operand after failed apply_change_group.
19168 2018-04-12 Tom de Vries <tom@codesourcery.com>
19171 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
19172 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
19173 array with flexible array member as array without given dimension.
19174 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
19175 argument for undefined param to true.
19177 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
19180 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
19181 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
19182 from PowerPC section.
19183 * config/rs6000/sysv4.opt (mcall-): Improve help text.
19184 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
19185 help text that is too long.
19186 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
19187 help text that is too long.
19188 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
19189 help text that is too long.
19191 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
19193 * config/alpha/alpha.md (stack_probe_internal): Rename
19194 from "probe_stack". Update all callers.
19196 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19198 PR rtl-optimization/84566
19199 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
19200 sched_macro_fuse_insns.
19202 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19205 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
19206 (compute_block_dependences): ... from here.
19208 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19210 PR tree-optimization/85331
19211 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
19212 from int to HOST_WIDE_INT.
19214 2018-04-11 Martin Jambor <mjambor@suse.cz>
19217 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
19218 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
19219 not the same as the source val.
19220 (cgraph_edge_brings_value_p): New parameter.
19221 (gather_edges_for_value): Pass destination value to
19222 cgraph_edge_brings_value_p.
19223 (perhaps_add_new_callers): Likewise.
19224 (get_info_about_necessary_edges): Likewise and exclude values brought
19225 only by self-recursive edges.
19226 (create_specialized_node): Redirect only clones of self-calling edges.
19227 (+self_recursive_pass_through_p): New function.
19228 (find_more_scalar_values_for_callers_subset): Use it.
19229 (find_aggregate_values_for_callers_subset): Likewise.
19230 (known_aggs_to_agg_replacement_list): Removed.
19231 (decide_whether_version_node): Re-calculate known constants for all
19232 remaining context clones.
19234 2018-04-11 Richard Biener <rguenther@suse.de>
19237 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
19238 from early DWARF output.
19239 (dwarf2out_early_finish): Output line info unconditionally into
19240 early DWARF and add reference to it.
19242 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19245 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
19246 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
19247 other than V2DFmode using iptr mode attribute.
19248 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
19250 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19252 PR rtl-optimization/84659
19253 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
19255 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19258 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
19260 (output_loc_list): Pass address of a dummy size variable even in the
19261 locview handling loop.
19262 (index_location_lists): Add comment on why skip_loc_list_entry can't
19265 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
19268 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
19271 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
19274 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
19275 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
19276 and -mstring-compare-inline-limit.
19278 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19281 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
19282 for stack clash protection in a register whenever we need it to be in
19285 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19287 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
19288 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
19290 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19293 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
19295 (mlong-double-): Ditto.
19296 * config/rs6000/sysv4.opt (msdata=): Ditto.
19297 (mtls-size=): Ditto.
19299 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19301 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
19302 erroneous entries for
19303 "vector int vec_ldl (int, long int *)", and
19304 "vector unsigned int vec_ldl (int, unsigned long int *)".
19305 Add comments and entries for
19306 "vector bool char vec_ldl (int, bool char *)",
19307 "vector bool short vec_ldl (int, bool short *)",
19308 "vector bool int vec_ldl (int, bool int *)",
19309 "vector bool long long vec_ldl (int, bool long long *)",
19310 "vector pixel vec_ldl (int, pixel *)",
19311 "vector long long vec_ldl (int, long long *)",
19312 "vector unsigned long long vec_ldl (int, unsigned long long *)".
19313 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
19314 type tree bool_long_long_type_node and correct definition of
19315 bool_V2DI_type_node to make reference to this new type tree.
19316 (rs6000_mangle_type): Replace erroneous reference to
19317 bool_long_type_node with bool_long_long_type_node.
19318 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
19319 comments to emphasize sign distinctions for char and int types and
19320 replace RS6000_BTI_bool_long constant with
19321 RS6000_BTI_bool_long_long constant. Also add comment to restrict
19322 use of RS6000_BTI_pixel.
19323 (bool_long_type_node): Remove this macro definition.
19324 (bool_long_long_type_node): New macro definition
19326 2018-04-10 Jakub Jelinek <jakub@redhat.com>
19328 PR rtl-optimization/85300
19329 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
19330 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
19331 simplify_unary_operation fails.
19333 2018-04-10 Martin Liska <mliska@suse.cz>
19335 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
19336 cgraph_edge and ipa_ref.
19338 2018-04-10 Jakub Jelinek <jakub@redhat.com>
19342 * config/i386/sse.md
19343 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
19344 computation of the VEC_MERGE selector from mask.
19345 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
19346 Fix decoding of the VEC_MERGE selector into mask.
19348 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
19350 PR tree-optimization/85286
19351 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
19353 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
19355 * final.c (final_1): Set insn_last_address as well as
19356 insn_current_address.
19358 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19361 * explow.c (emit_stack_probe): Call validize_mem on memory location
19362 before passing it to gen_probe_stack. Create address operand and
19363 legitimize it for the probe_stack_address case.
19365 2018-04-09 Jan Hubicka <jh@suse.cz>
19368 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
19369 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
19370 * tree.c (free_lang_data_in_type): Fix handling of binfos;
19372 (free_lang_data): Rebuild type inheritance graph.
19374 2018-04-09 Martin Sebor <msebor@redhat.com>
19376 * invoke.texi (-finline-small-functions): Mention other optimization
19378 (-findirect-inlining, -fpartial-inlining): Same.
19379 (-finline-functions-called-once): Same.
19380 (-freorder-blocks-and-partition): Same.
19382 2018-04-09 Jan Hubicka <jh@suse.cz>
19385 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
19386 jumps; choose last target that matches the criteria (i.e.
19387 no partition changes for non-crossing jumps).
19388 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
19389 support for redirecting crossing jumps to non-crossing.
19391 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
19393 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
19394 also for naked functions.
19396 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
19398 * config/arc/arc.md (add_shift): New pattern.
19399 (add_shift2): Likewise.
19400 (sub_shift): Likewise.
19401 (sub_shift_cmp0_noout): Likewise.
19402 (compare_si_ashiftsi): Likewise.
19403 (xbfu_cmp0_noout): New combine pattern.
19404 (xbfu_cmp0"): Likewise.
19405 (movsi_set_cc_insn): Place the predicable variant first.
19406 (commutative_binary_cmp0_noout): Remove clobber.
19407 (commutative_binary_cmp0): New pattern.
19408 (noncommutative_binary_cmp0): Likewise.
19409 (noncommutative_binary_cmp0_noout): Likewise.
19410 (noncommutative_binary_comparison_result_used): Removed.
19411 (rsub_cmp0): New pattern.
19412 (rsub_cmp0_noout): Likewise.
19413 (extzvsi): Changed, keep only meaningful variants.
19414 (SQH, SEZ): New iterators.
19415 (SQH_postfix): New mode attribute.
19416 (SEZ_prefix): New code attribute.
19417 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
19418 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
19419 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
19420 of numerical value.
19421 (noncommutative_operator): Check the availability of barrel
19424 2018-04-09 Richard Biener <rguenther@suse.de>
19426 PR tree-optimization/85284
19427 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
19428 Only use the niter constraining form of simple_iv when the exit
19429 is always executed.
19431 2018-04-09 Tom de Vries <tom@codesourcery.com>
19434 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
19435 (define_expand "*memory_barrier"): New define_expand.
19436 (define_insn "memory_barrier"): New insn.
19438 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19440 PR rtl-optimization/80463
19441 PR rtl-optimization/83972
19442 PR rtl-optimization/83480
19444 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
19445 correct producer for the insn.
19446 (tidy_control_flow): Fixup seqnos in case of debug insns.
19448 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19450 PR rtl-optimization/83913
19452 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
19453 different sched-times when merging exprs.
19455 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19457 PR rtl-optimization/83962
19459 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
19460 tidy_fallthru_edge and tidy_control_flow.
19462 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19464 PR rtl-optimization/83530
19466 * sel-sched.c (force_next_insn): New global variable.
19467 (remove_insn_for_debug): When force_next_insn is true, also leave only
19468 next insn in the ready list.
19469 (sel_sched_region): When the region wasn't scheduled, make another pass
19470 over it with force_next_insn set to 1.
19472 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
19474 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
19476 * config/nds32/constants.md (unspec_volatile_element): Add enum values
19477 for interrupt control.
19478 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
19479 functions for interrupt control.
19480 * config/nds32/nds32-intrinsic.md: Likewise.
19481 * config/nds32/nds32_intrinsic.h: Likewise.
19482 * config/nds32/nds32.h (nds32_builtins): Likewise.
19484 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
19486 * config/nds32/nds32.c (nds32_init_machine_status,
19487 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
19488 strict_aligned_p field.
19489 (nds32_expand_to_rtl_hook): New function.
19490 (TARGET_EXPAND_TO_RTL_HOOK): Define.
19491 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
19493 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19494 Chung-Ju Wu <jasonwucj@gmail.com>
19496 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
19497 * config/nds32/nds32-n7.md: New file.
19498 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
19499 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
19501 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
19502 * config/nds32/nds32.md (pipeline_model): Add n7.
19503 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
19504 * config/nds32/pipelines.md: Include n7 settings.
19506 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19507 Chung-Ju Wu <jasonwucj@gmail.com>
19509 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
19510 * config/nds32/nds32-e8.md: New file.
19511 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
19512 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
19514 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
19515 * config/nds32/nds32.md (pipeline_model): Add e8.
19516 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
19517 * config/nds32/pipelines.md: Include e8 settings.
19519 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19520 Chung-Ju Wu <jasonwucj@gmail.com>
19522 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
19523 * config/nds32/nds32-n8.md: New file.
19524 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
19525 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
19527 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
19528 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
19529 * config/nds32/nds32.md (pipeline_model): Add n8.
19530 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
19531 * config/nds32/pipelines.md: Include n8 settings.
19533 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19534 Chung-Ju Wu <jasonwucj@gmail.com>
19536 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
19537 * config/nds32/nds32-n9-2r1w.md: New file.
19538 * config/nds32/nds32-n9-3r2w.md: New file.
19539 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
19540 nds32_register_ports): New or modify for cpu n9.
19541 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
19543 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
19544 * config/nds32/nds32-utils.c: New file.
19545 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
19546 TARGET_MUL_SLOW): Define.
19547 * config/nds32/nds32.md (pipeline_model): New attribute.
19548 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
19549 New options that support cpu n9.
19550 * config/nds32/pipelines.md: Include n9 settings.
19551 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
19553 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
19555 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
19556 information if necessary.
19557 (output_cond_branch_compare_zero): Likewise.
19558 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
19559 (nds32_target_alignment): Refine for alignment.
19560 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
19561 (FUNCTION_BOUNDARY): Modify.
19562 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
19564 * config/nds32/nds32.opt (malways-align, malign-functions): New.
19566 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
19568 * config/nds32/constants.md (unspec_volatile_element): Add values for
19569 TLB operation and data prefetch.
19570 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
19571 functions for TLB operation and data prefetch.
19572 * config/nds32/nds32-intrinsic.md: Likewise.
19573 * config/nds32/nds32_intrinsic.h: Likewise.
19574 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
19575 (nds32_print_operand): Likewise.
19576 * config/nds32/nds32.h (nds32_builtins): Likewise.
19578 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
19579 Andrew Pinski <pinsika@gcc.gnu.org>
19581 PR middle-end/82976
19582 * match.pd: Use constant_boolean_node of correct type instead of
19583 boolean_true_node or boolean_false_node for simplifying
19584 pointer comparisons to zero.
19586 2018-04-07 Jakub Jelinek <jakub@redhat.com>
19588 PR tree-optimization/80021
19589 * tree.c (verify_type_variant): Make error call in verify_variant_match
19590 translatable and remove final full stop.
19592 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19594 * config/nds32/constants.md (unspec_volatile_element): Add
19595 UNSPEC_VOLATILE_EH_RETURN.
19596 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
19597 nds32_output_stack_pop): Support dwarf exception handling process.
19598 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
19599 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
19600 exception handling process.
19601 (nds32_compute_stack_frame): Likewise.
19602 (nds32_return_addr_rtx): Likewise.
19603 (nds32_initial_elimination_offset): Likewise.
19604 (nds32_expand_prologue): Likewise.
19605 (nds32_expand_epilogue): Likewise.
19606 (nds32_dynamic_chain_address): New function.
19607 * config/nds32/nds32.h (machine_function): Add fields for dwarf
19608 exception handling.
19609 (DYNAMIC_CHAIN_ADDRESS): Define.
19610 (EH_RETURN_DATA_REGNO): Define.
19611 (EH_RETURN_STACKADJ_RTX): Define.
19612 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
19613 patterns for dwarf exception handling.
19615 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19617 * config/nds32/nds32.h: Clean up obsolete macros.
19619 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19621 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
19622 Add enum values for particular instructions.
19623 * config/nds32/nds32-intrinsic.c: Implementation of expanding
19624 particular intrinsic functions.
19625 * config/nds32/nds32-intrinsic.md: Likewise.
19626 * config/nds32/nds32_intrinsic.h: Likewise.
19627 * config/nds32/nds32.h (nds32_builtins): Likewise.
19628 * config/nds32/nds32.md (type): Add pbsad and pbsada.
19629 (btst, ave): New patterns for particular instructions.
19631 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19633 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
19634 Add enum values for atomic load/store and memory sync.
19635 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
19637 * config/nds32/nds32-intrinsic.md: Likewise.
19638 * config/nds32/nds32_intrinsic.h: Likewise.
19639 * config/nds32/nds32.h (nds32_builtins): Likewise.
19641 2018-04-07 Jakub Jelinek <jakub@redhat.com>
19643 PR tree-optimization/85257
19644 * fold-const.c (native_encode_vector): If not all elts could fit
19645 and off is -1, return 0 rather than offset.
19646 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
19647 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
19648 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
19649 adjust buffer in native_interpret_expr call.
19651 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19653 * config/nds32/constants.md (unspec_volatile_element): Add cache
19654 control enum values.
19655 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
19656 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
19657 * config/nds32/nds32.c (nds32_cctl_names): New.
19658 (nds32_print_operand): Handle cache control register names.
19659 * config/nds32/nds32.h (nds32_builtins): New enum values.
19660 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
19662 * config/nds32/nds32.md (type): Add mmu.
19663 * config/nds32/pipelines.md (simple_insn): Add mmu.
19665 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19667 * config/nds32/nds32.md (type): Remove call.
19668 * config/nds32/pipelines.md (simple_insn): Likewise.
19670 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19672 * config/nds32/constants.md (unspec_volatile_element): Add
19673 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
19674 UNSPEC_VOLATILE_FMFCFG.
19675 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
19676 description for fmfcfg and fmfcsr.
19677 (bdesc_1arg): Add fmtcsr.
19678 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
19679 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
19680 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
19681 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
19682 unspec_fmfcfg): New patterns.
19683 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
19684 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
19685 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
19686 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
19687 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
19688 __nds32__fmfcfg): Define.
19690 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19692 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
19693 intrinsic register names.
19694 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
19695 intrinsic register enum values and macros.
19697 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19699 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
19700 for load/store addressing form.
19701 (nds32_print_operand_address): Likewise.
19703 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
19706 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
19707 based on LABEL_REF. Remove useless assertion.
19708 (pic_address_needs_scratch): Fix formatting.
19709 (sparc_legitimize_pic_address): Minor tweaks.
19710 (sparc_delegitimize_address): Adjust assertion accordingly.
19711 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
19712 into symbolic_operand.
19713 (movsi_high_pic_label_ref): Likewise.
19714 (movsi_lo_sum_pic_label_ref): Likewise.
19715 (movdi_pic_label_ref): Likewise.
19716 (movdi_high_pic_label_ref): Likewise.
19717 (movdi_lo_sum_pic_label_ref): Likewise.
19719 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
19721 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
19722 custom LIB_SPEC setup.
19724 2018-04-06 Ruslan Bukin <br@bsdpad.com>
19725 Kito Cheng <kito.cheng@gmail.com>
19727 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
19728 * config/riscv/freebsd.h: New.
19730 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
19732 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
19733 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
19736 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
19737 Kito Cheng <kito.cheng@gmail.com>
19739 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
19740 nds32_output_call, nds32_symbol_binds_local_p): New functions.
19741 * config/nds32/nds32-protos.h (nds32_output_call,
19742 nds32_output_return): Declare.
19743 * config/nds32/nds32.md: Refine all the call and return patterns.
19745 2018-04-06 Jakub Jelinek <jakub@redhat.com>
19748 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
19749 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
19751 PR rtl-optimization/84872
19752 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
19753 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
19754 EDGE_CROSSING edge.
19756 2018-04-06 Tamar Christina <tamar.christina@arm.com>
19758 * expr.c (copy_blkmode_to_reg): Revert 254862.
19759 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
19761 2018-04-06 Richard Biener <rguenther@suse.de>
19763 PR middle-end/85244
19764 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
19765 after seeing a component reference with an adjacent field. Treat
19766 refs to arrays at struct end of external decls similar to
19767 refs to unconstrained commons.
19769 2018-04-06 Jakub Jelinek <jakub@redhat.com>
19772 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
19773 look through SAVE_EXPRs with non-side-effects argument. Adjust
19775 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
19778 2018-04-06 Richard Biener <rguenther@suse.de>
19780 PR middle-end/85180
19781 * alias.c (find_base_term): New wrapper around find_base_term
19782 unwinding CSELIB_VAL_PTR changes.
19783 (find_base_term): Do not restore CSELIB_VAL_PTR during the
19786 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19788 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
19790 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
19791 constant definitions.
19792 ("nop"): lr 0,0 -> nopr r0
19793 ("nop_lr0", "nop_lr1"): New insn definitions.
19795 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
19797 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
19798 NDS32_V3PUSH_AVAILABLE_P macro.
19800 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
19801 Chung-Ju Wu <jasonwucj@gmail.com>
19803 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
19804 (nds32*-*-*): Add float and fpu_config into supported_defaults.
19805 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
19806 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
19807 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
19808 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
19809 * config/nds32/constraints.md: New constraints and checking for hard
19810 float configuration.
19811 * config/nds32/iterators.md: New mode iterator and attribute for hard
19812 float configuration.
19813 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
19815 * config/nds32/nds32-fpu.md: New file.
19816 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
19817 deal with hard float code generation.
19818 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
19820 (abi_type, float_reg_number): New enum type.
19821 * config/nds32/nds32-predicates.c: New predicates for hard float.
19822 * config/nds32/nds32-protos.h: Declare functions for hard float.
19823 * config/nds32/nds32.c: Implementation for hard float configuration.
19824 * config/nds32/nds32.h: Definitions for hard float configuration.
19825 * config/nds32/nds32.md: Include hard float machine description and
19826 modify patterns for hard float configuration.
19827 * config/nds32/nds32.opt: New options for hard float configuration.
19828 * config/nds32/predicates.md: New predicates for hard float
19831 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
19833 * common/config/nds32/nds32-common.c
19834 (nds32_option_optimization_table): Enable -mreleax-hint by default.
19836 2018-04-05 Jakub Jelinek <jakub@redhat.com>
19838 PR middle-end/85195
19839 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
19840 CONSTRUCTOR_ELT (ctor, ...)->value.
19842 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
19845 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
19847 2018-04-05 Tom de Vries <tom@codesourcery.com>
19850 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
19853 2018-04-05 Shiva Chen <shiva0217@gmail.com>
19854 Kito Cheng <kito.cheng@gmail.com>
19856 * config/nds32/constraints.md (U33): Fine-tune checking condition.
19857 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
19858 * config/nds32/nds32.h (nds32_16bit_address_type): Add
19859 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
19861 2018-04-05 Shiva Chen <shiva0217@gmail.com>
19862 Kito Cheng <kito.cheng@gmail.com>
19864 * config/nds32/constraints.md (Ufe): New memory constraint.
19865 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
19866 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
19867 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
19869 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
19870 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
19872 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19874 * config/nds32/nds32.md: Use optimize_size in the condition for
19875 alu-shift instructions.
19877 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19879 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
19881 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19883 * config/nds32/nds32.md (negsi2): Refine pattern.
19885 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
19886 Chung-Ju Wu <jasonwucj@gmail.com>
19888 * config/nds32/iterators.md (shift_rotate): New code iterator.
19889 (shift): New code attribute.
19890 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
19891 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
19892 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
19893 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
19894 bit-wise operations.
19895 (andsi3, *andsi3): Ditto.
19896 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
19897 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
19898 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
19899 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
19900 nds32_ior_operand, nds32_xor_operand): New predicates.
19902 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19904 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
19905 (addsi3, subsi3): ... this.
19907 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19909 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
19911 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19913 * config/nds32/nds32.md: Adjust indention.
19915 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
19917 * config/nds32/nds32.md (feature): New attribute.
19919 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
19921 * config/nds32/nds32.md (subtype): New attribute.
19923 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19926 * config/arm/arm-builtins.c (arm_expand_builtin): Change
19927 expansion to perform a bitwise AND of the argument followed by a
19928 boolean negation of the result.
19930 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
19932 PR rtl-optimization/84878
19933 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
19934 the basic block. Assert the use reference is not artificial and that
19935 it has an associated insn.
19937 2018-04-04 Michael Matz <matz@suse.de>
19939 * builtins.c (compute_objsize): Pass correct operand
19940 to array_at_struct_end_p.
19942 2018-04-04 Richard Biener <rguenther@suse.de>
19945 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
19946 from contexts for DINFO_LEVEL_TERSE and below.
19948 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
19950 * config/nds32/nds32-doubleword.md (move_<mode>): Require
19951 resiter_operand condition.
19952 * config/nds32/nds32.md (*move<mode>): Ditto.
19954 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
19955 Monk Chiang <sh.chiang04@gmail.com>
19957 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
19959 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
19961 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
19963 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
19964 Kito Cheng <kito.cheng@gmail.com>
19966 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
19967 nds32_cond_code_str, output_cond_branch,
19968 output_cond_branch_compare_zero, nds32_expand_cbranch,
19969 nds32_expand_cstore, nds32_expand_movcc,
19970 nds32_output_cbranchsi4_equality_zero,
19971 nds32_output_cbranchsi4_equality_reg,
19972 nds32_output_cbranchsi4_equality_reg_or_const_int,
19973 nds32_output_cbranchsi4_greater_less_zero: New functions.
19974 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
19975 nds32_expand_cstore, nds32_expand_movcc,
19976 nds32_output_cbranchsi4_equality_zero,
19977 nds32_output_cbranchsi4_equality_reg,
19978 nds32_output_cbranchsi4_equality_reg_or_const_int,
19979 nds32_output_cbranchsi4_greater_less_zero): Declare.
19980 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
19981 nds32_rimm11s_operand): New predicates.
19982 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
19983 * config/nds32/nds32.md: Rewrite all the branch and conditional move
19986 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
19988 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
19989 * config/nds32/nds32.md: Ditto.
19990 * config/nds32/pipelines.md: Ditto.
19992 2018-04-04 Richard Biener <rguenther@suse.de>
19994 PR tree-optimization/85168
19995 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
19996 propagating abnormals.
19998 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
20000 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
20002 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
20003 Kito Cheng <kito.cheng@gmail.com>
20005 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
20006 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
20007 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
20008 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
20009 * config/nds32/nds32.md (sibcall_internal): New.
20010 (sibcall_register): Remove.
20011 (sibcall_immediate): Remove.
20012 (sibcall_value_internal): New.
20013 (sibcall_value_register): Remove.
20014 (sibcall_value_immediate): Remove.
20015 * config/nds32/predicates.md (nds32_general_register_operand): New.
20016 (nds32_call_address_operand): New.
20018 2018-04-03 Jakub Jelinek <jakub@redhat.com>
20020 PR rtl-optimization/85167
20021 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
20022 bb_defs if *split_p, instead preinitialize it to NULL.
20024 PR tree-optimization/85156
20025 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
20026 evaluating the argument multiple times.
20028 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
20030 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
20032 (_mm_cvtpd_ps): Likewise.
20033 (_mm_cvttpd_epi32): Likewise.
20034 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
20035 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
20036 vector, pixel, and bool following altivec.h include.
20038 2018-04-03 Martin Sebor <msebor@redhat.com>
20040 * doc/extend.texi (Common Function Attributes): Clarify.
20041 (const attribute): Likewise.
20042 (pure attribute): Likewise.
20044 2018-04-03 Jakub Jelinek <jakub@redhat.com>
20047 * config/i386/i386.c (ix86_expand_vector_set): Use
20048 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
20050 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
20052 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
20053 instructions when changing rounding bits to preserve precision bits
20054 in the x87 control word.
20056 2018-04-03 Martin Liska <mliska@suse.cz>
20058 PR tree-optimization/82491
20059 * rtl.h (strip_offset_and_add): Replace += suboffset with
20060 poly_uint64 () + suboffset.
20062 2018-03-29 Martin Liska <mliska@suse.cz>
20063 Martin Jambor <mjambor@suse.cz>
20066 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
20067 param_type is not an integral or pointer type.
20069 2018-04-03 Richard Biener <rguenther@suse.de>
20071 * sese.h (recompute_all_dominators): Remove.
20073 2018-04-02 Martin Sebor <msebor@redhat.com>
20075 * doc/invoke.texi (-Wrestrict): Fix typos.
20077 2018-04-02 Jim Wilson <jimw@sifive.com>
20079 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
20080 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
20081 (<optab>di3, <optab>si3_extend): Likewise.
20082 (<optab>si3_mask, <optab>si3_mask_1): New.
20083 (<optab>di3_mask, <optab>di3_mask_1): New.
20084 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
20085 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
20086 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
20088 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
20090 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
20093 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
20095 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
20096 (nds32_canonicalize_comparison): New function.
20098 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
20099 Kito Cheng <kito.cheng@gmail.com>
20100 Kuan-Lin Chen <kuanlinchentw@gmail.com>
20102 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
20103 * config/nds32/constants.md (unspec_volatile_element): Add
20104 UNSPEC_VOLATILE_RELAX_GROUP.
20105 * config/nds32/nds32-relax-opt.c: New file.
20106 * config/nds32/nds32-predicates.c
20107 (nds32_symbol_load_store_p): New function.
20108 * config/nds32/nds32-protos.h
20109 (nds32_symbol_load_store_p): Declare function.
20110 (make_pass_nds32_relax_opt): Declare new rtl pass function.
20111 * config/nds32/nds32.c
20112 (nds32_register_pass): New function to register pass.
20113 (nds32_register_passes): New function to register passes.
20114 * config/nds32/nds32.md (relax_group): New pattern.
20115 * config/nds32/nds32.opt (mrelax-hint): New option.
20116 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
20118 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
20120 * config/nds32/t-nds32: Modify files dependency.
20122 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
20124 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
20125 (PROFILE_HOOK): Define its implementation.
20127 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
20129 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
20130 type and 32-bit size.
20132 2018-04-01 Jakub Jelinek <jakub@redhat.com>
20134 PR middle-end/85090
20135 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
20136 (V_128_256): New mode iterator.
20137 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
20138 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
20139 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
20141 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
20142 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
20144 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
20147 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
20148 NaN inputs correctly.
20150 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
20153 * config/rs6000/vsx.md (??r): New mode attribute.
20154 (*vsx_mov<mode>_64bit): Use it.
20155 (*vsx_mov<mode>_32bit): Likewise.
20157 2018-03-30 Martin Sebor <msebor@redhat.com>
20159 PR tree-optimization/84818
20160 * builtins.c (check_access): Use warning_n.
20162 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20165 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
20167 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
20170 2018-03-30 Julia Koval <julia.koval@intel.com>
20173 * x86-tune.def (movx, partial_reg_dependency): Enable for
20176 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
20178 PR inline-asm/84985
20179 * lra-constraints.c (process_alt_operands): Move setting
20180 this_alternative_matches below.
20182 2018-03-29 Martin Liska <mliska@suse.cz>
20185 * doc/invoke.texi: Document how LTO works with debug info.
20186 Describe auto-load support of binutils. Mention 'x86-64'
20187 as valid option value of -march option.
20189 2018-03-29 Jakub Jelinek <jakub@redhat.com>
20191 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
20194 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
20195 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
20196 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
20199 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
20202 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
20203 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
20204 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
20205 for RS6000_BTM_POWERPC64.
20206 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
20207 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
20208 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
20211 (DIVDEU): Likewise.
20213 2018-03-28 Carl Love <cel@us.ibm.com>
20216 2017-09-27 Carl Love <cel@us.ibm.com>
20218 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
20219 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
20220 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
20223 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20225 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
20226 instead of __vector bool.
20227 (_mm_max_pu8): Likewise.
20228 (_mm_min_pi16): Likewise.
20230 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
20233 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
20234 (DIVWEUO): Likewise.
20235 (DIVDEO): Likewise.
20236 (DIVDEUO): Likewise.
20237 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
20238 DIVWEUO and DIVDEUO.
20239 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
20240 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
20241 (div_extend): Likewise.
20242 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
20244 (__builtin_divweuo): Likewise.
20245 (__builtin_divdeo): Likewise.
20246 (__builtin_divdeuo): Likewise.
20248 2018-03-28 Jakub Jelinek <jakub@redhat.com>
20251 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
20252 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
20254 PR tree-optimization/82004
20255 * gimple-match-head.c (optimize_pow_to_exp): New function.
20256 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
20257 Don't fold to exp if optimize_pow_to_exp is false.
20259 2018-03-28 Martin Liska <mliska@suse.cz>
20262 * calls.c (initialize_argument_information): Fix trailing space.
20263 * common.opt: Fix typo and provide better explanation for
20264 -fsanitize-coverage option.
20265 * config/i386/i386.opt: Fix typo.
20267 2018-03-28 Jakub Jelinek <jakub@redhat.com>
20268 Martin Liska <mliska@suse.cz>
20271 * gimplify.c (asan_poison_variable): Don't do the check for
20272 gimplify_omp_ctxp here.
20273 (gimplify_decl_expr): Do it here.
20274 (gimplify_target_expr): Likewise.
20276 2018-03-28 Martin Liska <mliska@suse.cz>
20279 * config/i386/i386.c (ix86_function_arg_advance): Do not call
20280 chkp_type_bounds_count if MPX is not enabled.
20282 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
20284 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
20286 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20289 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
20290 function to create the function decl for complex long double
20291 multiply and divide for -mabi=ieeelongdouble.
20292 (init_float128_ieee): Call it.
20294 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
20297 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
20298 -fcf-protection=branch -mibt.
20299 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
20301 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20304 * config/arm/arm.c (arm_valid_symbolic_address): Handle
20305 arm_word_relocations.
20307 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
20310 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
20311 extern array declarations.
20313 2018-03-27 Richard Biener <rguenther@suse.de>
20315 PR middle-end/84067
20316 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
20317 explicit single_use checks.
20319 2018-03-27 Richard Biener <rguenther@suse.de>
20321 PR tree-optimization/85082
20322 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
20325 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20327 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
20328 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
20329 Turn on fasynchronous-unwind-tables and funwind-tables.
20331 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
20334 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
20335 (*bmi_blsr_<mode>_ccz): Ditto.
20337 2018-03-26 Tom de Vries <tom@codesourcery.com>
20339 PR tree-optimization/85063
20340 * omp-general.c (offloading_function_p): New function. Factor out
20342 * omp-offload.c (pass_omp_target_link::gate): ... here.
20343 * omp-general.h (offloading_function_p): Declare.
20344 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
20345 with attribute omp declare target for offloading functions.
20347 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
20349 PR tree-optimization/84005
20350 * tree-data-ref.h (get_base_for_alignment): Declare.
20351 * tree-data-ref.c (get_base_for_alignment_1): New function.
20352 (get_base_for_alignment): Likewise.
20353 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
20354 get_base_for_alignment to find a suitable base object, instead
20355 of always using drb->base_address.
20357 2018-03-23 Jakub Jelinek <jakub@redhat.com>
20359 PR inline-asm/85022
20360 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
20361 known size by default.
20363 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
20365 PR inline-asm/85030
20366 * lra-constraints.c (process_alt_operands): Don't match BLKmode
20367 and non BLKmode operands.
20369 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20372 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
20373 Clean up attributes.
20375 2018-03-23 Richard Biener <rguenther@suse.de>
20378 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
20379 we are going to emit early debug for LTO.
20381 2018-03-23 Jakub Jelinek <jakub@redhat.com>
20383 PR inline-asm/85034
20384 * function.c (match_asm_constraints_1): Don't optimize if input
20385 doesn't satisfy general_operand predicate for output's mode.
20387 PR inline-asm/85022
20388 * alias.c (write_dependence_p): Don't require for x_canonicalized
20389 non-VOIDmode if x has VOIDmode.
20392 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
20393 just don't try to optimize it rather than assert it never happens.
20395 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
20397 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
20398 macro expansions for definition of ST_INTERNAL_<mode> and
20399 LD_INTERNAL_<mode> builtins.
20400 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
20402 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
20404 (altivec_expand_st_builtin): Likewise.
20405 (altivec_expand_builtin): Remove calls to deleted functions.
20406 (rs6000_address_for_altivec): Delete this function.
20407 * config/rs6000/vector.md: Remove expands for
20408 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
20410 2018-03-22 Sudakshina Das <sudi.das@arm.com>
20413 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
20414 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
20415 re-computing once computed.
20416 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
20417 (arm_init_machine_status): Initialize
20418 machine->static_chain_stack_bytes.
20420 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
20423 * doc/extend.texi: Add four new prototypes for vec_ld.
20424 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
20425 definitions for more logical presentation.
20426 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
20427 entries for V1TI variants of __builtin_altivec_ld builtin.
20428 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
20429 handling of V1TI variant of LVX icode pattern.
20430 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
20431 (rs6000_gimple_fold_builtin): Likewise.
20432 (altivec_init_builtins): Add code to define
20433 __builtin_altivec_lvx_v1ti function.
20435 2018-03-22 Jakub Jelinek <jakub@redhat.com>
20437 PR inline-asm/84941
20438 * function.c (match_asm_constraints_1): Don't do the optimization
20439 if input isn't a REG, SUBREG, MEM or constant.
20441 2018-03-22 Tom de Vries <tom@codesourcery.com>
20443 PR tree-optimization/84956
20444 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
20445 bb_has_abnormal_pred.
20447 2018-03-22 Jakub Jelinek <jakub@redhat.com>
20450 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
20451 DECL_INITIAL (decl) to decl at the end.
20452 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
20453 adjust the comment.
20455 2018-03-21 Joseph Myers <joseph@codesourcery.com>
20457 * doc/extend.texi (__builtin_tgmath): Document when complex
20458 integer types are treated as _Complex _Float64.
20460 2018-03-21 Tom de Vries <tom@codesourcery.com>
20462 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
20464 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20466 PR tree-optimization/84960
20467 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
20468 if it is ENTRY block, move them into single succ of ENTRY in that case.
20470 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
20472 PR tree-optimization/84811
20473 * poly-int.h (poly_span_traits): Remove the T3 parameter and
20474 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
20475 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
20476 (known_subrange_p): Update accordingly. Cast each value involved
20477 in the size comparison, rather than casting the result of the
20480 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20482 PR tree-optimization/84982
20483 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
20484 by flipping the least significant bit rather than all bits from
20485 bitpos to bitpos + bitsize - 1.
20487 2018-03-21 Nathan Sidwell <nathan@acm.org>
20489 * doc/extend.texi (Deprecated Features): Remove mention of
20490 long-deleted deprecations.
20492 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20495 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
20496 * configure: Regenerate.
20498 2018-03-21 Tom de Vries <tom@codesourcery.com>
20500 PR tree-optimization/83126
20501 * tree-parloops.c (num_phis): New function.
20502 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
20504 2018-03-21 Nathan Sidwell <nathan@acm.org>
20506 * doc/extend.texi (Deprecated Features): Update deprecated flags,
20507 mention anon-struct/union members and trailing attributes.
20509 2018-03-21 Bin Cheng <bin.cheng@arm.com>
20511 PR tree-optimization/84969
20512 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
20513 builtin memset partitions if they set different rhs values.
20515 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20517 PR rtl-optimization/84989
20518 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
20519 VEC_DUPLICATE with scalar result mode.
20521 2018-03-21 Martin Liska <mliska@suse.cz>
20524 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
20525 not intended return statement.
20527 2018-03-21 Martin Liska <mliska@suse.cz>
20530 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
20531 (chkp_find_bound_slots_1): Limit number of iterations.
20533 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
20536 * Minor grammar fixes for x86 options.
20538 2018-03-20 Jakub Jelinek <jakub@redhat.com>
20541 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
20542 holding REG_CFA_RESTORE notes, instead turn them into a USE.
20544 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
20547 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
20548 (altivec_lvx_<mode>_1op): Likewise.
20549 (altivec_stvx_<mode>_2op): Likewise.
20550 (altivec_stvx_<mode>_1op): Likewise.
20551 (altivec_lvx_<VM2:mode>): New define_expand.
20552 (altivec_stvx_<VM2:mode>): Likewise.
20553 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
20554 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
20555 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
20556 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
20557 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
20558 (rs6000_gen_lvx): Likewise.
20559 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
20560 (altivec_expand_stv_builtin): Likewise.
20561 (altivec_expand_builtin): Likewise.
20562 * config/rs6000/vector.md: Likewise.
20564 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20567 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
20570 2018-03-20 Richard Biener <rguenther@suse.de>
20573 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
20574 sign-conversions as zero, fall back to standard scalar_stmt
20577 2018-03-20 Martin Liska <mliska@suse.cz>
20580 * predict.c (rebuild_frequencies): Handle case when we have
20581 PROFILE_ABSENT, but flag_guess_branch_prob is false.
20583 2018-03-20 Jakub Jelinek <jakub@redhat.com>
20586 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
20587 flag_section_anchors.
20588 * varasm.c (use_blocks_for_decl_p): Remove hack for
20589 dw2_force_const_mem.
20592 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
20594 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
20595 be created, use lowpart_subreg of operands[0] rather than operands[0]
20597 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
20598 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
20599 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
20600 and n constraint instead of aarch64_shift_imm_di and Usd.
20601 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
20602 (*aarch64_<optab>_reg_minus<mode>3): ... this.
20604 2018-03-20 Sudakshina Das <sudi.das@arm.com>
20607 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
20608 to favor GPR over NEON registers.
20609 (<shift>di3_neon): Likewise.
20611 2018-03-20 Tom de Vries <tom@codesourcery.com>
20614 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
20615 (nvptx_process_pars): Emit bar.sync asap and alap.
20617 2018-03-20 Tom de Vries <tom@codesourcery.com>
20620 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
20621 seen_label if seen_label is already set.
20623 2018-03-20 Jakub Jelinek <jakub@redhat.com>
20626 * config/i386/i386.c (fold_builtin_cpu): For features above 31
20627 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
20628 Use 1U instead of 1. Formatting fixes.
20631 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
20632 instead of TREE_TYPE (s1) for the return value.
20634 2018-03-19 Jakub Jelinek <jakub@redhat.com>
20636 PR tree-optimization/84946
20637 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
20638 bitsize + bitsize in poly_uint64 rather than poly_int64.
20641 * dwarf2asm.c: Include fold-const.c.
20642 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
20643 of decl rather than decl itself.
20645 PR rtl-optimization/84643
20646 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
20648 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
20651 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
20652 calling assemble_variable.
20654 2018-03-19 Sudakshina Das <sudi.das@arm.com>
20657 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
20658 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
20660 2018-03-19 Jim Wilson <jimw@sifive.com>
20663 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
20664 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
20665 (riscv_first_stack_step): Likewise.
20666 (riscv_option_override): Use STACK_BOUNDARY instead of
20667 MIN_STACK_BOUNDARY.
20668 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
20669 MIN_STACK_BOUNDARY.
20670 (BIGGEST_ALIGNMENT): Set to 128.
20671 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
20672 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
20675 2018-03-19 Richard Biener <rguenther@suse.de>
20677 PR tree-optimization/84933
20678 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
20679 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
20681 2018-03-19 Richard Biener <rguenther@suse.de>
20683 PR tree-optimization/84859
20684 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
20685 (cond_if_else_store_replacement): Perform sinking operation on
20686 single-store BBs regardless of MAX_STORES_TO_SINK setting.
20687 Generalize what a BB with a single eligible store is.
20689 2018-03-19 Richard Biener <rguenther@suse.de>
20691 PR tree-optimization/84929
20692 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
20693 chrec_is_positive against non-chrec arg.
20695 2018-03-19 Tamar Christina <tamar.christina@arm.com>
20698 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
20700 2018-03-18 Martin Liska <mliska@suse.cz>
20702 PR rtl-optimization/84635
20703 * regrename.c (build_def_use): Use matches_mode only when
20706 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
20708 PR tree-optimization/84913
20709 * tree-vect-loop.c (vectorizable_reduction): Don't try to
20710 vectorize chains of COND_EXPRs.
20712 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
20714 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
20716 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
20718 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
20720 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
20722 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
20724 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
20725 Kito Cheng <kito.cheng@gmail.com>
20727 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
20728 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
20729 (nds32_adjust_reg_alloc_order): New function.
20730 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
20732 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
20734 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
20735 nds32_print_operand, nds32_print_operand_address): Use
20736 HOST_WIDE_INT_PRINT_DEC instead.
20738 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
20740 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
20742 2018-03-17 Jakub Jelinek <jakub@redhat.com>
20745 * config/i386/i386.c (initial_ix86_tune_features,
20746 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
20747 unsigned long long.
20748 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
20749 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
20750 rather than 1u << ix86_tune. Formatting fix.
20751 (ix86_option_override_internal): Change ix86_arch_mask from
20752 unsigned int to unsigned HOST_WIDE_INT, initialize to
20753 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
20754 (ix86_function_specific_restore): Likewise.
20756 2018-03-16 Jakub Jelinek <jakub@redhat.com>
20759 * postreload.c (reload_combine_recognize_pattern): Perform
20760 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
20761 truncate_int_for_mode the result for the destination's mode.
20764 * hsa-gen.c (mem_type_for_type): Fix comment typo.
20765 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
20767 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
20770 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
20773 * lra-assigns.c (lra_split_hard_reg_for): Don't use
20774 regno_allocno_class_array and sorted_pseudos.
20775 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
20776 insns where regno is used.
20778 2018-03-16 Martin Liska <mliska@suse.cz>
20781 * multiple_target.c (create_dispatcher_calls): Redirect
20782 reference in the symbol table.
20784 2018-03-16 Martin Liska <mliska@suse.cz>
20787 * multiple_target.c (create_dispatcher_calls): Redirect also
20790 2018-03-16 Jakub Jelinek <jakub@redhat.com>
20794 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
20795 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
20796 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
20798 2018-03-16 Julia Koval <julia.koval@intel.com>
20800 * doc/invoke.texi (Skylake Server): Add CLWB.
20801 Cannonlake): Remove CLWB.
20803 2018-03-16 Jakub Jelinek <jakub@redhat.com>
20805 PR tree-optimization/84841
20806 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
20808 (FLOAT_ONE_CONST_TYPE): Define.
20809 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
20810 (sort_by_operand_rank): Put entries with higher constant_type last
20811 rather than first to match comments.
20813 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
20815 * config/nios2/nios2.md (movsi_internal): Fix thinko in
20818 2018-03-15 Jakub Jelinek <jakub@redhat.com>
20821 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
20822 check and use address of target always.
20824 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
20827 * config/i386/i386.c (indirect_thunk_needed): Update comments.
20828 (indirect_thunk_bnd_needed): Likewise.
20829 (indirect_thunks_used): Likewise.
20830 (indirect_thunks_bnd_used): Likewise.
20831 (indirect_return_needed): New.
20832 (indirect_return_bnd_needed): Likewise.
20833 (output_indirect_thunk_function): Add a bool argument for
20835 (output_indirect_thunk_function): Don't generate alias for
20836 function return thunk.
20837 (ix86_code_end): Call output_indirect_thunk_function to generate
20838 function return thunks.
20839 (ix86_output_function_return): Set indirect_return_bnd_needed
20840 and indirect_return_needed instead of indirect_thunk_bnd_needed
20841 and indirect_thunk_needed.
20843 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
20845 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
20846 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
20847 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
20849 2018-03-15 David Malcolm <dmalcolm@redhat.com>
20850 Paul Hua <paul.hua.gm@gmail.com>
20853 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
20855 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
20857 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
20858 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
20859 resp. SFmode cases.
20861 2018-03-15 Tamar Christina <tamar.christina@arm.com>
20864 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
20865 instead of GET_MODE_SIZE when comparing Units.
20867 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
20870 * varasm.c (hash_section): Return an unchangeble hash value
20871 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
20872 Return !aarch64_can_use_per_function_literal_pools_p ().
20874 2018-03-15 Jakub Jelinek <jakub@redhat.com>
20877 * optabs.c (emit_conditional_move): Pass address of cmode's copy
20878 rather than address of cmode as last argument to prepare_cmp_insn.
20880 2018-03-15 Julia Koval <julia.koval@intel.com>
20882 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
20883 F_AVX512VNNI, F_AVX512BITALG): New.
20885 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
20888 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
20889 insn for floating-point loads and stores.
20891 2018-03-14 Carl Love <cel@us.ibm.com>
20893 * config/rs6000/rs6000-c.c: Add macro definitions for
20894 ALTIVEC_BUILTIN_VEC_PERMXOR.
20895 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
20896 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
20897 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
20898 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
20900 * config/doc/extend.texi: Add prototypes for vec_permxor.
20902 2018-03-14 David Malcolm <dmalcolm@redhat.com>
20905 * diagnostic-show-locus.c (class layout_point): Convert m_line
20906 from int to linenum_type.
20907 (line_span::comparator): Use linenum "compare" function when
20908 comparing line numbers.
20909 (test_line_span): New function.
20910 (layout_range::contains_point): Convert param "row" from int to
20912 (layout_range::intersects_line_p): Likewise.
20913 (layout::will_show_line_p): Likewise.
20914 (layout::print_source_line): Likewise.
20915 (layout::should_print_annotation_line_p): Likewise.
20916 (layout::print_annotation_line): Likewise.
20917 (layout::print_leading_fixits): Likewise.
20918 (layout::annotation_line_showed_range_p): Likewise.
20919 (struct line_corrections): Likewise for field m_row.
20920 (line_corrections::line_corrections): Likewise for param "row".
20921 (layout::print_trailing_fixits): Likewise.
20922 (layout::get_state_at_point): Likewise.
20923 (layout::get_x_bound_for_row): Likewise.
20924 (layout::print_line): Likewise.
20925 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
20926 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
20927 * input.c (selftest::test_linenum_comparisons): New function.
20928 (selftest::input_c_tests): Call it.
20929 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
20930 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
20931 * selftest.h (ASSERT_GT): New macro.
20932 (ASSERT_GT_AT): New macro.
20933 (ASSERT_LT): New macro.
20934 (ASSERT_LT_AT): New macro.
20936 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
20938 PR rtl-optimization/84780
20939 * combine.c (distribute_links): Don't make a link based on pc_rtx.
20941 2018-03-14 Martin Liska <mliska@suse.cz>
20943 * tree.c (record_node_allocation_statistics): Use
20944 get_stats_node_kind.
20945 (get_stats_node_kind): New function extracted from
20946 record_node_allocation_statistics.
20947 (free_node): Use get_stats_node_kind.
20949 2018-03-14 Richard Biener <rguenther@suse.de>
20951 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
20952 that the value-set of ANTIC_IN doesn't grow.
20955 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
20957 (BB_VISITED_WITH_VISITED_SUCCS): New define.
20958 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
20960 2018-03-14 Julia Koval <julia.koval@intel.com>
20962 * config.gcc (icelake-client, icelake-server): New.
20964 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
20965 (initial_ix86_arch_features): Ditto.
20966 (PTA_SKYLAKE): Add SGX.
20967 (PTA_ICELAKE): Remove.
20968 (PTA_ICELAKE_CLIENT): New.
20969 (PTA_ICELAKE_SERVER): New.
20970 (ix86_option_override_internal): Split up icelake on icelake client and
20972 (get_builtin_code_for_version): Ditto.
20973 (fold_builtin_cpu): Ditto.
20974 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
20975 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
20976 * config/i386/i386.h (processor_type): Ditto.
20977 * doc/invoke.texi: Ditto.
20979 2018-03-14 Jakub Jelinek <jakub@redhat.com>
20982 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
20983 INTEGER_CST offset, add it together with bitpos / 8 and
20984 sign extend based on POINTER_SIZE.
20988 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20991 * config/i386/constraints.md (Yc): New register constraint.
20992 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
20993 Use Yc constraint for alternative 2 of operand 0. Remove
20994 preferred_for_speed attribute.
20996 2018-03-14 Richard Biener <rguenther@suse.de>
20998 PR tree-optimization/84830
20999 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
21000 with the old one to avoid oscillations.
21002 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
21005 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
21007 (assign_by_spills): Return a flag of reload assignment failure.
21008 Do not process the reload assignment failures. Do not spill other
21009 reload pseudos if they has the same reg class. Update n if
21011 (lra_assign): Add a return arg. Set up from the result of
21012 assign_by_spills call.
21013 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
21014 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
21015 usage_insns if it is not NULL.
21016 (spill_hard_reg_in_range): New function.
21017 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
21018 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
21019 function prototypes.
21020 (lra_assign): Change prototype.
21021 * lra.c (lra): Add code to deal with fails by splitting hard reg
21024 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
21026 * config/riscv/riscv.opt (mrelax): New option.
21027 * config/riscv/riscv.c (riscv_file_start): Emit ".option
21028 "norelax" when riscv_mrelax is disabled.
21029 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
21031 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21034 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
21035 reassociation for int modes.
21037 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
21039 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
21040 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
21042 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
21043 * config/aarch64/aarch64-sve.md
21044 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
21045 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
21046 (*extend<mode><Vwide>2): Rename to...
21047 (aarch64_sve_extend<mode><Vwide>2): ...this.
21048 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
21049 renaming the old pattern to...
21050 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
21052 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
21053 define_expand, renaming the old pattern to...
21054 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
21055 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
21056 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
21057 account when deciding which SVE instruction the optab should use.
21058 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
21060 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
21062 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
21063 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
21064 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
21065 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
21066 (tlsdesc_small_<mode>): Turn a define_expand and use
21067 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
21068 (tlsdesc_small_advsimd_<mode>): ...this.
21069 (tlsdesc_small_sve_<mode>): New pattern.
21071 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
21073 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
21074 (UNSPEC_UMUL_HIGHPART): New constants.
21075 (MUL_HIGHPART): New int iteraor.
21076 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
21077 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
21079 (*<su>mul<mode>3_highpart): New define_insn.
21081 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
21084 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
21087 2018-03-13 Martin Liska <mliska@suse.cz>
21090 * (sem_item_optimizer::sem_item_optimizer): Initialize new
21092 (sem_item_optimizer::~sem_item_optimizer): Release it.
21093 (sem_item_optimizer::merge_classes): Register variable aliases.
21094 (sem_item_optimizer::fixup_pt_set): New function.
21095 (sem_item_optimizer::fixup_points_to_sets): Likewise.
21096 * ipa-icf.h: Declare new variables and functions.
21098 2018-03-13 Jakub Jelinek <jakub@redhat.com>
21100 PR middle-end/84834
21101 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
21102 integer_pow2p@2 and test integer_pow2p in condition.
21103 (A < 0 ? C : 0): Similarly for @1.
21105 PR middle-end/84831
21106 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
21107 characters starting at p contain '\0' character, don't look beyond
21111 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
21112 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
21115 * reg-stack.c (change_stack): Change update_end var from int to
21116 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
21117 also call set_block_for_insn on the newly added insns and rescan.
21120 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
21121 on the last operand.
21124 * tree.c (stabilize_reference_1): Return save_expr (e) for
21125 STATEMENT_LIST even if it doesn't have side-effects.
21127 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
21129 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
21131 2018-03-12 Renlin Li <renlin.li@arm.com>
21133 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
21134 aarch64_output_scalar_simd_mov_immediate.
21136 2018-03-12 Martin Sebor <msebor@redhat.com>
21138 PR tree-optimization/83456
21139 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
21140 for perfectly overlapping calls to memcpy.
21141 (gimple_fold_builtin_memory_chk): Same.
21142 (gimple_fold_builtin_strcpy): Handle no-warning.
21143 (gimple_fold_builtin_stxcpy_chk): Same.
21144 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
21146 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
21148 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
21149 parameter. Use it for SFmode.
21150 (rs6000_function_arg_advance_1): Adjust.
21151 (rs6000_function_arg): Adjust.
21152 (rs6000_gimplify_va_arg): Pass false for that new parameter.
21154 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
21156 PR rtl-optimization/84169
21157 PR rtl-optimization/84780
21158 * combine.c (can_combine_p): Check for a 2-insn combination whether
21159 the destination register is used between the two insns, too.
21161 2018-03-12 Richard Biener <rguenther@suse.de>
21163 PR tree-optimization/84803
21164 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
21165 for refs DR analysis didn't process.
21167 2018-03-12 Richard Biener <rguenther@suse.de>
21169 PR tree-optimization/84777
21170 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
21171 force-vectorize loops ignore whether we are optimizing for size.
21173 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
21175 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
21176 (TARGET_MD_ASM_ADJUST): Define.
21178 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
21179 Kito Cheng <kito.cheng@gmail.com>
21180 Chung-Ju Wu <jasonwucj@gmail.com>
21182 * config/nds32/nds32.c (nds32_compute_stack_frame,
21183 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
21184 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
21185 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
21186 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
21187 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
21188 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
21189 * config/nds32/nds32.md (prologue, epilogue): Use macro
21190 NDS32_V3PUSH_AVAILABLE_P to do checking.
21192 2018-03-11 Jakub Jelinek <jakub@redhat.com>
21195 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
21196 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
21197 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
21198 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
21199 addition of most attributes on !orig_type_die or the attribute not
21200 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
21202 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21203 Chung-Ju Wu <jasonwucj@gmail.com>
21205 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
21206 __NDS32_VH__ macro.
21207 * config/nds32/nds32.opt (mvh): New option.
21209 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21210 Chung-Ju Wu <jasonwucj@gmail.com>
21212 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
21214 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
21215 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
21218 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21219 Chung-Ju Wu <jasonwucj@gmail.com>
21221 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
21223 * config/nds32/nds32-multiple.md (strlensi): New pattern.
21224 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
21226 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
21227 Kito Cheng <kito.cheng@gmail.com>
21228 Chung-Ju Wu <jasonwucj@gmail.com>
21230 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
21231 UNSPEC_FFMISM and UNSPEC_FLMISM.
21232 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
21233 for ffb, ffmism and flmism.
21234 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
21235 (unspec_ffmism): Ditto.
21236 (unspec_flmism): Ditto.
21237 (nds32_expand_builtin_impl): Check if string extension is available.
21238 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
21239 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
21241 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
21244 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
21247 * lra-assigns.c (assign_by_spills): Return a flag of reload
21248 assignment failure. Do not process the reload assignment
21249 failures. Do not spill other reload pseudos if they has the same
21251 (lra_assign): Add a return arg. Set up from the result of
21252 assign_by_spills call.
21253 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
21254 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
21255 usage_insns if it is not NULL.
21256 (spill_hard_reg_in_range): New function.
21257 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
21258 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
21259 function prototypes.
21260 (lra_assign): Change prototype.
21261 * lra.c (lra): Add code to deal with fails by splitting hard reg
21264 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
21267 * config/i386/i386.opt: Replace Enforcment with Enforcement.
21269 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
21272 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
21273 (dw_val_node): Add val_symbolic_view.
21274 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
21275 (symview_upper_bound): New.
21276 (new_line_info_table): Initialize symviews_since_reset.
21277 (dwarf2out_source_line): Count symviews_since_reset and set
21278 symview_upper_bound.
21279 (dw_val_equal_p): Handle symview.
21280 (add_AT_symview): New.
21281 (print_dw_val): Handle symview.
21282 (attr_checksum, attr_checksum_ordered): Likewise.
21283 (same_dw_val_p, size_of_die): Likewise.
21284 (value_format, output_die): Likewise.
21285 (add_high_low_attributes): Use add_AT_symview for entry_view.
21286 (dwarf2out_finish): Reset symview_upper_bound, clear
21289 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
21292 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
21293 Add strict argument and use it.
21294 (rs6000_split_multireg_move): Update for new strict argument.
21295 (mem_operand_gpr): Disallow all non-offsettable addresses.
21296 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
21298 2018-03-09 Jakub Jelinek <jakub@redhat.com>
21301 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
21302 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
21303 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
21306 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
21307 decl, use remap_type if we want to use the type.
21309 2018-03-09 Martin Sebor <msebor@redhat.com>
21311 PR tree-optimization/84526
21312 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
21314 (builtin_access::generic_overlap): Be prepared to handle non-array
21317 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
21319 PR rtl-optimization/84682
21320 * lra-constraints.c (process_address_1): Check is_address flag
21321 for address constraints.
21322 (process_alt_operands): Likewise.
21323 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
21324 preprocess_constraints.
21325 * recog.h (preprocess_constraints): Add oploc parameter.
21327 * recog.c (preprocess_constraints): Test address_operand for
21328 CT_ADDRESS constraints.
21330 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
21333 * lra-assigns.c (assign_by_spills): Return a flag of reload
21334 assignment failure. Do not process the reload assignment
21335 failures. Do not spill other reload pseudos if they has the same
21337 (lra_assign): Add a return arg. Set up from the result of
21338 assign_by_spills call.
21339 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
21340 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
21341 usage_insns if it is not NULL.
21342 (spill_hard_reg_in_range): New function.
21343 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
21344 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
21345 function prototypes.
21346 (lra_assign): Change prototype.
21347 * lra.c (lra): Add code to deal with fails by splitting hard reg
21350 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21353 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
21354 Accept complain bool parameter. Only emit errors if it is true.
21355 (arm_parse_cpu_option_name): Likewise.
21356 (arm_target_thumb_only): Adjust callers of the above.
21357 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
21358 prototype to take a default true bool parameter.
21359 (arm_parse_arch_option_name): Likewise.
21361 2018-03-09 David Malcolm <dmalcolm@redhat.com>
21362 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
21366 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
21367 * configure: Regenerate.
21368 * configure.ac ("linker --version-script option"): New.
21369 ("linker soname option"): New.
21371 2018-03-09 Richard Biener <rguenther@suse.de>
21373 PR tree-optimization/84775
21374 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
21375 immediate uses of predicate stmts and mark them modified.
21378 PR tree-optimization/84178
21379 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
21381 (version_loop_for_if_conversion): Delay update_ssa call.
21382 (tree_if_conversion): Delay update_ssa until after predicate
21385 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
21388 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
21389 when the function accesses prior frames.
21391 2018-03-08 Jakub Jelinek <jakub@redhat.com>
21394 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
21395 gen_llsym, otherwise call maybe_gen_llsym.
21397 PR inline-asm/84742
21398 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
21399 has ',' character inside of it.
21401 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21404 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
21405 as clobbering CC_REGNUM.
21407 2018-03-08 Richard Biener <rguenther@suse.de>
21409 PR middle-end/84552
21410 * tree-scalar-evolution.c: Include tree-into-ssa.h.
21411 (follow_copies_to_constant): Do not follow SSA names registered
21414 2018-03-08 Richard Biener <rguenther@suse.de>
21416 PR tree-optimization/84178
21417 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
21419 (version_loop_for_if_conversion): Delay update_ssa call.
21420 (tree_if_conversion): Delay update_ssa until after predicate
21423 2018-03-08 David Malcolm <dmalcolm@redhat.com>
21425 PR tree-optimization/84178
21426 * tree-if-conv.c (release_bb_predicate): Remove the
21427 the assertion that the stmts have NULL use_ops.
21428 Discard the statements, asserting that they haven't
21429 yet been added to a BB.
21431 2018-03-08 Richard Biener <rguenther@suse.de>
21433 PR tree-optimization/84746
21434 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
21435 (phi_translate): Pass in destination ANTIC_OUT set.
21436 (phi_translate_1): Likewise. For a simplified result lookup
21437 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
21438 (phi_translate_set): Adjust.
21439 (do_pre_regular_insertion): Likewise.
21440 (do_pre_partial_partial_insertion): Likewise.
21442 2018-03-08 Martin Liska <mliska@suse.cz>
21444 PR gcov-profile/84735
21445 * doc/gcov.texi: Document usage of profile files.
21446 * gcov-io.h: Document changes in the format.
21448 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
21452 * dwarf2out.c (struct dw_line_info_table): Update comments for
21454 (FORCE_RESET_NEXT_VIEW): New.
21455 (FORCE_RESETTING_VIEW_P): New.
21456 (RESETTING_VIEW_P): Check for -1 too.
21457 (ZERO_VIEW_P): Likewise.
21458 (new_line_info_table): Force-reset next view.
21459 (dwarf2out_begin_function): Likewise.
21460 (dwarf2out_source_line): Simplify zero_view_p initialization.
21461 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
21462 view directly. Omit view when omitting .loc at line 0.
21464 2018-03-08 Jakub Jelinek <jakub@redhat.com>
21466 PR tree-optimization/84740
21467 * tree-switch-conversion.c (process_switch): Call build_constructors
21468 only if info.phi_count is non-zero.
21470 PR tree-optimization/84739
21471 * tree-tailcall.c (find_tail_calls): Check call arguments against
21472 DECL_ARGUMENTS (current_function_decl) rather than
21473 DECL_ARGUMENTS (func) when checking for tail recursion.
21475 2018-03-07 Jakub Jelinek <jakub@redhat.com>
21477 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
21478 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
21479 Volker Reichelt's entry and add entries for people that perform
21480 GCC fuzzy testing and report numerous bugs.
21482 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21485 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
21486 readonly data in sdata, if that is disabled.
21487 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
21488 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
21489 -mreadonly-in-sdata option.
21491 2018-03-07 Martin Sebor <msebor@redhat.com>
21493 PR tree-optimization/84468
21494 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
21495 basic block when looking for nul assignment.
21497 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
21500 * except.h (output_function_exception_table): Adjust prototype.
21501 * except.c (output_function_exception_table): Remove FNNAME parameter
21502 and add SECTION parameter. Ouput one part of the table at a time.
21503 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
21504 the first part of the exception table and emit unwind directives.
21505 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
21506 (i386_pe_seh_cold_init): Likewise.
21507 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
21508 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
21509 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
21510 (ix86_output_call_insn): Emit a nop in one more case for SEH.
21511 * config/i386/winnt.c: Include except.h.
21512 (struct seh_frame_state): Add reg_offset, after_prologue and
21513 in_cold_section fields.
21514 (i386_pe_seh_end_prologue): Set seh->after_prologue.
21515 (i386_pe_seh_cold_init): New function.
21516 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
21517 to seh->in_cold_section.
21518 (seh_emit_push): Record the offset of the push.
21519 (seh_emit_save): Record the offet of the save.
21520 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
21521 Test seh->after_prologue to disregard the epilogue.
21522 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
21523 (i386_pe_end_cold_function): New function.
21525 2018-03-07 Jakub Jelinek <jakub@redhat.com>
21528 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
21529 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
21532 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
21534 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
21535 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
21537 PR middle-end/84723
21538 * multiple_target.c: Include tree-inline.h and intl.h.
21539 (expand_target_clones): Diagnose and fail if node->definition and
21540 !tree_versionable_function_p (node->decl).
21542 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
21544 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
21546 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
21547 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
21548 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
21550 2018-03-06 Jakub Jelinek <jakub@redhat.com>
21553 * combine.c (try_combine): Use reg_or_subregno instead of handling
21554 just paradoxical SUBREGs and REGs.
21556 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
21558 * config/arc/arc.c (arc_finalize_pic): Remove function.
21559 (arc_must_save_register): We use single base PIC register, remove
21560 checks to save/restore the PIC register.
21561 (arc_expand_prologue): Likewise.
21562 * config/arc/arc-protos.h (arc_set_default_type_attributes):
21564 (arc_verify_short): Likewise.
21565 (arc_attr_type): Likewise.
21566 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
21567 (walk_stores): Likewise.
21568 (arc_address_cost): Make it static.
21569 (arc_verify_short): Likewise.
21570 (branch_dest): Likewise.
21571 (arc_attr_type): Likewise.
21572 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
21573 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
21574 (arc_final_prescan_insn): Remove inserting the nops due to
21575 hardware hazards. It is done in reorg step.
21576 (insn_length_variant_t): Remove.
21577 (insn_length_parameters_t): Likewise.
21578 (arc_insn_length_parameters): Likewise.
21579 (arc_get_insn_variants): Likewise.
21580 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
21582 2018-03-06 Jakub Jelinek <jakub@redhat.com>
21584 PR inline-asm/84683
21585 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
21588 PR tree-optimization/84687
21589 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
21591 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
21593 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21595 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
21596 Rename to ppc_speculation_barrier.
21597 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
21598 __builtin_ppc_speculation_barrier.
21600 2018-03-05 Jakub Jelinek <jakub@redhat.com>
21603 * combine.c (combine_simplify_rtx): Don't try to simplify if
21604 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
21607 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
21609 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
21610 to 32 bytes when compiling for POWER9.
21612 2018-03-05 Jakub Jelinek <jakub@redhat.com>
21615 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
21616 regparm >= 3 with no arg reg available also for calls with
21617 flag_force_indirect_call. Pass decl to ix86_function_regparm.
21620 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
21622 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
21624 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
21627 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
21629 2018-03-05 Richard Biener <rguenther@suse.de>
21631 PR tree-optimization/84486
21632 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
21633 When inserting a __builtin_assume_aligned call set the LHS
21634 SSA name alignment info accordingly.
21636 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
21638 PR tree-optimization/84114
21639 * config/aarch64/aarch64.c (aarch64_reassociation_width)
21640 Avoid reassociation of FLOAT_MODE addition.
21642 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
21644 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
21645 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
21646 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
21647 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
21648 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
21649 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
21650 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
21652 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
21653 __builtin_ia32_wbinvd): New builtins.
21654 (SPECIAL_ARGS2): New.
21655 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
21656 (SPECIAL_ARGS2): New.
21657 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
21658 (ix86_valid_target_attribute_inner_p): Ditto.
21659 (ix86_init_mmx_sse_builtins): Add special_args2.
21660 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
21661 TARGET_WBNOINVD_P): New.
21662 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
21663 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
21664 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
21665 * config/i386/immintrin.h (_wbinvd): New intrinsic.
21666 * config/i386/pconfigintrin.h: New file.
21667 * config/i386/wbnoinvdintrin.h: Ditto.
21668 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
21670 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
21672 2018-03-05 Richard Biener <rguenther@suse.de>
21674 PR tree-optimization/84670
21675 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
21677 (BB_VISITED_WITH_VISITED_SUCCS): New define.
21678 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
21679 (compute_antic_aux): Only assert the number of values in ANTIC_IN
21680 doesn't grow if all successors (recursively) were visited at least
21683 2018-03-05 Richard Biener <rguenther@suse.de>
21685 PR tree-optimization/84650
21686 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
21687 if executed in the loop pipeline.
21689 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
21691 * doc/configfiles.texi (Configuration Files): Move info about
21692 conditionalizing $target-protos.h to...
21693 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
21694 differs from $target-protos.h.
21696 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
21697 Chung-Ju Wu <jasonwucj@gmail.com>
21699 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
21700 * config/nds32/nds32-multiple.md (setmemsi): Define.
21701 * config/nds32/nds32-memory-manipulation.c
21702 (nds32_gen_dup_4_byte_to_word_value): New.
21703 (emit_setmem_word_loop): New.
21704 (emit_setmem_byte_loop): New.
21705 (nds32_expand_setmem_loop): New.
21706 (nds32_expand_setmem_loop_v3m): New.
21707 (nds32_expand_setmem_unroll): New.
21708 (nds32_expand_setmem): New.
21710 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
21711 Chung-Ju Wu <jasonwucj@gmail.com>
21713 * config/nds32/nds32-memory-manipulation.c
21714 (nds32_emit_load_store): New.
21715 (nds32_emit_post_inc_load_store): New.
21716 (nds32_emit_mem_move): New.
21717 (nds32_emit_mem_move_block): New.
21718 (nds32_expand_movmemsi_loop_unknown_size): New.
21719 (nds32_expand_movmemsi_loop_known_size): New.
21720 (nds32_expand_movmemsi_loop): New.
21721 (nds32_expand_movmemsi_unroll): New.
21722 (nds32_expand_movmemqi): Rename ...
21723 (nds32_expand_movmemsi): ... to this.
21724 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
21725 (movmemsi): ... to this.
21726 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
21727 (nds32_expand_movmemsi): ... to this.
21729 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
21730 Monk Chiang <sh.chiang04@gmail.com>
21731 Chung-Ju Wu <jasonwucj@gmail.com>
21733 * config/nds32/nds32-protos.h
21734 (nds32_expand_load_multiple): New arguments.
21735 (nds32_expand_store_multiple): Ditto.
21736 (nds32_valid_multiple_load_store): Rename ...
21737 (nds32_valid_multiple_load_store_p): ... to this.
21738 * config/nds32/nds32-memory-manipulation.c
21739 (nds32_expand_load_multiple): Refine implementation.
21740 (nds32_expand_store_multiple): Ditto.
21741 * config/nds32/nds32-multiple.md
21742 (load_multiple): Update nds32_expand_load_multiple interface.
21743 (store_multiple): Update nds32_expand_store_multiple interface.
21744 * config/nds32/nds32-predicates.c
21745 (nds32_valid_multiple_load_store): Rename ...
21746 (nds32_valid_multiple_load_store_p): ... to this and refine
21748 * config/nds32/predicates.md
21749 (nds32_load_multiple_and_update_address_operation): New predicate.
21750 (nds32_store_multiple_and_update_address_operation): New predicate.
21752 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
21753 Chung-Ju Wu <jasonwucj@gmail.com>
21755 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
21756 (combo): New attribute.
21757 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
21759 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
21761 * config/nds32/nds32.opt: Change -mcmodel= default value.
21763 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
21764 Monk Chiang <sh.chiang04@gmail.com>
21765 Chung-Ju Wu <jasonwucj@gmail.com>
21767 * config/nds32/constants.md (unspec_element): New enum.
21768 * config/nds32/constraints.md (Umw): New constraint.
21769 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
21770 * config/nds32/nds32-intrinsic.md: Likewise.
21771 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
21772 (nds32_valid_smw_lwm_base_p): New.
21773 (nds32_output_smw_single_word): New.
21774 (nds32_output_lmw_single_word): New.
21775 (nds32_expand_unaligned_load): New.
21776 (nds32_expand_unaligned_store): New.
21777 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
21778 (nds32_output_smw_single_word): Declare.
21779 (nds32_output_lmw_single_word): Declare.
21780 (nds32_expand_unaligned_load): Declare.
21781 (nds32_expand_unaligned_store): Declare.
21782 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
21783 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
21784 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
21785 NDS32_BUILTIN_UASTORE_DW.
21786 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
21789 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
21790 Kito Cheng <kito.cheng@gmail.com>
21791 Chung-Ju Wu <jasonwucj@gmail.com>
21793 * config/nds32/nds32-intrinsic.c
21794 (nds32_expand_builtin_null_ftype_reg): Delete.
21795 (nds32_expand_builtin_reg_ftype_imm): Ditto.
21796 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
21797 (nds32_read_argument): New.
21798 (nds32_legitimize_target): Ditto.
21799 (nds32_legitimize_argument): Ditto.
21800 (nds32_check_constant_argument): Ditto.
21801 (nds32_expand_unop_builtin): Ditto.
21802 (nds32_expand_unopimm_builtin): Ditto.
21803 (nds32_expand_binop_builtin): Ditto.
21804 (nds32_builtin_decl_impl): Ditto.
21805 (builtin_description): Ditto.
21806 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
21807 (nds32_init_builtins_impl): Ditto.
21808 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
21809 (nds32_builtin_decl): New.
21810 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
21811 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
21813 2018-03-02 Jeff Law <law@redhat.com>
21815 * reorg.c (stop_search_p): Handle DEBUG_INSN.
21816 (redundant_insn, fill_simple_delay_slots): Likewise.
21817 (fill_slots_from_thread): Likewise.
21818 * resource.c (mark_referenced_resources): Likewise.
21819 (mark_set_resources, find_dead_or_set_registers): Likewise.
21821 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21823 * substring-locations.h (format_warning_va): Formatting fix for
21824 ATTRIBUTE_GCC_DIAG.
21825 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
21827 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
21828 * substring-locations.c: Include intl.h.
21829 (format_warning_va): Turned into small wrapper around
21830 format_warning_n_va, renamed to ...
21831 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
21832 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
21834 (format_warning_at_substring_n): New function.
21835 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
21836 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
21837 format_warning_at_substring with just a shorter name instead of
21838 const function pointer.
21839 (fmtwarn_n): New function.
21840 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
21841 appropriate, get rid of all the fmtstr temporaries, move conditionals
21842 with G_() wrapped string literals directly into fmtwarn arguments,
21843 cast dir.len to (int), formatting fixes.
21845 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
21847 * doc/invoke.texi: Remove "Cilk Plus" references.
21849 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21850 Richard Biener <rguenther@suse.de>
21853 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
21854 for error or warning attributes if CALL_FROM_THUNK_P is set.
21857 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21860 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
21861 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
21864 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
21865 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
21867 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
21868 instead of -1U in last predictors element's probability member.
21870 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
21873 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
21874 arguments if they are comparable.
21876 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
21878 PR tree-optimization/84634
21879 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
21880 masks and masked_loop_p with a single loop_masks, making sure it's
21881 null for bb vectorization.
21883 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
21885 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
21886 (vect_analyze_data_ref_access): Use loop->safe_len rather than
21887 loop->force_vectorize to check whether there is no alias.
21889 2018-03-02 Jakub Jelinek <jakub@redhat.com>
21892 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
21894 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
21896 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
21897 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
21898 instead of a loop around prev_real_insn.
21899 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
21902 PR inline-asm/84625
21903 * config/i386/i386.c (ix86_print_operand): Use conditional
21904 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
21907 2018-03-02 Richard Biener <rguenther@suse.de>
21909 PR tree-optimization/84427
21910 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
21911 (bitmap_set_subtract_values): Rewrite to handle multiple
21914 (prune_clobbered_mems): Likewise.
21915 (phi_translate): Take edge instead of pred/phiblock.
21916 (phi_translate_1): Likewise.
21917 (phi_translate_set): Likewise. Insert all translated
21918 exprs for a value into the set, keeping possibly multiple
21919 expressions per value.
21920 (compute_antic_aux): Adjust for phi_translate changes.
21921 When intersecting union the expressions and prune those
21922 not in the final value set, keeping possibly multiple
21923 expressions per value. Do not use value-insertion
21924 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
21925 all expressions. Add verification that the value-sets
21926 only shrink during iteration.
21927 (compute_partial_antic_aux): Adjust for the phi_translate changes.
21928 (do_pre_regular_insertion): Likewise.
21929 (do_pre_partial_partial_insertion): Likewise.
21931 2018-03-02 Richard Biener <rguenther@suse.de>
21934 * config/darwin.c (saved_debug_info_level): New static global.
21935 (darwin_asm_lto_start): Disable debug info generation for LTO out.
21936 (darwin_asm_lto_end): Restore debug info generation settings.
21938 2018-03-01 Martin Liska <mliska@suse.cz>
21941 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
21942 volatile arguments.
21944 2018-03-01 Richard Biener <rguenther@suse.de>
21947 * dwarf2out.c (gen_variable_die): Properly handle late VLA
21948 type annotation with LTO when debug was disabled at compile-time.
21950 2018-03-01 Matthew Fortune <mfortune@gmail.com>
21952 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
21954 (mips_final_postscan_insn): Likewise.
21956 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
21958 PR rtl-optimization/84528
21959 * alias.c (init_alias_target): Add commentary.
21960 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
21961 a unique base value if the frame pointer is not eliminated
21962 to the stack pointer.
21964 2018-03-01 Tom de Vries <tom@codesourcery.com>
21966 PR rtl-optimization/83327
21967 * lra-int.h (hard_regs_spilled_into): Declare.
21968 * lra.c (hard_regs_spilled_into): Define.
21969 (init_reg_info): Init hard_regs_spilled_into.
21970 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
21971 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
21972 (process_bb_lives): Handle hard_regs_spilled_into.
21973 (lra_create_live_ranges_1): Before doing liveness propagation, clear
21974 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
21976 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
21978 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
21979 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
21980 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
21981 * config/rs6000/aix72.h: New file.
21983 2018-02-28 Jakub Jelinek <jakub@redhat.com>
21985 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
21986 instead of warning_at with conditional singular and plural messages
21990 * stor-layout.c (update_alignment_for_field): For
21991 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
21992 && !DECL_PACKED (field), do the alignment update, just use
21993 only desired_align instead of MAX (type_align, desired_align)
21995 (place_field): Don't do known_align < desired_align handling
21996 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
21997 is non-NULL, instead do it after rli->prev_field handling and
21998 only if not within a bitfield word. For DECL_PACKED (field)
21999 use type_align of BITS_PER_UNIT.
22001 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
22003 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
22004 superfluous parentheses and trailing spaces.
22006 2018-02-28 Richard Biener <rguenther@suse.de>
22008 PR tree-optimization/84584
22009 * graphite-scop-detection.c (scop_detection::add_scop): Discard
22010 SCoPs with fake exit edge.
22012 2018-02-28 Martin Liska <mliska@suse.cz>
22015 * timevar.c (timer::print): Fix format to properly print 100%
22018 2018-02-28 Richard Biener <rguenther@suse.de>
22020 PR middle-end/84607
22021 * genmatch.c (capture_info::walk_match): Do not mark
22022 captured expressions without operands as expr_p given
22023 they act more like predicates and should be subject to
22024 "lost tail" side-effect preserving.
22026 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
22028 PR rtl-optimization/81611
22029 * auto-inc-dec.c (attempt_change): Move dead note from
22030 mem_insn if it's the next use of regno
22031 (find_address): Take address use of reg holding
22032 non-incremented value. Add parm to limit search to the named
22034 (merge_in_block): Attempt to use a mem insn that is the next
22035 use of the original regno.
22037 2018-02-27 Martin Sebor <msebor@redhat.com>
22040 * doc/invoke.texi (-Wmissing-attributes): New option.
22041 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
22043 2018-02-27 Martin Sebor <msebor@redhat.com>
22045 PR translation/84207
22046 * diagnostic-core.h (warning_n, error_n, inform_n): Change
22047 n argument to unsigned HOST_WIDE_INT.
22048 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
22049 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
22050 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
22051 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
22053 2018-02-27 Richard Biener <rguenther@suse.de>
22055 PR tree-optimization/84512
22056 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
22057 Do not use the estimate returned from record_stmt_cost for
22058 the scalar iteration cost but sum properly using add_stmt_cost.
22060 2018-02-27 Richard Biener <rguenther@suse.de>
22062 PR tree-optimization/84466
22063 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
22064 Adjust last change to less strictly validate use operands.
22066 2018-02-27 Martin Liska <mliska@suse.cz>
22068 PR gcov-profile/84548
22069 * gcov.c (process_file): Allow partial overlap and consider it
22070 also as group functions.
22071 (output_lines): Properly calculate range of lines for a group.
22073 2018-02-27 Martin Liska <mliska@suse.cz>
22075 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
22076 'ggc' suffixes. Change first column width.
22077 (timer::print): Fix formatting of the column.
22079 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
22081 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
22082 preserve inline entry blocks for the sake of debug inline
22083 entry point markers alone.
22084 (remove_unused_locals): Suggest in comments a better place to
22085 force the preservation of inline entry blocks that are
22086 otherwise unused, but do not preserve them.
22088 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
22090 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
22092 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
22095 * config/i386/constraints.md (Bs): Replace
22096 ix86_indirect_branch_register with
22097 TARGET_INDIRECT_BRANCH_REGISTER.
22099 * config/i386/i386.md (indirect_jump): Likewise.
22100 (tablejump): Likewise.
22101 (*sibcall_memory): Likewise.
22102 (*sibcall_value_memory): Likewise.
22103 Peepholes of indirect call and jump via memory: Likewise.
22104 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
22105 (*sibcall_value_GOT_32): Likewise.
22106 * config/i386/predicates.md (indirect_branch_operand): Likewise.
22107 (GOT_memory_operand): Likewise.
22108 (call_insn_operand): Likewise.
22109 (sibcall_insn_operand): Likewise.
22110 (GOT32_symbol_operand): Likewise.
22111 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
22113 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
22115 PR rtl-optimization/83496
22116 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
22117 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
22118 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
22119 redundant insn, if any.
22120 (relax_delay_slots): Likewise.
22121 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
22123 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
22125 PR tree-optimization/83965
22126 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
22127 that grouped statements are part of a reduction chain. Return
22128 true if the statement is not marked as a reduction itself but
22129 is part of a group.
22130 (vect_recog_dot_prod_pattern): Don't check whether the statement
22131 is part of a group here.
22132 (vect_recog_sad_pattern): Likewise.
22133 (vect_recog_widen_sum_pattern): Likewise.
22135 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
22138 * final.c (rest_of_clean_state): Also look for calls inside sequences.
22140 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
22143 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
22145 (ix86_output_indirect_function_return): New prototype.
22146 (ix86_split_simple_return_pop_internal): Likewise.
22147 * config/i386/i386.c (indirect_return_via_cx): New.
22148 (indirect_return_via_cx_bnd): Likewise.
22149 (indirect_thunk_name): Handle return va CX_REG.
22150 (output_indirect_thunk_function): Create alias for
22151 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
22152 (ix86_output_indirect_jmp): Remove the bool argument.
22153 (ix86_output_indirect_function_return): New function.
22154 (ix86_split_simple_return_pop_internal): Likewise.
22155 * config/i386/i386.md (*indirect_jump): Don't pass false
22156 to ix86_output_indirect_jmp.
22157 (*tablejump_1): Likewise.
22158 (simple_return_pop_internal): Change it to define_insn_and_split.
22159 Call ix86_split_simple_return_pop_internal to split it for
22160 -mfunction-return=.
22161 (simple_return_indirect_internal): Call
22162 ix86_output_indirect_function_return instead of
22163 ix86_output_indirect_jmp.
22165 2018-02-26 Jakub Jelinek <jakub@redhat.com>
22168 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
22169 memset and value initialization afterwards.
22171 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
22173 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
22175 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22178 * common/config/aarch64/aarch64-common.c
22179 (aarch_option_optimization_table[]): Switch
22180 off fomit-frame-pointer
22182 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
22183 Chung-Ju Wu <jasonwucj@gmail.com>
22185 * config/nds32/nds32-multiple.md (load_multiple): Disallow
22187 (store_multiple): Ditto.
22189 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
22191 * config.gcc: Add --with-cpu support for nds32 target.
22192 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
22193 * config/nds32/nds32.opt: Add -mcpu= option.
22195 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
22197 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
22198 isel=yes): Warn for these deprecated options.
22200 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
22202 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
22203 ISA_2_5_MASKS_EMBEDDED.
22205 2018-02-23 Jakub Jelinek <jakub@redhat.com>
22207 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
22208 p->max as pointers rather than using iterative_hash_expr.
22210 2018-02-23 Carl Love <cel@us.ibm.com>
22212 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
22213 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
22215 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
22216 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
22217 P8V_BUILTIN_VEC_VUNSIGNED2.
22219 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
22222 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
22223 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
22225 * lra-constraints.c (curr_insn_transform): Set up
22226 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
22228 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
22229 * lra-eliminations.c (spill_pseudos): Ditto.
22230 (process_insn_for_elimination): Ditto.
22231 * lra-lives.c (reg_early_clobber_p): Use the new macros.
22232 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
22233 LRA_NON_CLOBBERED_ALT.
22235 2018-02-22 Martin Sebor <msebor@redhat.com>
22237 PR tree-optimization/84480
22238 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
22239 to maybe_diag_stxncpy_trunc. Call it.
22240 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
22241 from gimple_fold_builtin_strcpy. Print inlining stack.
22242 (handle_builtin_stxncpy): Print inlining stack.
22243 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
22245 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
22248 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
22249 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
22250 and -fcheck-pointer-bounds are used together.
22251 (indirect_thunk_prefix): New enum.
22252 (indirect_thunk_need_prefix): New function.
22253 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
22254 "_nt" instead of "_bnd" for NOTRACK prefix.
22255 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
22256 (output_indirect_thunk_function): Likewise.
22258 (ix86_code_end): Update output_indirect_thunk_function calls.
22259 (ix86_output_indirect_branch_via_reg): Replace
22260 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
22261 (ix86_output_indirect_branch_via_push): Likewise.
22262 (ix86_output_function_return): Likewise.
22263 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
22264 incompatible with -fcf-protection=branch and
22265 -fcheck-pointer-bounds.
22267 2018-02-22 Steve Ellcey <sellcey@cavium.com>
22270 * config/aarch64/aarch64.c (aarch64_print_address_internal):
22271 Change gcc_assert call to output_operand_lossage.
22273 2018-02-22 Steve Ellcey <sellcey@cavium.com>
22275 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
22277 2018-02-22 DJ Delorie <dj@redhat.com>
22278 Sebastian Perta <sebastian.perta@renesas.com>
22279 Oleg Endo <olegendo@gcc.gnu.org>
22281 * config/rx/rx.c (rx_rtx_costs): New function.
22282 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
22284 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
22286 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
22288 2018-02-22 Martin Liska <mliska@suse.cz>
22291 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
22292 Add "native" as a possible value.
22294 2018-02-22 Martin Liska <mliska@suse.cz>
22297 * config/i386/i386.c (ix86_option_override_internal):
22298 Add "native" as a possible value for -march and -mtune.
22300 2018-02-22 Jakub Jelinek <jakub@redhat.com>
22303 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
22304 to all type variants.
22306 PR tree-optimization/84503
22307 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
22308 width as info->bitpos + info->bitsize - start.
22309 (merged_store_group::merge_overlapping): Simplify width computation.
22310 (check_no_overlap): New function.
22311 (imm_store_chain_info::try_coalesce_bswap): Compute expected
22312 start + width and last_order of the group, fail if check_no_overlap
22314 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
22315 to group if check_no_overlap fails.
22317 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
22319 * config/rs6000/altivec.md: Delete contraint arguments to
22320 define_expand, define_split, and define_peephole2, and in
22321 define_insn_and_split if always unused.
22322 * config/rs6000/darwin.md: Ditto.
22323 * config/rs6000/dfp.md: Ditto.
22324 * config/rs6000/rs6000.md: Ditto.
22325 * config/rs6000/sync.md: Ditto.
22326 * config/rs6000/vector.md: Ditto.
22327 * config/rs6000/vsx.md: Ditto.
22329 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
22331 * config/rs6000/altivec.md: Write output control strings as braced
22332 blocks instead of double-quoted strings.
22333 * config/rs6000/darwin.md: Ditto.
22334 * config/rs6000/rs6000.md: Ditto.
22335 * config/rs6000/vector.md: Ditto.
22336 * config/rs6000/vsx.md: Ditto.
22338 2018-02-21 Jason Merrill <jason@redhat.com>
22340 PR c++/84314 - ICE with templates and fastcall attribute.
22341 * attribs.c (build_type_attribute_qual_variant): Remove assert.
22343 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
22345 * ipa-cp.c (determine_versionability): Fix comment typos.
22347 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
22350 * ipa-cp.c (determine_versionability): Do not version functions caling
22353 2018-02-21 Martin Liska <mliska@suse.cz>
22356 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
22357 Add "native" as a possible value.
22358 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
22359 the macro when native cpu detection is available.
22361 2018-02-21 Martin Liska <mliska@suse.cz>
22364 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
22365 Add "native" as a possible value.
22366 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
22367 when native cpu detection is available.
22369 2018-02-21 Jakub Jelinek <jakub@redhat.com>
22370 Martin Sebor <msebor@redhat.com>
22372 PR tree-optimization/84478
22373 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
22375 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
22376 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
22377 support which is conservatively correct, for 2 only stay conservative
22378 for maxlen. Formatting and comment capitalization fixes. Add STRICT
22379 argument to the 2 argument get_range_strlen, adjust 6 arg
22380 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
22382 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
22383 (gimple_fold_builtin_strlen): Pass true as last argument to
22386 2018-02-20 Martin Sebor <msebor@redhat.com>
22388 PR middle-end/84095
22389 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
22390 (builtin_memref::set_base_and_offset): Same. Handle inner references.
22391 (builtin_memref::builtin_memref): Factor out parts into
22392 set_base_and_offset and call it.
22394 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
22396 PR middle-end/84406
22397 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
22398 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
22399 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
22400 search at the associated MODE_INT.
22402 2018-02-20 Jeff Law <law@redhat.com>
22404 PR middle-end/82123
22405 PR tree-optimization/81592
22406 PR middle-end/79257
22407 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
22408 for range data rather than using global data.
22409 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
22410 range data rather than using global data.
22411 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
22412 pass it to children as needed.
22413 (struct directive::fmtresult): Similarly.
22414 (struct directive::set_width): Similarly.
22415 (struct directive::set_precision): Similarly.
22416 (format_integer, format_directive, parse_directive): Similarly.
22417 (format_none): Accept unnamed vr_values parameter.
22418 (format_percent, format_floating, format_character): Similarly.
22419 (format_string, format_plain): Similarly.
22420 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
22421 the EVRP range analyzer for range data rather than using global data.
22422 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
22423 gimple-ssa-evrp-analyze.h
22424 (class sprintf_dom_walker): Add after_dom_children member function.
22425 Add evrp_range_analyzer member.
22426 (sprintf_dom_walker::before_dom_children): Call into the EVRP
22427 range analyzer as needed.
22428 (sprintf_dom_walker::after_dom_children): New member function.
22429 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
22431 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
22432 (evrp_range_analyzer::pop_to_marker): Likewise.
22434 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
22436 PR tree-optimization/84419
22437 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
22438 with the required type if its current type is compatible but
22441 2018-02-20 Jakub Jelinek <jakub@redhat.com>
22443 PR middle-end/82004
22444 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
22445 after vectorization.
22447 2018-02-20 Martin Liska <mliska@suse.cz>
22450 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
22451 possible values if we don't have a hint.
22453 2018-02-20 Martin Liska <mliska@suse.cz>
22457 * final.c (shorten_branches): Build align_tab array with one
22459 * opts.c (finish_options): Add alignment option limit check.
22460 (MAX_CODE_ALIGN): Likewise.
22461 (MAX_CODE_ALIGN_VALUE): Likewise.
22462 * doc/invoke.texi: Document maximum allowed option value for
22463 all -falign-* options.
22465 2018-02-19 Jakub Jelinek <jakub@redhat.com>
22468 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
22469 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
22470 * var-tracking.c (emit_note_insn_var_location): Remove all references
22471 to NOTE_INSN_CALL_ARG_LOCATION.
22472 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
22473 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
22474 Use copy_rtx_if_shared.
22475 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
22476 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
22477 (dwarf2out_var_location): Remove handling of
22478 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
22480 * final.c (final_scan_insn): Remove all references to
22481 NOTE_INSN_CALL_ARG_LOCATION.
22482 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
22483 before dumping final insns.
22484 * except.c (emit_note_eh_region_end): Remove all references to
22485 NOTE_INSN_CALL_ARG_LOCATION.
22486 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
22487 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
22488 * config/arc/arc.c (hwloop_optimize): Likewise.
22489 * config/arm/arm.c (create_fix_barrier): Likewise.
22490 * config/s390/s390.c (s390_chunkify_start): Likewise.
22491 * config/sh/sh.c (find_barrier): Likewise.
22492 * config/i386/i386.c (rest_of_insert_endbranch,
22493 ix86_seh_fixup_eh_fallthru): Likewise.
22494 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
22495 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
22496 * config/frv/frv.c (frv_function_prologue): Likewise.
22497 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
22499 (note_outside_basic_block_p): Remove all references to
22500 NOTE_INSN_CALL_ARG_LOCATION.
22501 * gengtype.c (adjust_field_rtx_def): Likewise.
22502 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
22504 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
22505 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
22508 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
22511 PR tree-optimization/84452
22512 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
22513 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
22516 2018-02-19 Martin Liska <mliska@suse.cz>
22519 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
22521 2018-02-19 Martin Liska <mliska@suse.cz>
22522 Richard Sandiford <richard.sandiford@linaro.org>
22524 PR tree-optimization/82491
22525 * gimple-fold.c (get_base_constructor): Make earlier bail out
22528 2018-02-19 Carl Love <cel@us.ibm.com>
22530 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
22531 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
22533 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
22534 P8V_BUILTIN_VEC_NEG.
22536 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
22538 * config/rl78/rl78.md (movdf): New define expand.
22540 2018-02-19 Martin Liska <mliska@suse.cz>
22543 * doc/invoke.texi: Fix typo.
22544 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
22546 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
22548 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
22549 handle rs6000_single_float and rs6000_double_float specially for
22552 2018-02-16 Jeff Law <law@redhat.com>
22554 * config/rx/rx.c (add_pop_cfi_notes): New function.;
22555 (pop_regs): Use it.
22557 2018-02-16 Jakub Jelinek <jakub@redhat.com>
22560 * ipa-inline.c (inline_small_functions): Fix a typo.
22562 2018-02-16 Nathan Sidwell <nathan@acm.org>
22564 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
22566 2018-02-16 Carl Love <cel@us.ibm.com>
22568 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
22569 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
22570 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
22571 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
22572 expansion to P8V_BUILTIN_VEC_FLOAT2.
22574 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
22576 PR rtl-optimization/70023
22577 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
22578 src_regno into account.
22580 2018-02-16 Carl Love <cel@us.ibm.com>
22582 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
22583 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
22584 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
22585 * config/rs6000/rs6000.c: Remove case statements for
22586 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
22587 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
22588 and P9V_BUILTIN_VEC_VINSERT4B.
22589 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
22590 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
22591 * config/rs6000/vsx.md:
22592 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
22595 2018-02-16 Carl Love <cel@us.ibm.com>
22597 * config/rs6000/altivec.h: Add builtin names vec_extract4b
22599 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
22601 * config/rs6000/rs6000-c.c: Add the definitions for
22602 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
22603 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
22604 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
22605 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
22606 definition for insert4b and define insn *insert3b_internal.
22607 * doc/extend.texi: Add documentation for vec_extract4b.
22609 2018-02-16 Nathan Sidwell <nathan@acm.org>
22611 * doc/extend.texi (Backwards Compatibility): Mention friend
22612 injection. Note for-scope is deprecated.
22613 * doc/invoke.texi (-ffriend-injection): Deprecate.
22615 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
22617 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
22618 that moved to I2, also allow destinations that are a paradoxical
22619 subreg (instead of a normal reg).
22621 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
22624 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
22627 2018-02-16 Richard Biener <rguenther@suse.de>
22629 PR tree-optimization/84037
22630 PR tree-optimization/84016
22632 * config/i386/i386.c (ix86_builtin_vectorization_cost):
22633 Adjust vec_construct for the fact we need additional higher latency
22634 128bit inserts for AVX256 and AVX512 vector builds.
22635 (ix86_add_stmt_cost): Scale vector construction cost for
22638 2018-02-16 Richard Biener <rguenther@suse.de>
22640 PR tree-optimization/84417
22641 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
22642 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
22643 (non_rewritable_lvalue_p): Likewise, use poly-ints.
22645 2018-02-16 Martin Liska <mliska@suse.cz>
22648 * internal-fn.def (ASAN_CHECK): Set proper flags.
22649 (ASAN_MARK): Likewise.
22651 2018-02-16 Julia Koval <julia.koval@intel.com>
22653 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
22654 from PTA_CANNONLAKE.
22656 2018-02-16 Jakub Jelinek <jakub@redhat.com>
22659 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
22660 Use ++iter rather than iter++ for std::list iterators.
22661 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
22662 defer deleting them until all nodes in the forest are processed. Do
22663 free even leaf nodes. Change to_process into auto_vec.
22666 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
22667 * vec.h (vec_default_construct): Use memset instead of placement new
22668 if BROKEN_VALUE_INITIALIZATION is defined.
22669 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
22670 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
22673 PR rtl-optimization/83723
22674 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
22675 * lra.c (lra_substitute_pseudo): Likewise. If true, use
22676 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
22678 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
22680 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
22682 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
22684 PR rtl-optimization/81443
22685 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
22686 from inner REGs to paradoxical SUBREGs.
22688 2018-02-16 Richard Biener <rguenther@suse.de>
22690 PR tree-optimization/84399
22691 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
22692 For operands we can analyze at their definition make sure we can
22693 analyze them at each use as well.
22695 2018-02-16 Richard Biener <rguenther@suse.de>
22697 PR tree-optimization/84190
22698 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
22699 volatile accesses if the decl isn't volatile.
22701 2018-02-15 Jason Merrill <jason@redhat.com>
22703 PR c++/84314 - ICE with templates and fastcall attribute.
22704 * attribs.c (build_type_attribute_qual_variant): Don't clobber
22705 TYPE_CANONICAL on an existing type.
22707 2018-02-15 Jakub Jelinek <jakub@redhat.com>
22709 PR tree-optimization/84383
22710 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
22711 dstoff nor call operand_equal_p if dstbase is NULL.
22713 PR tree-optimization/84334
22714 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
22715 also a CONSTANT_CLASS_P, punt.
22717 2018-02-14 Jim Wilson <jimw@sifive.com>
22719 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
22720 first SMALL_OPERAND check. New local min_second_step. Move assert
22721 to where locals are set. Add TARGET_RVC support.
22722 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
22724 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
22726 * doc/invoke.texi: Correct -Wformat-overflow code sample.
22728 2018-02-14 Martin Sebor <msebor@redhat.com>
22730 PR tree-optimization/83698
22731 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
22732 arrays constrain the offset range to their bounds.
22733 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
22734 (builtin_access::overlap): Avoid setting the size of overlap if it's
22736 (maybe_diag_overlap): Also consider arrays when deciding what values
22737 of offsets to include in diagnostics.
22739 2018-02-14 Martin Sebor <msebor@redhat.com>
22742 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
22743 that correspond to the kind of a declaration.
22745 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
22748 * config/pa/pa.md: Load address of PIC label using the linkage table
22749 if the label is nonlocal.
22751 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
22753 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
22754 warning message if user requests -maltivec=be.
22755 * doc/invoke.texi: Document deprecation of -maltivec=be.
22757 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
22760 * config/rs6000/rs6000-c.c: Update definitions for
22761 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
22762 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
22764 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
22767 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
22768 add _get_ssp intrinsics. Remove argument from
22769 __builtin_ia32_rdssp[d|q].
22770 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
22771 * config/i386/i386-builtin.def: Remove argument from
22772 __builtin_ia32_rdssp[d|q].
22773 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
22774 ix86_expand_special_args_builtin for _rdssp[d|q].
22775 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
22776 Clear register before usage.
22777 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
22778 Add documentation for new _get_ssp and _inc_ssp intrinsics.
22780 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
22782 PR tree-optimization/84357
22783 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
22784 operand 1 of an ARRAY_REF too.
22786 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
22789 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
22790 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
22792 (set_of_reg): New struct.
22793 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
22794 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
22795 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
22797 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
22798 Split into bitclr, bitset, bitinvert patterns if appropriate.
22799 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
22800 use rx_fuse_in_memory_bitop.
22801 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
22802 to named insn, correct maximum insn length.
22804 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
22807 * machmode.def: Define a complex mode for PARTIAL_INT.
22808 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
22810 * doc/rtl.texi: Document CSPImode.
22811 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
22813 (msp430_hard_regno_nregs_with_padding): Likewise.
22815 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
22818 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
22820 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
22822 PR rtl-optimization/84169
22823 * combine.c (try_combine): New variable split_i2i3. Set it to true if
22824 we generated a parallel as new i3 and we split that to new i2 and i3
22825 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
22826 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
22827 those to i2, not i1. Partially rewrite this scan code.
22829 2018-02-13 Jakub Jelinek <jakub@redhat.com>
22832 * stor-layout.c (place_field): For variable length fields, adjust
22833 offset_align afterwards not just based on the field's alignment,
22834 but also on the size.
22836 PR middle-end/84309
22837 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
22838 of exps and logs in the use_exp2 case.
22840 2018-02-13 Jeff Law <law@redhat.com>
22842 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
22843 entry for "vector".
22845 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
22848 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
22852 * common.opt (gas-loc-support, gas-locview-support): New.
22853 (ginline-points, ginternal-reset-location-views): New.
22854 * doc/invoke.texi: Document them. Use @itemx where intended.
22855 (gvariable-location-views): Adjust.
22856 * target.def (reset_location_view): New.
22857 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
22858 (TARGET_RESET_LOCATION_VIEW): New.
22859 * doc/tm.texi: Rebuilt.
22860 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
22861 (dwarf2out_default_as_locview_support): New.
22862 (output_asm_line_debug_info): Use option variables.
22863 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
22864 (output_loc_list): Likewise.
22865 (add_high_low_attributes): Check option variables.
22866 Don't output entry view attribute in strict mode.
22867 (gen_inlined_subroutine_die): Check option variables.
22868 (dwarf2out_inline_entry): Likewise.
22869 (init_sections_and_labels): Likewise.
22870 (dwarf2out_early_finish): Likewise.
22871 (maybe_reset_location_view): New, from...
22872 (dwarf2out_var_location): ... here. Call it.
22873 * debug.h (dwarf2out_default_as_loc_support): Declare.
22874 (dwarf2out_default_as_locview_support): Declare.
22875 * hooks.c (hook_int_rtx_insn_0): New.
22876 * hooks.h (hook_int_rtx_insn_0): Declare.
22877 * toplev.c (process_options): Take -gas-loc-support and
22878 -gas-locview-support from dwarf2out. Enable
22879 -gvariable-location-views by default only with locview
22880 assembler support. Enable -ginternal-reset-location-views by
22881 default only if the target defines the corresponding hook.
22882 Enable -ginline-points by default if location views are
22883 enabled; force it disabled if statement frontiers are
22885 * tree-inline.c (expand_call_inline): Check option variables.
22886 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
22888 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
22890 PR tree-optimization/84321
22891 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
22892 handling. Also check whether the anti-range contains any values
22893 that satisfy the mask; switch to a VR_RANGE if not.
22895 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
22898 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
22900 2018-02-13 Martin Jambor <mjambor@suse.cz>
22903 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
22904 of call statements, also set location of a load to a temporary.
22906 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
22908 * config/rl78/rl78.c (add_vector_labels): New function.
22909 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
22910 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
22911 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
22912 which checks that no arguments are passed.
22913 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
22914 * doc/extend.texi: Documentation for the new attribute.
22916 2018-02-13 Andreas Schwab <schwab@suse.de>
22918 * config/riscv/linux.h (CPP_SPEC): Define.
22920 2018-02-13 Jakub Jelinek <jakub@redhat.com>
22923 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
22924 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
22925 OPTION_MASK_ISA_AES as first argument to def_builtin_const
22926 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
22927 instead of OPTION_MASK_ISA_PCLMUL as first argument to
22928 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
22929 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
22930 temporarily for AES and PCLMUL builtins.
22932 PR tree-optimization/84339
22933 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
22934 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
22937 PR middle-end/84309
22938 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
22939 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
22940 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
22942 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
22944 * omp-simd-clone.h: New file.
22945 * omp-simd-clone.c: Include omp-simd-clone.h.
22946 (expand_simd_clones): No longer static.
22947 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
22948 cgraph.h and omp-simd-clone.h.
22949 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
22950 (vect_recog_widen_shift_pattern): Formatting fix.
22951 (vect_pattern_recog_1): Don't check optab for calls.
22954 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
22955 operands[2] into a REG before using gen_lowpart on it.
22957 2018-02-12 Jeff Law <law@redhat.com>
22960 * config/sh/sh.c (find_barrier): Consider a sibling call
22963 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
22964 successfully back substituting a reg.
22966 2018-02-12 Richard Biener <rguenther@suse.de>
22968 PR tree-optimization/84037
22969 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
22970 parameter, move visited init to caller.
22971 (vect_slp_analyze_operations): Separate cost from validity
22972 check, initialize visited once for all instances.
22973 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
22975 * tree-vect-stmts.c (vect_model_simple_cost): Make early
22977 (vect_model_promotion_demotion_cost): Likewise.
22978 (vectorizable_bswap): Guard cost modeling with !slp_node
22979 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
22981 (vectorizable_call): Likewise.
22982 (vectorizable_conversion): Likewise.
22983 (vectorizable_assignment): Likewise.
22984 (vectorizable_shift): Likewise.
22985 (vectorizable_operation): Likewise.
22986 (vectorizable_store): Likewise.
22987 (vectorizable_load): Likewise.
22988 (vectorizable_condition): Likewise.
22989 (vectorizable_comparison): Likewise.
22991 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
22994 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
22995 (ASAN_MARK): Fix fnspec to account for return value, change pointer
22996 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
22998 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
23000 PR middle-end/83665
23001 * params.def (inline-min-speedup): Increase from 8 to 15.
23002 (max-inline-insns-auto): Decrease from 40 to 30.
23003 * ipa-split.c (consider_split): Add some buffer for function to
23004 be considered inlining candidate.
23005 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
23008 2018-02-12 Richard Biener <rguenther@suse.de>
23010 PR tree-optimization/84037
23011 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
23012 matched stmts if we cannot swap the non-matched ones.
23014 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
23016 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
23017 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
23018 _mm_maskz_scalef_round_ss): New intrinsics.
23019 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
23020 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
23021 __builtin_ia32_scalefss_round): Remove.
23022 (__builtin_ia32_scalefsd_mask_round,
23023 __builtin_ia32_scalefss_mask_round): New intrinsics.
23024 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
23025 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
23026 ((match_operand:VF_128 2 "<round_nimm_predicate>"
23027 "<round_constraint>")): Changed to ...
23028 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
23029 "<round_scalar_constraint>")): ... this.
23030 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
23031 %0, %1, %2<round_op3>}"): Changed to ...
23032 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
23033 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
23034 %2<round_scalar_mask_op3>}"): ... this.
23035 * config/i386/subst.md (round_scalar_nimm_predicate): New.
23037 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
23039 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
23040 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
23041 (_mm_maskz_sqrt_round_ss): New intrinsics.
23042 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
23043 (__builtin_ia32_sqrtsd_mask_round)
23044 (__builtin_ia32_sqrtss_mask_round): New builtins.
23045 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
23046 (__builtin_ia32_sqrtss_round): Remove.
23047 (__builtin_ia32_sqrtsd_mask_round)
23048 (__builtin_ia32_sqrtss_mask_round): New builtins.
23049 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
23050 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
23051 ((match_operand:VF_128 1 "vector_operand"
23052 "xBm,<round_constraint>")): Changed to ...
23053 ((match_operand:VF_128 1 "vector_operand"
23054 "xBm,<round_scalar_constraint>")): ... this.
23055 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
23056 %0, %2, %<iptr>1<round_op3>}): Changed to ...
23057 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
23058 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
23059 %<iptr>1<round_scalar_mask_op3>}): ... this.
23060 ((set_attr "prefix" "<round_prefix>")): Changed to ...
23061 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
23063 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
23066 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
23067 Cast vec_cmpeq result to correct type.
23068 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
23069 Cast vec_cmpgt result to correct type.
23071 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
23073 * final.c (final_scan_insn_1): Renamed from...
23074 (final_scan_insn): ... this. New wrapper, to recover
23075 seen from the outermost call in recursive ones.
23076 * config/sparc/sparc.c (output_return): Drop seen from call.
23077 (output_sibcall): Likewise.
23078 * config/visium/visium.c (output_branch): Likewise.
23080 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
23082 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
23085 2018-02-10 Alan Modra <amodra@gmail.com>
23088 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
23089 Specify LR as an input.
23091 2018-02-10 Jakub Jelinek <jakub@redhat.com>
23094 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
23095 remove_member_access_dummy_vars): New functions.
23096 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
23097 lower_omp_1, execute_lower_omp): Use them.
23099 PR rtl-optimization/84308
23100 * shrink-wrap.c (spread_components): Release todo vector.
23102 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
23104 PR rtl-optimization/57193
23105 * ira-color.c (struct allocno_color_data): Add member
23106 conflict_allocno_hard_prefs.
23107 (update_conflict_allocno_hard_prefs): New.
23108 (bucket_allocno_compare_func): Add a preference based on
23109 conflict_allocno_hard_prefs.
23110 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
23111 (color_allocnos): Remove a dead code. Initiate
23112 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
23114 2018-02-09 Jakub Jelinek <jakub@redhat.com>
23117 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
23118 constraint from =wa to wa. Avoid a subreg on the output operand,
23119 instead use a pseudo and subreg it in a move.
23120 (p9_xxbrd_<mode>): Changed to ...
23121 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
23122 (p9_xxbrd_v2df): New expander.
23123 (p9_xxbrw_<mode>): Changed to ...
23124 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
23125 (p9_xxbrw_v4sf): New expander.
23127 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
23129 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
23131 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
23134 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
23135 multiply in 32-bit mode.
23136 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
23137 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
23140 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
23142 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
23143 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
23144 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
23145 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
23147 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
23150 * dwarf2out.c (is_trivial_indirect_ref): New function.
23151 (dwarf2out_late_global_decl): Do not generate a location
23152 attribute for variables that have a non-trivial DECL_VALUE_EXPR
23153 and that are not defined in the current unit.
23155 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
23157 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
23158 instead of a libcall for UNORDERED.
23160 2018-02-09 Tamar Christina <tamar.christina@arm.com>
23163 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
23164 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
23166 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23169 * config/s390/s390.c (s390_set_current_function): Invoke
23170 s390_indirect_branch_settings also if fndecl didn't change.
23172 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
23174 * config/rs6000/rs6000.md (blockage): Set length to zero.
23176 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
23178 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
23180 2018-02-09 Jakub Jelinek <jakub@redhat.com>
23183 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
23184 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
23188 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
23189 PARALLEL incoming that failed vt_get_decl_and_offset check.
23191 PR middle-end/84237
23192 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
23193 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
23195 (get_variable_section): For decls in named .bss* sections pass true as
23196 second argument to bss_initializer_p.
23198 2018-02-09 Marek Polacek <polacek@redhat.com>
23199 Jakub Jelinek <jakub@redhat.com>
23202 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
23203 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
23204 Sync some changes from cxx_fold_indirect_ref.
23206 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
23208 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
23210 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
23211 (BLOCK_INLINE_ENTRY_LABEL): New.
23212 (dwarf2out_var_location): Disregard inline entry markers.
23213 (inline_entry_data): New struct.
23214 (inline_entry_data_hasher): New hashtable type.
23215 (inline_entry_data_hasher::hash): New.
23216 (inline_entry_data_hasher::equal): New.
23217 (inline_entry_data_table): New variable.
23218 (add_high_low_attributes): Add DW_AT_entry_pc and
23219 DW_AT_GNU_entry_view attributes if a pending entry is found
23220 in inline_entry_data_table. Add old entry_pc attribute only
23221 if debug nonbinding markers are disabled.
23222 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
23223 markers are enabled.
23224 (block_within_block_p, dwarf2out_inline_entry): New.
23225 (dwarf2out_finish): Check that no entries remained in
23226 inline_entry_data_table.
23227 * final.c (reemit_insn_block_notes): Handle inline entry notes.
23228 (final_scan_insn, notice_source_line): Likewise.
23229 (rest_of_clean_state): Skip inline entry markers.
23230 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
23232 * gimple.c (gimple_build_debug_inline_entry): New.
23233 * gimple.h (enum gimple_debug_subcode): Add
23234 GIMPLE_DEBUG_INLINE_ENTRY.
23235 (gimple_build_debug_inline_entry): Declare.
23236 (gimple_debug_inline_entry_p): New.
23237 (gimple_debug_nonbind_marker_p): Adjust.
23238 * insn-notes.def (INLINE_ENTRY): New.
23239 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
23240 inline entry marker notes.
23241 (print_insn): Likewise.
23242 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
23243 (INSN_DEBUG_MARKER_KIND): Likewise.
23244 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
23245 * tree-inline.c (expand_call_inline): Build and insert
23246 debug_inline_entry stmt.
23247 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
23248 inline entry blocks early, if nonbind markers are enabled.
23249 (dump_scope_block): Dump fragment info.
23250 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
23251 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
23252 (gimple_build_debug_inline_entry): New.
23253 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
23254 Enable/disable inline entry points too.
23255 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
23256 (DEBUG_INSN): Describe inline entry markers.
23258 * common.opt (gvariable-location-views): New.
23259 (gvariable-location-views=incompat5): New.
23260 * config.in: Rebuilt.
23261 * configure: Rebuilt.
23262 * configure.ac: Test assembler for view support.
23263 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
23264 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
23265 * dwarf2out.c (var_loc_view): New typedef.
23266 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
23267 (dwarf2out_locviews_in_attribute): New.
23268 (dwarf2out_locviews_in_loclist): New.
23269 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
23270 (enum dw_line_info_opcode): Add LI_adv_address.
23271 (struct dw_line_info_table): Add view.
23272 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
23273 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
23274 (zero_view_p): New variable.
23275 (ZERO_VIEW_P): New macro.
23276 (output_asm_line_debug_info): New.
23277 (struct var_loc_node): Add view.
23278 (add_AT_view_list, AT_loc_list): New.
23279 (add_var_loc_to_decl): Add view param. Test it against last.
23280 (new_loc_list): Add view params. Record them.
23281 (AT_loc_list_ptr): Handle loc and view lists.
23282 (view_list_to_loc_list_val_node): New.
23283 (print_dw_val): Handle dw_val_class_view_list.
23284 (size_of_die): Likewise.
23285 (value_format): Likewise.
23286 (loc_list_has_views): New.
23287 (gen_llsym): Set vl_symbol too.
23288 (maybe_gen_llsym, skip_loc_list_entry): New.
23289 (dwarf2out_maybe_output_loclist_view_pair): New.
23290 (output_loc_list): Output view list or entries too.
23291 (output_view_list_offset): New.
23292 (output_die): Handle dw_val_class_view_list.
23293 (output_dwarf_version): New.
23294 (output_compilation_unit_header): Use it.
23295 (output_skeleton_debug_sections): Likewise.
23296 (output_rnglists, output_line_info): Likewise.
23297 (output_pubnames, output_aranges): Update version comments.
23298 (output_one_line_info_table): Output view numbers in asm comments.
23299 (dw_loc_list): Determine current endview, pass it to new_loc_list.
23300 Call maybe_gen_llsym.
23301 (loc_list_from_tree_1): Adjust.
23302 (add_AT_location_description): Create view list attribute if
23303 needed, check it's absent otherwise.
23304 (convert_cfa_to_fb_loc_list): Adjust.
23305 (maybe_emit_file): Call output_asm_line_debug_info for test.
23306 (dwarf2out_var_location): Reset views as needed. Precompute
23307 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
23308 attribute. Set view.
23309 (new_line_info_table): Reset next view.
23310 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
23311 (dwarf2out_source_line): Likewise. Output view resets and labels to
23312 the assembler, or select appropriate line info opcodes.
23313 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
23314 (optimize_string_length): Catch it. Adjust.
23315 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
23316 dw_val_class_view_list, and remove it if no longer needed.
23317 (hash_loc_list): Hash view numbers.
23318 (loc_list_hasher::equal): Compare them.
23319 (optimize_location_lists): Check whether a view list symbol is
23320 needed, and whether the locview attribute is present, and
23321 whether they match. Remove the locview attribute if no longer
23323 (index_location_lists): Call skip_loc_list_entry for test.
23324 (dwarf2out_finish): Call output_asm_line_debug_info for test.
23325 Use output_dwarf_version.
23326 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
23327 (struct dw_val_node): Add val_view_list.
23328 * final.c (SEEN_NEXT_VIEW): New.
23329 (set_next_view_needed): New.
23330 (clear_next_view_needed): New.
23331 (maybe_output_next_view): New.
23332 (final_start_function): Rename to...
23333 (final_start_function_1): ... this. Take pointer to FIRST,
23334 add SEEN parameter. Emit param bindings in the initial view.
23335 (final_start_function): Reintroduce SEEN-less interface.
23336 (final): Rename to...
23337 (final_1): ... this. Take SEEN parameter. Output final pending
23338 next view at the end.
23339 (final): Reintroduce seen-less interface.
23340 (final_scan_insn): Output pending next view before switching
23341 sections or ending a block. Mark the next view as needed when
23342 outputting variable locations. Notify debug backend of section
23343 changes, and of location view changes.
23344 (rest_of_handle_final): Adjust.
23345 * toplev.c (process_options): Autodetect value for debug variable
23346 location views option. Warn on incompat5 without -gdwarf-5.
23347 * doc/invoke.texi (gvariable-location-views): New.
23348 (gvariable-location-views=incompat5): New.
23349 (gno-variable-location-views): New.
23351 2018-02-08 David Malcolm <dmalcolm@redhat.com>
23353 PR tree-optimization/84136
23354 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
23355 that the result of find_edge is non-NULL.
23357 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
23360 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
23361 storing integer register in SImode. Fix cost of 256 and 512
23362 byte aligned SSE register store.
23364 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
23366 * config/i386/i386.c (ix86_multiplication_cost): Fix
23367 multiplication cost for TARGET_AVX512DQ.
23369 2018-02-08 Marek Polacek <polacek@redhat.com>
23371 PR tree-optimization/84238
23372 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
23375 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23377 PR tree-optimization/84265
23378 * tree-vect-stmts.c (vectorizable_store): Don't treat
23379 VMAT_CONTIGUOUS accesses as grouped.
23380 (vectorizable_load): Likewise.
23382 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23384 PR tree-optimization/81635
23385 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
23386 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
23387 (test_round_for_mask): New functions.
23388 (wide_int_cc_tests): Call test_round_for_mask.
23389 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
23390 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
23391 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
23392 range returned by get_range_info.
23394 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
23397 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
23398 * symtab.c: Include builtins.h
23399 (symtab_node::output_to_lto_symbol_table_p): Move here
23400 from lto-streamer-out.c:output_symbol_p.
23401 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
23402 (output_symbol_p): Move all logic to symtab.c
23403 (produce_symtab): Update.
23405 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23407 * config/s390/s390-opts.h (enum indirect_branch): Define.
23408 * config/s390/s390-protos.h (s390_return_addr_from_memory)
23409 (s390_indirect_branch_via_thunk)
23410 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
23411 (enum s390_indirect_branch_type): Define.
23412 * config/s390/s390.c (struct s390_frame_layout, struct
23413 machine_function): Remove.
23414 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
23415 (indirect_branch_table_label_no, indirect_branch_table_name):
23417 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
23418 (enum s390_indirect_branch_option): Define.
23419 (s390_return_addr_from_memory): New function.
23420 (s390_handle_string_attribute): New function.
23421 (s390_attribute_table): Add new attribute handler.
23422 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
23423 (s390_indirect_branch_via_thunk): New function.
23424 (s390_indirect_branch_via_inline_thunk): New function.
23425 (s390_function_ok_for_sibcall): When jumping via thunk disallow
23426 sibling call optimization for non z10 compiles.
23427 (s390_emit_call): Force indirect branch target to be a single
23428 register. Add r1 clobber for non-z10 compiles.
23429 (s390_emit_epilogue): Emit return jump via return_use expander.
23430 (s390_reorg): Handle JUMP_INSNs as execute targets.
23431 (s390_option_override_internal): Perform validity checks for the
23432 new command line options.
23433 (s390_indirect_branch_attrvalue): New function.
23434 (s390_indirect_branch_settings): New function.
23435 (s390_set_current_function): Invoke s390_indirect_branch_settings.
23436 (s390_output_indirect_thunk_function): New function.
23437 (s390_code_end): Implement target hook.
23438 (s390_case_values_threshold): Implement target hook.
23439 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
23441 * config/s390/s390.h (struct s390_frame_layout)
23442 (struct machine_function): Move here from s390.c.
23443 (TARGET_INDIRECT_BRANCH_NOBP_RET)
23444 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
23445 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
23446 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
23447 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
23448 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
23449 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
23450 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
23451 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
23452 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
23453 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
23454 (mnemonic attribute): Add values which aren't recognized
23456 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
23457 pattern for branch conversion. Fix mnemonic attribute.
23458 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
23459 indirect branch via thunk if requested.
23460 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
23461 ("*indirect_jump"): Disable for branch conversion using out of
23463 ("indirect_jump_via_thunk<mode>_z10")
23464 ("indirect_jump_via_thunk<mode>")
23465 ("indirect_jump_via_inlinethunk<mode>_z10")
23466 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
23467 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
23468 ("casesi_jump_via_inlinethunk<mode>_z10")
23469 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
23470 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
23471 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
23472 ("*indirect2_jump"): Disable for branch conversion.
23473 ("casesi_jump"): Turn into expander and expand patterns for branch
23475 ("return_use"): New expander.
23476 ("*return"): Emit return via thunk and rename it to ...
23477 ("*return<mode>"): ... this one.
23478 * config/s390/s390.opt: Add new options and and enum for the
23481 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23483 * lra-constraints.c (match_reload): Unconditionally use
23484 gen_lowpart_SUBREG, rather than selecting between that
23485 and equivalent gen_rtx_SUBREG code.
23487 2018-02-08 Richard Biener <rguenther@suse.de>
23489 PR tree-optimization/84233
23490 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
23491 changed flag instead of boguously re-using phi_inserted.
23493 2018-02-08 Martin Jambor <mjambor@suse.cz>
23495 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
23496 static local variables.
23498 2018-02-08 Richard Biener <rguenther@suse.de>
23500 PR tree-optimization/84278
23501 * tree-vect-stmts.c (vectorizable_store): When looking for
23502 smaller vector types to perform grouped strided loads/stores
23503 make sure the mode is supported by the target.
23504 (vectorizable_load): Likewise.
23506 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
23508 * config/aarch64/aarch64.c (aarch64_components_for_bb):
23509 Increase LDP/STP opportunities by adding adjacent callee-saves.
23511 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
23513 PR rtl-optimization/84068
23514 PR rtl-optimization/83459
23515 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
23517 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
23519 PR tree-optimization/84224
23520 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
23521 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
23522 non-zero arguments.
23524 2018-02-07 Iain Sandoe <iain@codesourcery.com>
23527 * config/rs6000/altivec.md (*restore_world): Remove LR use.
23528 * config/rs6000/predicates.md (restore_world_operation): Adjust op
23529 count, remove one USE.
23531 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23533 * doc/install.texi (Configuration): Document the
23534 --with-long-double-format={ibm,ieee} PowerPC configuration
23538 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
23539 Convert from define_expand to be define_insn_and_split. Rework
23540 float/double/_Float128 conversions to QI/HI/SImode to work with
23541 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
23542 conversions to QI/HImode types did a store and then a load to
23543 truncate the value. For conversions to VSX registers, don't split
23544 the insn, instead emit the code directly. Use the code iterator
23545 any_fix to combine signed and unsigned conversions.
23546 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
23547 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
23548 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
23549 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
23550 (fix_<mode>di2_hw): Likewise.
23551 (fixuns_<mode>di2_hw): Likewise.
23552 (fix_<mode>si2_hw): Likewise.
23553 (fixuns_<mode>si2_hw): Likewise.
23554 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
23555 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
23556 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
23557 fix<uns>_trunc<SFDF:mode>si2_p8.
23558 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
23559 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23560 (fix<uns>_<mode>_mem): Likewise.
23561 (fctiw<u>z_<mode>_mem): Likewise.
23562 (fix<uns>_<mode>_mem): Likewise.
23563 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
23564 the register allocator from doing a direct move to the GPRs to do
23565 a store, and instead use the ISA 3.0 store byte/half-word from
23566 vector register instruction. For IEEE 128-bit floating point,
23567 also optimize stores of 32-bit ints.
23568 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
23570 2018-02-07 Alan Hayward <alan.hayward@arm.com>
23572 * genextract.c (push_pathstr_operand): New function to support
23574 (walk_rtx): Call push_pathstr_operand.
23575 (print_path): Support [a-zA-Z].
23577 2018-02-07 Richard Biener <rguenther@suse.de>
23579 PR tree-optimization/84037
23580 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
23581 (cse_and_gimplify_to_preheader): Declare.
23582 (vect_get_place_in_interleaving_chain): Likewise.
23583 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
23585 (_loop_vec_info::~_loop_vec_info): Delete it.
23586 (cse_and_gimplify_to_preheader): New function.
23587 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
23588 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
23589 (vectorizable_load): Likewise. For grouped stores always base
23590 the IV on the first element.
23591 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
23592 condition before gimplifying.
23594 2018-02-07 Jakub Jelinek <jakub@redhat.com>
23596 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
23597 *DIV_EXPR and *MOD_EXPR.
23599 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
23602 * config/i386/i386.c (ix86_option_override_internal): Mask out
23603 the CF_SET bit when checking -fcf-protection.
23605 2018-02-07 Tom de Vries <tom@codesourcery.com>
23608 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
23611 2018-02-07 Richard Biener <rguenther@suse.de>
23613 PR tree-optimization/84204
23614 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
23617 PR tree-optimization/84205
23618 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
23619 special-case isl_ast_op_zdiv_r.
23621 PR tree-optimization/84223
23622 * graphite-scop-detection.c (gather_bbs::before_dom_children):
23623 Only add conditions from within the region.
23624 (gather_bbs::after_dom_children): Adjust.
23626 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
23629 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
23630 * config/avr/avr.md: Only post-reload split REG-REG moves if
23631 either register is GENERAL_REG_P.
23633 2018-02-07 Jakub Jelinek <jakub@redhat.com>
23635 PR tree-optimization/84235
23636 * tree-ssa-scopedtables.c
23637 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
23638 if the subtraction is performed in floating point type where NaNs are
23639 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
23640 build 1. Formatting fix.
23642 2018-02-06 Jakub Jelinek <jakub@redhat.com>
23645 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
23646 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
23647 and skip it regardless of bb boundaries. Use CALL_P macro,
23648 don't test INSN_P (insn) together with CALL_P or JUMP_P check
23649 unnecessarily, formatting fix.
23651 2018-02-06 Michael Collison <michael.collison@arm.com>
23653 * config/arm/thumb2.md:
23654 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
23655 (*thumb_mov_notscc): Ditto.
23657 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
23660 * config/rs6000/rs6000.md (su code attribute): Use "u" for
23661 unsigned_fix, not "s".
23663 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23665 * configure.ac (gcc_fn_eh_frame_ro): New function.
23666 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
23667 correct .eh_frame permissions.
23668 * configure: Regenerate.
23670 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
23672 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
23673 irrelevant options.
23675 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23677 * config/rs6000/rs6000.c (rs6000_option_override_internal):
23678 Display warning message for -mno-speculate-indirect-jumps.
23680 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
23682 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
23684 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
23686 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
23688 PR tree-optimization/84225
23689 * tree-eh.c (find_trapping_overflow): Only call
23690 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
23692 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
23695 * config/i386/i386.c: Reimplement the check of possible options
23696 -mibt/-mshstk conbination. Change error messages.
23697 * doc/invoke.texi: Fix a typo: remove extra '='.
23699 2018-02-06 Marek Polacek <polacek@redhat.com>
23701 PR tree-optimization/84228
23702 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
23704 2018-02-06 Tamar Christina <tamar.christina@arm.com>
23707 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
23708 emitted arch directives.
23709 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
23710 __ARM_FEATURE_COPROC before changing architectures.
23712 2018-02-06 Richard Biener <rguenther@suse.de>
23714 * config/i386/i386.c (print_reg): Fix typo.
23715 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
23717 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
23719 * configure: Regenerate.
23721 2018-02-05 Martin Sebor <msebor@redhat.com>
23723 PR tree-optimization/83369
23724 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
23727 2018-02-05 Martin Liska <mliska@suse.cz>
23729 * doc/invoke.texi: Cherry-pick upstream r323995.
23731 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
23733 * ira.c (ira_init_register_move_cost): Adjust comment.
23735 2018-02-05 Martin Liska <mliska@suse.cz>
23737 PR gcov-profile/84137
23738 * doc/gcov.texi: Fix typo in documentation.
23740 2018-02-05 Martin Liska <mliska@suse.cz>
23742 PR gcov-profile/83879
23743 * doc/gcov.texi: Document necessity of --dynamic-list-data when
23744 using dlopen functionality.
23746 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
23748 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
23749 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
23750 _mm_maskz_range_ss, _mm_mask_range_round_ss,
23751 _mm_maskz_range_round_ss): New intrinsics.
23752 (__builtin_ia32_rangesd128_round)
23753 (__builtin_ia32_rangess128_round): Remove.
23754 (__builtin_ia32_rangesd128_mask_round,
23755 __builtin_ia32_rangess128_mask_round): New builtins.
23756 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
23757 __builtin_ia32_rangess128_round): Remove.
23758 (__builtin_ia32_rangesd128_mask_round,
23759 __builtin_ia32_rangess128_mask_round): New builtins.
23760 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
23761 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
23762 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
23763 "<round_saeonly_constraint>")): Changed to ...
23764 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
23765 "<round_saeonly_scalar_constraint>")): ... this.
23766 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
23767 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
23768 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
23769 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
23770 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
23772 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
23774 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
23776 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
23777 Remove all values except native, 8540 and 8548.
23779 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
23781 * config/i386/i386.c (ix86_output_function_return): Pass
23782 INVALID_REGNUM, instead of -1, as invalid register number to
23783 indirect_thunk_name and output_indirect_thunk.
23785 2018-02-02 Julia Koval <julia.koval@intel.com>
23787 * config.gcc: Add -march=icelake.
23788 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
23789 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
23790 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
23791 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
23792 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
23793 (processor_target_table): Add icelake.
23794 (ix86_option_override_internal): Handle new PTAs.
23795 (get_builtin_code_for_version): Handle icelake.
23796 (M_INTEL_COREI7_ICELAKE): New.
23797 (fold_builtin_cpu): Handle icelake.
23798 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
23799 * doc/invoke.texi: Add -march=icelake.
23801 2018-02-02 Julia Koval <julia.koval@intel.com>
23803 * config/i386/i386.c (ix86_option_override_internal): Change flags type
23804 to wide_int_bitmask.
23805 * wide-int-bitmask.h: New.
23807 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
23810 * config/i386/i386.md: Replace Pmode with word_mode in
23811 builtin_setjmp_setup and builtin_longjmp to support x32.
23813 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
23817 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
23819 (rs6000_supported_cpu_names): New static variable.
23820 (linux_cpu_translation_table): Likewise.
23821 (elf_platform) <cpu>: Define new static variable and use it.
23822 Translate kernel AT_PLATFORM name to canonical name if needed.
23823 Error if platform name is unknown.
23825 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
23828 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
23830 2018-02-01 Jeff Law <law@redhat.com>
23833 * config/i386/i386.c (release_scratch_register_on_entry): Add new
23834 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
23835 the scratch if RELEASE_VIA_POP is false.
23836 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
23837 If we have to save a temporary register, decrement SIZE appropriately.
23838 Pass new arguments to release_scratch_register_on_entry.
23839 (ix86_adjust_stack_and_probe): Likewise.
23840 (ix86_emit_probe_stack_range): Pass new arguments to
23841 release_scratch_register_on_entry.
23843 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
23845 PR rtl-optimization/84157
23846 * combine.c (change_zero_ext): Use REG_P predicate in
23847 front of HARD_REGISTER_P predicate.
23849 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
23851 * config/avr/avr.c (avr_option_override): Move disabling of
23852 -fdelete-null-pointer-checks to...
23853 * common/config/avr/avr-common.c (avr_option_optimization_table):
23856 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23858 PR tree-optimization/81635
23859 * tree-data-ref.c (split_constant_offset_1): For types that
23860 wrap on overflow, try to use range info to prove that wrapping
23863 2018-02-01 Renlin Li <renlin.li@arm.com>
23866 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
23867 TAILCALL_ADDR_REGS.
23868 (aarch64_register_move_cost): Likewise.
23869 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
23870 TAILCALL_ADDR_REGS.
23871 (REG_CLASS_NAMES): Likewise.
23872 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
23873 TAILCALL_ADDR_REGS. Remove IP registers.
23874 * config/aarch64/aarch64.md (Ucs): Update register constraint.
23876 2018-02-01 Richard Biener <rguenther@suse.de>
23878 * domwalk.h (dom_walker::dom_walker): Add additional constructor
23879 for specifying RPO order and allow NULL for that.
23880 * domwalk.c (dom_walker::dom_walker): Likewise.
23881 (dom_walker::walk): Handle NULL RPO order.
23882 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
23884 (rewrite_update_dom_walker): Likewise.
23885 (mark_def_dom_walker): Likewise.
23887 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23889 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
23890 (aarch64_maybe_expand_sve_subreg_move): Declare.
23891 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
23892 * config/aarch64/predicates.md (aarch64_any_register_operand): New
23894 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
23895 that are semantically a reverse operation.
23896 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
23897 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
23898 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
23900 (aarch64_can_change_mode_class): For big-endian, forbid changes
23901 between two SVE modes if they have different element sizes.
23903 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23905 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
23906 the TImode handling for big-endian targets.
23908 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23910 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
23911 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
23913 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
23914 Remove BSWAP handing for big-endian targets and use the form of
23915 LD1RQ appropariate for the mode.
23917 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23919 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
23920 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
23921 duplicated element.
23923 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
23926 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
23927 check for operands that need to go through aarch64_sve_reload_be.
23929 2018-02-01 Jakub Jelinek <jakub@redhat.com>
23931 PR tree-optimization/81661
23932 PR tree-optimization/84117
23933 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
23934 * tree-eh.c: Include gimplify.h.
23935 (find_trapping_overflow, replace_trapping_overflow,
23936 rewrite_to_non_trapping_overflow): New functions.
23937 * tree-vect-loop.c: Include tree-eh.h.
23938 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
23939 * tree-data-ref.c: Include tree-eh.h.
23940 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
23942 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
23944 PR rtl-optimization/84123
23945 * combine.c (change_zero_ext): Check if hard register satisfies
23946 can_change_dest_mode before calling gen_lowpart_SUBREG.
23948 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
23951 * ira.c (ira_init_register_move_cost): Remove assert.
23953 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
23955 PR rtl-optimization/84071
23956 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
23957 * doc/tm.texi: Regenerate.
23959 2018-01-31 Richard Biener <rguenther@suse.de>
23961 PR tree-optimization/84132
23962 * tree-data-ref.c (analyze_miv_subscript): Properly
23963 check whether evolution_function_is_affine_multivariate_p
23964 before calling gcd_of_steps_may_divide_p.
23966 2018-01-31 Julia Koval <julia.koval@intel.com>
23969 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
23970 * config/i386/i386.md (rdpid_rex64) New.
23971 (rdpid): Make 32bit only.
23973 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
23976 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
23977 an IDENTIFIER_NODE for FUNCTION_TYPE's.
23979 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
23982 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
23984 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
23986 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
23988 PR rtl-optimization/84071
23989 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
23990 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
23992 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
23994 * config/arc/arc.c (arc_handle_aux_attribute): New function.
23995 (arc_attribute_table): Add 'aux' attribute.
23996 (arc_in_small_data_p): Consider aux like variables.
23997 (arc_is_aux_reg_p): New function.
23998 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
23999 (arc_get_aux_arg): New function.
24000 (prepare_move_operands): Handle aux-register access.
24001 (arc_handle_aux_attribute): New function.
24002 * doc/extend.texi (ARC Variable attributes): Add subsection.
24004 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
24006 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
24007 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
24008 (arc_attribute_table): Add 'uncached' attribute.
24009 (arc_print_operand): Print '.di' flag for uncached memory
24011 (arc_in_small_data_p): Do not consider for small data the uncached
24013 (arc_is_uncached_mem_p): New function.
24014 * config/arc/predicates.md (compact_store_memory_operand): Check
24015 for uncached memory accesses.
24016 (nonvol_nonimm_operand): Likewise.
24017 * doc/extend.texi (ARC Type Attribute): New subsection.
24019 2018-01-31 Jakub Jelinek <jakub@redhat.com>
24022 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
24023 falign-loops=): Add Optimization flag.
24025 2018-01-30 Jeff Law <law@redhat.com>
24028 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
24029 INT_REGISTERS_SAVED. Check it prior to calling
24030 get_scratch_register_on_entry.
24031 (ix86_adjust_stack_and_probe): Similarly.
24032 (ix86_emit_probe_stack_range): Similarly.
24033 (ix86_expand_prologue): Corresponding changes.
24035 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24038 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
24039 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
24041 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
24044 * lra-constraints.c (curr_insn_transform): Process AND in the
24047 2018-01-30 Jakub Jelinek <jakub@redhat.com>
24049 PR rtl-optimization/83986
24050 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
24051 dependence against last_pending_memory_flush in addition to
24052 pending_jump_insns.
24054 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
24056 PR tree-optimization/81611
24057 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
24060 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24063 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
24066 2018-01-30 Richard Biener <rguenther@suse.de>
24067 Jakub Jelinek <jakub@redhat.com>
24069 PR tree-optimization/84111
24070 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
24071 inner loops added during recursion, as they don't have up-to-date
24074 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
24077 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
24078 (can_inline_edge_by_limits_p): ... here.
24079 (can_early_inline_edge_p, check_callers,
24080 update_caller_keys, update_callee_keys, recursive_inlining,
24081 add_new_edges_to_heap, speculation_useful_p,
24082 inline_small_functions,
24083 inline_small_functions, flatten_function,
24084 inline_to_all_callers_1): Update.
24086 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
24088 * profile-count.c (profile_count::combine_with_ipa_count): Handle
24091 2018-01-30 Richard Biener <rguenther@suse.de>
24093 PR tree-optimization/83008
24094 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
24095 invariant and constant vector uses in stmts when they need
24096 more than one stmt.
24098 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24101 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
24102 * configure: Regenerate.
24104 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24106 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
24108 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
24109 Use gen_rtx_REG rather than gen_lowpart.
24111 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24113 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
24114 rather than 0 when creating partial subregs.
24116 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24118 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
24121 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
24124 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
24125 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
24126 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
24127 flags. This restores the settings used before the 2017-07-24.
24128 Turning off pre increment/decrement/modify allows IVOPTS to
24129 optimize DF/SF loops where the index is an int.
24131 2018-01-29 Richard Biener <rguenther@suse.de>
24132 Kelvin Nilsen <kelvin@gcc.gnu.org>
24135 * tree-vect-stmts.c (vectorizable_call): Don't call
24136 targetm.vectorize_builtin_md_vectorized_function if callee is
24139 2018-01-22 Carl Love <cel@us.ibm.com>
24141 * doc/extend.tex: Fix typo in second arg in
24142 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
24144 2018-01-29 Richard Biener <rguenther@suse.de>
24146 PR tree-optimization/84086
24147 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
24148 (flush_ssaname_freelist): When SSA names were released reset
24149 the SCEV hash table.
24151 2018-01-29 Richard Biener <rguenther@suse.de>
24153 PR tree-optimization/84057
24154 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
24155 removed paths when removing edges.
24157 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
24159 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
24160 -mfunction-return=@var{choice}.
24162 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
24164 PR diagnostic/84034
24165 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
24166 Handle CR like TAB.
24167 (layout::print_source_line): Likewise.
24168 (test_get_line_width_without_trailing_whitespace): Add test cases.
24170 2018-01-27 Jakub Jelinek <jakub@redhat.com>
24172 PR middle-end/84040
24173 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
24176 2018-01-26 Jim Wilson <jimw@sifive.com>
24178 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
24180 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
24183 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24185 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
24186 and CMP + SUB-immediate -> SUBS.
24188 2018-01-26 Martin Sebor <msebor@redhat.com>
24190 PR tree-optimization/83896
24191 * tree-ssa-strlen.c (get_string_len): Rename...
24192 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
24193 Avoid assuming length is constant.
24194 (handle_char_store): Use HOST_WIDE_INT for string length.
24196 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
24199 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
24200 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
24202 2018-01-26 Richard Biener <rguenther@suse.de>
24204 PR rtl-optimization/84003
24205 * dse.c (record_store): Only record redundant stores when
24206 the earlier store aliases at least all accesses the later one does.
24208 2018-01-26 Jakub Jelinek <jakub@redhat.com>
24210 PR rtl-optimization/83985
24211 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
24212 REG_CFA_RESTORE insns.
24213 (delete_unmarked_insns): Don't ignore separate shrink wrapping
24214 REG_CFA_RESTORE insns here.
24217 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
24218 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
24220 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24222 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
24223 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
24224 (arc_init): Likewise.
24225 (arc_override_options): Likewise.
24226 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
24228 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
24230 * config/arc/arc.h (TARGET_DBNZ): Define.
24231 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
24232 properly set the tune attribute.
24233 (dbnz): Use TARGET_DBNZ guard.
24234 * config/arc/arc.opt (mtune): Add core3 option.
24236 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24238 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
24239 recognize new pic like addresses.
24240 (arc_delegitimize_address): Clean up.
24242 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24244 * config/arc/arc-arches.def: Option mrf16 valid for all
24246 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
24247 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
24248 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
24249 * config/arc/arc-tables.opt: Regenerate.
24250 * config/arc/arc.c (arc_conditional_register_usage): Handle
24251 reduced register file case.
24252 (arc_file_start): Set must have build attributes.
24253 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
24254 mrf16 option value.
24255 * config/arc/arc.opt (mrf16): Add new option.
24256 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
24257 * config/arc/genmultilib.awk: Handle new mrf16 option.
24258 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
24259 * config/arc/t-multilib: Regenerate.
24260 * doc/invoke.texi (ARC Options): Document mrf16 option.
24262 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24264 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
24265 * config/arc/arc.c (arc_handle_secure_attribute): New function.
24266 (arc_attribute_table): Add 'secure_call' attribute.
24267 (arc_print_operand): Print secure call operand.
24268 (arc_function_ok_for_sibcall): Don't optimize tail calls when
24270 (arc_is_secure_call_p): New function. * config/arc/arc.md
24271 (call_i): Add support for sjli instruction.
24272 (call_value_i): Likewise.
24273 * config/arc/constraints.md (Csc): New constraint.
24275 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24276 John Eric Martin <John.Martin@emmicro-us.com>
24278 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
24279 * config/arc/arc.c (_arc_jli_section): New struct.
24280 (arc_jli_section): New type.
24281 (rc_jli_sections): New static variable.
24282 (arc_handle_jli_attribute): New function.
24283 (arc_attribute_table): Add jli_always and jli_fixed attribute.
24284 (arc_file_end): New function.
24285 (TARGET_ASM_FILE_END): Define.
24286 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
24287 (arc_add_jli_section): New function.
24288 (jli_call_scan): Likewise.
24289 (arc_reorg): Call jli_call_scan.
24290 (arc_output_addsi): Remove 'S' from printing asm operand.
24291 (arc_is_jli_call_p): New function.
24292 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
24294 (movhi_insn): Likewise.
24295 (movsi_insn): Likewise.
24296 (movsi_set_cc_insn): Likewise.
24297 (loadqi_update): Likewise.
24298 (load_zeroextendqisi_update): Likewise.
24299 (load_signextendqisi_update): Likewise.
24300 (loadhi_update): Likewise.
24301 (load_zeroextendhisi_update): Likewise.
24302 (load_signextendhisi_update): Likewise.
24303 (loadsi_update): Likewise.
24304 (loadsf_update): Likewise.
24305 (movsicc_insn): Likewise.
24306 (bset_insn): Likewise.
24307 (bxor_insn): Likewise.
24308 (bclr_insn): Likewise.
24309 (bmsk_insn): Likewise.
24310 (bicsi3_insn): Likewise.
24311 (cmpsi_cc_c_insn): Likewise.
24312 (movsi_ne): Likewise.
24313 (movsi_cond_exec): Likewise.
24314 (clrsbsi2): Likewise.
24315 (norm_f): Likewise.
24325 (call_i): Remove 'S' asm letter, add jli instruction.
24326 (call_value_i): Likewise.
24327 * config/arc/arc.op (mjli-always): New option.
24328 * config/arc/constraints.md (Cji): New constraint.
24329 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
24331 (subsf3_fpx): Likewise.
24332 (mulsf3_fpx): Likewise.
24333 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
24335 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
24336 function attrbutes.
24337 * doc/invoke.texi (ARC): Document mjli-always option.
24339 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
24341 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
24342 avoid addition with 0 and use incw and decw where possible.
24344 2018-01-26 Richard Biener <rguenther@suse.de>
24346 PR tree-optimization/81082
24347 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
24348 association if it requires casting to unsigned.
24349 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
24350 from fold_plusminus_mult_expr to catch important cases late when
24351 range info is available.
24353 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24355 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
24356 * configure.ac (hidden_linkonce): New test.
24357 * configure: Regenerate.
24358 * config.in: Regenerate.
24360 2018-01-26 Julia Koval <julia.koval@intel.com>
24362 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
24363 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
24364 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
24365 _mm_mask_bitshuffle_epi64_mask): Fix type.
24366 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
24367 USI_FTYPE_V4DI_V4DI_USI): Remove.
24368 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
24369 __builtin_ia32_vpshufbitqmb256_mask,
24370 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
24371 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
24372 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
24374 2018-01-26 Alan Modra <amodra@gmail.com>
24377 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
24378 UNSPEC_VBPERMQ. Sort other unspecs.
24380 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
24382 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
24384 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
24386 PR middle-end/83055
24387 * predict.c (drop_profile): Do not push/pop cfun; update also
24389 (handle_missing_profiles): Fix logic looking for zero profiles.
24391 2018-01-25 Jakub Jelinek <jakub@redhat.com>
24393 PR middle-end/83977
24394 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
24395 on functions with #pragma omp declare simd or functions with simd
24397 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
24398 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
24399 Remove trailing \n from warning_at calls.
24401 2018-01-25 Tom de Vries <tom@codesourcery.com>
24404 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
24405 for neutered workers.
24407 2018-01-24 Joseph Myers <joseph@codesourcery.com>
24410 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
24411 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
24413 2018-01-24 Jeff Law <law@redhat.com>
24416 * i386.c (get_probe_interval): Move to earlier point.
24417 (ix86_compute_frame_layout): If -fstack-clash-protection and
24418 the frame is larger than the probe interval, then use pushes
24419 to save registers rather than reg->mem moves.
24420 (ix86_expand_prologue): Remove conditional for int_registers_saved
24423 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
24426 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
24427 min/max for never referenced object.
24429 2018-01-24 Jakub Jelinek <jakub@redhat.com>
24431 PR middle-end/83977
24432 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
24434 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
24435 attributes from DECL_ATTRIBUTES (decl) without affecting
24436 DECL_ATTRIBUTES (current_function_decl).
24437 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
24438 functions with non-NULL DECL_ABSTRACT_ORIGIN.
24440 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
24442 PR tree-optimization/83979
24443 * fold-const.c (fold_comparison): Use constant_boolean_node
24444 instead of boolean_{true,false}_node.
24446 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
24448 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
24451 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24453 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
24454 Simplify the clause that sets the length attribute.
24455 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
24456 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
24457 clause that sets the length attribute.
24458 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
24460 2018-01-24 Tom de Vries <tom@codesourcery.com>
24463 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
24464 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
24465 Add strict parameter.
24466 (prevent_branch_around_nothing): Insert dummy insn between branch to
24467 label and label with no ptx insn inbetween.
24468 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
24470 2018-01-24 Tom de Vries <tom@codesourcery.com>
24473 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
24474 for neutered threads in warp.
24475 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
24477 2018-01-24 Richard Biener <rguenther@suse.de>
24479 PR tree-optimization/83176
24480 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
24483 2018-01-24 Richard Biener <rguenther@suse.de>
24485 PR tree-optimization/82819
24486 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
24487 code generating pluses that are no-ops in the target precision.
24489 2018-01-24 Richard Biener <rguenther@suse.de>
24491 PR middle-end/84000
24492 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
24494 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
24496 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
24497 to merge probabilities.
24498 * predict.c (probably_never_executed): Also mark as cold functions
24499 with global 0 profile and guessed local profile.
24500 * profile-count.c (profile_probability::combine_with_count): New
24502 * profile-count.h (profile_probability::operator*,
24503 profile_probability::operator*=, profile_probability::operator/,
24504 profile_probability::operator/=): Reduce precision to adjusted
24505 and set value to guessed on contradictory divisions.
24506 (profile_probability::combine_with_freq): Remove.
24507 (profile_probability::combine_wiht_count): Declare.
24508 (profile_count::force_nonzero):: Set to adjusted.
24509 (profile_count::probability_in):: Set quality to adjusted.
24510 * tree-ssa-tail-merge.c (replace_block_by): Use
24511 combine_with_count.
24513 2018-01-23 Andrew Waterman <andrew@sifive.com>
24514 Jim Wilson <jimw@sifive.com>
24516 * config/riscv/riscv.c (riscv_stack_boundary): New.
24517 (riscv_option_override): Set riscv_stack_boundary. Handle
24518 riscv_preferred_stack_boundary_arg.
24519 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
24520 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
24521 (STACK_BOUNDARY): Set to riscv_stack_boundary.
24522 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
24523 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
24524 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
24526 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
24529 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
24530 of struct ix86_frame.
24531 (ix86_expand_epilogue): Likewise. Add a local variable for
24532 the reg_save_offset field in struct ix86_frame.
24534 2018-01-23 Bin Cheng <bin.cheng@arm.com>
24536 PR tree-optimization/82604
24537 * tree-loop-distribution.c (enum partition_kind): New enum item
24538 PKIND_PARTIAL_MEMSET.
24539 (partition_builtin_p): Support above new enum item.
24540 (generate_code_for_partition): Ditto.
24541 (compute_access_range): Differentiate cases that equality can be
24542 proven at all loops, the innermost loops or no loops.
24543 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
24544 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
24545 (finalize_partitions, distribute_loop): Don't fuse partition of
24546 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
24547 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
24548 parloop is enabled.
24550 2018-01-23 Martin Liska <mliska@suse.cz>
24552 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
24553 order to ignore the predictor.
24554 (PRED_POLYMORPHIC_CALL): Likewise.
24555 (PRED_RECURSIVE_CALL): Likewise.
24557 2018-01-23 Martin Liska <mliska@suse.cz>
24559 * tree-profile.c (tree_profiling): Print function header to
24560 aware reader which function we are working on.
24561 * value-prof.c (gimple_find_values_to_profile): Do not print
24562 not interesting value histograms.
24564 2018-01-23 Martin Liska <mliska@suse.cz>
24566 * profile-count.h (enum profile_quality): Add
24567 profile_uninitialized as the first value. Do not number values
24568 as they are zero based.
24569 (profile_count::verify): Update sanity check.
24570 (profile_probability::verify): Likewise.
24572 2018-01-23 Nathan Sidwell <nathan@acm.org>
24574 * doc/invoke.texi (ffor-scope): Deprecate.
24576 2018-01-23 David Malcolm <dmalcolm@redhat.com>
24578 PR tree-optimization/83510
24579 * domwalk.c (set_all_edges_as_executable): New function.
24580 (dom_walker::dom_walker): Convert bool param
24581 "skip_unreachable_blocks" to enum reachability. Move setup of
24582 edge flags to set_all_edges_as_executable and only do it when
24583 reachability is REACHABLE_BLOCKS.
24584 * domwalk.h (enum dom_walker::reachability): New enum.
24585 (dom_walker::dom_walker): Convert bool param
24586 "skip_unreachable_blocks" to enum reachability.
24587 (set_all_edges_as_executable): New decl.
24588 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
24589 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
24591 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
24592 but converting true to REACHABLE_BLOCKS.
24593 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
24595 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
24596 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
24597 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
24599 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
24600 if check_all_array_refs will be called.
24602 2018-01-23 David Malcolm <dmalcolm@redhat.com>
24604 * tree.c (selftest::test_location_wrappers): Add more test
24607 2018-01-23 David Malcolm <dmalcolm@redhat.com>
24609 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
24610 (selftest::test_bit_in_range): Likewise.
24612 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
24615 * doc/sourcebuild.texi (vect_float): Say that the selector
24616 only describes the situation when -funsafe-math-optimizations is on.
24617 (vect_float_strict): Document.
24619 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
24621 PR tree-optimization/83965
24622 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
24623 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
24624 instead of checking only for a reduction.
24625 (vect_recog_widen_sum_pattern): Likewise.
24627 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
24629 * predict.c (probably_never_executed): Only use precise profile info.
24630 (compute_function_frequency): Skip after inlining hack since we now
24631 have quality checking.
24633 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
24635 * profile-count.h (profile_probability::very_unlikely,
24636 profile_probability::unlikely, profile_probability::even): Set
24637 precision to guessed.
24639 2018-01-23 Richard Biener <rguenther@suse.de>
24641 PR tree-optimization/83963
24642 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
24643 Properly terminate dominator walk when crossing the exit edge not
24644 when visiting its source block.
24646 2018-01-23 Jakub Jelinek <jakub@redhat.com>
24649 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
24650 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
24652 2018-01-22 Jakub Jelinek <jakub@redhat.com>
24654 PR tree-optimization/83957
24655 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
24656 semicolon after for body surrounded by braces.
24658 PR tree-optimization/83081
24659 * profile-count.h (profile_probability::split): New method.
24660 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
24661 Use profile_probability::split.
24662 (do_compare_rtx_and_jump): Fix adjustment of probabilities
24663 when splitting a single conditional jump into 2.
24665 2018-01-22 David Malcolm <dmalcolm@redhat.com>
24667 PR tree-optimization/69452
24668 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
24671 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24673 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
24674 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
24675 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
24677 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24679 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
24681 * config/rl78/rl78.md (movdi): New define_expand.
24682 * config/rl78/rl78.c (rl78_split_movdi): New function.
24684 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
24687 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
24689 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
24690 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
24691 128-bit to produce an UNSPEC move to get the double word with the
24692 signbit and then a shift directly to do signbit.
24693 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
24694 implementation with a new version that just does either a direct
24695 move or a regular move. Move memory interface to separate insns.
24696 Move insns so they are next to the expander.
24697 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
24698 with signbit move. Split big and little endian case.
24699 (signbit<mode>2_dm_mem_le): Likewise.
24700 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
24701 (signbit<mode>2_dm2): Likewise.
24703 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24705 * config/rl78/rl78.md (anddi3): New define_expand.
24707 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24709 * config/rl78/rl78.md (umindi3): New define_expand.
24711 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24713 * config/rl78/rl78.md (smindi3): New define_expand.
24715 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24717 * config/rl78/rl78.md (smaxdi3): New define_expand.
24719 2018-01-22 Carl Love <cel@us.ibm.com>
24721 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
24722 LVX_V1TI): Add macro expansion.
24723 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
24724 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
24725 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
24726 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
24727 Change check to determine if the instruction is a byte reversing
24728 entry. Fix typo in comment.
24729 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
24730 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
24731 Add def_builtin calls for new builtins.
24732 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
24733 Add define_insn expansion.
24735 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24737 * config/rl78/rl78.md (umaxdi3): New define_expand.
24739 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24741 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
24742 for non-QImode registers.
24744 2018-01-22 Richard Biener <rguenther@suse.de>
24746 PR tree-optimization/83963
24747 * graphite-scop-detection.c (scop_detection::get_sese): Delay
24748 including the loop exit block.
24749 (scop_detection::merge_sese): Likewise.
24750 (scop_detection::add_scop): Do it here instead.
24752 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24754 * doc/sourcebuild.texi (arm_softfloat): Document.
24756 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
24759 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
24760 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
24761 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
24763 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24764 David Edelsohn <dje.gcc@gmail.com>
24767 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
24768 Change "crset eq" to "crset 2".
24769 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
24770 (*call_indirect_aix<mode>_nospec): Likewise.
24771 (*call_value_indirect_aix<mode>_nospec): Likewise.
24772 (*call_indirect_elfv2<mode>_nospec): Likewise.
24773 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
24774 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
24775 change assembly output from . to $.
24776 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
24777 (indirect_jump<mode>_nospec): Change assembly output from . to $.
24778 (*tablejump<mode>_internal1_nospec): Likewise.
24780 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
24783 * config/sh/sh_optimize_sett_clrt.cc:
24784 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
24786 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
24788 PR tree-optimization/83940
24789 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
24790 offset_dt to vect_constant_def rather than vect_unknown_def_type.
24791 (vect_check_load_store_mask): Add a mask_dt_out parameter and
24792 use it to pass back the definition type.
24793 (vect_check_store_rhs): Likewise rhs_dt_out.
24794 (vect_build_gather_load_calls): Add a mask_dt argument and use
24795 it instead of a call to vect_is_simple_use.
24796 (vectorizable_store): Update calls to vect_check_load_store_mask
24797 and vect_check_store_rhs. Use the dt returned by the latter instead
24798 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
24799 instead of calls to vect_is_simple_use. Pass the scalar rather
24800 than the vector operand to vect_is_simple_use when handling
24801 second and subsequent copies of an rhs value.
24802 (vectorizable_load): Update calls to vect_check_load_store_mask
24803 and vect_build_gather_load_calls. Use the cached mask_dt and
24804 gs_info.offset_dt instead of calls to vect_is_simple_use.
24806 2018-01-20 Jakub Jelinek <jakub@redhat.com>
24808 PR middle-end/83945
24809 * tree-emutls.c: Include gimplify.h.
24810 (lower_emutls_2): New function.
24811 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
24812 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
24813 it before further processing.
24816 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
24817 UINTVAL (trueop1) instead of INTVAL (op1).
24819 2018-01-19 Jakub Jelinek <jakub@redhat.com>
24823 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
24824 INCOMING_FRAME_SP_OFFSET if not defined.
24825 (scan_trace): Add ENTRY argument. If true and
24826 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
24827 emit a note to adjust the CFA offset.
24828 (create_cfi_notes): Adjust scan_trace callers.
24829 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
24830 INCOMING_FRAME_SP_OFFSET in the CIE.
24831 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
24832 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
24834 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
24835 * doc/tm.texi: Regenerated.
24837 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24839 PR rtl-optimization/83147
24840 * lra-constraints.c (remove_inheritance_pseudos): Use
24841 lra_substitute_pseudo_within_insn.
24843 2018-01-19 Tom de Vries <tom@codesourcery.com>
24844 Cesar Philippidis <cesar@codesourcery.com>
24847 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
24849 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
24852 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
24853 spaces for function labels.
24855 2018-01-19 Martin Liska <mliska@suse.cz>
24857 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
24858 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
24859 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
24860 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
24861 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
24862 (PRED_CONST_RETURN): Change from 69 to 65.
24863 (PRED_NULL_RETURN): Change from 91 to 71.
24864 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
24865 (PRED_LOOP_GUARD): Change from 66 to 73.
24867 2018-01-19 Martin Liska <mliska@suse.cz>
24869 * predict.c (predict_insn_def): Add new assert.
24870 (struct branch_predictor): Change type to signed integer.
24871 (test_prediction_value_range): Amend test to cover
24872 PROB_UNINITIALIZED.
24873 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
24874 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
24875 (PRED_LOOP_ITERATIONS_MAX): Likewise.
24876 (PRED_LOOP_IV_COMPARE): Likewise.
24877 * predict.h (PROB_UNINITIALIZED): Define new constant.
24879 2018-01-19 Martin Liska <mliska@suse.cz>
24881 * predict.c (dump_prediction): Add new format for
24882 analyze_brprob.py script which is enabled with -details
24884 * profile-count.h (precise_p): New function.
24886 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
24888 PR tree-optimization/83922
24889 * tree-vect-loop.c (vect_verify_full_masking): Return false if
24890 there are no statements that need masking.
24891 (vect_active_double_reduction_p): New function.
24892 (vect_analyze_loop_operations): Use it when handling phis that
24893 are not in the loop header.
24895 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
24897 PR tree-optimization/83914
24898 * tree-vect-loop.c (vectorizable_induction): Don't convert
24899 init_expr or apply the peeling adjustment for inductions
24900 that are nested within the vectorized loop.
24902 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24904 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
24907 2018-01-18 Jakub Jelinek <jakub@redhat.com>
24911 * function.h (gimplify_parameters): Add gimple_seq * argument.
24912 * function.c: Include gimple.h and options.h.
24913 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
24914 for the added local temporaries if needed.
24915 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
24916 if there are any parameter cleanups, wrap whole body into a
24917 try/finally with the cleanups.
24919 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
24922 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
24923 Use GET_MODE_CLASS for scalar floating point.
24925 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
24929 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
24930 Fix call of call_cgraph_insertion_hooks.
24932 2018-01-18 Martin Sebor <msebor@redhat.com>
24934 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
24936 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
24939 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
24942 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
24945 * common.opt: (-ffile-prefix-map): New option.
24946 * opts.c (common_handle_option): Defer it.
24947 * opts-global.c (handle_common_deferred_options): Handle it.
24948 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
24949 * file-prefix-map.h: New file.
24950 (remap_debug_filename, add_debug_prefix_map): ...here.
24951 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
24952 * final.c (debug_prefix_map, add_debug_prefix_map
24953 remap_debug_filename): Move to...
24954 * file-prefix-map.c: New file.
24955 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
24956 generalize, get rid of alloca(), use strrchr() instead of strchr().
24957 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
24958 Implement in terms of add_prefix_map().
24959 (remap_macro_filename, remap_debug_filename): Implement in term of
24961 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
24962 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
24963 * dbxout.c: Include file-prefix-map.h.
24964 * varasm.c: Likewise.
24965 * vmsdbgout.c: Likewise.
24966 * xcoffout.c: Likewise.
24967 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
24968 * doc/cppopts.texi (-fmacro-prefix-map): Document.
24969 * doc/invoke.texi (-ffile-prefix-map): Document.
24970 (-fdebug-prefix-map): Update description.
24972 2018-01-18 Martin Liska <mliska@suse.cz>
24974 * config/i386/i386.c (indirect_thunk_name): Document that also
24976 (output_indirect_thunk): Document why both instructions
24977 (pause and lfence) are generated.
24979 2018-01-18 Richard Biener <rguenther@suse.de>
24981 PR tree-optimization/83887
24982 * graphite-scop-detection.c
24983 (scop_detection::get_nearest_dom_with_single_entry): Remove.
24984 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
24985 (scop_detection::merge_sese): Re-implement with a flood-fill
24986 algorithm that properly finds a SESE region if it exists.
24988 2018-01-18 Jakub Jelinek <jakub@redhat.com>
24991 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
24992 pointer_diff optimizations use view_convert instead of convert.
24994 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24996 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
24997 Generate different code for -mno-speculate-indirect-jumps.
24998 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
24999 (*call_indirect_aix<mode>): Disable for
25000 -mno-speculate-indirect-jumps.
25001 (*call_indirect_aix<mode>_nospec): New define_insn.
25002 (*call_value_indirect_aix<mode>): Disable for
25003 -mno-speculate-indirect-jumps.
25004 (*call_value_indirect_aix<mode>_nospec): New define_insn.
25005 (*sibcall_nonlocal_sysv<mode>): Generate different code for
25006 -mno-speculate-indirect-jumps.
25007 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
25009 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
25011 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
25012 long double type, set the flags for noting the default long double
25013 type, even if we don't pass or return a long double type.
25015 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
25018 * ipa-inline.c (flatten_function): Do not overwrite final inlining
25021 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
25023 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
25024 support for merge[hl].
25025 (fold_mergehl_helper): New helper function.
25026 (tree-vector-builder.h): New #include for tree_vector_builder usage.
25027 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
25028 (altivec_vmrglw_direct): Add xxmrglw insn.
25030 2018-01-17 Andrew Waterman <andrew@sifive.com>
25032 * config/riscv/riscv.c (riscv_conditional_register_usage): If
25033 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
25035 2018-01-17 David Malcolm <dmalcolm@redhat.com>
25038 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
25039 call the lto_location_cache before reading the
25040 DECL_SOURCE_LOCATION of the types.
25042 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
25043 Richard Sandiford <richard.sandiford@linaro.org>
25045 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
25046 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
25047 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
25048 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
25049 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
25051 * config/aarch64/constraints.md (aarch64_movti_operand):
25053 * config/aarch64/predicates.md (Uti): Add new constraint.
25055 2018-01-17 Carl Love <cel@us.ibm.com>
25057 * config/rs6000/vsx.md (define_expand xl_len_r,
25058 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
25059 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
25061 (define_expand, define_insn): Move the shift left from the
25062 define_insn to the define_expand for lxvl and stxvl instructions.
25063 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
25064 and XL_LEN_R definitions to PURE.
25066 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
25068 * config/i386/i386.c (indirect_thunk_name): Declare regno
25069 as unsigned int. Compare regno with INVALID_REGNUM.
25070 (output_indirect_thunk): Ditto.
25071 (output_indirect_thunk_function): Ditto.
25072 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
25073 in the call to output_indirect_thunk_function.
25075 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
25077 PR middle-end/83884
25078 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
25079 rather than the size of inner_type to determine the stack slot size
25080 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
25082 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
25085 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
25088 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
25090 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
25091 endian Linux systems to optionally enable multilibs for selecting
25092 the long double type if the user configured an explicit type.
25093 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
25094 have no long double multilibs if not defined.
25095 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
25096 warn if the user used -mabi={ieee,ibm}longdouble and we built
25097 multilibs for long double.
25098 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
25099 appropriate multilib option.
25100 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
25102 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
25103 for building long double multilibs.
25104 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
25106 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
25108 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
25111 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
25113 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
25116 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
25119 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
25122 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
25124 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
25125 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
25127 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
25129 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
25130 different rtl trees depending on TARGET_64BIT.
25131 (rs6000_gen_lvx): Likewise.
25133 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
25135 * config/visium/visium.md (nop): Tweak comment.
25136 (hazard_nop): Likewise.
25138 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25140 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
25141 -mspeculate-indirect-jumps.
25142 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
25143 for -mno-speculate-indirect-jumps.
25144 (*call_indirect_elfv2<mode>_nospec): New define_insn.
25145 (*call_value_indirect_elfv2<mode>): Disable for
25146 -mno-speculate-indirect-jumps.
25147 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
25148 (indirect_jump): Emit different RTL for
25149 -mno-speculate-indirect-jumps.
25150 (*indirect_jump<mode>): Disable for
25151 -mno-speculate-indirect-jumps.
25152 (*indirect_jump<mode>_nospec): New define_insn.
25153 (tablejump): Emit different RTL for
25154 -mno-speculate-indirect-jumps.
25155 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
25156 (tablejumpsi_nospec): New define_expand.
25157 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
25158 (tablejumpdi_nospec): New define_expand.
25159 (*tablejump<mode>_internal1): Disable for
25160 -mno-speculate-indirect-jumps.
25161 (*tablejump<mode>_internal1_nospec): New define_insn.
25162 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
25165 2018-01-16 Artyom Skrobov tyomitch@gmail.com
25167 * caller-save.c (insert_save): Drop unnecessary parameter. All
25170 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25171 Richard Biener <rguenth@suse.de>
25174 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
25175 return early, inline manually is_gimple_sizepos. Make sure if we
25176 call gimplify_expr we don't end up with a gimple constant.
25177 * tree.c (variably_modified_type_p): Don't return true for
25178 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
25179 * gimplify.h (is_gimple_sizepos): Remove.
25181 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25183 PR tree-optimization/83857
25184 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
25185 vectorizable_live_operation for pure SLP statements.
25186 (vectorizable_live_operation): Handle PHIs.
25188 2018-01-16 Richard Biener <rguenther@suse.de>
25190 PR tree-optimization/83867
25191 * tree-vect-stmts.c (vect_transform_stmt): Precompute
25192 nested_in_vect_loop_p since the scalar stmt may get invalidated.
25194 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25197 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
25198 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
25199 If off is not INTEGER_CST, issue a may not be aligned warning
25200 rather than isn't aligned. Use isn%'t rather than isn't.
25201 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
25203 <case MULT_EXPR>: Improve the case when bottom and one of the
25204 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
25205 operand, in that case check if the other operand is multiple of
25206 bottom divided by the INTEGER_CST operand.
25208 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25211 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
25212 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
25213 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
25214 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
25215 * config/pa/pa.c (pa_function_arg_advance): Likewise.
25216 (pa_function_arg, pa_arg_partial_bytes): Likewise.
25217 (pa_function_arg_size): New function.
25219 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25221 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
25222 in a separate statement.
25224 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25226 PR tree-optimization/83847
25227 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
25228 group gathers and scatters.
25230 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25232 PR rtl-optimization/86620
25233 * params.def (max-sched-ready-insns): Bump minimum value to 1.
25235 PR rtl-optimization/83213
25236 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
25237 to last if both are JUMP_INSNs.
25239 PR tree-optimization/83843
25240 * gimple-ssa-store-merging.c
25241 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
25242 store_immediate_info for bswap/nop orig_stores.
25244 2018-01-15 Andrew Waterman <andrew@sifive.com>
25246 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
25248 <UDIV>: Increase cost if !TARGET_DIV.
25250 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25252 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
25253 (define_attr "cr_logical_3op"): New.
25254 (cceq_ior_compare): Adjust.
25255 (cceq_ior_compare_complement): Adjust.
25256 (*cceq_rev_compare): Adjust.
25257 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
25258 (is_cracked_insn): Adjust.
25259 (insn_must_be_first_in_group): Adjust.
25260 * config/rs6000/40x.md: Adjust.
25261 * config/rs6000/440.md: Adjust.
25262 * config/rs6000/476.md: Adjust.
25263 * config/rs6000/601.md: Adjust.
25264 * config/rs6000/603.md: Adjust.
25265 * config/rs6000/6xx.md: Adjust.
25266 * config/rs6000/7450.md: Adjust.
25267 * config/rs6000/7xx.md: Adjust.
25268 * config/rs6000/8540.md: Adjust.
25269 * config/rs6000/cell.md: Adjust.
25270 * config/rs6000/e300c2c3.md: Adjust.
25271 * config/rs6000/e500mc.md: Adjust.
25272 * config/rs6000/e500mc64.md: Adjust.
25273 * config/rs6000/e5500.md: Adjust.
25274 * config/rs6000/e6500.md: Adjust.
25275 * config/rs6000/mpc.md: Adjust.
25276 * config/rs6000/power4.md: Adjust.
25277 * config/rs6000/power5.md: Adjust.
25278 * config/rs6000/power6.md: Adjust.
25279 * config/rs6000/power7.md: Adjust.
25280 * config/rs6000/power8.md: Adjust.
25281 * config/rs6000/power9.md: Adjust.
25282 * config/rs6000/rs64.md: Adjust.
25283 * config/rs6000/titan.md: Adjust.
25285 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25287 * config/i386/predicates.md (indirect_branch_operand): Rewrite
25288 ix86_indirect_branch_register logic.
25290 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25292 * config/i386/constraints.md (Bs): Update
25293 ix86_indirect_branch_register check. Don't check
25294 ix86_indirect_branch_register with GOT_memory_operand.
25296 * config/i386/predicates.md (GOT_memory_operand): Don't check
25297 ix86_indirect_branch_register here.
25298 (GOT32_symbol_operand): Likewise.
25300 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25302 * config/i386/predicates.md (constant_call_address_operand):
25303 Rewrite ix86_indirect_branch_register logic.
25304 (sibcall_insn_operand): Likewise.
25306 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25308 * config/i386/constraints.md (Bs): Replace
25309 ix86_indirect_branch_thunk_register with
25310 ix86_indirect_branch_register.
25312 * config/i386/i386.md (indirect_jump): Likewise.
25313 (tablejump): Likewise.
25314 (*sibcall_memory): Likewise.
25315 (*sibcall_value_memory): Likewise.
25316 Peepholes of indirect call and jump via memory: Likewise.
25317 * config/i386/i386.opt: Likewise.
25318 * config/i386/predicates.md (indirect_branch_operand): Likewise.
25319 (GOT_memory_operand): Likewise.
25320 (call_insn_operand): Likewise.
25321 (sibcall_insn_operand): Likewise.
25322 (GOT32_symbol_operand): Likewise.
25324 2018-01-15 Jakub Jelinek <jakub@redhat.com>
25326 PR middle-end/83837
25327 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
25328 type rather than type addr's type points to.
25329 (expand_omp_atomic_mutex): Likewise.
25330 (expand_omp_atomic): Likewise.
25332 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25335 * config/i386/i386.c (output_indirect_thunk_function): Use
25336 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
25337 for __x86_return_thunk.
25339 2018-01-15 Richard Biener <rguenther@suse.de>
25341 PR middle-end/83850
25342 * expmed.c (extract_bit_field_1): Fix typo.
25344 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25347 * config/arm/iterators.md (VF): New mode iterator.
25348 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
25349 Remove integer-related logic from pattern.
25350 (neon_vabd<mode>_3): Likewise.
25352 2018-01-15 Jakub Jelinek <jakub@redhat.com>
25354 PR middle-end/82694
25355 * common.opt (fstrict-overflow): No longer an alias.
25356 (fwrapv-pointer): New option.
25357 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
25358 also for pointer types based on flag_wrapv_pointer.
25359 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
25360 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
25361 opts->x_flag_wrapv got set.
25362 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
25363 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
25364 POINTER_TYPE_OVERFLOW_UNDEFINED.
25365 * match.pd: Likewise in address comparison pattern.
25366 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
25368 2018-01-15 Richard Biener <rguenther@suse.de>
25371 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
25372 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
25373 Reset type names to their identifier if their TYPE_DECL doesn't
25374 have linkage (and thus is used for ODR and devirt).
25375 (save_debug_info_for_decl): Remove.
25376 (save_debug_info_for_type): Likewise.
25377 (add_tree_to_fld_list): Adjust.
25378 * tree-pretty-print.c (dump_generic_node): Make dumping of
25379 type names more robust.
25381 2018-01-15 Richard Biener <rguenther@suse.de>
25383 * BASE-VER: Bump to 8.0.1.
25385 2018-01-14 Martin Sebor <msebor@redhat.com>
25388 * builtins.c (check_access): Avoid warning when the no-warning bit
25391 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
25393 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
25394 * ira-color (allocno_hard_regs_compare): Likewise.
25396 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
25399 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
25400 Use .pushsection/.popsection.
25402 2018-01-14 Martin Sebor <msebor@redhat.com>
25405 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
25407 2018-01-14 Jakub Jelinek <jakub@redhat.com>
25409 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
25410 entry from extra_headers.
25411 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
25412 extra_headers, make the list bitwise identical to the i?86-*-* one.
25414 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25416 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
25417 -mcmodel=large with -mindirect-branch=thunk,
25418 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
25419 -mfunction-return=thunk-extern.
25420 * doc/invoke.texi: Document -mcmodel=large is incompatible with
25421 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
25422 -mfunction-return=thunk and -mfunction-return=thunk-extern.
25424 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25426 * config/i386/i386.c (print_reg): Print the name of the full
25427 integer register without '%'.
25428 (ix86_print_operand): Handle 'V'.
25429 * doc/extend.texi: Document 'V' modifier.
25431 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25433 * config/i386/constraints.md (Bs): Disallow memory operand for
25434 -mindirect-branch-register.
25436 * config/i386/predicates.md (indirect_branch_operand): Likewise.
25437 (GOT_memory_operand): Likewise.
25438 (call_insn_operand): Likewise.
25439 (sibcall_insn_operand): Likewise.
25440 (GOT32_symbol_operand): Likewise.
25441 * config/i386/i386.md (indirect_jump): Call convert_memory_address
25442 for -mindirect-branch-register.
25443 (tablejump): Likewise.
25444 (*sibcall_memory): Likewise.
25445 (*sibcall_value_memory): Likewise.
25446 Disallow peepholes of indirect call and jump via memory for
25447 -mindirect-branch-register.
25448 (*call_pop): Replace m with Bw.
25449 (*call_value_pop): Likewise.
25450 (*sibcall_pop_memory): Replace m with Bs.
25451 * config/i386/i386.opt (mindirect-branch-register): New option.
25452 * doc/invoke.texi: Document -mindirect-branch-register option.
25454 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25456 * config/i386/i386-protos.h (ix86_output_function_return): New.
25457 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
25458 set function_return_type.
25459 (indirect_thunk_name): Add ret_p to indicate thunk for function
25461 (output_indirect_thunk_function): Pass false to
25462 indirect_thunk_name.
25463 (ix86_output_indirect_branch_via_reg): Likewise.
25464 (ix86_output_indirect_branch_via_push): Likewise.
25465 (output_indirect_thunk_function): Create alias for function
25466 return thunk if regno < 0.
25467 (ix86_output_function_return): New function.
25468 (ix86_handle_fndecl_attribute): Handle function_return.
25469 (ix86_attribute_table): Add function_return.
25470 * config/i386/i386.h (machine_function): Add
25471 function_return_type.
25472 * config/i386/i386.md (simple_return_internal): Use
25473 ix86_output_function_return.
25474 (simple_return_internal_long): Likewise.
25475 * config/i386/i386.opt (mfunction-return=): New option.
25476 (indirect_branch): Mention -mfunction-return=.
25477 * doc/extend.texi: Document function_return function attribute.
25478 * doc/invoke.texi: Document -mfunction-return= option.
25480 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25482 * config/i386/i386-opts.h (indirect_branch): New.
25483 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
25484 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
25485 with local indirect jump when converting indirect call and jump.
25486 (ix86_set_indirect_branch_type): New.
25487 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
25488 (indirectlabelno): New.
25489 (indirect_thunk_needed): Likewise.
25490 (indirect_thunk_bnd_needed): Likewise.
25491 (indirect_thunks_used): Likewise.
25492 (indirect_thunks_bnd_used): Likewise.
25493 (INDIRECT_LABEL): Likewise.
25494 (indirect_thunk_name): Likewise.
25495 (output_indirect_thunk): Likewise.
25496 (output_indirect_thunk_function): Likewise.
25497 (ix86_output_indirect_branch_via_reg): Likewise.
25498 (ix86_output_indirect_branch_via_push): Likewise.
25499 (ix86_output_indirect_branch): Likewise.
25500 (ix86_output_indirect_jmp): Likewise.
25501 (ix86_code_end): Call output_indirect_thunk_function if needed.
25502 (ix86_output_call_insn): Call ix86_output_indirect_branch if
25504 (ix86_handle_fndecl_attribute): Handle indirect_branch.
25505 (ix86_attribute_table): Add indirect_branch.
25506 * config/i386/i386.h (machine_function): Add indirect_branch_type
25507 and has_local_indirect_jump.
25508 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
25510 (tablejump): Likewise.
25511 (*indirect_jump): Use ix86_output_indirect_jmp.
25512 (*tablejump_1): Likewise.
25513 (simple_return_indirect_internal): Likewise.
25514 * config/i386/i386.opt (mindirect-branch=): New option.
25515 (indirect_branch): New.
25518 (thunk-inline): Likewise.
25519 (thunk-extern): Likewise.
25520 * doc/extend.texi: Document indirect_branch function attribute.
25521 * doc/invoke.texi: Document -mindirect-branch= option.
25523 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
25526 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
25528 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
25530 * ipa-inline.c (want_inline_small_function_p): Return false if
25531 inlining has already failed with CIF_FINAL_ERROR.
25532 (update_caller_keys): Call want_inline_small_function_p before
25534 (update_callee_keys): Likewise.
25536 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25538 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
25540 (rs6000_quadword_masked_address_p): Likewise.
25541 (quad_aligned_load_p): Likewise.
25542 (quad_aligned_store_p): Likewise.
25543 (const_load_sequence_p): Add comment to describe the outer-most loop.
25544 (mimic_memory_attributes_and_flags): New function.
25545 (rs6000_gen_stvx): Likewise.
25546 (replace_swapped_aligned_store): Likewise.
25547 (rs6000_gen_lvx): Likewise.
25548 (replace_swapped_aligned_load): Likewise.
25549 (replace_swapped_load_constant): Capitalize argument name in
25550 comment describing this function.
25551 (rs6000_analyze_swaps): Add a third pass to search for vector loads
25552 and stores that access quad-word aligned addresses and replace
25553 with stvx or lvx instructions when appropriate.
25554 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
25555 New function prototype.
25556 (rs6000_quadword_masked_address_p): Likewise.
25557 (rs6000_gen_lvx): Likewise.
25558 (rs6000_gen_stvx): Likewise.
25559 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
25560 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
25561 when memory address is aligned.
25562 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
25563 this split to select lvx instruction when memory address is aligned.
25564 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
25565 instruction when memory address is aligned.
25566 (*vsx_le_perm_load_v16qi): Likewise.
25567 (four unnamed splitters): Modify to select the stvx instruction
25568 when memory is aligned.
25570 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
25572 * predict.c (determine_unlikely_bbs): Handle correctly BBs
25573 which appears in the queue multiple times.
25575 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25576 Alan Hayward <alan.hayward@arm.com>
25577 David Sherwood <david.sherwood@arm.com>
25579 * tree-vectorizer.h (vec_lower_bound): New structure.
25580 (_loop_vec_info): Add check_nonzero and lower_bounds.
25581 (LOOP_VINFO_CHECK_NONZERO): New macro.
25582 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
25583 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
25584 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
25585 fields. Make seg_len the distance travelled, not including the
25587 (dr_direction_indicator): Declare.
25588 (dr_zero_step_indicator): Likewise.
25589 (dr_known_forward_stride_p): Likewise.
25590 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
25592 (runtime_alias_check_p): Allow runtime alias checks with
25594 (operator ==): Compare access_size and align.
25595 (prune_runtime_alias_test_list): Rework for new distinction between
25596 the access_size and seg_len.
25597 (create_intersect_range_checks_index): Likewise. Cope with polynomial
25599 (get_segment_min_max): New function.
25600 (create_intersect_range_checks): Use it.
25601 (dr_step_indicator): New function.
25602 (dr_direction_indicator): Likewise.
25603 (dr_zero_step_indicator): Likewise.
25604 (dr_known_forward_stride_p): Likewise.
25605 * tree-loop-distribution.c (data_ref_segment_size): Return
25606 DR_STEP * (niters - 1).
25607 (compute_alias_check_pairs): Update call to the dr_with_seg_len
25609 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
25610 (vect_preserves_scalar_order_p): New function, split out from...
25611 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
25612 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
25613 (vect_vfa_access_size): New function.
25614 (vect_vfa_align): Likewise.
25615 (vect_compile_time_alias): Take access_size_a and access_b arguments.
25616 (dump_lower_bound): New function.
25617 (vect_check_lower_bound): Likewise.
25618 (vect_small_gap_p): Likewise.
25619 (vectorizable_with_step_bound_p): Likewise.
25620 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
25621 depencies if the vectorization factor is 1. Convert the checks
25622 for nonzero steps into checks on the bounds of DR_STEP. Try using
25623 a bunds check for variable steps if the minimum required step is
25624 relatively small. Update calls to the dr_with_seg_len
25625 constructor and to vect_compile_time_alias.
25626 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
25628 (vect_loop_versioning): Call it.
25629 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
25631 (vect_estimate_min_profitable_iters): Account for any bounds checks.
25633 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25634 Alan Hayward <alan.hayward@arm.com>
25635 David Sherwood <david.sherwood@arm.com>
25637 * doc/sourcebuild.texi (vect_scatter_store): Document.
25638 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
25640 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
25642 * genopinit.c (main): Add supports_vec_scatter_store and
25643 supports_vec_scatter_store_cached to target_optabs.
25644 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
25645 IFN_MASK_SCATTER_STORE.
25646 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
25648 * internal-fn.h (internal_store_fn_p): Declare.
25649 (internal_fn_stored_value_index): Likewise.
25650 * internal-fn.c (scatter_store_direct): New macro.
25651 (expand_scatter_store_optab_fn): New function.
25652 (direct_scatter_store_optab_supported_p): New macro.
25653 (internal_store_fn_p): New function.
25654 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
25655 IFN_MASK_SCATTER_STORE.
25656 (internal_fn_mask_index): Likewise.
25657 (internal_fn_stored_value_index): New function.
25658 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
25659 for scatter stores.
25660 * optabs-query.h (supports_vec_scatter_store_p): Declare.
25661 * optabs-query.c (supports_vec_scatter_store_p): New function.
25662 * tree-vectorizer.h (vect_get_store_rhs): Declare.
25663 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
25664 true for scatter stores.
25665 (vect_gather_scatter_fn_p): Handle scatter stores too.
25666 (vect_check_gather_scatter): Consider using scatter stores if
25667 supports_vec_scatter_store_p.
25668 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
25669 scatter stores too.
25670 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
25671 internal_fn_stored_value_index.
25672 (check_load_store_masking): Handle scatter stores too.
25673 (vect_get_store_rhs): Make public.
25674 (vectorizable_call): Use internal_store_fn_p.
25675 (vectorizable_store): Handle scatter store internal functions.
25676 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
25677 when deciding whether the end of the group has been reached.
25678 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
25679 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
25680 (mask_scatter_store<mode>): New insns.
25682 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25683 Alan Hayward <alan.hayward@arm.com>
25684 David Sherwood <david.sherwood@arm.com>
25686 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
25687 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
25688 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
25690 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
25691 Use vect_truncate_gather_scatter_offset if we can't treat the
25692 operation as a normal gather load or scatter store.
25693 (get_group_load_store_type): Take the gather_scatter_info
25694 as argument. Try using a gather load or scatter store for
25695 single-element groups.
25696 (get_load_store_type): Update calls to get_group_load_store_type
25697 and vect_use_strided_gather_scatters_p.
25699 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25700 Alan Hayward <alan.hayward@arm.com>
25701 David Sherwood <david.sherwood@arm.com>
25703 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
25704 optional tree argument.
25705 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
25707 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
25708 but continue to use the current value as a fallback.
25709 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
25710 to compare the updates.
25711 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
25712 (get_load_store_type): Use it when handling a strided access.
25713 (vect_get_strided_load_store_ops): New function.
25714 (vect_get_data_ptr_increment): Likewise.
25715 (vectorizable_load): Handle strided gather loads. Always pass
25716 a step to vect_create_data_ref_ptr and bump_vector_ptr.
25718 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25719 Alan Hayward <alan.hayward@arm.com>
25720 David Sherwood <david.sherwood@arm.com>
25722 * doc/md.texi (gather_load@var{m}): Document.
25723 (mask_gather_load@var{m}): Likewise.
25724 * genopinit.c (main): Add supports_vec_gather_load and
25725 supports_vec_gather_load_cached to target_optabs.
25726 * optabs-tree.c (init_tree_optimization_optabs): Use
25727 ggc_cleared_alloc to allocate target_optabs.
25728 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
25729 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
25731 * internal-fn.h (internal_load_fn_p): Declare.
25732 (internal_gather_scatter_fn_p): Likewise.
25733 (internal_fn_mask_index): Likewise.
25734 (internal_gather_scatter_fn_supported_p): Likewise.
25735 * internal-fn.c (gather_load_direct): New macro.
25736 (expand_gather_load_optab_fn): New function.
25737 (direct_gather_load_optab_supported_p): New macro.
25738 (direct_internal_fn_optab): New function.
25739 (internal_load_fn_p): Likewise.
25740 (internal_gather_scatter_fn_p): Likewise.
25741 (internal_fn_mask_index): Likewise.
25742 (internal_gather_scatter_fn_supported_p): Likewise.
25743 * optabs-query.c (supports_at_least_one_mode_p): New function.
25744 (supports_vec_gather_load_p): Likewise.
25745 * optabs-query.h (supports_vec_gather_load_p): Declare.
25746 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
25747 and memory_type field.
25748 (NUM_PATTERNS): Bump to 15.
25749 * tree-vect-data-refs.c: Include internal-fn.h.
25750 (vect_gather_scatter_fn_p): New function.
25751 (vect_describe_gather_scatter_call): Likewise.
25752 (vect_check_gather_scatter): Try using internal functions for
25753 gather loads. Recognize existing calls to a gather load function.
25754 (vect_analyze_data_refs): Consider using gather loads if
25755 supports_vec_gather_load_p.
25756 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
25757 (vect_get_gather_scatter_offset_type): Likewise.
25758 (vect_convert_mask_for_vectype): Likewise.
25759 (vect_add_conversion_to_patterm): Likewise.
25760 (vect_try_gather_scatter_pattern): Likewise.
25761 (vect_recog_gather_scatter_pattern): New pattern recognizer.
25762 (vect_vect_recog_func_ptrs): Add it.
25763 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
25764 internal_fn_mask_index and internal_gather_scatter_fn_p.
25765 (check_load_store_masking): Take the gather_scatter_info as an
25766 argument and handle gather loads.
25767 (vect_get_gather_scatter_ops): New function.
25768 (vectorizable_call): Check internal_load_fn_p.
25769 (vectorizable_load): Likewise. Handle gather load internal
25771 (vectorizable_store): Update call to check_load_store_masking.
25772 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
25773 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
25774 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
25775 (aarch64_gather_scale_operand_d): New predicates.
25776 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
25777 (mask_gather_load<mode>): New insns.
25779 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25780 Alan Hayward <alan.hayward@arm.com>
25781 David Sherwood <david.sherwood@arm.com>
25783 * optabs.def (fold_left_plus_optab): New optab.
25784 * doc/md.texi (fold_left_plus_@var{m}): Document.
25785 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
25786 * internal-fn.c (fold_left_direct): Define.
25787 (expand_fold_left_optab_fn): Likewise.
25788 (direct_fold_left_optab_supported_p): Likewise.
25789 * fold-const-call.c (fold_const_fold_left): New function.
25790 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
25791 * tree-parloops.c (valid_reduction_p): New function.
25792 (gather_scalar_reductions): Use it.
25793 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
25794 (vect_finish_replace_stmt): Declare.
25795 * tree-vect-loop.c (fold_left_reduction_fn): New function.
25796 (needs_fold_left_reduction_p): New function, split out from...
25797 (vect_is_simple_reduction): ...here. Accept reductions that
25798 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
25799 (vect_force_simple_reduction): Also store the reduction type in
25800 the assignment's STMT_VINFO_REDUC_TYPE.
25801 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
25802 (merge_with_identity): New function.
25803 (vect_expand_fold_left): Likewise.
25804 (vectorize_fold_left_reduction): Likewise.
25805 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
25806 scalar phi in place for it. Check for target support and reject
25807 cases that would reassociate the operation. Defer the transform
25808 phase to vectorize_fold_left_reduction.
25809 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
25810 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
25811 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
25813 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25815 * tree-if-conv.c (predicate_mem_writes): Remove redundant
25816 call to ifc_temp_var.
25818 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25819 Alan Hayward <alan.hayward@arm.com>
25820 David Sherwood <david.sherwood@arm.com>
25822 * target.def (legitimize_address_displacement): Take the original
25823 offset as a poly_int.
25824 * targhooks.h (default_legitimize_address_displacement): Update
25826 * targhooks.c (default_legitimize_address_displacement): Likewise.
25827 * doc/tm.texi: Regenerate.
25828 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
25829 as an argument, moving assert of ad->disp == ad->disp_term to...
25830 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
25831 Try calling targetm.legitimize_address_displacement before expanding
25832 the address rather than afterwards, and adjust for the new interface.
25833 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
25834 Match the new hook interface. Handle SVE addresses.
25835 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
25836 new hook interface.
25838 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25840 * Makefile.in (OBJS): Add early-remat.o.
25841 * target.def (select_early_remat_modes): New hook.
25842 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
25843 * doc/tm.texi: Regenerate.
25844 * targhooks.h (default_select_early_remat_modes): Declare.
25845 * targhooks.c (default_select_early_remat_modes): New function.
25846 * timevar.def (TV_EARLY_REMAT): New timevar.
25847 * passes.def (pass_early_remat): New pass.
25848 * tree-pass.h (make_pass_early_remat): Declare.
25849 * early-remat.c: New file.
25850 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
25852 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
25854 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25855 Alan Hayward <alan.hayward@arm.com>
25856 David Sherwood <david.sherwood@arm.com>
25858 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
25859 vfm1 with a bound_epilog parameter.
25860 (vect_do_peeling): Update calls accordingly, and move the prologue
25861 call earlier in the function. Treat the base bound_epilog as 0 for
25862 fully-masked loops and retain vf - 1 for other loops. Add 1 to
25863 this base when peeling for gaps.
25864 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
25865 with fully-masked loops.
25866 (vect_estimate_min_profitable_iters): Handle the single peeled
25867 iteration in that case.
25869 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25870 Alan Hayward <alan.hayward@arm.com>
25871 David Sherwood <david.sherwood@arm.com>
25873 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
25874 single-element interleaving even if the size is not a power of 2.
25875 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
25876 accesses for single-element interleaving if the group size is
25879 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25880 Alan Hayward <alan.hayward@arm.com>
25881 David Sherwood <david.sherwood@arm.com>
25883 * doc/md.texi (fold_extract_last_@var{m}): Document.
25884 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
25885 * optabs.def (fold_extract_last_optab): New optab.
25886 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
25887 * internal-fn.c (fold_extract_direct): New macro.
25888 (expand_fold_extract_optab_fn): Likewise.
25889 (direct_fold_extract_optab_supported_p): Likewise.
25890 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
25891 * tree-vect-loop.c (vect_model_reduction_cost): Handle
25892 EXTRACT_LAST_REDUCTION.
25893 (get_initial_def_for_reduction): Do not create an initial vector
25894 for EXTRACT_LAST_REDUCTION reductions.
25895 (vectorizable_reduction): Leave the scalar phi in place for
25896 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
25897 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
25898 epilogue code for EXTRACT_LAST_REDUCTION and defer the
25899 transform phase to vectorizable_condition.
25900 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
25902 (vect_finish_stmt_generation): ...here.
25903 (vect_finish_replace_stmt): New function.
25904 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
25905 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
25907 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
25909 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25910 Alan Hayward <alan.hayward@arm.com>
25911 David Sherwood <david.sherwood@arm.com>
25913 * doc/md.texi (extract_last_@var{m}): Document.
25914 * optabs.def (extract_last_optab): New optab.
25915 * internal-fn.def (EXTRACT_LAST): New internal function.
25916 * internal-fn.c (cond_unary_direct): New macro.
25917 (expand_cond_unary_optab_fn): Likewise.
25918 (direct_cond_unary_optab_supported_p): Likewise.
25919 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
25920 loops using EXTRACT_LAST.
25921 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
25922 (extract_last_<mode>): ...this optab.
25923 (vec_extract<mode><Vel>): Update accordingly.
25925 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25926 Alan Hayward <alan.hayward@arm.com>
25927 David Sherwood <david.sherwood@arm.com>
25929 * target.def (empty_mask_is_expensive): New hook.
25930 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
25931 * doc/tm.texi: Regenerate.
25932 * targhooks.h (default_empty_mask_is_expensive): Declare.
25933 * targhooks.c (default_empty_mask_is_expensive): New function.
25934 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
25935 if the target says that empty masks are expensive.
25936 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
25938 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
25940 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25941 Alan Hayward <alan.hayward@arm.com>
25942 David Sherwood <david.sherwood@arm.com>
25944 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
25945 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
25946 (vect_use_loop_mask_for_alignment_p): New function.
25947 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
25948 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
25949 niters_skip argument. Make sure that the first niters_skip elements
25950 of the first iteration are inactive.
25951 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
25952 Update call to vect_set_loop_masks_directly.
25953 (get_misalign_in_elems): New function, split out from...
25954 (vect_gen_prolog_loop_niters): ...here.
25955 (vect_update_init_of_dr): Take a code argument that specifies whether
25956 the adjustment should be added or subtracted.
25957 (vect_update_init_of_drs): Likewise.
25958 (vect_prepare_for_masked_peels): New function.
25959 (vect_do_peeling): Skip prologue peeling if we're using a mask
25960 instead. Update call to vect_update_inits_of_drs.
25961 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
25963 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
25964 alignment. Do not include the number of peeled iterations in
25965 the minimum threshold in that case.
25966 (vectorizable_induction): Adjust the start value down by
25967 LOOP_VINFO_MASK_SKIP_NITERS iterations.
25968 (vect_transform_loop): Call vect_prepare_for_masked_peels.
25969 Take the number of skipped iterations into account when calculating
25971 * tree-vect-stmts.c (vect_gen_while_not): New function.
25973 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25974 Alan Hayward <alan.hayward@arm.com>
25975 David Sherwood <david.sherwood@arm.com>
25977 * doc/sourcebuild.texi (vect_fully_masked): Document.
25978 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
25979 default value to 0.
25980 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
25982 (vect_analyze_loop_2): ...here. Don't check the vectorization
25983 factor against the number of loop iterations if the loop is
25986 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25987 Alan Hayward <alan.hayward@arm.com>
25988 David Sherwood <david.sherwood@arm.com>
25990 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
25991 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
25992 (dump_groups): Update accordingly.
25993 (iv_use::mem_type): New member variable.
25994 (address_p): New function.
25995 (record_use): Add a mem_type argument and initialize the new
25997 (record_group_use): Add a mem_type argument. Use address_p.
25998 Remove obsolete null checks of base_object. Update call to record_use.
25999 (find_interesting_uses_op): Update call to record_group_use.
26000 (find_interesting_uses_cond): Likewise.
26001 (find_interesting_uses_address): Likewise.
26002 (get_mem_type_for_internal_fn): New function.
26003 (find_address_like_use): Likewise.
26004 (find_interesting_uses_stmt): Try find_address_like_use before
26005 calling find_interesting_uses_op.
26006 (addr_offset_valid_p): Use the iv mem_type field as the type
26007 of the addressed memory.
26008 (add_autoinc_candidates): Likewise.
26009 (get_address_cost): Likewise.
26010 (split_small_address_groups_p): Use address_p.
26011 (split_address_groups): Likewise.
26012 (add_iv_candidate_for_use): Likewise.
26013 (autoinc_possible_for_pair): Likewise.
26014 (rewrite_groups): Likewise.
26015 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
26016 (determine_group_iv_cost): Update after split of USE_ADDRESS.
26017 (get_alias_ptr_type_for_ptr_address): New function.
26018 (rewrite_use_address): Rewrite address uses in calls that were
26019 identified by find_address_like_use.
26021 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26022 Alan Hayward <alan.hayward@arm.com>
26023 David Sherwood <david.sherwood@arm.com>
26025 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
26027 * gimple-expr.h (is_gimple_addressable: Likewise.
26028 * gimple-expr.c (is_gimple_address): Likewise.
26029 * internal-fn.c (expand_call_mem_ref): New function.
26030 (expand_mask_load_optab_fn): Use it.
26031 (expand_mask_store_optab_fn): Likewise.
26033 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26034 Alan Hayward <alan.hayward@arm.com>
26035 David Sherwood <david.sherwood@arm.com>
26037 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
26038 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
26039 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
26040 (cond_umax@var{mode}): Document.
26041 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
26042 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
26043 (cond_umin_optab, cond_umax_optab): New optabs.
26044 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
26045 (COND_IOR, COND_XOR): New internal functions.
26046 * internal-fn.h (get_conditional_internal_fn): Declare.
26047 * internal-fn.c (cond_binary_direct): New macro.
26048 (expand_cond_binary_optab_fn): Likewise.
26049 (direct_cond_binary_optab_supported_p): Likewise.
26050 (get_conditional_internal_fn): New function.
26051 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
26052 Cope with reduction statements that are vectorized as calls rather
26054 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
26055 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
26056 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
26057 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
26058 (UNSPEC_COND_EOR): New unspecs.
26059 (optab): Add mappings for them.
26060 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
26061 (sve_int_op, sve_fp_op): New int attributes.
26063 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26064 Alan Hayward <alan.hayward@arm.com>
26065 David Sherwood <david.sherwood@arm.com>
26067 * optabs.def (while_ult_optab): New optab.
26068 * doc/md.texi (while_ult@var{m}@var{n}): Document.
26069 * internal-fn.def (WHILE_ULT): New internal function.
26070 * internal-fn.h (direct_internal_fn_supported_p): New override
26071 that takes two types as argument.
26072 * internal-fn.c (while_direct): New macro.
26073 (expand_while_optab_fn): New function.
26074 (convert_optab_supported_p): Likewise.
26075 (direct_while_optab_supported_p): New macro.
26076 * wide-int.h (wi::udiv_ceil): New function.
26077 * tree-vectorizer.h (rgroup_masks): New structure.
26078 (vec_loop_masks): New typedef.
26079 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
26080 and fully_masked_p.
26081 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
26082 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
26083 (vect_max_vf): New function.
26084 (slpeel_make_loop_iterate_ntimes): Delete.
26085 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
26086 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
26087 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
26088 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
26089 internal-fn.h, stor-layout.h and optabs-query.h.
26090 (vect_set_loop_mask): New function.
26091 (add_preheader_seq): Likewise.
26092 (add_header_seq): Likewise.
26093 (interleave_supported_p): Likewise.
26094 (vect_maybe_permute_loop_masks): Likewise.
26095 (vect_set_loop_masks_directly): Likewise.
26096 (vect_set_loop_condition_masked): Likewise.
26097 (vect_set_loop_condition_unmasked): New function, split out from
26098 slpeel_make_loop_iterate_ntimes.
26099 (slpeel_make_loop_iterate_ntimes): Rename to..
26100 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
26101 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
26102 (vect_do_peeling): Update call accordingly.
26103 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
26105 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
26106 mask_compare_type, can_fully_mask_p and fully_masked_p.
26107 (release_vec_loop_masks): New function.
26108 (_loop_vec_info): Use it to free the loop masks.
26109 (can_produce_all_loop_masks_p): New function.
26110 (vect_get_max_nscalars_per_iter): Likewise.
26111 (vect_verify_full_masking): Likewise.
26112 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
26113 retries, and free the mask rgroups before retrying. Check loop-wide
26114 reasons for disallowing fully-masked loops. Make the final decision
26115 about whether use a fully-masked loop or not.
26116 (vect_estimate_min_profitable_iters): Do not assume that peeling
26117 for the number of iterations will be needed for fully-masked loops.
26118 (vectorizable_reduction): Disable fully-masked loops.
26119 (vectorizable_live_operation): Likewise.
26120 (vect_halve_mask_nunits): New function.
26121 (vect_double_mask_nunits): Likewise.
26122 (vect_record_loop_mask): Likewise.
26123 (vect_get_loop_mask): Likewise.
26124 (vect_transform_loop): Handle the case in which the final loop
26125 iteration might handle a partial vector. Call vect_set_loop_condition
26126 instead of slpeel_make_loop_iterate_ntimes.
26127 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
26128 (check_load_store_masking): New function.
26129 (prepare_load_store_mask): Likewise.
26130 (vectorizable_store): Handle fully-masked loops.
26131 (vectorizable_load): Likewise.
26132 (supportable_widening_operation): Use vect_halve_mask_nunits for
26134 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
26135 (vect_gen_while): New function.
26136 * config/aarch64/aarch64.md (umax<mode>3): New expander.
26137 (aarch64_uqdec<mode>): New insn.
26139 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26140 Alan Hayward <alan.hayward@arm.com>
26141 David Sherwood <david.sherwood@arm.com>
26143 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
26144 (reduc_xor_scal_optab): New optabs.
26145 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
26146 (reduc_xor_scal_@var{m}): Document.
26147 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
26148 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
26149 internal functions.
26150 * fold-const-call.c (fold_const_call): Handle them.
26151 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
26152 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
26153 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
26154 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
26155 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
26156 (UNSPEC_XORV): New unspecs.
26157 (optab): Add entries for them.
26158 (BITWISEV): New int iterator.
26159 (bit_reduc_op): New int attributes.
26161 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26162 Alan Hayward <alan.hayward@arm.com>
26163 David Sherwood <david.sherwood@arm.com>
26165 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
26166 * internal-fn.def (VEC_SHL_INSERT): New internal function.
26167 * optabs.def (vec_shl_insert_optab): New optab.
26168 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
26169 (duplicate_and_interleave): Likewise.
26170 * tree-vect-loop.c: Include internal-fn.h.
26171 (neutral_op_for_slp_reduction): New function, split out from
26172 get_initial_defs_for_reduction.
26173 (get_initial_def_for_reduction): Handle option 2 for variable-length
26174 vectors by loading the neutral value into a vector and then shifting
26175 the initial value into element 0.
26176 (get_initial_defs_for_reduction): Replace the code argument with
26177 the neutral value calculated by neutral_op_for_slp_reduction.
26178 Use gimple_build_vector for constant-length vectors.
26179 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
26180 but the first group_size elements have a neutral value.
26181 Use duplicate_and_interleave otherwise.
26182 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
26183 Update call to get_initial_defs_for_reduction. Handle SLP
26184 reductions for variable-length vectors by creating one vector
26185 result for each scalar result, with the elements associated
26186 with other scalar results stubbed out with the neutral value.
26187 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
26188 Require IFN_VEC_SHL_INSERT for double reductions on
26189 variable-length vectors, or SLP reductions that have
26190 a neutral value. Require can_duplicate_and_interleave_p
26191 support for variable-length unchained SLP reductions if there
26192 is no neutral value, such as for MIN/MAX reductions. Also require
26193 the number of vector elements to be a multiple of the number of
26194 SLP statements when doing variable-length unchained SLP reductions.
26195 Update call to vect_create_epilog_for_reduction.
26196 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
26197 and remove initial values.
26198 (duplicate_and_interleave): Make public.
26199 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
26200 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
26202 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26203 Alan Hayward <alan.hayward@arm.com>
26204 David Sherwood <david.sherwood@arm.com>
26206 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
26207 (can_duplicate_and_interleave_p): New function.
26208 (vect_get_and_check_slp_defs): Take the vector of statements
26209 rather than just the current one. Remove excess parentheses.
26210 Restriction rejectinon of vect_constant_def and vect_external_def
26211 for variable-length vectors to boolean types, or types for which
26212 can_duplicate_and_interleave_p is false.
26213 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
26214 (duplicate_and_interleave): New function.
26215 (vect_get_constant_vectors): Use gimple_build_vector for
26216 constant-length vectors and suitable variable-length constant
26217 vectors. Use duplicate_and_interleave for other variable-length
26218 vectors. Don't defer the update when inserting new statements.
26220 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26221 Alan Hayward <alan.hayward@arm.com>
26222 David Sherwood <david.sherwood@arm.com>
26224 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
26225 min_profitable_iters doesn't go negative.
26227 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26228 Alan Hayward <alan.hayward@arm.com>
26229 David Sherwood <david.sherwood@arm.com>
26231 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
26232 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
26233 * optabs.def (vec_mask_load_lanes_optab): New optab.
26234 (vec_mask_store_lanes_optab): Likewise.
26235 * internal-fn.def (MASK_LOAD_LANES): New internal function.
26236 (MASK_STORE_LANES): Likewise.
26237 * internal-fn.c (mask_load_lanes_direct): New macro.
26238 (mask_store_lanes_direct): Likewise.
26239 (expand_mask_load_optab_fn): Handle masked operations.
26240 (expand_mask_load_lanes_optab_fn): New macro.
26241 (expand_mask_store_optab_fn): Handle masked operations.
26242 (expand_mask_store_lanes_optab_fn): New macro.
26243 (direct_mask_load_lanes_optab_supported_p): Likewise.
26244 (direct_mask_store_lanes_optab_supported_p): Likewise.
26245 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
26247 (vect_load_lanes_supported): Likewise.
26248 * tree-vect-data-refs.c (strip_conversion): New function.
26249 (can_group_stmts_p): Likewise.
26250 (vect_analyze_data_ref_accesses): Use it instead of checking
26251 for a pair of assignments.
26252 (vect_store_lanes_supported): Take a masked_p parameter.
26253 (vect_load_lanes_supported): Likewise.
26254 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
26255 vect_store_lanes_supported and vect_load_lanes_supported.
26256 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
26257 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
26258 parameter. Don't allow gaps for masked accesses.
26259 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
26260 and vect_load_lanes_supported.
26261 (get_load_store_type): Take a masked_p parameter and update
26262 call to get_group_load_store_type.
26263 (vectorizable_store): Update call to get_load_store_type.
26264 Handle IFN_MASK_STORE_LANES.
26265 (vectorizable_load): Update call to get_load_store_type.
26266 Handle IFN_MASK_LOAD_LANES.
26268 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26269 Alan Hayward <alan.hayward@arm.com>
26270 David Sherwood <david.sherwood@arm.com>
26272 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
26274 * config/aarch64/aarch64-protos.h
26275 (aarch64_sve_struct_memory_operand_p): Declare.
26276 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
26277 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
26278 (VPRED, vpred): Handle SVE structure modes.
26279 * config/aarch64/constraints.md (Utx): New constraint.
26280 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
26281 (aarch64_sve_struct_nonimmediate_operand): New predicates.
26282 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
26283 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
26284 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
26285 structure modes. Split into pieces after RA.
26286 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
26287 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
26289 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
26290 SVE structure modes.
26291 (aarch64_classify_address): Likewise.
26292 (sizetochar): Move earlier in file.
26293 (aarch64_print_operand): Handle SVE register lists.
26294 (aarch64_array_mode): New function.
26295 (aarch64_sve_struct_memory_operand_p): Likewise.
26296 (TARGET_ARRAY_MODE): Redefine.
26298 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26299 Alan Hayward <alan.hayward@arm.com>
26300 David Sherwood <david.sherwood@arm.com>
26302 * target.def (array_mode): New target hook.
26303 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
26304 * doc/tm.texi: Regenerate.
26305 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
26306 * hooks.c (hook_optmode_mode_uhwi_none): New function.
26307 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
26308 targetm.array_mode.
26309 * stor-layout.c (mode_for_array): Likewise. Support polynomial
26312 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26313 Alan Hayward <alan.hayward@arm.com>
26314 David Sherwood <david.sherwood@arm.com>
26316 * fold-const.c (fold_binary_loc): Check the argument types
26317 rather than the result type when testing for a vector operation.
26319 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26321 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
26322 * doc/tm.texi: Regenerate.
26324 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26325 Alan Hayward <alan.hayward@arm.com>
26326 David Sherwood <david.sherwood@arm.com>
26328 * doc/invoke.texi (-msve-vector-bits=): Document new option.
26329 (sve): Document new AArch64 extension.
26330 * doc/md.texi (w): Extend the description of the AArch64
26331 constraint to include SVE vectors.
26332 (Upl, Upa): Document new AArch64 predicate constraints.
26333 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
26335 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
26336 (msve-vector-bits=): New option.
26337 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
26338 SVE when these are disabled.
26339 (sve): New extension.
26340 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
26341 modes. Adjust their number of units based on aarch64_sve_vg.
26342 (MAX_BITSIZE_MODE_ANY_MODE): Define.
26343 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
26344 aarch64_addr_query_type.
26345 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
26346 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
26347 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
26348 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
26349 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
26350 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
26351 (aarch64_simd_imm_zero_p): Delete.
26352 (aarch64_check_zero_based_sve_index_immediate): Declare.
26353 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
26354 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
26355 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
26356 (aarch64_sve_float_mul_immediate_p): Likewise.
26357 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
26358 rather than an rtx.
26359 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
26360 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
26361 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
26362 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
26363 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
26364 (aarch64_regmode_natural_size): Likewise.
26365 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
26366 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
26368 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
26369 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
26370 for VG and the SVE predicate registers.
26371 (V_ALIASES): Add a "z"-prefixed alias.
26372 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
26373 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
26374 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
26375 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
26376 (REG_CLASS_NAMES): Add entries for them.
26377 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
26378 and the predicate registers.
26379 (aarch64_sve_vg): Declare.
26380 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
26381 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
26382 (REGMODE_NATURAL_SIZE): Define.
26383 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
26385 * config/aarch64/aarch64.c: Include cfgrtl.h.
26386 (simd_immediate_info): Add a constructor for series vectors,
26387 and an associated step field.
26388 (aarch64_sve_vg): New variable.
26389 (aarch64_dbx_register_number): Handle VG and the predicate registers.
26390 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
26391 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
26392 (VEC_ANY_DATA, VEC_STRUCT): New constants.
26393 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
26394 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
26395 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
26396 (aarch64_get_mask_mode): New functions.
26397 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
26398 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
26399 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
26400 predicate modes and predicate registers. Explicitly restrict
26401 GPRs to modes of 16 bytes or smaller. Only allow FP registers
26402 to store a vector mode if it is recognized by
26403 aarch64_classify_vector_mode.
26404 (aarch64_regmode_natural_size): New function.
26405 (aarch64_hard_regno_caller_save_mode): Return the original mode
26407 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
26408 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
26409 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
26410 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
26412 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
26413 does not overlap dest if the function is frame-related. Handle
26415 (aarch64_split_add_offset): New function.
26416 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
26417 them aarch64_add_offset.
26418 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
26419 and update call to aarch64_sub_sp.
26420 (aarch64_add_cfa_expression): New function.
26421 (aarch64_expand_prologue): Pass extra temporary registers to the
26422 functions above. Handle the case in which we need to emit new
26423 DW_CFA_expressions for registers that were originally saved
26424 relative to the stack pointer, but now have to be expressed
26425 relative to the frame pointer.
26426 (aarch64_output_mi_thunk): Pass extra temporary registers to the
26428 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
26429 IP0 and IP1 values for SVE frames.
26430 (aarch64_expand_vec_series): New function.
26431 (aarch64_expand_sve_widened_duplicate): Likewise.
26432 (aarch64_expand_sve_const_vector): Likewise.
26433 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
26434 Handle SVE constants. Use emit_move_insn to move a force_const_mem
26435 into the register, rather than emitting a SET directly.
26436 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
26437 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
26438 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
26439 (offset_9bit_signed_scaled_p): New functions.
26440 (aarch64_replicate_bitmask_imm): New function.
26441 (aarch64_bitmask_imm): Use it.
26442 (aarch64_cannot_force_const_mem): Reject expressions involving
26443 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
26444 (aarch64_classify_index): Handle SVE indices, by requiring
26445 a plain register index with a scale that matches the element size.
26446 (aarch64_classify_address): Handle SVE addresses. Assert that
26447 the mode of the address is VOIDmode or an integer mode.
26448 Update call to aarch64_classify_symbol.
26449 (aarch64_classify_symbolic_expression): Update call to
26450 aarch64_classify_symbol.
26451 (aarch64_const_vec_all_in_range_p): New function.
26452 (aarch64_print_vector_float_operand): Likewise.
26453 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
26454 "vN" for FP registers with SVE modes. Handle (const ...) vectors
26455 and the FP immediates 1.0 and 0.5.
26456 (aarch64_print_address_internal): Handle SVE addresses.
26457 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
26458 (aarch64_regno_regclass): Handle predicate registers.
26459 (aarch64_secondary_reload): Handle big-endian reloads of SVE
26461 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
26462 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
26463 (aarch64_convert_sve_vector_bits): New function.
26464 (aarch64_override_options): Use it to handle -msve-vector-bits=.
26465 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
26466 rather than an rtx.
26467 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
26468 Handle SVE vector and predicate modes. Accept VL-based constants
26469 that need only one temporary register, and VL offsets that require
26470 no temporary registers.
26471 (aarch64_conditional_register_usage): Mark the predicate registers
26472 as fixed if SVE isn't available.
26473 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
26474 Return true for SVE vector and predicate modes.
26475 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
26476 rather than an unsigned int. Handle SVE modes.
26477 (aarch64_preferred_simd_mode): Update call accordingly. Handle
26479 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
26481 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
26482 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
26483 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
26484 (aarch64_sve_float_mul_immediate_p): New functions.
26485 (aarch64_sve_valid_immediate): New function.
26486 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
26487 Explicitly reject structure modes. Check for INDEX constants.
26488 Handle PTRUE and PFALSE constants.
26489 (aarch64_check_zero_based_sve_index_immediate): New function.
26490 (aarch64_simd_imm_zero_p): Delete.
26491 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
26492 vector modes. Accept constants in the range of CNT[BHWD].
26493 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
26494 ask for an Advanced SIMD mode.
26495 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
26496 (aarch64_simd_vector_alignment): Handle SVE predicates.
26497 (aarch64_vectorize_preferred_vector_alignment): New function.
26498 (aarch64_simd_vector_alignment_reachable): Use it instead of
26500 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
26501 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
26503 (MAX_VECT_LEN): Delete.
26504 (expand_vec_perm_d): Add a vec_flags field.
26505 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
26506 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
26507 (aarch64_evpc_ext): Don't apply a big-endian lane correction
26509 (aarch64_evpc_rev): Rename to...
26510 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
26511 (aarch64_evpc_rev_global): New function.
26512 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
26513 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
26515 (aarch64_evpc_sve_tbl): New function.
26516 (aarch64_expand_vec_perm_const_1): Update after rename of
26517 aarch64_evpc_rev. Handle SVE permutes too, trying
26518 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
26519 than aarch64_evpc_tbl.
26520 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
26521 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
26522 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
26523 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
26524 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
26525 (aarch64_expand_sve_vcond): New functions.
26526 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
26527 of aarch64_vector_mode_p.
26528 (aarch64_dwarf_poly_indeterminate_value): New function.
26529 (aarch64_compute_pressure_classes): Likewise.
26530 (aarch64_can_change_mode_class): Likewise.
26531 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
26532 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
26533 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
26534 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
26535 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
26536 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
26537 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
26538 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
26540 (Dn, Dl, Dr): Accept const as well as const_vector.
26541 (Dz): Likewise. Compare against CONST0_RTX.
26542 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
26543 of "vector" where appropriate.
26544 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
26545 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
26546 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
26547 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
26548 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
26549 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
26550 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
26551 (v_int_equiv): Extend to SVE modes.
26552 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
26554 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
26555 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
26556 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
26557 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
26558 (SVE_COND_FP_CMP): New int iterators.
26559 (perm_hilo): Handle the new unpack unspecs.
26560 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
26562 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
26563 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
26564 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
26565 (aarch64_equality_operator, aarch64_constant_vector_operand)
26566 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
26567 (aarch64_sve_nonimmediate_operand): Likewise.
26568 (aarch64_sve_general_operand): Likewise.
26569 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
26570 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
26571 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
26572 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
26573 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
26574 (aarch64_sve_float_arith_immediate): Likewise.
26575 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
26576 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
26577 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
26578 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
26579 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
26580 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
26581 (aarch64_sve_float_arith_operand): Likewise.
26582 (aarch64_sve_float_arith_with_sub_operand): Likewise.
26583 (aarch64_sve_float_mul_operand): Likewise.
26584 (aarch64_sve_vec_perm_operand): Likewise.
26585 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
26586 (aarch64_mov_operand): Accept const_poly_int and const_vector.
26587 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
26588 as well as const_vector.
26589 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
26590 in file. Use CONST0_RTX and CONSTM1_RTX.
26591 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
26592 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
26593 Use aarch64_simd_imm_zero.
26594 * config/aarch64/aarch64-sve.md: New file.
26595 * config/aarch64/aarch64.md: Include it.
26596 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
26597 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
26598 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
26599 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
26600 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
26601 (sve): New attribute.
26602 (enabled): Disable instructions with the sve attribute unless
26604 (movqi, movhi): Pass CONST_POLY_INT operaneds through
26605 aarch64_expand_mov_immediate.
26606 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
26607 CNT[BHSD] immediates.
26608 (movti): Split CONST_POLY_INT moves into two halves.
26609 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
26610 Split additions that need a temporary here if the destination
26611 is the stack pointer.
26612 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
26613 (*add<mode>3_poly_1): New instruction.
26614 (set_clobber_cc): New expander.
26616 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26618 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
26619 parameter and use it instead of GET_MODE_SIZE (innermode). Use
26620 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
26621 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
26622 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
26623 Change innermode from fixed_mode_size to machine_mode.
26624 (simplify_subreg): Update call accordingly. Handle a constant-sized
26625 subreg of a variable-length CONST_VECTOR.
26627 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26628 Alan Hayward <alan.hayward@arm.com>
26629 David Sherwood <david.sherwood@arm.com>
26631 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
26632 (add_offset_to_base): New function, split out from...
26633 (create_mem_ref): ...here. When handling a scale other than 1,
26634 check first whether the address is valid without the offset.
26635 Add it into the base if so, leaving the index and scale as-is.
26637 2018-01-12 Jakub Jelinek <jakub@redhat.com>
26640 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
26641 fold_for_warn before checking if arg2 is INTEGER_CST.
26643 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
26645 * config/rs6000/predicates.md (load_multiple_operation): Delete.
26646 (store_multiple_operation): Delete.
26647 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
26648 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
26649 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
26650 guarded by TARGET_STRING.
26651 (rs6000_output_load_multiple): Delete.
26652 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26653 OPTION_MASK_STRING / TARGET_STRING handling.
26654 (print_operand) <'N', 'O'>: Add comment that these are unused now.
26655 (const rs6000_opt_masks) <"string">: Change mask to 0.
26656 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
26657 (MASK_STRING): Delete.
26658 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
26660 (load_multiple): Delete.
26667 (store_multiple): Delete.
26674 (movmemsi_8reg): Delete.
26675 (corresponding unnamed define_insn): Delete.
26676 (movmemsi_6reg): Delete.
26677 (corresponding unnamed define_insn): Delete.
26678 (movmemsi_4reg): Delete.
26679 (corresponding unnamed define_insn): Delete.
26680 (movmemsi_2reg): Delete.
26681 (corresponding unnamed define_insn): Delete.
26682 (movmemsi_1reg): Delete.
26683 (corresponding unnamed define_insn): Delete.
26684 * config/rs6000/rs6000.opt (mno-string): New.
26685 (mstring): Replace by deprecation warning stub.
26686 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
26688 2018-01-12 Jakub Jelinek <jakub@redhat.com>
26690 * regrename.c (regrename_do_replace): If replacing the same
26691 reg multiple times, try to reuse last created gen_raw_REG.
26694 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
26695 main to workaround a bug in GDB.
26697 2018-01-12 Tom de Vries <tom@codesourcery.com>
26700 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
26702 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
26704 PR rtl-optimization/80481
26705 * ira-color.c (get_cap_member): New function.
26706 (allocnos_conflict_by_live_ranges_p): Use it.
26707 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
26708 (setup_slot_coalesced_allocno_live_ranges): Ditto.
26710 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
26713 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
26714 (*saddl_se_1): Ditto.
26715 (*ssubsi_1): Ditto.
26716 (*ssubl_se_1): Ditto.
26718 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
26720 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
26721 rather than wi::to_widest for DR_INITs.
26722 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
26723 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
26724 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
26726 (vect_analyze_group_access_1): Note that here.
26728 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
26730 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
26731 polynomial type sizes.
26733 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
26735 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
26736 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
26737 (gimple_add_tmp_var): Likewise.
26739 2018-01-12 Martin Liska <mliska@suse.cz>
26741 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
26742 (gimple_alloc_sizes): Likewise.
26743 (dump_gimple_statistics): Use PRIu64 in printf format.
26744 * gimple.h: Change uint64_t to int.
26746 2018-01-12 Martin Liska <mliska@suse.cz>
26748 * tree-core.h: Use uint64_t instead of int.
26749 * tree.c (tree_node_counts): Likewise.
26750 (tree_node_sizes): Likewise.
26751 (dump_tree_statistics): Use PRIu64 in printf format.
26753 2018-01-12 Martin Liska <mliska@suse.cz>
26755 * Makefile.in: As qsort_chk is implemented in vec.c, add
26756 vec.o to linkage of gencfn-macros.
26757 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
26758 passing the info to record_node_allocation_statistics.
26759 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
26761 * ggc-common.c (struct ggc_usage): Add operator== and use
26762 it in operator< and compare function.
26763 * mem-stats.h (struct mem_usage): Likewise.
26764 * vec.c (struct vec_usage): Remove operator< and compare
26765 function. Can be simply inherited.
26767 2018-01-12 Martin Jambor <mjambor@suse.cz>
26770 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
26771 * tree-ssa-math-opts.c: Include domwalk.h.
26772 (convert_mult_to_fma_1): New function.
26773 (fma_transformation_info): New type.
26774 (fma_deferring_state): Likewise.
26775 (cancel_fma_deferring): New function.
26776 (result_of_phi): Likewise.
26777 (last_fma_candidate_feeds_initial_phi): Likewise.
26778 (convert_mult_to_fma): Added deferring logic, split actual
26779 transformation to convert_mult_to_fma_1.
26780 (math_opts_dom_walker): New type.
26781 (math_opts_dom_walker::after_dom_children): New method, body moved
26782 here from pass_optimize_widening_mul::execute, added deferring logic
26784 (pass_optimize_widening_mul::execute): Moved most of code to
26785 math_opts_dom_walker::after_dom_children.
26786 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
26787 * config/i386/i386.c (ix86_option_override_internal): Added
26788 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
26790 2018-01-12 Richard Biener <rguenther@suse.de>
26793 * dwarf2out.c (gen_variable_die): Do not reset old_die for
26794 inline instance vars.
26796 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
26799 * config/rx/rx.c (rx_is_restricted_memory_address):
26800 Handle SUBREG case.
26802 2018-01-12 Richard Biener <rguenther@suse.de>
26804 PR tree-optimization/80846
26805 * target.def (split_reduction): New target hook.
26806 * targhooks.c (default_split_reduction): New function.
26807 * targhooks.h (default_split_reduction): Declare.
26808 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
26809 target requests first reduce vectors by combining low and high
26811 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
26812 (get_vectype_for_scalar_type_and_size): Export.
26813 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
26814 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
26815 * doc/tm.texi: Regenerate.
26816 * config/i386/i386.c (ix86_split_reduction): Implement
26817 TARGET_VECTORIZE_SPLIT_REDUCTION.
26819 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
26822 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
26823 in PIC mode except for TARGET_VXWORKS_RTP.
26824 * config/sparc/sparc.c: Include cfgrtl.h.
26825 (TARGET_INIT_PIC_REG): Define.
26826 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
26827 (sparc_pic_register_p): New predicate.
26828 (sparc_legitimate_address_p): Use it.
26829 (sparc_legitimize_pic_address): Likewise.
26830 (sparc_delegitimize_address): Likewise.
26831 (sparc_mode_dependent_address_p): Likewise.
26832 (gen_load_pcrel_sym): Remove 4th parameter.
26833 (load_got_register): Adjust call to above. Remove obsolete stuff.
26834 (sparc_expand_prologue): Do not call load_got_register here.
26835 (sparc_flat_expand_prologue): Likewise.
26836 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
26837 (sparc_use_pseudo_pic_reg): New function.
26838 (sparc_init_pic_reg): Likewise.
26839 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
26840 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
26842 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
26844 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26845 Add item for branch_cost.
26847 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
26849 PR rtl-optimization/83565
26850 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
26851 not extend the result to a larger mode for rotate operations.
26852 (num_sign_bit_copies1): Likewise.
26854 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26857 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
26859 Use values-Xc.o for -pedantic.
26860 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
26862 2018-01-12 Martin Liska <mliska@suse.cz>
26865 * ipa-devirt.c (final_warning_record::grow_type_warnings):
26867 (possible_polymorphic_call_targets): Use it.
26868 (ipa_devirt): Likewise.
26870 2018-01-12 Martin Liska <mliska@suse.cz>
26872 * profile-count.h (enum profile_quality): Use 0 as invalid
26873 enum value of profile_quality.
26875 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
26877 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
26878 -mext-string options.
26880 2018-01-12 Richard Biener <rguenther@suse.de>
26882 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
26883 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
26884 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
26886 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
26888 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
26890 * configure.ac (--with-long-double-format): Add support for the
26891 configuration option to change the default long double format on
26893 * config.gcc (powerpc*-linux*-*): Likewise.
26894 * configure: Regenerate.
26895 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
26896 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
26897 used without modification.
26899 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26901 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
26902 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
26903 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
26904 MISC_BUILTIN_SPEC_BARRIER.
26905 (rs6000_init_builtins): Likewise.
26906 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
26908 (speculation_barrier): New define_insn.
26909 * doc/extend.texi: Document __builtin_speculation_barrier.
26911 2018-01-11 Jakub Jelinek <jakub@redhat.com>
26914 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
26915 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
26916 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
26918 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
26919 integral modes instead of "ss" and "sd".
26920 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
26921 vectors with 32-bit and 64-bit elements.
26922 (vecdupssescalarmodesuffix): New mode attribute.
26923 (vec_dup<mode>): Use it.
26925 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
26928 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
26929 frame if argument is passed on stack.
26931 2018-01-11 Jakub Jelinek <jakub@redhat.com>
26934 * ree.c (combine_reaching_defs): Optimize also
26935 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
26936 reg2=any_extend(exp); reg1=reg2;, formatting fix.
26938 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
26940 PR middle-end/83189
26941 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
26943 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
26945 PR middle-end/83718
26946 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
26947 after they are computed.
26949 2018-01-11 Bin Cheng <bin.cheng@arm.com>
26951 PR tree-optimization/83695
26952 * gimple-loop-linterchange.cc
26953 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
26954 reset cached scev information after interchange.
26955 (pass_linterchange::execute): Remove call to scev_reset_htab.
26957 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26959 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
26960 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
26961 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
26962 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
26963 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
26964 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
26965 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
26966 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
26967 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
26968 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
26969 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
26970 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
26971 (V_lane_reg): Likewise.
26972 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
26974 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
26975 (vfmal_lane_low<mode>_intrinsic,
26976 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
26977 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
26978 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
26979 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
26980 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
26981 vfmsl_lane_high<mode>_intrinsic): New define_insns.
26983 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26985 * config/arm/arm-cpus.in (fp16fml): New feature.
26986 (ALL_SIMD): Add fp16fml.
26987 (armv8.2-a): Add fp16fml as an option.
26988 (armv8.3-a): Likewise.
26989 (armv8.4-a): Add fp16fml as part of fp16.
26990 * config/arm/arm.h (TARGET_FP16FML): Define.
26991 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
26993 * config/arm/arm-modes.def (V2HF): Define.
26994 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
26995 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
26996 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
26997 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
26998 vfmsl_low, vfmsl_high): New set of builtins.
26999 * config/arm/iterators.md (PLUSMINUS): New code iterator.
27000 (vfml_op): New code attribute.
27001 (VFMLHALVES): New int iterator.
27002 (VFML, VFMLSEL): New mode attributes.
27003 (V_reg): Define mapping for V2HF.
27004 (V_hi, V_lo): New mode attributes.
27005 (VF_constraint): Likewise.
27006 (vfml_half, vfml_half_selector): New int attributes.
27007 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
27009 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
27010 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
27012 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
27013 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
27014 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
27015 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
27017 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
27018 Document new effective target and option set.
27020 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27022 * config/arm/arm-cpus.in (armv8_4): New feature.
27023 (ARMv8_4a): New fgroup.
27024 (armv8.4-a): New arch.
27025 * config/arm/arm-tables.opt: Regenerate.
27026 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
27027 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
27028 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
27029 Add matching rules for -march=armv8.4-a and extensions.
27030 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
27032 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
27035 * config/rx/rx.md (BW): New mode attribute.
27036 (sync_lock_test_and_setsi): Add mode suffix to insn output.
27038 2018-01-11 Richard Biener <rguenther@suse.de>
27040 PR tree-optimization/83435
27041 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
27042 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
27043 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
27045 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27046 Alan Hayward <alan.hayward@arm.com>
27047 David Sherwood <david.sherwood@arm.com>
27049 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
27051 (aarch64_classify_address): Initialize it. Track polynomial offsets.
27052 (aarch64_print_address_internal): Use it to check for a zero offset.
27054 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27055 Alan Hayward <alan.hayward@arm.com>
27056 David Sherwood <david.sherwood@arm.com>
27058 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
27059 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
27060 Return a poly_int64 rather than a HOST_WIDE_INT.
27061 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
27062 rather than a HOST_WIDE_INT.
27063 * config/aarch64/aarch64.h (aarch64_frame): Protect with
27064 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
27065 hard_fp_offset, frame_size, initial_adjust, callee_offset and
27066 final_offset from HOST_WIDE_INT to poly_int64.
27067 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
27068 to_constant when getting the number of units in an Advanced SIMD
27070 (aarch64_builtin_vectorized_function): Check for a constant number
27072 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
27074 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
27075 attribute instead of GET_MODE_NUNITS.
27076 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
27077 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
27078 GET_MODE_SIZE for fixed-size registers.
27079 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
27080 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
27081 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
27082 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
27083 (aarch64_print_operand, aarch64_print_address_internal)
27084 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
27085 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
27086 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
27087 Handle polynomial GET_MODE_SIZE.
27088 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
27089 wider than SImode without modification.
27090 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
27091 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
27092 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
27093 passing and returning SVE modes.
27094 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
27095 rather than GEN_INT.
27096 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
27097 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
27098 (aarch64_allocate_and_probe_stack_space): Likewise.
27099 (aarch64_layout_frame): Cope with polynomial offsets.
27100 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
27101 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
27102 polynomial offsets.
27103 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
27104 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
27105 poly_int64 rather than a HOST_WIDE_INT.
27106 (aarch64_get_separate_components, aarch64_process_components)
27107 (aarch64_expand_prologue, aarch64_expand_epilogue)
27108 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
27109 (aarch64_anchor_offset): New function, split out from...
27110 (aarch64_legitimize_address): ...here.
27111 (aarch64_builtin_vectorization_cost): Handle polynomial
27112 TYPE_VECTOR_SUBPARTS.
27113 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
27115 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
27116 number of elements from the PARALLEL rather than the mode.
27117 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
27118 rather than GET_MODE_BITSIZE.
27119 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
27120 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
27121 (aarch64_expand_vec_perm_const_1): Handle polynomial
27122 d->perm.length () and d->perm elements.
27123 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
27124 Apply to_constant to d->perm elements.
27125 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
27126 polynomial CONST_VECTOR_NUNITS.
27127 (aarch64_move_pointer): Take amount as a poly_int64 rather
27129 (aarch64_progress_pointer): Avoid temporary variable.
27130 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
27131 the mode attribute instead of GET_MODE.
27133 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27134 Alan Hayward <alan.hayward@arm.com>
27135 David Sherwood <david.sherwood@arm.com>
27137 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
27138 x exists before using it.
27139 (aarch64_add_constant_internal): Rename to...
27140 (aarch64_add_offset_1): ...this. Replace regnum with separate
27141 src and dest rtxes. Handle the case in which they're different,
27142 including when the offset is zero. Replace scratchreg with an rtx.
27143 Use 2 additions if there is no spare register into which we can
27144 move a 16-bit constant.
27145 (aarch64_add_constant): Delete.
27146 (aarch64_add_offset): Replace reg with separate src and dest
27147 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
27148 Use aarch64_add_offset_1.
27149 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
27150 an rtx rather than an int. Take the delta as a poly_int64
27151 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
27152 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
27153 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
27154 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
27155 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
27156 and aarch64_add_sp.
27157 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
27158 aarch64_add_constant.
27160 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27162 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
27163 Use scalar_float_mode.
27165 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27167 * config/aarch64/aarch64-simd.md
27168 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
27169 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
27170 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
27171 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
27172 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
27173 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
27174 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
27175 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
27176 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
27177 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
27179 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27182 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
27183 targ_options->x_arm_arch_string is non NULL.
27185 2018-01-11 Tamar Christina <tamar.christina@arm.com>
27187 * config/aarch64/aarch64.h
27188 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
27190 2018-01-11 Sudakshina Das <sudi.das@arm.com>
27193 * expmed.c (emit_store_flag_force): Swap if const op0
27194 and change VOIDmode to mode of op0.
27196 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27198 PR rtl-optimization/83761
27199 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
27200 than bytes to mode_for_size.
27202 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
27204 PR middle-end/83189
27205 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
27206 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
27209 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
27211 PR middle-end/83575
27212 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
27213 when in layout mode.
27214 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
27215 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
27218 2018-01-10 Michael Collison <michael.collison@arm.com>
27220 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
27221 * config/aarch64/aarch64-option-extension.def: Add
27222 AARCH64_OPT_EXTENSION of 'fp16fml'.
27223 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27224 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
27225 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
27226 * config/aarch64/constraints.md (Ui7): New constraint.
27227 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
27228 (VFMLA_SEL_W): Ditto.
27231 (VFMLA16_LOW): New int iterator.
27232 (VFMLA16_HIGH): Ditto.
27233 (UNSPEC_FMLAL): New unspec.
27234 (UNSPEC_FMLSL): Ditto.
27235 (UNSPEC_FMLAL2): Ditto.
27236 (UNSPEC_FMLSL2): Ditto.
27237 (f16mac): New code attribute.
27238 * config/aarch64/aarch64-simd-builtins.def
27239 (aarch64_fmlal_lowv2sf): Ditto.
27240 (aarch64_fmlsl_lowv2sf): Ditto.
27241 (aarch64_fmlalq_lowv4sf): Ditto.
27242 (aarch64_fmlslq_lowv4sf): Ditto.
27243 (aarch64_fmlal_highv2sf): Ditto.
27244 (aarch64_fmlsl_highv2sf): Ditto.
27245 (aarch64_fmlalq_highv4sf): Ditto.
27246 (aarch64_fmlslq_highv4sf): Ditto.
27247 (aarch64_fmlal_lane_lowv2sf): Ditto.
27248 (aarch64_fmlsl_lane_lowv2sf): Ditto.
27249 (aarch64_fmlal_laneq_lowv2sf): Ditto.
27250 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
27251 (aarch64_fmlalq_lane_lowv4sf): Ditto.
27252 (aarch64_fmlsl_lane_lowv4sf): Ditto.
27253 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
27254 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
27255 (aarch64_fmlal_lane_highv2sf): Ditto.
27256 (aarch64_fmlsl_lane_highv2sf): Ditto.
27257 (aarch64_fmlal_laneq_highv2sf): Ditto.
27258 (aarch64_fmlsl_laneq_highv2sf): Ditto.
27259 (aarch64_fmlalq_lane_highv4sf): Ditto.
27260 (aarch64_fmlsl_lane_highv4sf): Ditto.
27261 (aarch64_fmlalq_laneq_highv4sf): Ditto.
27262 (aarch64_fmlsl_laneq_highv4sf): Ditto.
27263 * config/aarch64/aarch64-simd.md:
27264 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
27265 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
27266 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
27267 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
27268 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
27269 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
27270 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
27271 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
27272 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
27273 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
27274 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
27275 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
27276 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
27277 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
27278 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
27279 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
27280 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
27281 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
27282 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
27283 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
27284 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
27285 (vfmlsl_low_u32): Ditto.
27286 (vfmlalq_low_u32): Ditto.
27287 (vfmlslq_low_u32): Ditto.
27288 (vfmlal_high_u32): Ditto.
27289 (vfmlsl_high_u32): Ditto.
27290 (vfmlalq_high_u32): Ditto.
27291 (vfmlslq_high_u32): Ditto.
27292 (vfmlal_lane_low_u32): Ditto.
27293 (vfmlsl_lane_low_u32): Ditto.
27294 (vfmlal_laneq_low_u32): Ditto.
27295 (vfmlsl_laneq_low_u32): Ditto.
27296 (vfmlalq_lane_low_u32): Ditto.
27297 (vfmlslq_lane_low_u32): Ditto.
27298 (vfmlalq_laneq_low_u32): Ditto.
27299 (vfmlslq_laneq_low_u32): Ditto.
27300 (vfmlal_lane_high_u32): Ditto.
27301 (vfmlsl_lane_high_u32): Ditto.
27302 (vfmlal_laneq_high_u32): Ditto.
27303 (vfmlsl_laneq_high_u32): Ditto.
27304 (vfmlalq_lane_high_u32): Ditto.
27305 (vfmlslq_lane_high_u32): Ditto.
27306 (vfmlalq_laneq_high_u32): Ditto.
27307 (vfmlslq_laneq_high_u32): Ditto.
27308 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
27309 (AARCH64_FL_FOR_ARCH8_4): New.
27310 (AARCH64_ISA_F16FML): New ISA flag.
27311 (TARGET_F16FML): New feature flag for fp16fml.
27312 (doc/invoke.texi): Document new fp16fml option.
27314 2018-01-10 Michael Collison <michael.collison@arm.com>
27316 * config/aarch64/aarch64-builtins.c:
27317 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
27318 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27319 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
27320 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
27321 (AARCH64_ISA_SHA3): New ISA flag.
27322 (TARGET_SHA3): New feature flag for sha3.
27323 * config/aarch64/iterators.md (sha512_op): New int attribute.
27324 (CRYPTO_SHA512): New int iterator.
27325 (UNSPEC_SHA512H): New unspec.
27326 (UNSPEC_SHA512H2): Ditto.
27327 (UNSPEC_SHA512SU0): Ditto.
27328 (UNSPEC_SHA512SU1): Ditto.
27329 * config/aarch64/aarch64-simd-builtins.def
27330 (aarch64_crypto_sha512hqv2di): New builtin.
27331 (aarch64_crypto_sha512h2qv2di): Ditto.
27332 (aarch64_crypto_sha512su0qv2di): Ditto.
27333 (aarch64_crypto_sha512su1qv2di): Ditto.
27334 (aarch64_eor3qv8hi): Ditto.
27335 (aarch64_rax1qv2di): Ditto.
27336 (aarch64_xarqv2di): Ditto.
27337 (aarch64_bcaxqv8hi): Ditto.
27338 * config/aarch64/aarch64-simd.md:
27339 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
27340 (aarch64_crypto_sha512su0qv2di): Ditto.
27341 (aarch64_crypto_sha512su1qv2di): Ditto.
27342 (aarch64_eor3qv8hi): Ditto.
27343 (aarch64_rax1qv2di): Ditto.
27344 (aarch64_xarqv2di): Ditto.
27345 (aarch64_bcaxqv8hi): Ditto.
27346 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
27347 (vsha512h2q_u64): Ditto.
27348 (vsha512su0q_u64): Ditto.
27349 (vsha512su1q_u64): Ditto.
27350 (veor3q_u16): Ditto.
27351 (vrax1q_u64): Ditto.
27352 (vxarq_u64): Ditto.
27353 (vbcaxq_u16): Ditto.
27354 * config/arm/types.md (crypto_sha512): New type attribute.
27355 (crypto_sha3): Ditto.
27356 (doc/invoke.texi): Document new sha3 option.
27358 2018-01-10 Michael Collison <michael.collison@arm.com>
27360 * config/aarch64/aarch64-builtins.c:
27361 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
27362 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27363 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
27364 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
27365 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
27366 (AARCH64_ISA_SM4): New ISA flag.
27367 (TARGET_SM4): New feature flag for sm4.
27368 * config/aarch64/aarch64-simd-builtins.def
27369 (aarch64_sm3ss1qv4si): Ditto.
27370 (aarch64_sm3tt1aq4si): Ditto.
27371 (aarch64_sm3tt1bq4si): Ditto.
27372 (aarch64_sm3tt2aq4si): Ditto.
27373 (aarch64_sm3tt2bq4si): Ditto.
27374 (aarch64_sm3partw1qv4si): Ditto.
27375 (aarch64_sm3partw2qv4si): Ditto.
27376 (aarch64_sm4eqv4si): Ditto.
27377 (aarch64_sm4ekeyqv4si): Ditto.
27378 * config/aarch64/aarch64-simd.md:
27379 (aarch64_sm3ss1qv4si): Ditto.
27380 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
27381 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
27382 (aarch64_sm4eqv4si): Ditto.
27383 (aarch64_sm4ekeyqv4si): Ditto.
27384 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
27385 (sm3part_op): Ditto.
27386 (CRYPTO_SM3TT): Ditto.
27387 (CRYPTO_SM3PART): Ditto.
27388 (UNSPEC_SM3SS1): New unspec.
27389 (UNSPEC_SM3TT1A): Ditto.
27390 (UNSPEC_SM3TT1B): Ditto.
27391 (UNSPEC_SM3TT2A): Ditto.
27392 (UNSPEC_SM3TT2B): Ditto.
27393 (UNSPEC_SM3PARTW1): Ditto.
27394 (UNSPEC_SM3PARTW2): Ditto.
27395 (UNSPEC_SM4E): Ditto.
27396 (UNSPEC_SM4EKEY): Ditto.
27397 * config/aarch64/constraints.md (Ui2): New constraint.
27398 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
27399 * config/arm/types.md (crypto_sm3): New type attribute.
27400 (crypto_sm4): Ditto.
27401 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
27402 (vsm3tt1aq_u32): Ditto.
27403 (vsm3tt1bq_u32): Ditto.
27404 (vsm3tt2aq_u32): Ditto.
27405 (vsm3tt2bq_u32): Ditto.
27406 (vsm3partw1q_u32): Ditto.
27407 (vsm3partw2q_u32): Ditto.
27408 (vsm4eq_u32): Ditto.
27409 (vsm4ekeyq_u32): Ditto.
27410 (doc/invoke.texi): Document new sm4 option.
27412 2018-01-10 Michael Collison <michael.collison@arm.com>
27414 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
27415 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
27416 (AARCH64_FL_FOR_ARCH8_4): New.
27417 (AARCH64_FL_V8_4): New flag.
27418 (doc/invoke.texi): Document new armv8.4-a option.
27420 2018-01-10 Michael Collison <michael.collison@arm.com>
27422 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27423 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
27424 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
27425 * config/aarch64/aarch64-option-extension.def: Add
27426 AARCH64_OPT_EXTENSION of 'sha2'.
27427 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
27428 (crypto): Disable sha2 and aes if crypto disabled.
27429 (crypto): Enable aes and sha2 if enabled.
27430 (simd): Disable sha2 and aes if simd disabled.
27431 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
27433 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
27434 (TARGET_SHA2): New feature flag for sha2.
27435 (TARGET_AES): New feature flag for aes.
27436 * config/aarch64/aarch64-simd.md:
27437 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
27438 conditional on TARGET_AES.
27439 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
27440 (aarch64_crypto_sha1hsi): Make pattern conditional
27442 (aarch64_crypto_sha1hv4si): Ditto.
27443 (aarch64_be_crypto_sha1hv4si): Ditto.
27444 (aarch64_crypto_sha1su1v4si): Ditto.
27445 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
27446 (aarch64_crypto_sha1su0v4si): Ditto.
27447 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
27448 (aarch64_crypto_sha256su0v4si): Ditto.
27449 (aarch64_crypto_sha256su1v4si): Ditto.
27450 (doc/invoke.texi): Document new aes and sha2 options.
27452 2018-01-10 Martin Sebor <msebor@redhat.com>
27454 PR tree-optimization/83781
27455 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
27458 2018-01-11 Martin Sebor <msebor@gmail.com>
27459 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27461 PR tree-optimization/83501
27462 PR tree-optimization/81703
27464 * tree-ssa-strlen.c (get_string_cst): Rename...
27465 (get_string_len): ...to this. Handle global constants.
27466 (handle_char_store): Adjust.
27468 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
27469 Jim Wilson <jimw@sifive.com>
27471 * config/riscv/riscv-protos.h (riscv_output_return): New.
27472 * config/riscv/riscv.c (struct machine_function): New naked_p field.
27473 (riscv_attribute_table, riscv_output_return),
27474 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
27475 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
27476 (riscv_compute_frame_info): Only compute frame->mask if not a naked
27478 (riscv_expand_prologue): Add early return for naked function.
27479 (riscv_expand_epilogue): Likewise.
27480 (riscv_function_ok_for_sibcall): Return false for naked function.
27481 (riscv_set_current_function): New.
27482 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
27483 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
27484 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
27485 * doc/extend.texi (RISC-V Function Attributes): New.
27487 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
27489 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
27490 check for 128-bit long double before checking TCmode.
27491 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
27492 128-bit long doubles before checking TFmode or TCmode.
27493 (FLOAT128_IBM_P): Likewise.
27495 2018-01-10 Martin Sebor <msebor@redhat.com>
27497 PR tree-optimization/83671
27498 * builtins.c (c_strlen): Unconditionally return zero for the empty
27500 Use -Warray-bounds for warnings.
27501 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
27502 for non-constant array indices with COMPONENT_REF, arrays of
27503 arrays, and pointers to arrays.
27504 (gimple_fold_builtin_strlen): Determine and set length range for
27505 non-constant character arrays.
27507 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
27509 PR middle-end/81897
27510 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
27513 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
27515 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
27517 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
27520 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
27521 VECTOR_MEM_ALTIVEC_OR_VSX_P.
27522 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
27523 indexed_or_indirect_operand predicate.
27524 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
27525 (*vsx_le_perm_load_v8hi): Likewise.
27526 (*vsx_le_perm_load_v16qi): Likewise.
27527 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
27528 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
27529 (*vsx_le_perm_store_v8hi): Likewise.
27530 (*vsx_le_perm_store_v16qi): Likewise.
27531 (eight unnamed splitters): Likewise.
27533 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
27535 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
27536 * config/rs6000/emmintrin.h: Likewise.
27537 * config/rs6000/mmintrin.h: Likewise.
27538 * config/rs6000/xmmintrin.h: Likewise.
27540 2018-01-10 David Malcolm <dmalcolm@redhat.com>
27543 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
27545 * tree.c (tree_nop_conversion): Return true for location wrapper
27547 (maybe_wrap_with_location): New function.
27548 (selftest::check_strip_nops): New function.
27549 (selftest::test_location_wrappers): New function.
27550 (selftest::tree_c_tests): Call it.
27551 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
27552 (maybe_wrap_with_location): New decl.
27553 (EXPR_LOCATION_WRAPPER_P): New macro.
27554 (location_wrapper_p): New inline function.
27555 (tree_strip_any_location_wrapper): New inline function.
27557 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
27560 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
27561 stack_realign_offset for the largest alignment of stack slot
27563 (ix86_find_max_used_stack_alignment): New function.
27564 (ix86_finalize_stack_frame_flags): Use it. Set
27565 max_used_stack_alignment if we don't realign stack.
27566 * config/i386/i386.h (machine_function): Add
27567 max_used_stack_alignment.
27569 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
27571 * config/arm/arm.opt (-mbranch-cost): New option.
27572 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
27575 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
27578 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
27579 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
27581 2018-01-10 Richard Biener <rguenther@suse.de>
27584 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
27585 early out so it also covers the case where we have a non-NULL
27588 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
27590 PR tree-optimization/83753
27591 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
27592 for non-strided grouped accesses if the number of elements is 1.
27594 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
27597 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
27598 * i386.h (TARGET_USE_GATHER): Define.
27599 * x86-tune.def (X86_TUNE_USE_GATHER): New.
27601 2018-01-10 Martin Liska <mliska@suse.cz>
27604 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
27605 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
27607 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
27608 CLEANUP_NO_PARTITIONING is not set.
27610 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
27612 * doc/rtl.texi: Remove documentation of (const ...) wrappers
27613 for vectors, as a partial revert of r254296.
27614 * rtl.h (const_vec_p): Delete.
27615 (const_vec_duplicate_p): Don't test for vector CONSTs.
27616 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
27617 * expmed.c (make_tree): Likewise.
27620 * common.md (E, F): Use CONSTANT_P instead of checking for
27622 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
27623 checking for CONST_VECTOR.
27625 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
27627 PR middle-end/83575
27628 * predict.c (force_edge_cold): Handle in more sane way edges
27629 with no prediction.
27631 2018-01-09 Carl Love <cel@us.ibm.com>
27633 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
27635 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
27636 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
27637 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
27638 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
27639 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
27640 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
27641 * config/rs6000/rs6000-protos.h: Add extern defition for
27642 rs6000_generate_float2_double_code.
27643 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
27645 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
27646 (float2_v2df): Add define_expand.
27648 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
27651 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
27652 op_mode in the force_to_mode call.
27654 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
27656 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
27657 instead of checking each element individually.
27658 (aarch64_evpc_uzp): Likewise.
27659 (aarch64_evpc_zip): Likewise.
27660 (aarch64_evpc_ext): Likewise.
27661 (aarch64_evpc_rev): Likewise.
27662 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
27663 instead of checking each element individually. Return true without
27665 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
27666 whether all selected elements come from the same input, instead of
27667 checking each element individually. Remove calls to gen_rtx_REG,
27668 start_sequence and end_sequence and instead assert that no rtl is
27671 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
27673 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
27674 order of HIGH and CONST checks.
27676 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
27678 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
27679 if the destination isn't an SSA_NAME.
27681 2018-01-09 Richard Biener <rguenther@suse.de>
27683 PR tree-optimization/83668
27684 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
27686 (canonicalize_loop_form): ... here, renamed from ...
27687 (canonicalize_loop_closed_ssa_form): ... this and amended to
27688 swap successor edges for loop exit blocks to make us use
27689 the RPO order we need for initial schedule generation.
27691 2018-01-09 Joseph Myers <joseph@codesourcery.com>
27693 PR tree-optimization/64811
27694 * match.pd: When optimizing comparisons with Inf, avoid
27695 introducing or losing exceptions from comparisons with NaN.
27697 2018-01-09 Martin Liska <mliska@suse.cz>
27700 * asan.c (shadow_mem_size): Add gcc_assert.
27702 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
27704 Don't save registers in main().
27707 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
27708 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
27709 * config/avr/avr.c (avr_set_current_function): Don't error if
27710 naked, OS_task or OS_main are specified at the same time.
27711 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
27713 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
27715 * common/config/avr/avr-common.c (avr_option_optimization_table):
27716 Switch on -mmain-is-OS_task for optimizing compilations.
27718 2018-01-09 Richard Biener <rguenther@suse.de>
27720 PR tree-optimization/83572
27721 * graphite.c: Include cfganal.h.
27722 (graphite_transform_loops): Connect infinite loops to exit
27723 and remove fake edges at the end.
27725 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
27727 * ipa-inline.c (edge_badness): Revert accidental checkin.
27729 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
27732 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
27733 symbols; not inline clones.
27735 2018-01-09 Jakub Jelinek <jakub@redhat.com>
27738 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
27739 hard registers. Formatting fixes.
27741 PR preprocessor/83722
27742 * gcc.c (try_generate_repro): Pass
27743 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
27744 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
27747 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
27748 Kito Cheng <kito.cheng@gmail.com>
27750 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
27751 (riscv_leaf_function_p): Delete.
27752 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
27754 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27756 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
27758 (do_ifelse): New function.
27759 (do_isel): New function.
27760 (do_sub3): New function.
27761 (do_add3): New function.
27762 (do_load_mask_compare): New function.
27763 (do_overlap_load_compare): New function.
27764 (expand_compare_loop): New function.
27765 (expand_block_compare): Call expand_compare_loop() when appropriate.
27766 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
27767 option description.
27768 (-mblock-compare-inline-loop-limit): New option.
27770 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27773 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
27774 Reverse order of second and third operands in first alternative.
27775 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
27776 of first and second elements in UNSPEC_VPERMR vector.
27777 (altivec_expand_vec_perm_le): Likewise.
27779 2018-01-08 Jeff Law <law@redhat.com>
27781 PR rtl-optimizatin/81308
27782 * tree-switch-conversion.c (cfg_altered): New file scoped static.
27783 (process_switch): If group_case_labels makes a change, then set
27785 (pass_convert_switch::execute): If a switch is converted, then
27786 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
27788 PR rtl-optimization/81308
27789 * recog.c (split_all_insns): Conditionally cleanup the CFG after
27792 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
27794 PR target/83663 - Revert r255946
27795 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
27796 generation for cases where splatting a value is not useful.
27797 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
27798 across a vec_duplicate and a paradoxical subreg forming a vector
27799 mode to a vec_concat.
27801 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27803 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
27804 -march=armv8.3-a variants.
27805 * config/arm/t-multilib: Likewise.
27806 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
27808 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27810 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
27812 (cceq_ior_compare_complement): Give it a name so I can use it, and
27813 change boolean_or_operator predicate to boolean_operator so it can
27814 be used to generate a crand.
27815 (eqne): New code iterator.
27816 (bd/bd_neg): New code_attrs.
27817 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
27818 a single define_insn.
27819 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
27820 decrement (bdnzt/bdnzf/bdzt/bdzf).
27821 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
27822 with the new names of the branch decrement patterns, and added the
27823 names of the branch decrement conditional patterns.
27825 2018-01-08 Richard Biener <rguenther@suse.de>
27827 PR tree-optimization/83563
27828 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
27831 2018-01-08 Richard Biener <rguenther@suse.de>
27833 PR middle-end/83713
27834 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
27836 2018-01-08 Richard Biener <rguenther@suse.de>
27838 PR tree-optimization/83685
27839 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
27840 references to abnormals.
27842 2018-01-08 Richard Biener <rguenther@suse.de>
27845 * dwarf2out.c (output_indirect_strings): Handle empty
27846 skeleton_debug_str_hash.
27847 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
27849 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
27851 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
27852 (emit_store_direct): Likewise.
27853 (arc_trampoline_adjust_address): Likewise.
27854 (arc_asm_trampoline_template): New function.
27855 (arc_initialize_trampoline): Use asm_trampoline_template.
27856 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
27857 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
27858 * config/arc/arc.md (flush_icache): Delete pattern.
27860 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
27862 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
27863 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
27866 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
27869 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
27870 by not USED_FOR_TARGET.
27871 (make_pass_resolve_sw_modes): Likewise.
27873 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
27875 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
27878 2018-01-08 Richard Biener <rguenther@suse.de>
27880 PR middle-end/83580
27881 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
27883 2018-01-08 Richard Biener <rguenther@suse.de>
27885 PR middle-end/83517
27886 * match.pd ((t * 2) / 2) -> t): Add missing :c.
27888 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
27890 PR middle-end/81897
27891 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
27892 basic blocks with a small number of successors.
27893 (convert_control_dep_chain_into_preds): Improve handling of
27895 (dump_predicates): Split apart into...
27896 (dump_pred_chain): ...here...
27897 (dump_pred_info): ...and here.
27898 (can_one_predicate_be_invalidated_p): Add debugging printfs.
27899 (can_chain_union_be_invalidated_p): Improve check for invalidation
27901 (uninit_uses_cannot_happen): Avoid unnecessary if
27902 convert_control_dep_chain_into_preds yielded nothing.
27904 2018-01-06 Martin Sebor <msebor@redhat.com>
27906 PR tree-optimization/83640
27907 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
27908 subtracting negative offset from size.
27909 (builtin_access::overlap): Adjust offset bounds of the access to fall
27910 within the size of the object if possible.
27912 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
27914 PR rtl-optimization/83699
27915 * expmed.c (extract_bit_field_1): Restrict the vector usage of
27916 extract_bit_field_as_subreg to cases in which the extracted
27917 value is also a vector.
27919 * lra-constraints.c (process_alt_operands): Test for the equivalence
27920 substitutions when detecting a possible reload cycle.
27922 2018-01-06 Jakub Jelinek <jakub@redhat.com>
27925 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
27926 by default if flag_selective_schedling{,2}. Formatting fixes.
27928 PR rtl-optimization/83682
27929 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
27930 if it has non-VECTOR_MODE element mode.
27931 (vec_duplicate_p): Likewise.
27933 PR middle-end/83694
27934 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
27935 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
27937 2018-01-05 Jakub Jelinek <jakub@redhat.com>
27940 * config/i386/i386-builtin.def
27941 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
27942 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
27943 Require also OPTION_MASK_ISA_AVX512F in addition to
27944 OPTION_MASK_ISA_GFNI.
27945 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
27946 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
27947 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
27948 to OPTION_MASK_ISA_GFNI.
27949 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
27950 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
27951 OPTION_MASK_ISA_AVX512BW.
27952 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
27953 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
27954 addition to OPTION_MASK_ISA_GFNI.
27955 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
27956 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
27957 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
27958 to OPTION_MASK_ISA_GFNI.
27959 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
27960 a requirement for all ISAs rather than any of them with a few
27962 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
27964 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
27965 bitmasks to be enabled with 3 exceptions, instead of requiring any
27966 enabled ISA with lots of exceptions.
27967 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
27968 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
27969 Change avx512bw in isa attribute to avx512f.
27970 * config/i386/sgxintrin.h: Add license boilerplate.
27971 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
27972 to __AVX512F__ and __AVX512VL to __AVX512VL__.
27973 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
27974 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
27976 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
27977 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
27978 temporarily sse2 rather than sse if not enabled already.
27981 * config/i386/sse.md (VI248_VLBW): Rename to ...
27982 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
27983 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
27984 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
27985 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
27986 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
27987 mode iterator instead of VI248_VLBW.
27989 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
27991 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
27992 (record_modified): Skip clobbers; add debug output.
27993 (param_change_prob): Use sreal frequencies.
27995 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
27997 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
27998 punt for user-aligned variables.
28000 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
28002 * tree-chrec.c (chrec_contains_symbols): Return true for
28005 2018-01-05 Sudakshina Das <sudi.das@arm.com>
28008 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
28009 of (x|y) == x for BICS pattern.
28011 2018-01-05 Jakub Jelinek <jakub@redhat.com>
28013 PR tree-optimization/83605
28014 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
28015 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
28018 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
28020 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
28021 * config/epiphany/rtems.h: New file.
28023 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28024 Uros Bizjak <ubizjak@gmail.com>
28027 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
28028 QIreg_operand instead of register_operand predicate.
28029 * config/i386/i386.c (ix86_rop_should_change_byte_p,
28030 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
28031 comments instead of -fmitigate[-_]rop.
28033 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28036 * cgraphunit.c (symbol_table::compile): Switch to text_section
28037 before calling assembly_start debug hook.
28038 * run-rtl-passes.c (run_rtl_passes): Likewise.
28041 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28043 * tree-vrp.c (extract_range_from_binary_expr_1): Check
28044 range_int_cst_p rather than !symbolic_range_p before calling
28045 extract_range_from_multiplicative_op_1.
28047 2018-01-04 Jeff Law <law@redhat.com>
28049 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
28050 redundant test in assertion.
28052 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28054 * doc/rtl.texi: Document machine_mode wrapper classes.
28056 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28058 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
28059 using tree_to_uhwi.
28061 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28063 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
28064 the VEC_PERM_EXPR fold to fail.
28066 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28069 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
28070 to switched_sections.
28072 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28075 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
28076 test for d.testing.
28078 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
28081 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
28082 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
28084 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28087 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
28088 is BLKmode and bitpos not zero or mode change is needed.
28090 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28093 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
28096 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
28099 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
28100 instead of MULT rtx. Update all corresponding splitters.
28101 (*saddl_se): Ditto.
28102 (*ssub<modesuffix>): Ditto.
28103 (*ssubl_se): Ditto.
28104 (*cmp_sadd_di): Update split patterns.
28105 (*cmp_sadd_si): Ditto.
28106 (*cmp_sadd_sidi): Ditto.
28107 (*cmp_ssub_di): Ditto.
28108 (*cmp_ssub_si): Ditto.
28109 (*cmp_ssub_sidi): Ditto.
28110 * config/alpha/predicates.md (const23_operand): New predicate.
28111 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
28112 Look for ASHIFT, not MULT inner operand.
28113 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
28115 2018-01-04 Martin Liska <mliska@suse.cz>
28117 PR gcov-profile/83669
28118 * gcov.c (output_intermediate_file): Add version to intermediate
28120 * doc/gcov.texi: Document new field 'version' in intermediate
28121 file format. Fix location of '-k' option of gcov command.
28123 2018-01-04 Martin Liska <mliska@suse.cz>
28126 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
28128 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28130 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
28132 2018-01-03 Martin Sebor <msebor@redhat.com>
28134 PR tree-optimization/83655
28135 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
28136 checking calls with invalid arguments.
28138 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28140 * tree-vect-stmts.c (vect_get_store_rhs): New function.
28141 (vectorizable_mask_load_store): Delete.
28142 (vectorizable_call): Return false for masked loads and stores.
28143 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
28144 instead of gimple_assign_rhs1.
28145 (vectorizable_load): Handle IFN_MASK_LOAD.
28146 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
28148 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28150 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
28152 (vectorizable_mask_load_store): ...here.
28153 (vectorizable_load): ...and here.
28155 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28157 * tree-vect-stmts.c (vect_build_all_ones_mask)
28158 (vect_build_zero_merge_argument): New functions, split out from...
28159 (vectorizable_load): ...here.
28161 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28163 * tree-vect-stmts.c (vect_check_store_rhs): New function,
28165 (vectorizable_mask_load_store): ...here.
28166 (vectorizable_store): ...and here.
28168 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28170 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
28172 (vectorizable_mask_load_store): ...here.
28174 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28176 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
28177 (vect_model_store_cost): Take a vec_load_store_type instead of a
28179 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
28180 (vect_model_store_cost): Take a vec_load_store_type instead of a
28182 (vectorizable_mask_load_store): Update accordingly.
28183 (vectorizable_store): Likewise.
28184 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
28186 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28188 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
28189 IFN_MASK_LOAD calls here rather than...
28190 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
28192 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28193 Alan Hayward <alan.hayward@arm.com>
28194 David Sherwood <david.sherwood@arm.com>
28196 * expmed.c (extract_bit_field_1): For vector extracts,
28197 fall back to extract_bit_field_as_subreg if vec_extract
28200 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28201 Alan Hayward <alan.hayward@arm.com>
28202 David Sherwood <david.sherwood@arm.com>
28204 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
28205 they are variable or constant sized.
28206 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
28207 slots for constant-sized data.
28209 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28210 Alan Hayward <alan.hayward@arm.com>
28211 David Sherwood <david.sherwood@arm.com>
28213 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
28214 handling COND_EXPRs with boolean comparisons, try to find a better
28215 basis for the mask type than the boolean itself.
28217 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28219 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
28220 is calculated and how it can be overridden.
28221 * genmodes.c (max_bitsize_mode_any_mode): New variable.
28222 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
28224 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
28227 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28228 Alan Hayward <alan.hayward@arm.com>
28229 David Sherwood <david.sherwood@arm.com>
28231 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
28232 Remove the mode argument.
28233 (aarch64_simd_valid_immediate): Remove the mode and inverse
28235 * config/aarch64/iterators.md (bitsize): New iterator.
28236 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
28237 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
28238 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
28239 aarch64_simd_valid_immediate.
28240 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
28241 (aarch64_reg_or_bic_imm): Likewise.
28242 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
28243 with an insn_type enum and msl with a modifier_type enum.
28244 Replace element_width with a scalar_mode. Change the shift
28245 to unsigned int. Add constructors for scalar_float_mode and
28246 scalar_int_mode elements.
28247 (aarch64_vect_float_const_representable_p): Delete.
28248 (aarch64_can_const_movi_rtx_p)
28249 (aarch64_simd_scalar_immediate_valid_for_move)
28250 (aarch64_simd_make_constant): Update call to
28251 aarch64_simd_valid_immediate.
28252 (aarch64_advsimd_valid_immediate_hs): New function.
28253 (aarch64_advsimd_valid_immediate): Likewise.
28254 (aarch64_simd_valid_immediate): Remove mode and inverse
28255 arguments. Rewrite to use the above. Use const_vec_duplicate_p
28256 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
28257 and aarch64_float_const_representable_p on the result.
28258 (aarch64_output_simd_mov_immediate): Remove mode argument.
28259 Update call to aarch64_simd_valid_immediate and use of
28260 simd_immediate_info.
28261 (aarch64_output_scalar_simd_mov_immediate): Update call
28264 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28265 Alan Hayward <alan.hayward@arm.com>
28266 David Sherwood <david.sherwood@arm.com>
28268 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
28269 (mode_nunits): Likewise CONST_MODE_NUNITS.
28270 * machmode.def (ADJUST_NUNITS): Document.
28271 * genmodes.c (mode_data::need_nunits_adj): New field.
28272 (blank_mode): Update accordingly.
28273 (adj_nunits): New variable.
28274 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
28276 (emit_mode_size_inline): Set need_bytesize_adj for all modes
28277 listed in adj_nunits.
28278 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
28279 listed in adj_nunits. Don't emit case statements for such modes.
28280 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
28281 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
28282 nothing if adj_nunits is nonnull.
28283 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
28284 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
28285 (emit_mode_fbit): Update use of print_maybe_const_decl.
28286 (emit_move_size): Likewise. Treat the array as non-const
28288 (emit_mode_adjustments): Handle adj_nunits.
28290 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28292 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
28293 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
28294 (VECTOR_MODES): Use it.
28295 (make_vector_modes): Take the prefix as an argument.
28297 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28298 Alan Hayward <alan.hayward@arm.com>
28299 David Sherwood <david.sherwood@arm.com>
28301 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
28302 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
28303 for MODE_VECTOR_BOOL.
28304 * machmode.def (VECTOR_BOOL_MODE): Document.
28305 * genmodes.c (VECTOR_BOOL_MODE): New macro.
28306 (make_vector_bool_mode): New function.
28307 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
28309 * lto-streamer-in.c (lto_input_mode_table): Likewise.
28310 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
28312 * stor-layout.c (int_mode_for_mode): Likewise.
28313 * tree.c (build_vector_type_for_mode): Likewise.
28314 * varasm.c (output_constant_pool_2): Likewise.
28315 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
28316 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
28317 for MODE_VECTOR_BOOL.
28318 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
28319 of mode class checks.
28320 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
28321 instead of a list of mode class checks.
28322 (expand_vector_scalar_condition): Likewise.
28323 (type_for_widest_vector_mode): Handle BImode as an inner mode.
28325 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28326 Alan Hayward <alan.hayward@arm.com>
28327 David Sherwood <david.sherwood@arm.com>
28329 * machmode.h (mode_size): Change from unsigned short to
28331 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
28332 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
28333 or if measurement_type is not polynomial.
28334 (fixed_size_mode::includes_p): Check for constant-sized modes.
28335 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
28336 return a poly_uint16 rather than an unsigned short.
28337 (emit_mode_size): Change the type of mode_size from unsigned short
28338 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
28339 (emit_mode_adjustments): Cope with polynomial vector sizes.
28340 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
28342 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
28344 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
28345 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
28346 * caller-save.c (setup_save_areas): Likewise.
28347 (replace_reg_with_saved_mem): Likewise.
28348 * calls.c (emit_library_call_value_1): Likewise.
28349 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
28350 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
28351 (gen_lowpart_for_combine): Likewise.
28352 * convert.c (convert_to_integer_1): Likewise.
28353 * cse.c (equiv_constant, cse_insn): Likewise.
28354 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
28355 (cselib_subst_to_values): Likewise.
28356 * dce.c (word_dce_process_block): Likewise.
28357 * df-problems.c (df_word_lr_mark_ref): Likewise.
28358 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
28359 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
28360 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
28361 (rtl_for_decl_location): Likewise.
28362 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
28363 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
28364 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
28365 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
28366 (expand_expr_real_1): Likewise.
28367 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
28368 (pad_below): Likewise.
28369 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
28370 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
28371 * ira.c (get_subreg_tracking_sizes): Likewise.
28372 * ira-build.c (ira_create_allocno_objects): Likewise.
28373 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
28374 (ira_sort_regnos_for_alter_reg): Likewise.
28375 * ira-costs.c (record_operand_costs): Likewise.
28376 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
28377 (resolve_simple_move): Likewise.
28378 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
28379 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
28380 (lra_constraints): Likewise.
28381 (CONST_POOL_OK_P): Reject variable-sized modes.
28382 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
28383 (add_pseudo_to_slot, lra_spill): Likewise.
28384 * omp-low.c (omp_clause_aligned_alignment): Likewise.
28385 * optabs-query.c (get_best_extraction_insn): Likewise.
28386 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
28387 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
28388 (expand_mult_highpart, valid_multiword_target_p): Likewise.
28389 * recog.c (offsettable_address_addr_space_p): Likewise.
28390 * regcprop.c (maybe_mode_change): Likewise.
28391 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
28392 * regrename.c (build_def_use): Likewise.
28393 * regstat.c (dump_reg_info): Likewise.
28394 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
28395 (find_reloads, find_reloads_subreg_address): Likewise.
28396 * reload1.c (eliminate_regs_1): Likewise.
28397 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
28398 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
28399 (simplify_binary_operation_1, simplify_subreg): Likewise.
28400 * targhooks.c (default_function_arg_padding): Likewise.
28401 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
28402 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
28403 (verify_gimple_assign_ternary): Likewise.
28404 * tree-inline.c (estimate_move_cost): Likewise.
28405 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
28406 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
28407 (get_address_cost_ainc): Likewise.
28408 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28409 (vect_supportable_dr_alignment): Likewise.
28410 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28411 (vectorizable_reduction): Likewise.
28412 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
28413 (vectorizable_operation, vectorizable_load): Likewise.
28414 * tree.c (build_same_sized_truth_vector_type): Likewise.
28415 * valtrack.c (cleanup_auto_inc_dec): Likewise.
28416 * var-tracking.c (emit_note_insn_var_location): Likewise.
28417 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
28418 (ADDR_VEC_ALIGN): Likewise.
28420 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28421 Alan Hayward <alan.hayward@arm.com>
28422 David Sherwood <david.sherwood@arm.com>
28424 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
28426 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
28427 or if measurement_type is polynomial.
28428 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
28429 * combine.c (make_extraction): Likewise.
28430 * dse.c (find_shift_sequence): Likewise.
28431 * dwarf2out.c (mem_loc_descriptor): Likewise.
28432 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
28433 (extract_bit_field, extract_low_bits): Likewise.
28434 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
28435 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
28436 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
28437 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
28438 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
28439 * reload.c (find_reloads): Likewise.
28440 * reload1.c (alter_reg): Likewise.
28441 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
28442 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
28443 * tree-if-conv.c (predicate_mem_writes): Likewise.
28444 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
28445 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
28446 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
28447 * valtrack.c (dead_debug_insert_temp): Likewise.
28448 * varasm.c (mergeable_constant_section): Likewise.
28449 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
28451 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28452 Alan Hayward <alan.hayward@arm.com>
28453 David Sherwood <david.sherwood@arm.com>
28455 * expr.c (expand_assignment): Cope with polynomial mode sizes
28456 when assigning to a CONCAT.
28458 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28459 Alan Hayward <alan.hayward@arm.com>
28460 David Sherwood <david.sherwood@arm.com>
28462 * machmode.h (mode_precision): Change from unsigned short to
28464 (mode_to_precision): Return a poly_uint16 rather than an unsigned
28466 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
28467 or if measurement_type is not polynomial.
28468 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
28469 in which the mode is already known to be a scalar_int_mode.
28470 * genmodes.c (emit_mode_precision): Change the type of mode_precision
28471 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
28473 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
28474 for GET_MODE_PRECISION.
28475 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
28476 for GET_MODE_PRECISION.
28477 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
28479 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
28480 (expand_field_assignment, make_extraction): Likewise.
28481 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
28482 (get_last_value): Likewise.
28483 * convert.c (convert_to_integer_1): Likewise.
28484 * cse.c (cse_insn): Likewise.
28485 * expr.c (expand_expr_real_1): Likewise.
28486 * lra-constraints.c (simplify_operand_subreg): Likewise.
28487 * optabs-query.c (can_atomic_load_p): Likewise.
28488 * optabs.c (expand_atomic_load): Likewise.
28489 (expand_atomic_store): Likewise.
28490 * ree.c (combine_reaching_defs): Likewise.
28491 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
28492 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
28493 * tree.h (type_has_mode_precision_p): Likewise.
28494 * ubsan.c (instrument_si_overflow): Likewise.
28496 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28497 Alan Hayward <alan.hayward@arm.com>
28498 David Sherwood <david.sherwood@arm.com>
28500 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
28501 polynomial numbers of units.
28502 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
28503 (valid_vector_subparts_p): New function.
28504 (build_vector_type): Remove temporary shim and take the number
28505 of units as a poly_uint64 rather than an int.
28506 (build_opaque_vector_type): Take the number of units as a
28507 poly_uint64 rather than an int.
28508 * tree.c (build_vector_from_ctor): Handle polynomial
28509 TYPE_VECTOR_SUBPARTS.
28510 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
28511 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
28512 (build_vector_from_val): If the number of units is variable,
28513 use build_vec_duplicate_cst for constant operands and
28514 VEC_DUPLICATE_EXPR otherwise.
28515 (make_vector_type): Remove temporary is_constant ().
28516 (build_vector_type, build_opaque_vector_type): Take the number of
28517 units as a poly_uint64 rather than an int.
28518 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
28520 * cfgexpand.c (expand_debug_expr): Likewise.
28521 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
28522 (store_constructor, expand_expr_real_1): Likewise.
28523 (const_scalar_mask_from_tree): Likewise.
28524 * fold-const-call.c (fold_const_reduction): Likewise.
28525 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
28526 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
28527 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
28528 (fold_relational_const): Likewise.
28529 (native_interpret_vector): Likewise. Change the size from an
28530 int to an unsigned int.
28531 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
28532 TYPE_VECTOR_SUBPARTS.
28533 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
28534 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
28535 duplicating a non-constant operand into a variable-length vector.
28536 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
28537 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
28538 * ipa-icf.c (sem_variable::equals): Likewise.
28539 * match.pd: Likewise.
28540 * omp-simd-clone.c (simd_clone_subparts): Likewise.
28541 * print-tree.c (print_node): Likewise.
28542 * stor-layout.c (layout_type): Likewise.
28543 * targhooks.c (default_builtin_vectorization_cost): Likewise.
28544 * tree-cfg.c (verify_gimple_comparison): Likewise.
28545 (verify_gimple_assign_binary): Likewise.
28546 (verify_gimple_assign_ternary): Likewise.
28547 (verify_gimple_assign_single): Likewise.
28548 * tree-pretty-print.c (dump_generic_node): Likewise.
28549 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
28550 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
28551 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
28552 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
28553 (vect_shift_permute_load_chain): Likewise.
28554 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
28555 (expand_vector_condition, optimize_vector_constructor): Likewise.
28556 (lower_vec_perm, get_compute_type): Likewise.
28557 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28558 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
28559 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
28560 (vect_recog_mask_conversion_pattern): Likewise.
28561 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
28562 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
28563 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
28564 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
28565 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
28566 (vectorizable_shift, vectorizable_operation, vectorizable_store)
28567 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
28568 (supportable_widening_operation): Likewise.
28569 (supportable_narrowing_operation): Likewise.
28570 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
28572 * varasm.c (output_constant): Likewise.
28574 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28575 Alan Hayward <alan.hayward@arm.com>
28576 David Sherwood <david.sherwood@arm.com>
28578 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
28579 so that both the length == 3 and length != 3 cases set up their
28580 own permute vectors. Add comments explaining why we know the
28581 number of elements is constant.
28582 (vect_permute_load_chain): Likewise.
28584 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28585 Alan Hayward <alan.hayward@arm.com>
28586 David Sherwood <david.sherwood@arm.com>
28588 * machmode.h (mode_nunits): Change from unsigned char to
28590 (ONLY_FIXED_SIZE_MODES): New macro.
28591 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
28592 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
28593 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
28595 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
28596 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
28597 or if measurement_type is not polynomial.
28598 * genmodes.c (ZERO_COEFFS): New macro.
28599 (emit_mode_nunits_inline): Make mode_nunits_inline return a
28601 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
28602 Use ZERO_COEFFS when emitting initializers.
28603 * data-streamer.h (bp_pack_poly_value): New function.
28604 (bp_unpack_poly_value): Likewise.
28605 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
28606 for GET_MODE_NUNITS.
28607 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
28608 for GET_MODE_NUNITS.
28609 * tree.c (make_vector_type): Remove temporary shim and make
28610 the real function take the number of units as a poly_uint64
28611 rather than an int.
28612 (build_vector_type_for_mode): Handle polynomial nunits.
28613 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
28614 * emit-rtl.c (const_vec_series_p_1): Likewise.
28615 (gen_rtx_CONST_VECTOR): Likewise.
28616 * fold-const.c (test_vec_duplicate_folding): Likewise.
28617 * genrecog.c (validate_pattern): Likewise.
28618 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
28619 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
28620 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
28621 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
28622 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
28623 * rtlanal.c (subreg_get_info): Likewise.
28624 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
28625 (vect_grouped_load_supported): Likewise.
28626 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
28627 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
28628 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
28629 (simplify_const_unary_operation, simplify_binary_operation_1)
28630 (simplify_const_binary_operation, simplify_ternary_operation)
28631 (test_vector_ops_duplicate, test_vector_ops): Likewise.
28632 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
28633 instead of CONST_VECTOR_NUNITS.
28634 * varasm.c (output_constant_pool_2): Likewise.
28635 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
28636 explicit-encoded elements in the XVEC for variable-length vectors.
28638 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28640 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
28642 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28643 Alan Hayward <alan.hayward@arm.com>
28644 David Sherwood <david.sherwood@arm.com>
28646 * coretypes.h (fixed_size_mode): Declare.
28647 (fixed_size_mode_pod): New typedef.
28648 * builtins.h (target_builtins::x_apply_args_mode)
28649 (target_builtins::x_apply_result_mode): Change type to
28650 fixed_size_mode_pod.
28651 * builtins.c (apply_args_size, apply_result_size, result_vector)
28652 (expand_builtin_apply_args_1, expand_builtin_apply)
28653 (expand_builtin_return): Update accordingly.
28655 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28657 * cse.c (hash_rtx_cb): Hash only the encoded elements.
28658 * cselib.c (cselib_hash_rtx): Likewise.
28659 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
28660 CONST_VECTOR encoding.
28662 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28663 Jeff Law <law@redhat.com>
28666 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
28667 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
28668 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
28669 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
28672 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
28673 explicitly probe *sp in a noreturn function if there were any callee
28674 register saves or frame pointer is needed.
28676 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28679 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
28680 BLKmode for ternary, binary or unary expressions.
28683 * var-tracking.c (delete_vta_debug_insn): New inline function.
28684 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
28685 insns from get_insns () to NULL instead of each bb separately.
28686 Use delete_vta_debug_insn. No longer static.
28687 (vt_debug_insns_local, variable_tracking_main_1): Adjust
28688 delete_vta_debug_insns callers.
28689 * rtl.h (delete_vta_debug_insns): Declare.
28690 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
28691 instead of variable_tracking_main.
28693 2018-01-03 Martin Sebor <msebor@redhat.com>
28695 PR tree-optimization/83603
28696 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
28697 arguments past the endof the argument list in functions declared
28698 without a prototype.
28699 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
28700 Avoid checking when arguments are null.
28702 2018-01-03 Martin Sebor <msebor@redhat.com>
28705 * doc/extend.texi (attribute const): Fix a typo.
28706 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
28707 issuing -Wsuggest-attribute for void functions.
28709 2018-01-03 Martin Sebor <msebor@redhat.com>
28711 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
28712 offset_int::from instead of wide_int::to_shwi.
28713 (maybe_diag_overlap): Remove assertion.
28714 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
28715 * gimple-ssa-sprintf.c (format_directive): Same.
28716 (parse_directive): Same.
28717 (sprintf_dom_walker::compute_format_length): Same.
28718 (try_substitute_return_value): Same.
28720 2018-01-03 Jeff Law <law@redhat.com>
28722 PR middle-end/83654
28723 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
28724 non-constant residual for zero at runtime and avoid probing in
28725 that case. Reorganize code for trailing problem to mirror handling
28728 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28730 PR tree-optimization/83501
28731 * tree-ssa-strlen.c (get_string_cst): New.
28732 (handle_char_store): Call get_string_cst.
28734 2018-01-03 Martin Liska <mliska@suse.cz>
28736 PR tree-optimization/83593
28737 * tree-ssa-strlen.c: Include tree-cfg.h.
28738 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
28739 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
28740 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
28742 (strlen_dom_walker::before_dom_children): Call
28743 gimple_purge_dead_eh_edges. Dump tranformation with details
28745 (strlen_dom_walker::before_dom_children): Update call by adding
28746 new argument cleanup_eh.
28747 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
28749 2018-01-03 Martin Liska <mliska@suse.cz>
28752 * cif-code.def (VARIADIC_THUNK): New enum value.
28753 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
28756 2018-01-03 Jan Beulich <jbeulich@suse.com>
28758 * sse.md (mov<mode>_internal): Tighten condition for when to use
28759 vmovdqu<ssescalarsize> for TI and OI modes.
28761 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28763 Update copyright years.
28765 2018-01-03 Martin Liska <mliska@suse.cz>
28768 * ipa-visibility.c (function_and_variable_visibility): Skip
28769 functions with noipa attribure.
28771 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28773 * gcc.c (process_command): Update copyright notice dates.
28774 * gcov-dump.c (print_version): Ditto.
28775 * gcov.c (print_version): Ditto.
28776 * gcov-tool.c (print_version): Ditto.
28777 * gengtype.c (create_file): Ditto.
28778 * doc/cpp.texi: Bump @copying's copyright year.
28779 * doc/cppinternals.texi: Ditto.
28780 * doc/gcc.texi: Ditto.
28781 * doc/gccint.texi: Ditto.
28782 * doc/gcov.texi: Ditto.
28783 * doc/install.texi: Ditto.
28784 * doc/invoke.texi: Ditto.
28786 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28788 * vector-builder.h (vector_builder::m_full_nelts): Change from
28789 unsigned int to poly_uint64.
28790 (vector_builder::full_nelts): Update prototype accordingly.
28791 (vector_builder::new_vector): Likewise.
28792 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
28793 (vector_builder::operator ==): Likewise.
28794 (vector_builder::finalize): Likewise.
28795 * int-vector-builder.h (int_vector_builder::int_vector_builder):
28796 Take the number of elements as a poly_uint64 rather than an
28798 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
28799 from unsigned int to poly_uint64.
28800 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
28801 (vec_perm_indices::new_vector): Likewise.
28802 (vec_perm_indices::length): Likewise.
28803 (vec_perm_indices::nelts_per_input): Likewise.
28804 (vec_perm_indices::input_nelts): Likewise.
28805 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
28806 number of elements per input as a poly_uint64 rather than an
28807 unsigned int. Use the original encoding for variable-length
28808 vectors, rather than clamping each individual element.
28809 For the second and subsequent elements in each pattern,
28810 clamp the step and base before clamping their sum.
28811 (vec_perm_indices::series_p): Handle polynomial element counts.
28812 (vec_perm_indices::all_in_range_p): Likewise.
28813 (vec_perm_indices_to_tree): Likewise.
28814 (vec_perm_indices_to_rtx): Likewise.
28815 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
28816 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
28817 (tree_vector_builder::new_binary_operation): Handle polynomial
28818 element counts. Return false if we need to know the number
28819 of elements at compile time.
28820 * fold-const.c (fold_vec_perm): Punt if the number of elements
28821 isn't known at compile time.
28823 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28825 * vec-perm-indices.h (vec_perm_builder): Change element type
28826 from HOST_WIDE_INT to poly_int64.
28827 (vec_perm_indices::element_type): Update accordingly.
28828 (vec_perm_indices::clamp): Handle polynomial element_types.
28829 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
28830 (vec_perm_indices::all_in_range_p): Likewise.
28831 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
28833 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
28834 polynomial vec_perm_indices element types.
28835 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
28836 * fold-const.c (fold_vec_perm): Likewise.
28837 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
28838 * tree-vect-generic.c (lower_vec_perm): Likewise.
28839 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
28840 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
28841 element type to HOST_WIDE_INT.
28843 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28844 Alan Hayward <alan.hayward@arm.com>
28845 David Sherwood <david.sherwood@arm.com>
28847 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
28848 rather than an int. Use plus_constant.
28849 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
28850 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
28852 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28853 Alan Hayward <alan.hayward@arm.com>
28854 David Sherwood <david.sherwood@arm.com>
28856 * calls.c (emit_call_1, expand_call): Change struct_value_size from
28857 a HOST_WIDE_INT to a poly_int64.
28859 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28860 Alan Hayward <alan.hayward@arm.com>
28861 David Sherwood <david.sherwood@arm.com>
28863 * calls.c (load_register_parameters): Cope with polynomial
28864 mode sizes. Require a constant size for BLKmode parameters
28865 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
28866 forces a parameter to be padded at the lsb end in order to
28867 fill a complete number of words, require the parameter size
28868 to be ordered wrt UNITS_PER_WORD.
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 * reload1.c (spill_stack_slot_width): Change element type
28875 from unsigned int to poly_uint64_pod.
28876 (alter_reg): Treat mode sizes as polynomial.
28878 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28879 Alan Hayward <alan.hayward@arm.com>
28880 David Sherwood <david.sherwood@arm.com>
28882 * reload.c (complex_word_subreg_p): New function.
28883 (reload_inner_reg_of_subreg, push_reload): Use it.
28885 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28886 Alan Hayward <alan.hayward@arm.com>
28887 David Sherwood <david.sherwood@arm.com>
28889 * lra-constraints.c (process_alt_operands): Reject matched
28890 operands whose sizes aren't ordered.
28891 (match_reload): Refer to this check here.
28893 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28894 Alan Hayward <alan.hayward@arm.com>
28895 David Sherwood <david.sherwood@arm.com>
28897 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
28898 that the mode size is in the set {1, 2, 4, 8, 16}.
28900 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28901 Alan Hayward <alan.hayward@arm.com>
28902 David Sherwood <david.sherwood@arm.com>
28904 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
28905 Use plus_constant instead of gen_rtx_PLUS.
28907 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28908 Alan Hayward <alan.hayward@arm.com>
28909 David Sherwood <david.sherwood@arm.com>
28911 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
28912 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
28913 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
28914 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
28915 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
28916 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
28917 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
28918 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
28919 * config/i386/i386.c (ix86_push_rounding): ...this new function.
28920 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
28922 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
28923 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
28924 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
28925 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
28926 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
28927 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
28928 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
28929 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
28930 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
28931 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
28933 * expr.c (emit_move_resolve_push): Treat the input and result
28934 of PUSH_ROUNDING as a poly_int64.
28935 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
28936 (emit_push_insn): Likewise.
28937 * lra-eliminations.c (mark_not_eliminable): Likewise.
28938 * recog.c (push_operand): Likewise.
28939 * reload1.c (elimination_effects): Likewise.
28940 * rtlanal.c (nonzero_bits1): Likewise.
28941 * calls.c (store_one_arg): Likewise. Require the padding to be
28942 known at compile time.
28944 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28945 Alan Hayward <alan.hayward@arm.com>
28946 David Sherwood <david.sherwood@arm.com>
28948 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
28949 Use plus_constant instead of gen_rtx_PLUS.
28951 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28952 Alan Hayward <alan.hayward@arm.com>
28953 David Sherwood <david.sherwood@arm.com>
28955 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
28956 rather than an int.
28958 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28959 Alan Hayward <alan.hayward@arm.com>
28960 David Sherwood <david.sherwood@arm.com>
28962 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
28963 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
28964 via stack temporaries. Treat the mode size as polynomial too.
28966 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28967 Alan Hayward <alan.hayward@arm.com>
28968 David Sherwood <david.sherwood@arm.com>
28970 * expr.c (expand_expr_real_2): When handling conversions involving
28971 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
28972 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
28973 as a poly_uint64 too.
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 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
28981 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28982 Alan Hayward <alan.hayward@arm.com>
28983 David Sherwood <david.sherwood@arm.com>
28985 * combine.c (can_change_dest_mode): Handle polynomial
28986 REGMODE_NATURAL_SIZE.
28987 * expmed.c (store_bit_field_1): Likewise.
28988 * expr.c (store_constructor): Likewise.
28989 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
28990 and polynomial REGMODE_NATURAL_SIZE.
28991 (gen_lowpart_common): Likewise.
28992 * reginfo.c (record_subregs_of_mode): Likewise.
28993 * rtlanal.c (read_modify_subreg_p): Likewise.
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 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
29000 numbers of elements.
29002 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29003 Alan Hayward <alan.hayward@arm.com>
29004 David Sherwood <david.sherwood@arm.com>
29006 * match.pd: Cope with polynomial numbers of vector elements.
29008 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29009 Alan Hayward <alan.hayward@arm.com>
29010 David Sherwood <david.sherwood@arm.com>
29012 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
29013 in a POINTER_PLUS_EXPR.
29015 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29016 Alan Hayward <alan.hayward@arm.com>
29017 David Sherwood <david.sherwood@arm.com>
29019 * omp-simd-clone.c (simd_clone_subparts): New function.
29020 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
29021 (ipa_simd_modify_function_body): Likewise.
29023 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29024 Alan Hayward <alan.hayward@arm.com>
29025 David Sherwood <david.sherwood@arm.com>
29027 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
29028 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
29029 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
29030 (expand_vector_condition, vector_element): Likewise.
29031 (subparts_gt): New function.
29032 (get_compute_type): Use subparts_gt.
29033 (count_type_subparts): Delete.
29034 (expand_vector_operations_1): Use subparts_gt instead of
29035 count_type_subparts.
29037 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29038 Alan Hayward <alan.hayward@arm.com>
29039 David Sherwood <david.sherwood@arm.com>
29041 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
29042 (vect_compile_time_alias): ...this new function. Do the calculation
29043 on poly_ints rather than trees.
29044 (vect_prune_runtime_alias_test_list): Update call accordingly.
29046 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29047 Alan Hayward <alan.hayward@arm.com>
29048 David Sherwood <david.sherwood@arm.com>
29050 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
29052 (vect_schedule_slp_instance): Likewise.
29054 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29055 Alan Hayward <alan.hayward@arm.com>
29056 David Sherwood <david.sherwood@arm.com>
29058 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
29059 constant and extern definitions for variable-length vectors.
29060 (vect_get_constant_vectors): Note that the number of units
29061 is known to be constant.
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 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
29068 of units as polynomial. Choose between WIDE and NARROW based
29071 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29072 Alan Hayward <alan.hayward@arm.com>
29073 David Sherwood <david.sherwood@arm.com>
29075 * tree-vect-stmts.c (simd_clone_subparts): New function.
29076 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
29078 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29079 Alan Hayward <alan.hayward@arm.com>
29080 David Sherwood <david.sherwood@arm.com>
29082 * tree-vect-stmts.c (vectorizable_call): Treat the number of
29083 vectors as polynomial. Use build_index_vector for
29084 IFN_GOMP_SIMD_LANE.
29086 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29087 Alan Hayward <alan.hayward@arm.com>
29088 David Sherwood <david.sherwood@arm.com>
29090 * tree-vect-stmts.c (get_load_store_type): Treat the number of
29091 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
29092 for variable-length vectors.
29093 (vectorizable_mask_load_store): Treat the number of units as
29094 polynomial, asserting that it is constant if the condition has
29095 already been enforced.
29096 (vectorizable_store, vectorizable_load): 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-loop.c (vectorizable_live_operation): Treat the number
29103 of units as polynomial. Punt if we can't tell at compile time
29104 which vector contains the final result.
29106 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29107 Alan Hayward <alan.hayward@arm.com>
29108 David Sherwood <david.sherwood@arm.com>
29110 * tree-vect-loop.c (vectorizable_induction): Treat the number
29111 of units as polynomial. Punt on SLP inductions. Use an integer
29112 VEC_SERIES_EXPR for variable-length integer reductions. Use a
29113 cast of such a series for variable-length floating-point
29116 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29117 Alan Hayward <alan.hayward@arm.com>
29118 David Sherwood <david.sherwood@arm.com>
29120 * tree.h (build_index_vector): Declare.
29121 * tree.c (build_index_vector): New function.
29122 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
29123 of units as polynomial, forcibly converting it to a constant if
29124 vectorizable_reduction has already enforced the condition.
29125 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
29126 to create a {1,2,3,...} vector.
29127 (vectorizable_reduction): Treat the number of units as polynomial.
29128 Choose vectype_in based on the largest scalar element size rather
29129 than the smallest number of units. Enforce the restrictions
29132 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29133 Alan Hayward <alan.hayward@arm.com>
29134 David Sherwood <david.sherwood@arm.com>
29136 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
29137 number of units as polynomial.
29139 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29140 Alan Hayward <alan.hayward@arm.com>
29141 David Sherwood <david.sherwood@arm.com>
29143 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
29144 * target.def (autovectorize_vector_sizes): Return the vector sizes
29145 by pointer, using vector_sizes rather than a bitmask.
29146 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
29147 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
29148 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
29150 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
29151 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
29152 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
29153 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
29154 * omp-general.c (omp_max_vf): Likewise.
29155 * omp-low.c (omp_clause_aligned_alignment): Likewise.
29156 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
29157 * tree-vect-loop.c (vect_analyze_loop): Likewise.
29158 * tree-vect-slp.c (vect_slp_bb): Likewise.
29159 * doc/tm.texi: Regenerate.
29160 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
29162 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
29163 the vector size as a poly_uint64 rather than an unsigned int.
29164 (current_vector_size): Change from an unsigned int to a poly_uint64.
29165 (get_vectype_for_scalar_type): Update accordingly.
29166 * tree.h (build_truth_vector_type): Take the size and number of
29167 units as a poly_uint64 rather than an unsigned int.
29168 (build_vector_type): Add a temporary overload that takes
29169 the number of units as a poly_uint64 rather than an unsigned int.
29170 * tree.c (make_vector_type): Likewise.
29171 (build_truth_vector_type): Take the number of units as a poly_uint64
29172 rather than an unsigned int.
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 * target.def (get_mask_mode): Take the number of units and length
29179 as poly_uint64s rather than unsigned ints.
29180 * targhooks.h (default_get_mask_mode): Update accordingly.
29181 * targhooks.c (default_get_mask_mode): Likewise.
29182 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
29183 * doc/tm.texi: Regenerate.
29185 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29186 Alan Hayward <alan.hayward@arm.com>
29187 David Sherwood <david.sherwood@arm.com>
29189 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
29190 * omp-general.c (omp_max_vf): Likewise.
29191 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
29192 (expand_omp_simd): Handle polynomial safelen.
29193 * omp-low.c (omplow_simd_context): Add a default constructor.
29194 (omplow_simd_context::max_vf): Change from int to poly_uint64.
29195 (lower_rec_simd_input_clauses): Update accordingly.
29196 (lower_rec_input_clauses): Likewise.
29198 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29199 Alan Hayward <alan.hayward@arm.com>
29200 David Sherwood <david.sherwood@arm.com>
29202 * tree-vectorizer.h (vect_nunits_for_cost): New function.
29203 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
29204 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
29205 (vect_analyze_slp_cost): Likewise.
29206 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
29207 (vect_model_load_cost): Likewise.
29209 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29210 Alan Hayward <alan.hayward@arm.com>
29211 David Sherwood <david.sherwood@arm.com>
29213 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
29214 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
29215 from an unsigned int * to a poly_uint64_pod *.
29216 (calculate_unrolling_factor): New function.
29217 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
29219 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29220 Alan Hayward <alan.hayward@arm.com>
29221 David Sherwood <david.sherwood@arm.com>
29223 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
29224 from an unsigned int to a poly_uint64.
29225 (_loop_vec_info::slp_unrolling_factor): Likewise.
29226 (_loop_vec_info::vectorization_factor): Change from an int
29228 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
29229 (vect_get_num_vectors): New function.
29230 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
29231 (vect_get_num_copies): Use vect_get_num_vectors.
29232 (vect_analyze_data_ref_dependences): Change max_vf from an int *
29233 to an unsigned int *.
29234 (vect_analyze_data_refs): Change min_vf from an int * to a
29236 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
29237 than an unsigned HOST_WIDE_INT.
29238 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
29239 (vect_analyze_data_ref_dependence): Change max_vf from an int *
29240 to an unsigned int *.
29241 (vect_analyze_data_ref_dependences): Likewise.
29242 (vect_compute_data_ref_alignment): Handle polynomial vf.
29243 (vect_enhance_data_refs_alignment): Likewise.
29244 (vect_prune_runtime_alias_test_list): Likewise.
29245 (vect_shift_permute_load_chain): Likewise.
29246 (vect_supportable_dr_alignment): Likewise.
29247 (dependence_distance_ge_vf): Take the vectorization factor as a
29248 poly_uint64 rather than an unsigned HOST_WIDE_INT.
29249 (vect_analyze_data_refs): Change min_vf from an int * to a
29251 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
29252 vfm1 as a poly_uint64 rather than an int. Make the same change
29253 for the returned bound_scalar.
29254 (vect_gen_vector_loop_niters): Handle polynomial vf.
29255 (vect_do_peeling): Likewise. Update call to
29256 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
29257 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
29259 * tree-vect-loop.c (vect_determine_vectorization_factor)
29260 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
29261 (vect_get_known_peeling_cost): Likewise.
29262 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
29263 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
29264 (vect_transform_loop): Likewise. Use the lowest possible VF when
29265 updating the upper bounds of the loop.
29266 (vect_min_worthwhile_factor): Make static. Return an unsigned int
29267 rather than an int.
29268 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
29269 polynomial unroll factors.
29270 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
29271 (vect_make_slp_decision): Likewise.
29272 (vect_supported_load_permutation_p): Likewise, and polynomial
29274 (vect_analyze_slp_cost): Handle polynomial vf.
29275 (vect_slp_analyze_node_operations): Likewise.
29276 (vect_slp_analyze_bb_1): Likewise.
29277 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
29278 than an unsigned HOST_WIDE_INT.
29279 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
29280 (vectorizable_load): Handle polynomial vf.
29281 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
29283 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
29285 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29286 Alan Hayward <alan.hayward@arm.com>
29287 David Sherwood <david.sherwood@arm.com>
29289 * match.pd: Handle bit operations involving three constants
29290 and try to fold one pair.
29292 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29294 * tree-vect-loop-manip.c: Include gimple-fold.h.
29295 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
29296 niters_maybe_zero parameters. Handle other cases besides a step of 1.
29297 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
29298 Add a path that uses a step of VF instead of 1, but disable it
29300 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
29301 and niters_no_overflow parameters. Update calls to
29302 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
29303 Create a new SSA name if the latter choses to use a ste other
29304 than zero, and return it via niters_vector_mult_vf_var.
29305 * tree-vect-loop.c (vect_transform_loop): Update calls to
29306 vect_do_peeling, vect_gen_vector_loop_niters and
29307 slpeel_make_loop_iterate_ntimes.
29308 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
29309 (vect_gen_vector_loop_niters): Update declarations after above changes.
29311 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
29313 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
29314 128-bit round to integer instructions.
29315 (ceil<mode>2): Likewise.
29316 (btrunc<mode>2): Likewise.
29317 (round<mode>2): Likewise.
29319 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29321 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
29322 unaligned VSX load/store on P8/P9.
29323 (expand_block_clear): Allow the use of unaligned VSX
29324 load/store on P8/P9.
29326 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29328 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
29330 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
29331 swap associated with both a load and a store.
29333 2018-01-02 Andrew Waterman <andrew@sifive.com>
29335 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
29336 * config/riscv/riscv.md (clear_cache): Use it.
29338 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
29340 * web.c: Remove out-of-date comment.
29342 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29344 * expr.c (fixup_args_size_notes): Check that any existing
29345 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
29346 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
29347 (emit_single_push_insn): ...here.
29349 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29351 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
29352 (const_vector_encoded_nelts): New function.
29353 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
29354 (const_vector_int_elt, const_vector_elt): Declare.
29355 * emit-rtl.c (const_vector_int_elt_1): New function.
29356 (const_vector_elt): Likewise.
29357 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
29358 of CONST_VECTOR_ELT.
29360 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29362 * expr.c: Include rtx-vector-builder.h.
29363 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
29364 directly on the tree encoding.
29365 (const_vector_from_tree): Likewise.
29366 * optabs.c: Include rtx-vector-builder.h.
29367 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
29368 sequence of "u" values.
29369 * vec-perm-indices.c: Include rtx-vector-builder.h.
29370 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
29371 directly on the vec_perm_indices encoding.
29373 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29375 * doc/rtl.texi (const_vector): Describe new encoding scheme.
29376 * Makefile.in (OBJS): Add rtx-vector-builder.o.
29377 * rtx-vector-builder.h: New file.
29378 * rtx-vector-builder.c: Likewise.
29379 * rtl.h (rtx_def::u2): Add a const_vector field.
29380 (CONST_VECTOR_NPATTERNS): New macro.
29381 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
29382 (CONST_VECTOR_DUPLICATE_P): Likewise.
29383 (CONST_VECTOR_STEPPED_P): Likewise.
29384 (CONST_VECTOR_ENCODED_ELT): Likewise.
29385 (const_vec_duplicate_p): Check for a duplicated vector encoding.
29386 (unwrap_const_vec_duplicate): Likewise.
29387 (const_vec_series_p): Check for a non-duplicated vector encoding.
29388 Say that the function only returns true for integer vectors.
29389 * emit-rtl.c: Include rtx-vector-builder.h.
29390 (gen_const_vec_duplicate_1): Delete.
29391 (gen_const_vector): Call gen_const_vec_duplicate instead of
29392 gen_const_vec_duplicate_1.
29393 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
29394 (gen_const_vec_duplicate): Use rtx_vector_builder.
29395 (gen_const_vec_series): Likewise.
29396 (gen_rtx_CONST_VECTOR): Likewise.
29397 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
29398 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
29399 Build a new vector rather than modifying a CONST_VECTOR in-place.
29400 (handle_special_swappables): Update call accordingly.
29401 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
29402 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
29403 Build a new vector rather than modifying a CONST_VECTOR in-place.
29404 (handle_special_swappables): Update call accordingly.
29406 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29408 * simplify-rtx.c (simplify_const_binary_operation): Use
29409 CONST_VECTOR_ELT instead of XVECEXP.
29411 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29413 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
29414 the selector elements to be different from the data elements
29415 if the selector is a VECTOR_CST.
29416 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
29417 ssizetype for the selector.
29419 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29421 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
29422 before testing each element individually.
29423 * tree-vect-generic.c (lower_vec_perm): Likewise.
29425 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29427 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
29428 * selftest-run-tests.c (selftest::run_tests): Call it.
29429 * vector-builder.h (vector_builder::operator ==): New function.
29430 (vector_builder::operator !=): Likewise.
29431 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
29432 (vec_perm_indices::all_from_input_p): New function.
29433 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
29434 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
29435 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
29436 instead of reading the VECTOR_CST directly. Detect whether both
29437 vector inputs are the same before constructing the vec_perm_indices,
29438 and update the number of inputs argument accordingly. Use the
29439 utility functions added above. Only construct sel2 if we need to.
29441 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29443 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
29444 the broadcast of the low byte.
29445 (expand_mult_highpart): Use an explicit encoding for the permutes.
29446 * optabs-query.c (can_mult_highpart_p): Likewise.
29447 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
29448 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29449 (vectorizable_bswap): Likewise.
29450 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
29451 explicit encoding for the power-of-2 permutes.
29452 (vect_permute_store_chain): Likewise.
29453 (vect_grouped_load_supported): Likewise.
29454 (vect_permute_load_chain): Likewise.
29456 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29458 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
29459 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
29460 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
29461 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
29462 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
29463 (vect_gen_perm_mask_any): Likewise.
29465 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29467 * int-vector-builder.h: New file.
29468 * vec-perm-indices.h: Include int-vector-builder.h.
29469 (vec_perm_indices): Redefine as an int_vector_builder.
29470 (auto_vec_perm_indices): Delete.
29471 (vec_perm_builder): Redefine as a stand-alone class.
29472 (vec_perm_indices::vec_perm_indices): New function.
29473 (vec_perm_indices::clamp): Likewise.
29474 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
29475 (vec_perm_indices::new_vector): New function.
29476 (vec_perm_indices::new_expanded_vector): Update for new
29477 vec_perm_indices class.
29478 (vec_perm_indices::rotate_inputs): New function.
29479 (vec_perm_indices::all_in_range_p): Operate directly on the
29480 encoded form, without computing elided elements.
29481 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
29482 encoding. Update for new vec_perm_indices class.
29483 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
29484 the given vec_perm_builder.
29485 (expand_vec_perm_var): Update vec_perm_builder constructor.
29486 (expand_mult_highpart): Use vec_perm_builder instead of
29487 auto_vec_perm_indices.
29488 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
29489 vec_perm_indices instead of auto_vec_perm_indices. Use a single
29490 or double series encoding as appropriate.
29491 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
29492 vec_perm_indices instead of auto_vec_perm_indices.
29493 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
29494 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
29495 (vect_permute_store_chain): Likewise.
29496 (vect_grouped_load_supported): Likewise.
29497 (vect_permute_load_chain): Likewise.
29498 (vect_shift_permute_load_chain): Likewise.
29499 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
29500 (vect_transform_slp_perm_load): Likewise.
29501 (vect_schedule_slp_instance): Likewise.
29502 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29503 (vectorizable_mask_load_store): Likewise.
29504 (vectorizable_bswap): Likewise.
29505 (vectorizable_store): Likewise.
29506 (vectorizable_load): Likewise.
29507 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
29508 vec_perm_indices instead of auto_vec_perm_indices. Use
29509 tree_to_vec_perm_builder to read the vector from a tree.
29510 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
29511 vec_perm_builder instead of a vec_perm_indices.
29512 (have_whole_vector_shift): Use vec_perm_builder and
29513 vec_perm_indices instead of auto_vec_perm_indices. Leave the
29514 truncation to calc_vec_perm_mask_for_shift.
29515 (vect_create_epilog_for_reduction): Likewise.
29516 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
29517 from auto_vec_perm_indices to vec_perm_indices.
29518 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
29519 instead of changing individual elements.
29520 (aarch64_vectorize_vec_perm_const): Use new_vector to install
29521 the vector in d.perm.
29522 * config/arm/arm.c (expand_vec_perm_d::perm): Change
29523 from auto_vec_perm_indices to vec_perm_indices.
29524 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
29525 instead of changing individual elements.
29526 (arm_vectorize_vec_perm_const): Use new_vector to install
29527 the vector in d.perm.
29528 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
29529 Update vec_perm_builder constructor.
29530 (rs6000_expand_interleave): Likewise.
29531 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
29532 (rs6000_expand_interleave): Likewise.
29534 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29536 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
29537 to qimode could truncate the indices.
29538 * optabs.c (expand_vec_perm_var): Likewise.
29540 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29542 * Makefile.in (OBJS): Add vec-perm-indices.o.
29543 * vec-perm-indices.h: New file.
29544 * vec-perm-indices.c: Likewise.
29545 * target.h (vec_perm_indices): Replace with a forward class
29547 (auto_vec_perm_indices): Move to vec-perm-indices.h.
29548 * optabs.h: Include vec-perm-indices.h.
29549 (expand_vec_perm): Delete.
29550 (selector_fits_mode_p, expand_vec_perm_var): Declare.
29551 (expand_vec_perm_const): Declare.
29552 * target.def (vec_perm_const_ok): Replace with...
29553 (vec_perm_const): ...this new hook.
29554 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
29555 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
29556 * doc/tm.texi: Regenerate.
29557 * optabs.def (vec_perm_const): Delete.
29558 * doc/md.texi (vec_perm_const): Likewise.
29559 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
29560 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
29561 expand_vec_perm for constant permutation vectors. Assert that
29562 the mode of variable permutation vectors is the integer equivalent
29563 of the mode that is being permuted.
29564 * optabs-query.h (selector_fits_mode_p): Declare.
29565 * optabs-query.c: Include vec-perm-indices.h.
29566 (selector_fits_mode_p): New function.
29567 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
29568 is defined, instead of checking whether the vec_perm_const_optab
29569 exists. Use targetm.vectorize.vec_perm_const instead of
29570 targetm.vectorize.vec_perm_const_ok. Check whether the indices
29571 fit in the vector mode before using a variable permute.
29572 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
29573 vec_perm_indices instead of an rtx.
29574 (expand_vec_perm): Replace with...
29575 (expand_vec_perm_const): ...this new function. Take the selector
29576 as a vec_perm_indices rather than an rtx. Also take the mode of
29577 the selector. Update call to shift_amt_for_vec_perm_mask.
29578 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
29579 Use vec_perm_indices::new_expanded_vector to expand the original
29580 selector into bytes. Check whether the indices fit in the vector
29581 mode before using a variable permute.
29582 (expand_vec_perm_var): Make global.
29583 (expand_mult_highpart): Use expand_vec_perm_const.
29584 * fold-const.c: Includes vec-perm-indices.h.
29585 * tree-ssa-forwprop.c: Likewise.
29586 * tree-vect-data-refs.c: Likewise.
29587 * tree-vect-generic.c: Likewise.
29588 * tree-vect-loop.c: Likewise.
29589 * tree-vect-slp.c: Likewise.
29590 * tree-vect-stmts.c: Likewise.
29591 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
29593 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
29594 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
29595 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
29596 (aarch64_vectorize_vec_perm_const): ...this new function.
29597 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
29598 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29599 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
29600 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
29601 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
29602 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29603 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
29605 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
29606 check for NEON modes.
29607 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
29608 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
29609 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
29610 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
29612 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
29613 the old VEC_PERM_CONST conditions.
29614 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
29615 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
29616 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
29617 (ia64_vectorize_vec_perm_const_ok): Merge into...
29618 (ia64_vectorize_vec_perm_const): ...this new function.
29619 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
29620 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
29621 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
29622 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
29623 * config/mips/mips.c (mips_expand_vec_perm_const)
29624 (mips_vectorize_vec_perm_const_ok): Merge into...
29625 (mips_vectorize_vec_perm_const): ...this new function.
29626 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
29627 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
29628 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
29629 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
29630 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
29631 (rs6000_expand_vec_perm_const): Delete.
29632 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
29634 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29635 (altivec_expand_vec_perm_const_le): Take each operand individually.
29636 Operate on constant selectors rather than rtxes.
29637 (altivec_expand_vec_perm_const): Likewise. Update call to
29638 altivec_expand_vec_perm_const_le.
29639 (rs6000_expand_vec_perm_const): Delete.
29640 (rs6000_vectorize_vec_perm_const_ok): Delete.
29641 (rs6000_vectorize_vec_perm_const): New function.
29642 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
29643 an element count and rtx array.
29644 (rs6000_expand_extract_even): Update call accordingly.
29645 (rs6000_expand_interleave): Likewise.
29646 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
29647 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
29648 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
29649 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
29650 (rs6000_expand_vec_perm_const): Delete.
29651 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
29652 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29653 (altivec_expand_vec_perm_const_le): Take each operand individually.
29654 Operate on constant selectors rather than rtxes.
29655 (altivec_expand_vec_perm_const): Likewise. Update call to
29656 altivec_expand_vec_perm_const_le.
29657 (rs6000_expand_vec_perm_const): Delete.
29658 (rs6000_vectorize_vec_perm_const_ok): Delete.
29659 (rs6000_vectorize_vec_perm_const): New function. Remove stray
29660 reference to the SPE evmerge intructions.
29661 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
29662 an element count and rtx array.
29663 (rs6000_expand_extract_even): Update call accordingly.
29664 (rs6000_expand_interleave): Likewise.
29665 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
29666 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
29668 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29670 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29672 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
29673 vector mode and that that mode matches the mode of the data
29675 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
29676 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
29677 directly using expand_vec_perm_1 when forcing selectors into
29679 (expand_vec_perm_var): New function, split out from expand_vec_perm.
29681 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29683 * optabs-query.h (can_vec_perm_p): Delete.
29684 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
29685 * optabs-query.c (can_vec_perm_p): Split into...
29686 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
29687 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
29688 particular selector is valid.
29689 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
29690 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
29691 (vect_grouped_load_supported): Likewise.
29692 (vect_shift_permute_load_chain): Likewise.
29693 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
29694 (vect_transform_slp_perm_load): Likewise.
29695 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29696 (vectorizable_bswap): Likewise.
29697 (vect_gen_perm_mask_checked): Likewise.
29698 * fold-const.c (fold_ternary_loc): Likewise. Don't take
29699 implementations of variable permutation vectors into account
29700 when deciding which selector to use.
29701 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
29702 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
29703 with a false third argument.
29704 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
29705 to test whether the constant selector is valid and can_vec_perm_var_p
29706 to test whether a variable selector is valid.
29708 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29710 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
29711 * optabs-query.c (can_vec_perm_p): Likewise.
29712 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
29713 instead of vec_perm_indices.
29714 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
29715 (vect_gen_perm_mask_checked): Likewise,
29716 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
29717 (vect_gen_perm_mask_checked): Likewise,
29719 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29721 * optabs-query.h (qimode_for_vec_perm): Declare.
29722 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
29723 (qimode_for_vec_perm): ...this new function.
29724 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
29726 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29728 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
29729 does not have a conditional at the top.
29731 2018-01-02 Richard Biener <rguenther@suse.de>
29733 * ipa-inline.c (big_speedup_p): Fix expression.
29735 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
29738 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
29741 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
29745 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
29746 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
29747 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
29748 cond_taken_branch_cost 3->4.
29750 2018-01-01 Jakub Jelinek <jakub@redhat.com>
29752 PR tree-optimization/83581
29753 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
29754 TODO_cleanup_cfg if any changes have been made.
29756 PR middle-end/83608
29757 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
29758 convert_modes if target mode has the right side, but different mode
29761 PR middle-end/83609
29762 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
29763 last argument when extracting from CONCAT. If either from_real or
29764 from_imag is NULL, use expansion through memory. If result is not
29765 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
29766 the parts directly to inner mode, if even that fails, use expansion
29769 PR middle-end/83623
29770 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
29771 check for bswap in mode rather than HImode and use that in expand_unop
29774 Copyright (C) 2018 Free Software Foundation, Inc.
29776 Copying and distribution of this file, with or without modification,
29777 are permitted in any medium without royalty provided the copyright
29778 notice and this notice are preserved.