1 2018-09-14 Richard Biener <rguenther@suse.de>
4 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
5 bits for the merged partition.
7 2018-09-13 Martin Sebor <msebor@redhat.com>
8 Bernd Edlinger <bernd.edlinger@hotmail.de>
10 * builtins.h (c_srlen): Add argument.
11 * builtins.c (warn_string_no_nul): New function.
12 (c_strlen): Add argument and use it. Update recursive calls.
13 Pass DECL argument to string_constant to get info on non
14 terminated strings. Update *NONSTR as needed.
15 (fold_builtin_strlen): Add argument to calls to c_strlen.
16 Warn for unterminated arrays.
17 (warn_string_no_null): Add prototype.
18 * expr.c (string_constant): Update arguments. Update recursive
19 calls appropriately. Detect missing NUL terminator and outermost
20 declaration its missing in.
21 Improve checks for arrays with nonzero lower bound or elements
22 that are not a single byte. Simplify offset computation.
23 Simplify checks for non-NUL terminated strings.
24 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
25 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
27 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
29 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
31 * fold-const.c (c_getstr): Fix function comment. Remove unused third
32 argument. Fix range checks.
33 * fold-const.h (c_getstr): Adjust protoype.
34 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
35 string is constant but contains no NUL byte.
37 * expr.c (string_constant): Adjust function comment.
38 Remove bogus check for zero termination.
40 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
42 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
43 (get_constant_size): Don't make STRING_CSTs larger than they are.
44 (check_string_literal): New check function for STRING_CSTs.
45 (output_constant): Use it.
47 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
50 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
52 2018-09-13 Richard Biener <rguenther@suse.de>
54 PR tree-optimization/87263
55 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
56 (struct unwind_state): Add max_rpo field.
57 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
58 Compute max_rpo, the max RPO number a block can be backwards reached
59 from. Re-write non-iterating mode to a RPO ordered worklist approach,
60 separating it from the iterating mode.
62 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
64 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
65 (rfs_decision): New scheduling decision.
67 2018-09-13 Richard Biener <rguenther@suse.de>
70 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
71 (vn_nary_op_insert_pieces_predicated): Do not write useless
72 valid_dominated_by_p entry outside of the allocated storage.
74 2018-09-13 Omar Sandoval <osandov@osandov.com>
75 Tom de Vries <tdevries@suse.de>
78 * dwarf2out.c (is_c): New function.
79 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
81 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
84 * config/aarch64/aarch64.md (*aarch64_bfxil):
86 * config/aarch64/constraints.md (Ulc): Define.
87 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
89 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
92 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
94 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
95 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
96 aarch64_layout_frame call.
97 (aarch64_expand_epilogue): Likewise.
98 (aarch64_initial_elimination_offset): Likewise.
99 (aarch64_get_separate_components): Likewise.
100 (aarch64_use_return_insn_p): Likewise.
101 (aarch64_layout_frame): Remove unneeded check.
103 2018-09-13 Jakub Jelinek <jakub@redhat.com>
105 * configure.ac: Only append
106 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
107 gcc_config_arguments if it was never reconfigured or last reconfigure
108 was with different arguments.
109 * configure: Regenerated.
111 2018-09-13 Jakub Jelinek <jakub@redhat.com>
112 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
115 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
116 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
118 2018-09-13 Jakub Jelinek <jakub@redhat.com>
120 PR tree-optimization/87287
121 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
122 X % C == 0 to X % (unsigned) C == 0 optimization to ...
123 * match.pd (X % C == 0): ... here. New optimization.
125 2018-09-12 Jakub Jelinek <jakub@redhat.com>
128 * expr.h (maybe_optimize_mod_cmp): Declare.
129 * expr.c (mod_inv): New function.
130 (maybe_optimize_mod_cmp): New function.
131 (do_store_flag): Use it.
132 * cfgexpand.c (expand_gimple_cond): Likewise.
134 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
135 Julian Brown <julian@codesourcery.com>
138 * gimplify.c (gimplify_scan_omp_clauses): Set
139 target_firstprivatize_array_bases in OpenACC parallel and kernels
140 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
141 OpenACC data regions.
143 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
145 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
146 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
148 2018-09-12 Richard Biener <rguenther@suse.de>
150 PR tree-optimization/87280
151 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
152 edge but unreachable target.
153 (do_rpo_vn): For conservatively handling a PHI only mark
154 the backedge executable but not the block reachable.
156 2018-09-12 Richard Biener <rguenther@suse.de>
158 PR tree-optimization/87266
159 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
162 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
164 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
166 ("trunc<BFP:mode><DFP_ALL:mode>2")
167 ("trunc<DFP_ALL:mode><BFP:mode>2")
168 ("extend<BFP:mode><DFP_ALL:mode>2")
169 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
170 according to the target operand type.
172 2018-09-12 Jakub Jelinek <jakub@redhat.com>
173 Andreas Krebbel <krebbel@linux.ibm.com>
175 PR tree-optimization/86844
176 * gimple-ssa-store-merging.c
177 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
178 there are any overlapping stores in between them, make sure they are
179 also coalesced or we give up completely.
181 2018-09-12 Jakub Jelinek <jakub@redhat.com>
184 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
185 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
187 2018-09-12 Tom de Vries <tdevries@suse.de>
189 * common.opt (gdescribe-dies): Add option.
190 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
191 attribute for artifical and nameless decls.
192 (dwarf2out_register_external_die): Add description attribute to
193 external reference die.
194 (add_desc_attribute): New functions.
195 (gen_subprogram_die): Add description attribute to
196 DW_TAG_call_site_parameter.
197 * tree-pretty-print.c (print_generic_expr_to_str): New function.
198 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
199 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
201 (@item -gdescribe-dies): Add.
203 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
205 * tree-vrp.c (vrp_shift_undefined_p): Remove.
206 (extract_range_from_binary_expr_1: Call
207 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
208 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
211 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
213 * gimple-ssa-warn-alloca.c
214 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
215 field for ALLOCA_BOUND_*_LARGE.
217 2018-09-11 Nathan Sidwell <nathan@acm.org>
219 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
221 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
223 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
224 for clobbers. Remove obsolete comment.
226 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
228 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
229 mpxchk, mpxld and mpxst types.
230 (define_attr length_immediate): Remove all processing of mpx types.
231 (define_attr prefix_0f): Ditto.
232 (define_attr memory): Ditto.
234 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
236 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
237 (log<mode>2): Change operand 1 predicate to general_operand.
238 Extend operand 1 to XFmode and generate logxf3 insn.
239 (log10<mode>2): Change operand 1 predicate to general_operand.
240 Extend operand 1 to XFmode and generate log10xf3 insn.
241 (log2<mode>2): Change operand 1 predicate to general_operand.
242 Extend operand 1 to XFmode and generate log2xf3 insn.
243 (fyl2xp1_extend<mode>xf3_i387): Remove.
244 (log1p<mode>2): Change operand 1 predicate to general_operand.
245 Extend operand 1 to XFmode and generate log1pxf3 insn.
246 (fxtract_extend<mode>xf3_i387): Remove.
247 (logb<mode>2): Change operand 1 predicate to general_operand.
248 Extend operand 1 to XFmode and generate logbxf3 insn.
249 (ilogb<mode>2): Change operand 1 predicate to general_operand.
250 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
251 (significand<mode>2): Change operand 1 predicate to general_operand.
252 Extend operand 1 to XFmode and generate significandxf3 insn.
254 2018-09-11 Nathan Sidwell <nathan@acm.org>
256 * gcc.c (perror_with_name, pfatal_with_name): Delete.
257 (load_specs): Use fatal_error.
258 (DELETE_IF_ORDINARY, process_command): Use error.
259 (execute, run_attempt): Use fatal_error.
261 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
263 * diagnostic-core.h (sorry_at): New prototype.
264 * diagnostic.c (sorry_at): New function.
266 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
268 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
269 by zero as VR_UNDEFINED.
271 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
273 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
274 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
275 (<sincos>mode2): New expander.
276 (sincos_extend<mode>xf3_i387): Remove insn pattern.
277 (sincos -> sin, cos splitters): Remove splitter patterns.
278 (sincos<mode>3): Change operand 2 predicate to general_operand.
279 Extend operand 2 to XFmode and generate sincosxf3 insn.
280 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
281 Change operand 3 predicate to const1_operand.
282 (fptan_extend<mode>xf4_i387): Remove insn pattern.
283 (tanxf2): Update operands in the call to fptanxf4_i387.
284 (tan<mode>2): Change operand 1 predicate to general_operand.
285 Extend operand 1 to XFmode and generate tanxf3 insn.
286 (atan2xf3): Rename from *fpatanxf3_i387.
287 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
288 (atan2xf3): Remove expander.
289 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
290 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
291 (atan<mode>2): Change operand 1 predicate to general_operand.
292 Extend operand 1 to XFmode and generate atanxf3 insn.
294 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
296 * config/i386/i386.md (x87/SSE constant load splitter): Use
297 memory_operand instead of nonimmediate_operand for input operand
300 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
302 * config/i386/i386.md (float partial SSE register stall splitter): Move
303 splitter near its instruction pattern.
304 (float_extend partial SSE register stall splitter): Ditto.
305 (float_truncate partial SSE register stall splitter): Ditto.
307 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
310 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
311 to speculation_safe_value_not_needed.
314 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
315 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
316 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
317 leaf_function_p, instead use has_hard_reg_initial_val.
319 2018-09-09 Nathan Sidwell <nathan@acm.org>
321 * gcc.h (pfatal_with_name): Don't declare here.
322 * gcc.c (pfatal_with_name): Make static.
324 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
326 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
329 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
331 PR rtl-optimization/85458
332 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
333 priority hook to reduce the priority of EXPR.
335 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
337 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
338 DImode for x87 on 32bit targets. Conditionally disable x87 modes
339 with X87_ENABLE_FLOAT. Remove preparation code.
340 (*float<SWI48:mode><MODEF:mode>2): Rename from
341 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
342 math using "enabled" attribute.
343 (*floatdi<MODEF:mode>2_i387): Rename from
344 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
345 enable for 32bit targets only.
346 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
348 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
349 as operand 1 predicate. Rewrite as define_insn_and_split.
350 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
352 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
354 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
355 to fallthru to FLOAT case.
357 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
360 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
361 around folding of vec_sl to handle out of range shift values.
363 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
365 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
366 Update callers to gen_fix_trunc<mode>_i387_fisttp
367 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
368 nonimmediate_operand.
369 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
370 and corresponding splitters.
371 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
372 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
373 (fix_truncdi_i387_with_temp): Remove insn pattern
374 and corresponding splitters.
375 (fix_trunc<mode>_i387): Change operand 0 predicate to
376 nonimmediate_operand.
377 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
378 and corresponding splitters.
379 (*fistdi2_1): Remove.
381 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
382 (lrintxfdi2): Remove expander. Reimplement as define_insn.
383 (*fist<mode>2_1): Remove.
384 (fist<mode>2): Ditto.
385 (fist<mode>2_with_temp): Remove insn pattern and corresponding
387 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
388 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
389 (fistdi2_<rounding>): Change operand 0 predicate to
390 nonimmediate_operand.
391 (fistdi2_<rounding>_with_temp): Remove insn pattern
392 and corresponding splitters.
393 (fist<mode>2_<rounding>): Change operand 0 predicate to
394 nonimmediate_operand.
395 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
396 and corresponding splitters.
398 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
400 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
402 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
405 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
407 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
408 early gimple folding of vec_splat().
409 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
410 * gimple-fold.h: Add an extern define for tree_vec_extract().
412 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
414 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
415 wrappers around TREE_TYPE comparisons.
417 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
420 * config/s390/predicates.md: Add nonsym_memory_operand.
421 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
422 contains a SYMBOL_REF, load it into an intermediate pseudo.
423 (s390_emit_compare_and_swap): Legitimize operand.
424 * config/s390/s390.md: Use the new nonsym_memory_operand
425 with UNSPECV_CAS patterns.
427 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
430 * config/s390/s390-passes.def: New file.
431 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
433 (make_pass_s390_early_mach): Add declaration.
434 * config/s390/s390.c (make_pass_s390_early_mach):
435 (s390_option_override): Remove dynamic registration.
436 * config/s390/t-s390: Add s390-passes.def.
438 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
440 * config/s390/s390.c (s390_decompose_constant_pool_ref):
441 Remove UNSPEC_LTREL_BASE check.
442 (annotate_constant_pool_refs): Likewise.
443 (find_constant_pool_ref): Likewise.
444 (find_ltrel_base): Removed.
445 (replace_ltrel_base): Removed.
446 (s390_mainpool_finish): Remove replace_ltrel_base call.
447 (s390_chunkify_start): Remove pending LTREL_BASE logic.
448 (s390_chunkify_finish): Remove replace_ltrel_base call.
449 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
451 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
454 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
455 to speculation_safe_value_not_needed.
457 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
458 Bernd Schmidt <bernds_cb1@t-online.de>
460 * config/nvptx/nvptx-opts.h: New file.
461 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
462 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
464 (TARGET_SM35): New macro.
465 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
467 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
470 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
472 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
474 * config/i386/i386.md (truncdfsf2): Remove expander.
475 (truncdfsf2_with_temp): Ditto.
476 (truncxf<mode>2): Ditto.
477 (*truncdfsf_fast_mixed): Remove insn pattern.
478 (*truncdfsf_fast_i387): Ditto.
479 (*truncdfsf_mixed): Ditto.
480 (*truncdfsf_i387): Ditto.
481 (*truncdfsf2_i387_1): Ditto.
482 (*truncxfsf2_mixed): Ditto.
483 (*truncxfdf2_mixed): Ditto.
484 (*truncxf<mode>2_i387_noop): Ditto. Update callers
485 to call gen_truncxf<mode>2 instead.
486 (*truncxf<mode>2_i387): Remove.
487 (reg->reg splitters): Remove splitter pattern.
488 (reg->mem splitters): Ditto.
490 (truncdfsf2): New insn pattern.
491 (truncxf<mode>2): Ditto.
493 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
495 * tree-ssa-math-opts.c (is_mult_by): New function.
496 (is_square_of): Use the above.
497 (optimize_recip_sqrt): New function.
498 (pass_cse_reciprocals::execute): Use the above.
500 2018-09-05 Richard Biener <rguenther@suse.de>
503 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
504 to zero-init the emplaced vec.
506 2018-09-05 Martin Liska <mliska@suse.cz>
508 PR tree-optimization/87205
509 * tree-switch-conversion.c (pass_lower_switch::execute):
510 Group cases for switch statements.
512 2018-09-05 Richard Biener <rguenther@suse.de>
514 PR tree-optimization/87217
515 * tree-ssa-sccvn.c (vuse_valueize): New.
516 (vn_reference_lookup_pieces): Use it.
517 (vn_reference_lookup): Likewise.
519 2018-09-05 Nathan Sidwell <nathan@acm.org>
522 * stor-layout.c (place_field): Scan forwards to check last
523 bitfield when ms_bitfield_placement is in effect.
525 2018-09-05 Richard Biener <rguenther@suse.de>
528 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
531 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
532 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
534 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
535 * config.gcc (extra_objs): Build it.
536 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
538 * config/aarch64/aarch64-passes.def
539 (pass_tag_collision_avoidance): New pass.
540 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
541 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
542 (aarch64_classify_address): Remove static qualifier.
543 (aarch64_address_info, aarch64_address_type): Move to...
544 * config/aarch64/aarch64-protos.h: ... here.
545 (make_pass_tag_collision_avoidance): New function.
546 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
549 2018-09-05 Martin Liska <mliska@suse.cz>
551 * doc/gcov.texi: Update documentation of humar
553 * gcov.c (format_count): Print one decimal place, it provides
554 more fine number of situations like '1G' vs. '1.4G'.
556 2018-09-05 Martin Liska <mliska@suse.cz>
559 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
560 * optc-gen.awk: Allow 'Var' for Deprecated options in order
561 to generate a MASK value.
563 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
566 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
567 if frame pointer isn't used.
568 (compute_frame_pointer_to_fb_displacement): Likewise.
570 2018-09-04 Jakub Jelinek <jakub@redhat.com>
573 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
574 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
575 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
576 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
577 and OPTION_MASK_ISA_XSAVEC_UNSET.
579 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
581 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
582 XOR operations in NAND case.
584 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
586 * wide-int-range.cc (wide_int_range_convert): New.
587 * wide-int-range.h (wide_int_range_convert): New.
588 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
589 code into wide_int_range_convert.
590 (extract_range_into_wide_ints): Do not munge anti range constants
591 into the entire domain. Just return the range back.
593 2018-09-04 Martin Liska <mliska@suse.cz>
595 * genmatch.c (output_line_directive): Add new argument
597 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
599 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
601 * doc/invoke.texi (Option Summary): Add whitespace.
603 * doc/invoke.texi (Option Summary): Add -Waligned-new.
605 2018-09-04 Richard Biener <rguenther@suse.de>
607 PR tree-optimization/87211
608 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
609 backedge value we're supposed to treat as VARYING also number
610 the PHI to VARYING in case it got a different value-number already.
612 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
614 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
615 (extract_range_from_binary_expr_1): Do not call
616 vrp_can_optimize_bit_op.
617 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
619 (wide_int_range_get_mask_and_bounds): New.
620 (wide_int_range_optimize_bit_op): New.
621 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
622 (wide_int_range_bit_and): Same.
623 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
624 (wide_int_range_optimize_bit_op): New.
625 (wide_int_range_get_mask_and_bounds): New.
627 2018-09-04 Richard Biener <rguenther@suse.de>
629 PR tree-optimization/87176
630 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
631 variable. When value-numbering a virtual PHI node make sure
632 to not value-number to the backedge value.
634 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
636 * doc/extend.texi (Long Long, Hex Floats): Document support for
637 long long and hex floats in more recent versions of ISO C++.
639 2018-09-03 Richard Biener <rguenther@suse.de>
641 PR tree-optimization/87177
642 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
645 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
647 * bb-reorder.c (edge_order): Convert to C-qsort-style
648 tri-state comparator.
649 (reorder_basic_blocks_simple): Change std::stable_sort to gcc_stablesort.
651 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
653 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
654 tri-state comparator.
655 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
657 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
659 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
660 (mergesort): ... here as maximum count for using netsort.
661 (gcc_qsort): Set nlim to 3 if stable sort is requested.
662 (gcc_stablesort): New.
663 * system.h (gcc_stablesort): Declare.
665 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
667 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
668 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
669 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
671 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
673 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
674 lxsdx and stxsdx alternatives.
675 (*mov<mode>_hardfloat64): Ditto.
676 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
678 2018-09-03 Richard Biener <rguenther@suse.de>
680 PR tree-optimization/87200
681 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
684 2018-09-03 Martin Liska <mliska@suse.cz>
686 PR tree-optimization/87201
687 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
688 Fix parenthesis in an expression.
690 2018-09-03 Richard Biener <rguenther@suse.de>
692 PR tree-optimization/87197
693 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
694 visited. CSE the VN_INFO hashtable lookup.
696 PR tree-optimization/87169
697 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
698 iterating make sure there's no extra backedges from irreducible
699 regions feeding the header. Mark the destination block
702 2018-09-03 Martin Liska <mliska@suse.cz>
705 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
706 * common/common-targhooks.c (default_get_valid_option_values):
708 * common/common-targhooks.h (default_get_valid_option_values):
710 * common/config/i386/i386-common.c: Move processor_target_table
712 (ix86_get_valid_option_values): New function.
713 (TARGET_GET_VALID_OPTION_VALUES): New macro.
714 * config/i386/i386.c (struct ptt): Move to i386-common.c.
715 (PTA_*): Move all defined masks into i386-common.c.
716 (ix86_function_specific_restore): Use new processor_cost_table.
717 * config/i386/i386.h (struct ptt): Moved from i386.c.
718 (struct pta): Likewise.
719 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
720 * doc/tm.texi.in: Likewise.
721 * opt-suggestions.c (option_proposer::suggest_option):
722 Pass prefix to build_option_suggestions.
723 (option_proposer::get_completions): Likewise.
724 (option_proposer::build_option_suggestions): Use the new target
726 * opts.c (struct option_help_tuple): New struct.
727 (print_filtered_help): Use the new target hook.
729 2018-09-03 Martin Liska <mliska@suse.cz>
732 * predict.c (set_even_probabilities): Add likely_edges
733 argument and handle cases where we have precisely one
735 (combine_predictions_for_bb): Catch also likely_edges.
736 (tree_predict_by_opcode): Handle gswitch statements.
737 * tree-cfg.h (find_case_label_for_value): New declaration.
738 (find_taken_edge_switch_expr): Likewise.
739 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
740 Find pivot in decision tree based on probabily, not by number of
743 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
745 * doc/standards.texi (Standards): Update Objective-C reference.
747 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
749 * doc/install.texi (Prerequisites): Update link for MPC.
751 2018-09-01 Michael Matz <matz@suse.de>
753 PR tree-optimization/87074
754 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
755 PHIs for outer-loop uses.
757 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
759 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
760 * doc/invoke.texi (C Dialect Options): Ditto.
762 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
764 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
766 2018-08-31 Richard Biener <rguenther@suse.de>
768 PR tree-optimization/87168
769 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
770 (rpo_elim::eliminate_avail): When OP was not visited it must
773 2018-08-31 David Malcolm <dmalcolm@redhat.com>
775 * tree-vrp.c (copy_value_range): Convert param "from" from
776 "value_range *" to "const value_range *".
777 (range_is_null): Likewise for param "vr".
778 (range_int_cst_p): Likewise.
779 (range_int_cst_singleton_p): Likewise.
780 (symbolic_range_p): Likewise.
781 (value_ranges_intersect_p): Likewise for both params.
782 (value_range_nonnegative_p): Likewise for param "vr".
783 (value_range_constant_singleton): Likewise.
784 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
785 (extract_range_into_wide_ints): Likewise for param "vr".
786 (extract_range_from_multiplicative_op): Likewise for params "vr0"
788 (vrp_can_optimize_bit_op): Likewise.
789 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
791 (extract_range_from_unary_expr): Likewise.
792 (debug_value_range): Likewise for param "vr".
793 (value_range::dump): Add "const" qualifier.
794 (vrp_prop::check_array_ref): Convert local "vr" from
795 "value_range *" to "const value_range *".
796 (vrp_prop::check_mem_ref): Likewise.
797 (vrp_prop::visit_stmt): Likewise for local "old_vr".
798 (vrp_intersect_ranges_1): Likewise for param "vr_1".
799 (vrp_intersect_ranges): Likewise.
800 (simplify_stmt_for_jump_threading): Likewise for local "vr".
801 (vrp_prop::vrp_finalize): Likewise.
802 * tree-vrp.h (value_range::dump): Add "const" qualifier.
803 (vrp_intersect_ranges): Add "const" qualifier to params as above.
804 (extract_range_from_unary_expr): Likewise.
805 (value_range_constant_singleton): Likewise.
806 (symbolic_range_p): Likewise.
807 (copy_value_range): Likewise.
808 (extract_range_from_binary_expr_1): Likewise.
809 (range_int_cst_p): Likewise.
810 (vrp_set_zero_nonzero_bits): Likewise.
811 (range_int_cst_singleton_p): Likewise.
813 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
815 * config/aarch64/arm_neon.h (vabsd_s64): New.
816 (vnegd_s64): Likewise.
818 2018-08-31 Martin Jambor <mjambor@suse.cz>
820 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
822 2018-08-31 Martin Liska <mliska@suse.cz>
824 * ipa-icf.c (sem_item::add_type): Use
825 sem_item::m_type_hash_cache.
826 * ipa-icf.h: Move the cache from sem_item_optimizer
829 2018-08-31 Nathan Sidwell <nathan@acm.org>
831 * doc/extend.texi (Backwards Compatibility): Remove implicit
832 extern C leeway of () being (...).
834 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
836 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
838 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
842 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
844 2018-08-31 Jakub Jelinek <jakub@redhat.com>
847 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
848 gen_int_mode. Formatting fixes.
850 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
852 * target.def (custom_function_descriptors): Improve documentation.
853 * doc/tm.texi.in (Trampolines): Expand discussion of function
854 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
855 beginning of the section.
856 * doc/tm.texi: Regenerated.
858 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
860 * cfg.h (class auto_edge_flag): Spell out the template-id of the
861 base class in the initializer list. This is a workaround for
862 building with older GCC.
863 (class auto_bb_flag): Likewise.
865 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
867 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
868 (altivec_vcmpequ<VI_char>_p): Remove star.
869 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
871 (expand_strncmp_vec_sequence): New function.
872 (emit_final_str_compare_vec): New function.
873 (expand_strn_compare): Add support for vector strncmp.
874 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
875 length specification to bytes.
876 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
877 (vcmpnezb_p): New pattern.
878 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
879 for option -mstring-compare-inline-limit.
881 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
883 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
884 (PTA_SKYLAKE): Add PTA_AES.
885 (PTA_GOLDMONT): Likewise.
887 2018-08-29 Jan Hubicka <jh@suse.cz>
890 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
891 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
893 2018-08-29 Jan Hubicka <jh@suse.cz>
895 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
897 (hash_tree): Do not visit TYPE_STUB_DECL.
898 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
899 stream TYPE_STUB_DECL.
900 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
901 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
902 after free_lang_data.
903 (type_in_anonymous_namespace_p): Likewise.
905 2018-08-29 Jan Hubicka <jh@suse.cz>
907 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
908 comment that it has to be even number.
909 (class sreal): Change m_sig type to int32_t.
910 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
911 int64_t for temporary calculations.
912 (sreal_verify_basics): Drop one bit from minimum and maximum.
914 2018-08-30 Richard Biener <rguenther@suse.de>
916 PR tree-optimization/87147
917 * tree-ssa-sccvn.c (SSA_VISITED): New function.
918 (visit_phi): When the degenerate result is from the backedge and
919 we didn't visit its definition yet drop to VARYING.
920 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
922 2018-08-29 Jan Hubicka <jh@suse.cz>
924 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
926 (hash_tree): Likewise.
928 2018-08-29 Jan Hubicka <jh@suse.cz>
930 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
931 and TYPE_NEXT_REF_TO.
933 2018-08-29 Jan Hubicka <jh@suse.cz>
935 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
936 comment that it has to be even number.
937 (class sreal): Change m_sig type to int32_t.
938 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
939 int64_t for temporary calculations.
940 (sreal_verify_basics): Drop one bit from minimum and maximum.
942 2018-08-30 Tamar Christina <tamar.christina@arm.com>
944 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
946 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
949 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
950 if to_add is negative.
952 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
955 * builtins.c (c_strlen): Improve range checks.
957 2018-08-29 Martin Sebor <msebor@redhat.com>
958 Jeff Law <law@redhat.com>
960 PR tree-optimization/86714
961 PR tree-optimization/86711
962 * builtins.c (c_strlen): Add arguments to call to string_constant.
963 * expr.c (string_constant): Add argument. Detect missing nul
964 terminator and outermost declaration it's missing in.
965 * expr.h (string_constant): Add argument.
966 * fold-const.c (read_from_constant_string): Add arguments to call to
968 (c_getstr): Likewise.
969 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
971 * tree-ssa-strlen.c (get_stridx): Likewise.
973 2018-08-29 Jan Hubicka <jh@suse.cz>
975 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
976 Do not stream DECL_VINDEX.
977 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
978 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
979 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
981 2018-08-29 Richard Biener <rguenther@suse.de>
983 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
984 virtual operands that are not default defs to honor region
986 (rpo_vn_valueize): Remove ineffective code here.
988 2018-08-29 Richard Biener <rguenther@suse.de>
990 PR tree-optimization/87132
991 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
992 when skipping defs reachable over backedges.
994 2018-08-29 Richard Biener <rguenther@suse.de>
996 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
997 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
998 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
999 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
1000 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
1001 (vn_reference_lookup_3): Remove use of const_parms.
1002 (free_rpo_vn): Do not free const_parms.
1003 (do_rpo_vn): Do not call init_const_parms.
1004 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
1005 SSA_NAME_POINTS_TO_READONLY_MEMORY.
1006 (call_may_clobber_ref_p_1): Likewise.
1008 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
1011 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
1013 (-ftree-scev-cprop): Document.
1015 2018-08-29 Jan Hubicka <jh@suse.cz>
1017 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
1019 (sreal constructor): Update.
1020 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
1021 sreal:operator/): Update.
1023 2018-08-29 Martin Liska <mliska@suse.cz>
1025 * tree-switch-conversion.c (switch_conversion::expand):
1026 Strenghten assumption about gswitch statements.
1028 2018-08-29 Richard Biener <rguenther@suse.de>
1030 PR tree-optimization/87117
1031 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
1032 re-value-number released SSA VDEFs.
1034 2018-08-29 Richard Biener <rguenther@suse.de>
1036 PR tree-optimization/87126
1037 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
1039 2018-08-28 Jim Wilson <jimw@sifive.com>
1041 * config/riscv/pic.md: Rewrite.
1042 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
1044 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
1045 (SOFTF, default_load, softload, softstore): New.
1047 2018-08-28 Jeff Law <law@redhat.com>
1049 * fold-const.c (fold_binary_loc): Remove recently added assert.
1051 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
1053 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
1054 to OP parmeter of generated function.
1056 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
1058 PR tree-optimization/87009
1059 * match.pd: Add boolean optimizations.
1061 2018-08-28 Martin Sebor <msebor@redhat.com>
1064 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
1065 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
1066 (pass_walloca::gate): Use it.
1067 (alloca_call_type): Same.
1068 (pass_walloca::execute): Same.
1069 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
1071 2018-08-28 David Malcolm <dmalcolm@redhat.com>
1073 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
1074 GCC_VERSION for usage of "__gcc_dump_printf__" format from
1077 2018-08-28 Richard Biener <rguenther@suse.de>
1079 PR tree-optimization/87124
1080 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
1081 constants before looking up avail.
1083 2018-08-28 Jakub Jelinek <jakub@redhat.com>
1086 * calls.c (maybe_warn_nonstring_arg): Punt early if
1087 warn_stringop_overflow is zero. Don't call get_range_strlen
1088 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
1089 Swap comparison operands to have constants on rhs. Only use
1090 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
1091 increment lenrng[0].
1093 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
1095 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
1096 use of tree_to_shwi. Remove duplicated test for the size being
1097 a whole number of bytes.
1099 2018-08-28 Richard Biener <rguenther@suse.de>
1101 PR tree-optimization/87117
1102 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
1103 Handle removed stmt without LHS (GIMPLE_NOP).
1105 2018-08-28 Richard Biener <rguenther@suse.de>
1107 PR tree-optimization/87117
1108 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
1109 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
1111 2018-08-28 Richard Biener <rguenther@suse.de>
1113 PR tree-optimization/87117
1114 * tree-ssa-pre.c (compute_avail): Do not make expressions
1115 with predicated values available.
1116 (get_expr_value_id): Assert we do not run into predicated value
1119 2018-08-28 Richard Biener <rguenther@suse.de>
1121 PR tree-optimization/87117
1122 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
1123 get virtual operands.
1124 (get_expr_operands): Handle STRING_CST like other decls.
1126 2018-08-28 Martin Liska <mliska@suse.cz>
1128 * tree.h: Update documentation of fndecl_built_in_p
1132 2018-08-27 Jeff Law <law@redhat.com>
1133 PR tree-optimization/87110
1134 * tree-ssa-dse.c (compute_trims): Handle non-constant
1137 2018-08-27 Martin Sebor <msebor@redhat.com>
1139 PR tree-optimization/86914
1140 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
1142 2018-08-27 Martin Sebor <msebor@redhat.com>
1144 PR tree-optimization/87112
1145 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
1146 the type of the bound argument.
1148 2018-08-27 Jeff Law <law@redhat.com>
1150 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
1151 type does not have a TYPE_SIZE_UNIT.
1153 2018-08-27 Steve Ellcey <sellcey@cavium.com>
1155 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
1156 with include of backend.h.
1158 2018-08-27 Richard Biener <rguenther@suse.de>
1160 PR tree-optimization/86927
1161 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
1162 use const cond reduction code.
1164 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
1166 PR tree-optimization/85758
1167 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
1169 2018-08-27 David Malcolm <dmalcolm@redhat.com>
1172 * diagnostic-show-locus.c (class layout_range): Update for
1173 conversion of show_caret_p to a tri-state.
1174 (layout_range::layout_range): Likewise.
1175 (make_range): Likewise.
1176 (layout::maybe_add_location_range): Likewise.
1177 (layout::should_print_annotation_line_p): Don't show annotation
1178 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
1179 (layout::get_state_at_point): Update for conversion of
1180 show_caret_p to a tri-state. Bail out early for
1181 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
1182 underlining or source colorization.
1183 (gcc_rich_location::add_location_if_nearby): Update for conversion
1184 of show_caret_p to a tri-state.
1185 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
1186 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
1188 (selftest::test_one_liner_labels): Likewise.
1189 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
1190 conversion of show_caret_p to a tri-state.
1191 * pretty-print.c (text_info::set_location): Likewise.
1192 * pretty-print.h (text_info::set_location): Likewise.
1193 * substring-locations.c (format_warning_n_va): Likewise.
1194 * tree-diagnostic.c (default_tree_printer): Likewise.
1195 * tree-pretty-print.c (newline_and_indent): Likewise.
1197 2018-08-27 David Malcolm <dmalcolm@redhat.com>
1200 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
1201 line above for line-insertion fix-it hints.
1202 (selftest::test_fixit_insert_containing_newline): Update the
1203 expected results, and add a test with line-numbering enabled.
1205 2018-08-27 Martin Liska <mliska@suse.cz>
1208 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
1209 params with DECL_HAS_VALUE_EXPR_P.
1211 2018-08-27 Martin Liska <mliska@suse.cz>
1213 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
1214 selected expansion strategy.
1216 2018-08-27 Martin Liska <mliska@suse.cz>
1218 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
1219 * builtins.c (is_builtin_fn): Likewise.
1220 * attribs.c (diag_attr_exclusions): Use new function
1221 fndecl_built_in_p and remove check for FUNCTION_DECL if
1223 (builtin_mathfn_code): Likewise.
1224 (fold_builtin_expect): Likewise.
1225 (fold_call_expr): Likewise.
1226 (fold_builtin_call_array): Likewise.
1227 (fold_call_stmt): Likewise.
1228 (set_builtin_user_assembler_name): Likewise.
1229 (is_simple_builtin): Likewise.
1230 * calls.c (gimple_alloca_call_p): Likewise.
1231 (maybe_warn_nonstring_arg): Likewise.
1232 * cfgexpand.c (expand_call_stmt): Likewise.
1233 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
1234 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
1235 (cgraph_node::verify_node): Likewise.
1236 * cgraphclones.c (build_function_decl_skip_args): Likewise.
1237 (cgraph_node::create_clone): Likewise.
1238 * config/arm/arm.c (arm_insert_attributes): Likewise.
1239 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
1240 * dse.c (scan_insn): Likewise.
1241 * expr.c (expand_expr_real_1): Likewise.
1242 * fold-const.c (operand_equal_p): Likewise.
1243 (fold_binary_loc): Likewise.
1244 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
1245 * gimple-low.c (lower_stmt): Likewise.
1246 * gimple-pretty-print.c (dump_gimple_call): Likewise.
1247 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Likewise.
1248 * gimple.c (gimple_build_call_from_tree): Likewise.
1249 (gimple_call_builtin_p): Likewise.
1250 (gimple_call_combined_fn): Likewise.
1251 * gimplify.c (gimplify_call_expr): Likewise.
1252 (gimple_boolify): Likewise.
1253 (gimplify_modify_expr): Likewise.
1254 (gimplify_addr_expr): Likewise.
1255 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
1256 * ipa-cp.c (determine_versionability): Likewise.
1257 * ipa-fnsummary.c (compute_fn_summary): Likewise.
1258 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
1259 * ipa-split.c (visit_bb): Likewise.
1260 (split_function): Likewise.
1261 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
1262 * lto-cgraph.c (input_node): Likewise.
1263 * lto-streamer-out.c (write_symbol): Likewise.
1264 * omp-low.c (setjmp_or_longjmp_p): Likewise.
1265 (lower_omp_1): Likewise.
1266 * predict.c (strip_predict_hints): Likewise.
1267 * print-tree.c (print_node): Likewise.
1268 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
1269 * trans-mem.c (is_tm_irrevocable): Likewise.
1270 (is_tm_load): Likewise.
1271 (is_tm_simple_load): Likewise.
1272 (is_tm_store): Likewise.
1273 (is_tm_simple_store): Likewise.
1274 (is_tm_abort): Likewise.
1275 (tm_region_init_1): Likewise.
1276 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
1277 * tree-cfg.c (verify_gimple_call): Likewise.
1278 (move_stmt_r): Likewise.
1279 (stmt_can_terminate_bb_p): Likewise.
1280 * tree-eh.c (lower_eh_constructs_2): Likewise.
1281 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
1282 * tree-inline.c (remap_gimple_stmt): Likewise.
1283 (copy_bb): Likewise.
1284 (estimate_num_insns): Likewise.
1285 (fold_marked_statements): Likewise.
1286 * tree-sra.c (scan_function): Likewise.
1287 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
1288 (optimize_stack_restore): Likewise.
1289 (pass_fold_builtins::execute): Likewise.
1290 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
1291 (mark_all_reaching_defs_necessary_1): Likewise.
1292 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
1293 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
1294 (pass_forwprop::execute): Likewise.
1295 * tree-ssa-loop-im.c (stmt_cost): Likewise.
1296 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1297 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
1298 * tree-ssa-strlen.c (get_string_length): Likewise.
1299 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
1300 (find_func_aliases_for_call): Likewise.
1301 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
1302 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
1303 * tree-tailcall.c (find_tail_calls): Likewise.
1304 * tree.c (need_assembler_name_p): Likewise.
1305 (free_lang_data_in_decl): Likewise.
1306 (get_call_combined_fn): Likewise.
1307 * ubsan.c (is_ubsan_builtin_p): Likewise.
1308 * varasm.c (incorporeal_function_p): Likewise.
1309 * tree.h (DECL_BUILT_IN): Remove and replace with
1311 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
1312 (fndecl_built_in_p): New.
1314 2018-08-27 Martin Liska <mliska@suse.cz>
1316 PR tree-optimization/86847
1317 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
1318 Dump also subtree probability.
1319 (switch_decision_tree::do_jump_if_equal): New function.
1320 (switch_decision_tree::emit_case_nodes): Handle special
1321 situations in balanced tree that can be emitted much simpler.
1322 Fix calculation of probabilities that happen in tree expansion.
1323 * tree-switch-conversion.h (struct cluster): Add
1325 (struct simple_cluster): Likewise.
1326 (struct case_tree_node): Add new function has_child.
1327 (do_jump_if_equal): New.
1329 2018-08-27 Martin Liska <mliska@suse.cz>
1331 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
1332 Add new argument to bit_test_cluster constructor.
1333 (bit_test_cluster::emit): Set bits really number of values
1335 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
1336 probability argument.
1337 * tree-switch-conversion.h (struct bit_test_cluster):
1338 Add m_handles_entire_switch.
1340 2018-08-27 Martin Liska <mliska@suse.cz>
1342 PR tree-optimization/86702
1343 * tree-switch-conversion.c (jump_table_cluster::emit):
1344 Make probabilities even for values in jump table
1345 according to number of cases handled.
1346 (switch_decision_tree::compute_cases_per_edge): Pass
1347 argument to reset_out_edges_aux function.
1348 (switch_decision_tree::analyze_switch_statement): Likewise.
1349 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
1352 2018-08-27 Martin Liska <mliska@suse.cz>
1354 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
1355 cfun argument explicitly.
1356 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
1357 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
1358 function gimple_switch_default_bb.
1359 (convert_switch_statements):
1361 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
1362 * stmt.c (label_to_block_fn): Use label_to_block and pass
1363 cfun argument explicitly and use gimple_switch_label_bb.
1364 (expand_case): Likewise.
1365 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
1366 cfun argument explicitly. Likewise.
1367 (make_edges_bb): Likewise.
1368 (make_cond_expr_edges): Likewise.
1369 (get_cases_for_edge): Likewise.
1370 (make_gimple_switch_edges): Likewise.
1371 (label_to_block_fn): Likewise.
1372 (label_to_block): Likewise.
1373 (make_goto_expr_edges): Likewise.
1374 (make_gimple_asm_edges): Likewise.
1375 (main_block_label): Likewise.
1376 (group_case_labels_stmt): Likewise.
1377 (find_taken_edge_computed_goto): Likewise.
1378 (find_taken_edge_switch_expr): Likewise.
1379 (gimple_verify_flow_info): Likewise.
1380 (gimple_redirect_edge_and_branch): Likewise.
1381 (gimple_switch_label_bb): New function.
1382 (gimple_switch_default_bb): Likewise.
1383 (gimple_switch_edge): Likewise.
1384 (gimple_switch_default_edge): Likewise.
1385 * tree-cfg.h (label_to_block_fn): Remove and replace ...
1386 (label_to_block): ... with this.
1387 (gimple_switch_label_bb): New.
1388 (gimple_switch_default_bb): Likewise.
1389 (gimple_switch_edge): Likewise.
1390 (gimple_switch_default_edge): Likewise.
1391 * tree-cfgcleanup.c (convert_single_case_switch): Use
1392 new gimple functions and pass new argument to label_to_block.
1393 (cleanup_control_flow_bb):
1394 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
1395 cfun argument explicitly.
1396 (make_eh_edges): Likewise.
1397 (redirect_eh_dispatch_edge): Likewise.
1398 (lower_resx): Likewise.
1399 (lower_eh_dispatch): Likewise.
1400 (maybe_remove_unreachable_handlers): Likewise.
1401 (unsplit_eh): Likewise.
1402 (cleanup_empty_eh): Likewise.
1403 (verify_eh_edges): Likewise.
1404 (verify_eh_dispatch_edge): Likewise.
1405 * tree-ssa-dom.c (record_edge_info): Likewise.
1406 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
1407 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
1408 (thread_through_normal_block): Likewise.
1409 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
1410 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
1411 * tree-switch-conversion.c (switch_conversion::collect): Use new
1413 (switch_conversion::check_final_bb): Likewise.
1414 (switch_conversion::gather_default_values): Pass new argument
1416 (switch_conversion::build_constructors): Likewise.
1417 (switch_decision_tree::compute_cases_per_edge): Use new
1418 gimple_switch_edge function.
1419 (switch_decision_tree::analyze_switch_statement): Pass new argument
1421 (switch_decision_tree::try_switch_expansion): Use
1422 gimple_switch_default_edge.
1423 * tree-vrp.c (find_switch_asserts): Pass new argument
1425 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
1426 (vr_values::simplify_switch_using_ranges): Likewise.
1428 2018-08-27 Richard Biener <rguenther@suse.de>
1430 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
1431 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
1433 * tree-ssa-sccvn.h (struct vn_pval): New structure.
1434 (struct vn_nary_op_s): Add unwind_to member. Add
1435 predicated_values flag and put result into a union together
1436 with a linked list of vn_pval.
1437 (struct vn_ssa_aux): Add name member to make maintaining
1438 a map of SSA name to vn_ssa_aux possible. Remove no longer
1439 needed info, dfsnum, low, visited, on_sccstack, use_processed
1440 and range_info_anti_range_p members.
1441 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
1442 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
1444 (vn_valueize): New global.
1445 (vn_context_bb): Likewise.
1446 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
1447 VN_INFO_PTR_INFO): Remove.
1448 * tree-ssa-sccvn.c: ... (rewrite)
1449 (pass_fre::execute): For -O2+ initialize loops and run
1450 RPO VN in optimistic mode (iterating). For -O1 and -Og
1451 run RPO VN in non-optimistic mode.
1452 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
1453 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
1454 * doc/invoke.texi (sccvn-max-scc-size): Remove.
1455 (rpo-vn-max-loop-depth): Document.
1456 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
1457 when valuezing the VUSE signals we walked out of the region.
1458 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
1459 (phi_translate): Set VN context block to use for availability
1461 (compute_avail): Likewise.
1462 (pre_valueize): New function.
1463 (pass_pre::execute): Adjust to the RPO VN API.
1465 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
1466 (propagate_constants_for_unrolling): Remove.
1467 (tree_unroll_loops_completely): Perform value-numbering
1468 on the unrolled bodies loop parent.
1470 2018-08-27 Richard Biener <rguenther@suse.de>
1472 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
1473 for partial antic compute.
1475 2018-08-27 Jakub Jelinek <jakub@redhat.com>
1477 PR rtl-optimization/87065
1478 * combine.c (simplify_if_then_else): Formatting fix.
1479 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
1481 (known_cond): Don't return const_true_rtx for vector modes. Use
1482 CONST0_RTX instead of const0_rtx. Formatting fixes.
1484 2018-08-27 Martin Liska <mliska@suse.cz>
1486 PR gcov-profile/87069
1487 * gcov.c (process_file): Record files already processed
1488 and warn about a file being processed multiple times.
1490 2018-08-27 Martin Liska <mliska@suse.cz>
1493 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1494 Set default values for x_aarch64_*_string strings.
1495 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
1496 prefix. For -mabi do not print '=ABI' in help and use
1497 <option_value> format for -msve-vector-bits and -moverride
1500 2018-08-26 Jeff Law <law@redhat.com>
1502 * config/mips/frame-header-opt.c: Include "backend.h" rather than
1505 2018-08-26 Marek Polacek <polacek@redhat.com>
1507 PR c++/87029, Implement -Wredundant-move.
1508 * doc/invoke.texi: Document -Wredundant-move.
1510 2018-08-25 Martin Sebor <msebor@redhat.com>
1512 PR tree-optimization/87059
1513 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
1514 to the same type as the other.
1515 * fold-const.c (fold_binary_loc): Assert expectation.
1517 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
1519 * config/darwin.c (machopic_legitimize_pic_address): Clean up
1520 extraneous parentheses, dead code section and formatting.
1522 2018-08-24 David Malcolm <dmalcolm@redhat.com>
1525 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
1526 wide enough for jumps in the line-numbering to be visible.
1527 (layout::print_gap_in_line_numbering): New member function.
1528 (layout::calculate_line_spans): When using line numbering, merge
1529 line spans that are only 1 line apart.
1530 (diagnostic_show_locus): When printing line numbers, show gaps in
1531 line numbering directly, rather than printing headers.
1532 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
1533 line-numbering with multiple line spans.
1534 (selftest::test_fixit_insert_containing_newline_2): Add test of
1535 line-numbering, in which the spans are close enough to be merged.
1537 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
1539 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
1540 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
1541 * tree-vrp.c (range_is_nonnull): Remove.
1542 (range_includes_zero_p): Accept value_range instead of min/max.
1543 (extract_range_from_binary_expr_1): Do not early bail on
1545 Use range_includes_zero_p instead of range_is_nonnull.
1546 (extract_range_from_unary_expr): Use range_includes_zero_p instead
1547 of range_is_nonnull.
1548 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
1549 special case VR_ANTI_RANGE.
1550 (vrp_finalize): Same.
1551 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
1553 (range_is_nonnull): Remove.
1554 * vr-values.c (vrp_stmt_computes_nonzero): Use
1555 range_includes_zero_p instead of range_is_nonnull.
1556 (extract_range_basic): Pass value_range to range_includes_zero_p
1557 instead of range_is_nonnull.
1559 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
1561 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
1562 * emit-rtl.h (rtl_data): Remove return_bnd.
1563 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
1564 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
1565 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
1566 (POINTER_BOUNDS_MODE): Remove definition.
1567 (make_pointer_bounds_mode): Remove.
1568 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
1569 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
1570 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
1571 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
1572 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
1573 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
1574 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
1576 * config/i386/i386-modes.def (BND32, BND64): Remove.
1577 * config/i386/i386.c (dbx_register_map): Remove bound registers.
1578 (dbx64_register_map): Ditto.
1579 (svr4_dbx_register_map): Ditto.
1580 (indirect_thunk_bnd_needed): Remove.
1581 (indirect_thunks_bnd_used): Ditto.
1582 (indirect_return_bnd_needed): Ditto.
1583 (indirect_return_via_cx_bnd): Ditto.
1584 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
1585 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
1586 (output_indirect_thunk): Ditto. Remove need_prefix argument.
1587 (output_indirect_thunk_function): Remove handling of
1588 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
1589 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
1590 (ix86_save_reg): Remove handling of crtl->return_bnd.
1591 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
1592 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
1593 and UNSPEC_BNDLX_ADDR.
1594 (ix86_output_indirect_branch_via_reg): Remove handling of
1595 indirect_thunk_prefix_bnd.
1596 (ix86_output_indirect_branch_via_push): Ditto.
1597 (ix86_output_function_return): Ditto.
1598 (ix86_output_indirect_function_return): Ditto.
1599 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
1600 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
1601 (CALL_USED_REGISTERS): Ditto.
1602 (REG_ALLOC_ORDER): Update for removal of bound registers.
1603 (HI_REGISTER_NAMES): Ditto.
1604 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
1605 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
1606 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
1607 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
1608 (FIRST_PSEUDO_REG): Update.
1609 (BND): Remove mode iterator.
1610 * config/i386/predicates.md (bnd_mem_operator): Remove.
1612 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
1614 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
1617 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
1619 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
1620 the case in which the permute needs only a single element and
1621 repeats for every vector of the result. Extend that case to
1622 handle variable-length vectors.
1623 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
1625 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
1628 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
1629 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
1631 2018-08-24 Richard Biener <rguenther@suse.de>
1633 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
1634 bb_flags_allocated members.
1635 (auto_flag): New RAII class for allocating flags.
1636 (auto_edge_flag): New RAII class for allocating edge flags.
1637 (auto_bb_flag): New RAII class for allocating bb flags.
1638 * cfgloop.c (verify_loop_structure): Allocate temporary edge
1640 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
1641 in favor of temporarily allocated BB flag.
1642 * hsa-brig.c: Re-order includes.
1643 * hsa-dump.c: Likewise.
1644 * hsa-regalloc.c: Likewise.
1645 * print-rtl.c: Likewise.
1646 * profile-count.c: Likewise.
1648 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
1651 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
1654 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
1657 * wide-int-range.cc (wide_int_range_div): Do not ignore result
1658 from wide_int_range_multiplicative_op.
1660 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1662 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
1665 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
1667 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
1670 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
1672 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
1673 full GENERIC expression used for replacement.
1675 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
1677 * tree-vrp.c (abs_extent_range): Remove.
1678 (extract_range_into_wide_ints): Pass wide ints by reference.
1679 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
1680 Pass wide ints by reference in all calls to
1681 extract_range_into_wide_ints.
1682 * wide-int-range.cc (wide_int_range_div): New.
1683 * wide-int-range.h (wide_int_range_div): New.
1684 (wide_int_range_includes_zero_p): New.
1685 (wide_int_range_zero_p): New.
1687 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
1689 * config/aarch64/aarch64.md (arches): New enum.
1690 (arch): New enum attr.
1691 (arch_enabled): New attr.
1692 (enabled): Now uses arch_enabled only.
1693 (simd, sve, fp16): Removed attribute.
1694 (fp): Attr now defined in terms of 'arch'.
1695 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
1696 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
1697 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
1698 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
1699 attributes into 'arch'.
1700 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
1701 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
1702 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
1703 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
1704 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
1705 'simd' attribute into 'arch'.
1706 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
1707 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
1708 Convert use of 'fp' attribute to 'arch'.
1709 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
1710 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
1712 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
1713 (different modes) Merge 'fp' and 'simd' into 'arch'.
1714 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
1717 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
1719 PR rtl-optimization/87026
1720 * expmed.c (canonicalize_comparison): If we can no longer create
1721 pseudoregisters, don't.
1723 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
1726 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
1728 * config/arm/arm.c (speculation_barrier_libfunc): New static
1730 (arm_init_libfuncs): Initialize it.
1731 (arm_emit_speculation_barrier): New function.
1732 * config/arm/arm.md (speculation_barrier): Call
1733 arm_emit_speculation_barrier for architectures that do not have
1735 (speculation_barrier_insn): Only match on Armv7 or later.
1737 2018-08-23 Richard Biener <rguenther@suse.de>
1740 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
1743 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
1745 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
1746 of single-vector TBLs.
1747 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
1750 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
1753 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
1754 aarch64_evpc_tbl guard.
1756 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
1758 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
1761 2018-08-22 Martin Sebor <msebor@redhat.com>
1764 * tree-pretty-print.c (pretty_print_string): Add argument.
1765 (dump_generic_node): Call to pretty_print_string with string size.
1767 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
1769 PR rtl-optimization/86771
1770 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
1771 of two SETs into those two SETs, one to be placed at i2, if that SETs
1772 destination is modified between i2 and i3.
1774 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1776 PR tree-optimization/86725
1777 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
1779 (vect_analyze_scalar_cycles_1): Check it.
1781 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1783 PR tree-optimization/86725
1784 * tree-vect-loop.c (vect_is_simple_reduction): When treating
1785 an outer loop phi as a double reduction, make sure that the
1786 single user of the phi result is an inner loop phi.
1788 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1790 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
1791 grouped stores with gaps to a strided group.
1793 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1795 * tree-vect-stmts.c (get_group_load_store_type)
1796 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
1797 first statement in a group.
1799 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1801 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
1802 the sequence used in gcc/gcc.c.
1804 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1807 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
1810 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1812 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
1813 Darwin10-specific unwinder-shim.
1814 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
1815 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
1816 New to cater for Darwin10 Rosetta.
1818 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1820 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
1821 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
1823 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1828 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
1829 (dwarf2out_switch_text_section): Generate a local label for the
1830 second function sub-section and apply it as the second FDE start
1832 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
1833 second sub-section start.
1835 2018-08-22 Richard Biener <rguenther@suse.de>
1837 PR tree-optimization/86988
1838 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
1840 2018-08-22 Richard Biener <rguenther@suse.de>
1842 PR tree-optimization/86945
1843 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
1845 2018-08-22 Alexandre Oliva <oliva@adacore.com>
1847 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
1848 a comment about how uses of r2 for .sdata2 come about.
1850 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
1852 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
1854 2018-08-21 Marek Polacek <polacek@redhat.com>
1856 PR c++/86981, Implement -Wpessimizing-move.
1857 * doc/invoke.texi: Document -Wpessimizing-move.
1859 2018-08-21 Jan Hubicka <jh@suse.cz>
1861 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
1862 * tree.h (is_redundant_typedef): Remove.
1863 * dwarf2out.c (is_redundant_typedef): Turn into static function.
1865 2018-08-21 Jan Hubicka <jh@suse.cz>
1867 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
1870 2018-08-21 Tamar Christina <tamar.christina@arm.com>
1872 * expmed.c (extract_low_bits): Reject invalid subregs early.
1874 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
1877 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
1880 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
1882 * config/vxworks.h: Guard vxworks_asm_out_constructor and
1883 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
1884 * config/vxworks.c: Likewise.
1886 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
1888 * config/vxworks.c: Set targetm.have_ctors_dtors
1889 if HAVE_INITFINI_ARRAY_SUPPORT.
1890 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
1891 if HAVE_INITFINI_ARRAY_SUPPORT.
1893 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
1895 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
1896 default search path for VxWorks < 7.
1898 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
1900 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
1901 (get_format_string): Refer to c_getstr.
1903 2018-08-21 Tom de Vries <tdevries@suse.de>
1905 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
1906 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
1907 (debuginfo_early_stop): Declare.
1908 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
1909 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
1910 (debuginfo_early_stop): New function.
1911 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
1912 and debuginfo_early_stop.
1913 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
1914 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
1915 (general_init): Call debuginfo_early_init.
1916 (finalize): Call debuginfo_fini.
1917 (do_compile): Call debuginfo_init.
1918 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
1920 (@item -fdump-debug, @item -fdump-earlydebug): Add.
1922 2018-08-21 Tom de Vries <tdevries@suse.de>
1924 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
1925 flag_dump_noaddr and flag_dump_unnumbered.
1927 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
1929 * wide-int-range.cc (wide_int_range_abs): New.
1930 (wide_int_range_order_set): Rename from wide_int_range_min_max.
1931 * wide-int-range.h (wide_int_range_abs): New.
1932 (wide_int_range_min_max): New.
1933 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
1934 case to call wide_int_range_abs.
1935 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
1936 (extract_range_from_abs_expr): Delete.
1938 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
1941 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
1942 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
1945 2018-08-20 Nathan Sidwell <nathan@acm.org>
1946 Jeff Law <law@redhat.com>
1948 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
1949 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
1951 2018-08-20 David Malcolm <dmalcolm@redhat.com>
1954 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
1955 (decl_attributes): Likewise.
1956 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
1958 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
1959 * diagnostic-core.h (class auto_diagnostic_group): New class.
1960 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
1961 (diagnostic_report_diagnostic): Handle the first diagnostics within
1963 (emit_diagnostic): Add auto_diagnostic_group instance.
1965 (inform_n): Likewise.
1966 (warning): Likewise.
1967 (warning_at): Likewise.
1968 (warning_n): Likewise.
1969 (pedwarn): Likewise.
1970 (permerror): Likewise.
1972 (error_n): Likewise.
1973 (error_at): Likewise.
1975 (fatal_error): Likewise.
1976 (internal_error): Likewise.
1977 (internal_error_no_backtrace): Likewise.
1978 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
1979 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
1980 * diagnostic.h (struct diagnostic_context): Add fields
1981 "diagnostic_group_nesting_depth",
1982 "diagnostic_group_emission_count", "begin_group_cb",
1984 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
1985 Add auto_diagnostic_group instance(s).
1986 (find_explicit_erroneous_behavior): Likewise.
1987 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
1988 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
1989 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
1990 (gimplify_va_arg_expr): Likewise.
1991 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
1992 (HSA_SORRY_AT): Likewise.
1993 * ipa-devirt.c (compare_virtual_tables): Likewise.
1994 (warn_odr): Likewise.
1995 * multiple_target.c (expand_target_clones): Likewise.
1996 * opts-common.c (cmdline_handle_error): Likewise.
1997 * reginfo.c (globalize_reg): Likewise.
1998 * substring-locations.c (format_warning_n_va): Likewise.
1999 * tree-inline.c (expand_call_inline): Likewise.
2000 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
2001 * tree-ssa-loop-niter.c
2002 (do_warn_aggressive_loop_optimizations): Likewise.
2003 * tree-ssa-uninit.c (warn_uninit): Likewise.
2004 * tree.c (warn_deprecated_use): Likewise.
2006 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
2009 * config/i386/i386.md (eh_return): Always update EH return
2010 address in word_mode.
2012 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
2014 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
2015 TARGET_SPLIT_COMPLEX_ARG is defined.
2017 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
2019 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
2021 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
2024 * expr.c (expand_assignment): Assert that bitpos is positive.
2025 (store_field): Likewise
2026 (expand_expr_real_1): Make sure that bitpos is positive.
2027 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
2030 2018-08-20 Nathan Sidwell <nathan@acm.org>
2032 * Makefile.in (CPP_ID_DATA_H): Delete.
2033 (CPP_INTERNAL_H): Don't add it.
2034 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
2035 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
2037 2018-08-20 Richard Biener <rguenther@suse.de>
2039 PR tree-optimization/78655
2040 * tree-vrp.c (extract_range_from_binary_expr_1): Make
2041 pointer + offset nonnull if either operand is nonnull work.
2043 2018-08-20 Tom de Vries <tdevries@suse.de>
2045 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
2046 unless the referenced die describes the added property using
2047 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
2048 Otherwise, add a DW_AT_location to the referenced die.
2050 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
2053 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
2054 register_operand when calling ix86_set_reg_reg_cost.
2055 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
2056 Set *total to 0 for operands that satisfy x86_64_immediate_operand
2057 predicate and to 1 otherwise.
2059 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
2061 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
2062 emit a diagnostic that it is not supported and reset the option.
2063 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
2064 supported and consume the option. (ASM_FINAL_SPEC): New.
2066 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
2068 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
2071 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
2073 C-SKY port: Documentation
2075 * doc/extend.texi (C-SKY Function Attributes): New section.
2076 * doc/invoke.texi (Option Summary): Add C-SKY options.
2077 (C-SKY Options): New section.
2078 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
2080 2018-08-17 Jojo <jijie_rong@c-sky.com>
2081 Huibin Wang <huibin_wang@c-sky.com>
2082 Sandra Loosemore <sandra@codesourcery.com>
2083 Chung-Lin Tang <cltang@codesourcery.com>
2085 C-SKY port: Backend implementation
2087 * config/csky/*: New.
2088 * common/config/csky/*: New.
2090 2018-08-17 Jojo <jijie_rong@c-sky.com>
2091 Huibin Wang <huibin_wang@c-sky.com>
2092 Sandra Loosemore <sandra@codesourcery.com>
2093 Chung-Lin Tang <cltang@codesourcery.com>
2094 Andrew Jenner <andrew@codesourcery.com>
2096 C-SKY port: Configury
2098 * config.gcc (csky-*-*): New.
2099 * configure.ac: Add csky to targets for dwarf2 debug_line support.
2100 * configure: Regenerated.
2102 2018-08-17 David Malcolm <dmalcolm@redhat.com>
2104 * dump-context.h: Include "dumpfile.h".
2105 (dump_context::dump_printf_va): Convert final param from va_list
2106 to va_list *. Convert from ATTRIBUTE_PRINTF to
2107 ATTRIBUTE_GCC_DUMP_PRINTF.
2108 (dump_context::dump_printf_loc_va): Likewise.
2109 * dumpfile.c: Include "stringpool.h".
2110 (make_item_for_dump_printf_va): Delete.
2111 (make_item_for_dump_printf): Delete.
2112 (class dump_pretty_printer): New class.
2113 (dump_pretty_printer::dump_pretty_printer): New ctor.
2114 (dump_pretty_printer::emit_items): New member function.
2115 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
2117 (dump_pretty_printer::emit_item): New member function.
2118 (dump_pretty_printer::stash_item): New member function.
2119 (dump_pretty_printer::format_decoder_cb): New member function.
2120 (dump_pretty_printer::decode_format): New member function.
2121 (dump_context::dump_printf_va): Reimplement in terms of
2122 dump_pretty_printer.
2123 (dump_context::dump_printf_loc_va): Convert final param from va_list
2125 (dump_context::begin_scope): Reimplement call to
2126 make_item_for_dump_printf.
2127 (dump_printf): Update for change to dump_printf_va.
2128 (dump_printf_loc): Likewise.
2129 (selftest::test_capture_of_dump_calls): Convert "stmt" from
2130 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
2131 with %T, %E, and %G.
2132 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
2133 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
2134 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
2135 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
2136 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
2137 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
2138 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
2139 within a dump_printf_loc call to "%wu".
2140 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
2141 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
2142 missing space after "=".
2143 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
2144 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
2145 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
2146 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
2147 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
2148 duplicate "vectorized" from message.
2150 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
2152 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
2153 polyNxK_t element's TYPE_STRING_FLAG.
2155 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
2157 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
2158 (they were unnamed before). Fix comments.
2160 2018-08-17 Nathan Sidwell <nathan@acm.org>
2162 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
2164 2018-08-17 Richard Biener <rguenther@suse.de>
2166 PR tree-optimization/86841
2167 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
2169 2018-08-17 Martin Liska <mliska@suse.cz>
2171 * common.opt: Remove Warn, Init and Report for options with
2172 Ignore/Deprecated flag. Warning is done automatically for
2174 * config/i386/i386.opt: Likewise.
2175 * config/ia64/ia64.opt: Likewise.
2176 * config/rs6000/rs6000.opt: Likewise.
2177 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
2178 Remove usage of flag_check_pointer_bounds.
2179 * lto-wrapper.c (merge_and_complain): Do not handle
2180 OPT_fcheck_pointer_bounds.
2181 (append_compiler_options): Likewise.
2182 * opt-functions.awk: Do not handle Deprecated.
2183 * optc-gen.awk: Check that Var, Report and Init are not
2184 used for an option with Ignore/Deprecated flag.
2185 * opts-common.c (decode_cmdline_option): Do not report
2187 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
2189 * opts.h (struct cl_option): Remove cl_deprecated flag.
2190 (CL_ERR_DEPRECATED): Remove error enum value.
2192 2018-08-17 Richard Biener <rguenther@suse.de>
2195 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
2196 across a va-arg-pack using call adjust its return value accordingly.
2198 2018-08-16 Martin Sebor <msebor@redhat.com>
2200 PR tree-optimization/86853
2201 * gimple-ssa-sprintf.c (struct format_result): Rename member.
2202 (struct fmtresult): Add member and initialize it in ctors.
2203 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
2204 (format_string): Handle %S the same as %ls. Set MAYFAIL.
2205 (format_directive): Set POSUNDER4K when MAYFAIL is set.
2206 (parse_directive): Handle %C same as %c.
2207 (sprintf_dom_walker::compute_format_length): Adjust.
2208 (is_call_safe): Adjust.
2210 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
2212 * builtins.c (c_strlen): Add new parameter eltsize. Use it
2213 for determining how to count the elements.
2214 * builtins.h (c_strlen): Adjust prototype.
2215 * expr.c (string_constant): Add new parameter mem_size.
2216 Set *mem_size appropriately.
2217 * expr.h (string_constant): Adjust protoype.
2218 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
2219 * gimple-fold.h (get_range_strlen): Adjust prototype.
2220 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
2221 (format_string): Call get_string_length with eltsize.
2223 2018-08-16 David Malcolm <dmalcolm@redhat.com>
2225 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
2226 to emit the span, rather than setting it as the prefix.
2228 2018-08-16 David Malcolm <dmalcolm@redhat.com>
2230 * diagnostic-show-locus.c (layout::start_annotation_line): Add
2231 "margin_char" parameter, defaulting to space. Use it in place
2232 of pp_space for the initial part of the margin.
2233 (layout::print_leading_fixits): Use '+' when filling the margin
2234 of line-insertion fix-it hints.
2236 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2238 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
2241 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2243 * config/rs6000/altivec.md: Don't set length attribute to the default
2245 * config/rs6000/darwin.md: Ditto.
2246 * config/rs6000/dfp.md: Ditto.
2247 * config/rs6000/htm.md: Ditto.
2248 * config/rs6000/rs6000.md: Ditto.
2249 * config/rs6000/sync.md: Ditto.
2250 * config/rs6000/vsx.md: Ditto.
2252 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2254 * config/rs6000/altivec.md: Don't set length attribute to the default
2255 value, for branch instructions.
2256 * config/rs6000/darwin.md: Ditto.
2257 * config/rs6000/rs6000.md: Ditto.
2259 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2261 * config/rs6000/rs6000.md (length): Always define as const_int 4.
2262 (unnamed conditional branch define_insn): Set length to 4 or 8
2263 depending on offset.
2264 (<bd>_<mode>): Similar, for alternative 0.
2265 (<bd>tf_<mode>): Ditto.
2267 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2269 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
2271 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
2273 * doc/rtl.texi: Replace old RTX class names with new names.
2276 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
2278 * expmed.h (canonicalize_comparison): New declaration.
2279 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
2280 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
2281 * optabs.c (prepare_cmp_insn): Likewise.
2282 * rtl.h (unsigned_condition_p): New function which checks if a
2283 comparison operator is unsigned.
2285 2018-08-16 Nathan Sidwell <nathan@acm.org>
2287 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
2288 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
2290 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2293 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
2294 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
2295 (mov<mov>): ..this and enable unconditionally.
2297 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2299 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
2301 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
2303 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
2304 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
2305 "Common" with "Target".
2307 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
2309 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
2310 * doc/invoke.texi (mmitigate-rop): Remove.
2311 * config/i386/i386.c: Do not include "regrename.h".
2312 (ix86_rop_should_change_byte_p, reg_encoded_number)
2313 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
2315 (ix86_reorg): Remove call to ix86_mitigate_rop.
2316 * config/i386/i386.md (attr "modrm_class"): Remove.
2317 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
2318 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
2319 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
2321 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
2323 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
2324 allow folding of mergeh() and mergel() for the float and double types.
2325 (fold_mergehl_helper): Rework to handle building a permute tree
2328 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
2330 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
2333 2018-08-15 David Malcolm <dmalcolm@redhat.com>
2335 * common.opt (fdiagnostics-show-labels): New option.
2336 * diagnostic-show-locus.c (class layout_range): Add field
2338 (class layout): Add field "m_show_labels_p".
2339 (layout_range::layout_range): Add param "label" and use it to
2341 (make_range): Pass in NULL for new "label" param of layout_range's
2343 (layout::layout): Initialize m_show_labels_p.
2344 (layout::maybe_add_location_range): Pass in loc_range->m_label
2345 when constructing layout_range instances.
2346 (struct line_label): New struct.
2347 (layout::print_any_labels): New member function.
2348 (layout::print_line): Call it if label-printing is enabled.
2349 (selftest::test_one_liner_labels): New test.
2350 (selftest::test_diagnostic_show_locus_one_liner): Call it.
2351 * diagnostic.c (diagnostic_initialize): Initialize
2352 context->show_labels_p.
2353 * diagnostic.h (struct diagnostic_context): Add field
2355 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
2356 -fno-diagnostics-show-labels.
2357 * dwarf2out.c (gen_producer_string): Add
2358 OPT_fdiagnostics_show_labels to the ignored options.
2359 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
2361 (gcc_rich_location::maybe_add_expr): Likewise.
2362 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
2363 label" param, defaulting to NULL.
2364 (gcc_rich_location::add_expr): Add "label" param.
2365 (gcc_rich_location::maybe_add_expr): Likewise.
2366 (class text_range_label): New class.
2367 (class range_label_for_type_mismatch): New class.
2368 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
2369 of format_warning_va.
2370 (fmtwarn_n): Likewise for new params of format_warning_n_va.
2371 * lto-wrapper.c (merge_and_complain): Add
2372 OPT_fdiagnostics_show_labels to the "pick one setting" options.
2373 (append_compiler_options): Likewise to the dropped options.
2374 (append_diag_options): Likewise to the passed-on options.
2375 * opts.c (common_handle_option): Handle the new option.
2376 * selftest-diagnostic.c
2377 (test_diagnostic_context::test_diagnostic_context): Enable
2379 * substring-locations.c: Include "gcc-rich-location.h".
2380 (format_warning_n_va): Add "fmt_label" and "param_label" params
2381 and use them as appropriate.
2382 (format_warning_va): Add "fmt_label" and "param_label" params,
2383 passing them on to format_warning_n_va.
2384 (format_warning_at_substring): Likewise.
2385 (format_warning_at_substring_n): Likewise.
2386 * substring-locations.h (format_warning_va): Add "fmt_label" and
2387 "param_label" params.
2388 (format_warning_n_va): Likewise.
2389 (format_warning_at_substring): Likewise.
2390 (format_warning_at_substring_n): Likewise.
2391 * toplev.c (general_init): Initialize global_dc->show_labels_p.
2393 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
2396 * builtins.c (expand_builtin_memcmp): Do not expand the call
2397 when overflow is detected.
2399 2018-08-15 Martin Sebor <msebor@redhat.com>
2401 PR tree-optimization/71625
2402 * config/aarch64/aarch64-builtins.c
2403 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
2405 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
2407 * config/s390/s390.c (s390_reorg): Remove loop.
2409 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
2412 (darwin_function_switched_text_sections): Delete.
2413 * gcc/config/darwin.h
2414 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
2416 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
2419 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
2420 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
2421 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
2423 2018-08-15 Martin Liska <mliska@suse.cz>
2425 PR tree-optimization/86925
2426 * predict.c (expr_expected_value_1): When taking
2427 later predictor, assign also probability.
2428 Use fold_build2_initializer_loc in order to fold
2429 the expression in -frounding-math.
2431 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
2433 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
2435 (expand_vec_perm_1): Try the new method.
2437 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
2440 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
2441 Check whether lra_live_max_point is 0 before dividing.
2443 2018-08-14 Martin Sebor <msebor@redhat.com>
2445 PR tree-optimization/86650
2446 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
2447 (vrp_prop::check_mem_ref): Same.
2449 2018-08-13 Liu Hao <lh_mouse@126.com>
2451 * pretty-print.c (eat_esc_sequence): Swap the foreground and
2452 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
2453 and clear it thereafter, as it only works for DBCS.
2455 2018-08-13 Liu Hao <lh_mouse@126.com>
2457 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
2458 handle returned by _get_osf_handle().
2460 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2462 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
2463 for folding vec_perm.
2465 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2467 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
2468 Add support for gimple-folding of vec_pack() and vec_unpack()
2471 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2473 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
2474 vec_xst variants to the list.
2475 (rs6000_gimple_fold_builtin): Add support for folding unaligned
2476 vector loads and stores.
2478 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
2480 * config.gcc (rs6000-ibm-aix4.x): Delete.
2481 (rs6000-ibm-aix5.1): Delete.
2482 (rs6000-ibm-aix5.2): Delete.
2483 (rs6000-ibm-aix5.3): Delete.
2484 * config/rs6000/aix43.h: Delete.
2485 * config/rs6000/aix51.h: Delete.
2486 * config/rs6000/aix52.h: Delete.
2487 * config/rs6000/t-aix43: Delete.
2489 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
2491 * config/s390/s390.c (s390_decompose_constant_pool_ref):
2493 (s390_decompose_address): Factor out constant pool ref
2496 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2498 * config/nds32/nds32-predicates.c
2499 (nds32_can_use_bclr_p): Change return type as bool.
2500 (nds32_can_use_bset_p): Ditto.
2501 (nds32_can_use_btgl_p): Ditto.
2502 (nds32_can_use_bitci_p): Ditto.
2503 * config/nds32/nds32-protos.h
2504 (nds32_can_use_bclr_p): Change declaration.
2505 (nds32_can_use_bset_p): Ditto.
2506 (nds32_can_use_btgl_p): Ditto.
2507 (nds32_can_use_bitci_p): Ditto.
2509 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2511 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
2512 Support -msched-prolog-epilog option.
2513 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
2515 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2517 * common/config/nds32/nds32-common.c
2518 (nds32_option_optimization_table): Enalbe -malways-align.
2520 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2522 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
2524 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
2525 OPT_misr_secure_ case.
2526 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
2527 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
2528 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
2530 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
2531 (nds32_isr_info): New field security_level.
2532 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
2533 * config/nds32/nds32.md (return_internal): Consider critical attribute.
2534 * config/nds32/nds32.opt (misr-secure): New option.
2535 * config/nds32/nds32_init.inc: New file.
2536 * config/nds32/nds32_isr.h: New file.
2538 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
2540 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
2541 Update comment for atomic instructions.
2542 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
2543 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
2545 (atomic_loaddi): Revise fence expansion to only emit fence prior to
2546 load for __ATOMIC_SEQ_CST model.
2547 (atomic_loaddi_1): Remove float register target.
2548 (atomic_storedi): Handle CONST_INT values.
2549 (atomic_storedi_1): Remove float register source. Add special case
2551 (memory_barrier): New expander and insn.
2553 2018-08-11 Jakub Jelinek <jakub@redhat.com>
2555 PR tree-optimization/86835
2556 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
2557 new_stmt after def_gsi, make sure to insert new_square_stmt after
2558 that stmt, not 2 stmts before it.
2560 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
2563 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
2566 2018-08-10 Martin Liska <mliska@suse.cz>
2569 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
2571 * builtins.c (expand_builtin_expect_with_probability):
2573 (expand_builtin_expect_with_probability): New function.
2574 (build_builtin_expect_predicate): Add new argumnet probability
2575 for BUILT_IN_EXPECT_WITH_PROBABILITY.
2576 (fold_builtin_expect):
2579 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
2580 * builtins.h (fold_builtin_expect): Set new argument.
2581 * doc/extend.texi: Document __builtin_expect_with_probability.
2582 * doc/invoke.texi: Likewise.
2583 * gimple-fold.c (gimple_fold_call): Pass new argument.
2584 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
2585 also BUILT_IN_EXPECT_WITH_PROBABILITY.
2586 * predict.c (get_predictor_value): New function.
2587 (expr_expected_value): Add new argument probability. Assume
2588 that predictor and probability are always non-null.
2589 (expr_expected_value_1): Likewise. For __builtin_expect and
2590 __builtin_expect_with_probability set probability. Handle
2591 combination in binary expressions.
2592 (tree_predict_by_opcode): Simplify code by simply calling
2593 get_predictor_value.
2594 (pass_strip_predict_hints::execute): Add handling of
2595 BUILT_IN_EXPECT_WITH_PROBABILITY.
2596 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
2598 * tree.h (DECL_BUILT_IN_P): New function.
2600 2018-08-10 Martin Liska <mliska@suse.cz>
2602 PR tree-optimization/85799
2603 * passes.def: Add argument for pass_strip_predict_hints.
2604 * predict.c (class pass_strip_predict_hints): Add new argument
2606 (strip_predictor_early): New function.
2607 (pass_strip_predict_hints::execute): Call the function to
2609 (strip_predict_hints): New function.
2610 * predict.def: Fix comment.
2612 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
2614 * Makefile.in: Clarify which tm.texi to copy over to assert the
2615 right to grant a GFDL license for all.
2617 2018-08-09 Jeff Law <law@redhat.com>
2619 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
2622 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
2624 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
2627 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
2629 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
2630 reductions for variable-length vectors.
2632 2018-08-09 David Malcolm <dmalcolm@redhat.com>
2635 * common.opt (fdiagnostics-show-line-numbers): New option.
2636 * diagnostic-show-locus.c (class layout): Add fields
2637 "m_show_line_numbers_p" and "m_linenum_width";
2638 (num_digits): New function.
2639 (test_num_digits): New function.
2640 (layout::layout): Initialize new fields. Update m_x_offset
2641 logic to handle any left margin.
2642 (layout::print_source_line): Print line number when requested.
2643 (layout::start_annotation_line): New member function.
2644 (layout::print_annotation_line): Call it.
2645 (layout::print_leading_fixits): Likewise.
2646 (layout::print_trailing_fixits): Likewise. Update calls to
2647 move_to_column for new parameter.
2648 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
2649 it to potentially call start_annotation_line.
2650 (layout::show_ruler): Call start_annotation_line.
2651 (selftest::test_line_numbers_multiline_range): New selftest.
2652 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
2653 and selftest::test_line_numbers_multiline_range.
2654 * diagnostic.c (diagnostic_initialize): Initialize
2655 show_line_numbers_p.
2656 * diagnostic.h (struct diagnostic_context): Add field
2657 "show_line_numbers_p".
2658 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
2659 -fno-diagnostics-show-line-numbers.
2660 * dwarf2out.c (gen_producer_string): Add
2661 OPT_fdiagnostics_show_line_numbers to the ignored options.
2662 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
2663 one setting" options.
2664 (append_compiler_options): Likewise to the dropped options.
2665 (append_diag_options): Likewise to the passed-on options.
2666 * opts.c (common_handle_option): Handle the new option.
2667 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
2669 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
2671 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
2672 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
2673 third argument of type "const signed char" to descriptions of
2674 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
2675 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
2676 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
2677 __builtin_bcdsub_ov functions.
2679 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
2681 PR tree-optimization/86858
2682 * tree-vect-loop.c (vect_is_simple_reduction): Restore
2683 flow_bb_inside_loop_p calls.
2685 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
2687 PR tree-optimization/86871
2688 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
2689 instead of gimple_assign_lhs.
2691 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
2694 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
2695 register constraint to operand 0.
2696 (add<mode>3_carryinC): Likewise.
2697 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
2699 2018-08-09 Martin Liska <mliska@suse.cz>
2702 * common.opt: Remove extra line.
2704 2018-08-09 Martin Liska <mliska@suse.cz>
2706 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
2707 at the end of a line, make first letter capital and end up
2708 a sentence with a dot.
2709 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
2710 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
2711 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
2712 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
2713 (PARAM_MAX_ISL_OPERATIONS): Likewise.
2714 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
2715 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
2716 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
2717 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
2718 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
2719 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
2720 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
2721 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
2722 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
2723 (PARAM_TREE_REASSOC_WIDTH): Likewise.
2724 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
2725 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
2726 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
2728 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
2731 * config/s390/s390.c (s390_option_override_internal): Reduce the
2732 stack-clash-protection-probe-interval param if it would be too big
2735 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
2738 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
2739 * config/m68k/m68k.c (handle_move_double): Don't call
2740 m68k_final_prescan_insn.
2741 (m68k_adjust_decorated_operand): Renamed from
2742 m68k_final_prescan_insn, remove first and third operand and
2744 (print_operand): Call it.
2745 (print_operand_address): Call it.
2747 2018-08-08 Nathan Sidwell <nathan@acm.org>
2749 * diagnostic.c (diagnostic_report_current_module): Use
2750 linemap_included_from & linemap_included_from_linemap.
2752 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
2754 * config/aarch64/aarch64-cores.def: Add phecda core.
2755 * config/aarch64/aarch64-tune.md: Regenerate.
2756 * doc/invoke.texi: Add phecda core.
2758 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
2761 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
2763 * config/s390/s390.md ("movti"): Add more alternatives for
2764 constant to GPR copies.
2766 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
2768 * config/s390/s390.c: Fix whitespace damage throughout the file.
2769 * config/s390/s390.h: Likewise.
2770 * config/s390/tpf.h: Likewise.
2772 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
2774 * config/s390/s390.c (s390_loadrelative_operand_p):
2775 Remove TARGET_CPU_ZARCH usages.
2776 (s390_rtx_costs): Likewise.
2777 (s390_legitimate_constant_p): Likewise.
2778 (s390_cannot_force_const_mem): Likewise.
2779 (legitimate_reload_constant_p): Likewise.
2780 (s390_preferred_reload_class): Likewise.
2781 (legitimize_pic_address): Likewise.
2782 (legitimize_tls_address): Likewise.
2783 (s390_split_branches): Removed.
2784 (s390_add_execute): Removed.
2785 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
2786 (s390_mainpool_start): Likewise.
2787 (s390_mainpool_finish): Likewise.
2788 (s390_mainpool_cancel): Removed.
2789 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
2790 (s390_chunkify_cancel): Likewise.
2791 (s390_return_addr_rtx): Likewise.
2792 (s390_register_info): Remove split_branches_pending_p uages.
2793 (s390_optimize_register_info): Likewise.
2794 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
2795 split_branches_pending_p usages.
2796 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
2797 (s390_load_got): Likewise.
2798 (s390_expand_split_stack_prologue): Likewise.
2799 (output_asm_nops): Likewise.
2800 (s390_function_profiler): Likewise.
2801 (s390_emit_call): Likewise.
2802 (s390_conditional_register_usage): Likewise.
2803 (s390_optimize_prologue): Likewise.
2804 (s390_reorg): Remove TARGET_CPU_ZARCH and
2805 split_branches_pending_p usages.
2806 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
2808 (s390_output_indirect_thunk_function): Likewise.
2809 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
2810 (TARGET_CPU_ZARCH_P): Removed.
2811 (struct machine_function): Remove split_branches_pending_p.
2812 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
2814 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
2816 * common/config/s390/s390-common.c (processor_flags_table):
2818 * config.gcc: Remove with_arch/with_tune support.
2819 * config/s390/2064.md: Remove cpu attribute comparisons.
2820 * config/s390/driver-native.c (s390_host_detect_local_cpu):
2822 * config/s390/linux.h (ASM_SPEC):
2823 Remove -march support.
2824 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
2825 Use a table to get an arch level.
2826 * config/s390/s390-opts.h (enum processor_type):
2828 * config/s390/s390.c
2829 (processor_table): Remove entries, add arch_level values.
2830 (s390_issue_rate): Remove cases.
2831 (s390_option_override): Adjust
2832 s390_option_override_internal() call.
2833 (s390_option_override_internal): Remove deprecation warning.
2834 (s390_valid_target_attribute_tree): Adjust
2835 s390_option_override_internal() call.
2836 * config/s390/s390.h (struct s390_processor):
2837 Share with s390-c.c, add arch_level field.
2838 * config/s390/s390.md:
2839 Remove occurrences in cpu attribute.
2840 * config/s390/s390.opt: Remove -march/-mtune support.
2841 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
2842 * doc/invoke.texi: Remove deprecation warning.
2844 2018-08-08 Luis Machado <luis.machado@linaro.org>
2846 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
2848 (qdf24xx_tunings): Set vector cost structure to
2849 qdf24xx_vector_cost.
2851 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
2852 <register_sextend>: Set to 3.
2854 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
2857 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
2858 * config/aarch64/aarch64-simd.md
2859 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
2860 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
2861 (aarch64_frecpx<mode>): ...this new pattern.
2862 * config/aarch64/aarch64-simd-builtins.def: Remove comment
2863 about aarch64_frecp<FRECP:frecp_suffix><mode>.
2865 2018-08-07 Martin Liska <mliska@suse.cz>
2868 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
2869 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
2870 * predict.def (PRED_MALLOC_NONNULL): New predictor.
2871 * doc/extend.texi: Document that malloc attribute adds
2874 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
2877 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2878 Define to speculation_safe_value_not_needed.
2880 2018-08-06 Jeff Law <law@redhat.com>
2882 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
2883 the vr_values instance to cprop_into_stmt.
2884 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
2885 (cprop_operand): Also query EVRP to determine if OP is a constant.
2887 2018-08-06 Nathan Sidwell <nathan@acm.org>
2889 * diagnostic.c (diagnostic_report_current_module): Reroll
2890 included-at loop. Translate text.
2892 2018-08-06 David Malcolm <dmalcolm@redhat.com>
2894 * function-tests.c (selftest::test_expansion_to_rtl): Call
2895 free_after_compilation.
2897 2018-08-06 Alan Hayward <alan.hayward@arm.com>
2899 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
2901 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
2903 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
2904 loops with memory block operations from getting unrolled.
2906 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
2909 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2910 Define to speculation_safe_value_not_needed.
2912 2018-08-06 Jeff Law <law@redhat.com>
2914 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
2917 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
2920 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
2921 with all enabled __intN types.
2923 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
2925 2018-08-06 Alan Hayward <alan.hayward@arm.com>
2927 * alias.c (record_set): Check for clobber high.
2928 * cfgexpand.c (expand_gimple_stmt): Likewise.
2929 * combine-stack-adj.c (single_set_for_csa): Likewise.
2930 * combine.c (find_single_use_1): Likewise.
2931 (set_nonzero_bits_and_sign_copies): Likewise.
2932 (get_combine_src_dest): Likewise.
2933 (is_parallel_of_n_reg_sets): Likewise.
2934 (try_combine): Likewise.
2935 (record_dead_and_set_regs_1): Likewise.
2936 (reg_dead_at_p_1): Likewise.
2937 (reg_dead_at_p): Likewise.
2938 * dce.c (deletable_insn_p): Likewise.
2939 (mark_nonreg_stores_1): Likewise.
2940 (mark_nonreg_stores_2): Likewise.
2941 * df-scan.c (df_find_hard_reg_defs): Likewise.
2942 (df_uses_record): Likewise.
2943 (df_get_call_refs): Likewise.
2944 * dwarf2out.c (mem_loc_descriptor): Likewise.
2945 * haifa-sched.c (haifa_classify_rtx): Likewise.
2946 * ira-build.c (create_insn_allocnos): Likewise.
2947 * ira-costs.c (scan_one_insn): Likewise.
2948 * ira.c (equiv_init_movable_p): Likewise.
2949 (rtx_moveable_p): Likewise.
2950 (interesting_dest_for_shprep): Likewise.
2951 * jump.c (mark_jump_label_1): Likewise.
2952 * postreload-gcse.c (record_opr_changes): Likewise.
2953 * postreload.c (reload_cse_simplify): Likewise.
2954 (struct reg_use): Add source expr.
2955 (reload_combine): Check for clobber high.
2956 (reload_combine_note_use): Likewise.
2957 (reload_cse_move2add): Likewise.
2958 (move2add_note_store): Likewise.
2959 * print-rtl.c (print_pattern): Likewise.
2960 * recog.c (decode_asm_operands): Likewise.
2961 (store_data_bypass_p): Likewise.
2962 (if_test_bypass_p): Likewise.
2963 * regcprop.c (kill_clobbered_value): Likewise.
2964 (kill_set_value): Likewise.
2965 * reginfo.c (reg_scan_mark_refs): Likewise.
2966 * reload1.c (maybe_fix_stack_asms): Likewise.
2967 (eliminate_regs_1): Likewise.
2968 (elimination_effects): Likewise.
2969 (mark_not_eliminable): Likewise.
2970 (scan_paradoxical_subregs): Likewise.
2971 (forget_old_reloads_1): Likewise.
2972 * reorg.c (find_end_label): Likewise.
2973 (try_merge_delay_insns): Likewise.
2974 (redundant_insn): Likewise.
2975 (own_thread_p): Likewise.
2976 (fill_simple_delay_slots): Likewise.
2977 (fill_slots_from_thread): Likewise.
2978 (dbr_schedule): Likewise.
2979 * resource.c (update_live_status): Likewise.
2980 (mark_referenced_resources): Likewise.
2981 (mark_set_resources): Likewise.
2982 * rtl.c (copy_rtx): Likewise.
2983 * rtlanal.c (reg_referenced_p): Likewise.
2984 (single_set_2): Likewise.
2985 (noop_move_p): Likewise.
2986 (note_stores): Likewise.
2987 * sched-deps.c (sched_analyze_reg): Likewise.
2988 (sched_analyze_insn): Likewise.
2990 2018-08-06 Alan Hayward <alan.hayward@arm.com>
2992 * cse.c (invalidate_reg): New function extracted from...
2993 (invalidate): ...here.
2994 (canonicalize_insn): Check for clobber high.
2995 (invalidate_from_clobbers): invalidate clobber highs.
2996 (invalidate_from_sets_and_clobbers): Likewise.
2997 (count_reg_usage): Check for clobber high.
2998 (insn_live_p): Likewise.
2999 * cselib.c (cselib_expand_value_rtx_1):Likewise.
3000 (cselib_invalidate_regno): Check for clobber in setter.
3001 (cselib_invalidate_rtx): Pass through setter.
3002 (cselib_invalidate_rtx_note_stores):
3003 (cselib_process_insn): Check for clobber high.
3004 * cselib.h (cselib_invalidate_rtx): Add operand.
3006 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3008 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
3009 (mark_not_eliminable): Likewise.
3010 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
3011 * lra-lives.c (process_bb_lives): Check for clobber high.
3012 * lra.c (new_insn_reg): Remember clobber highs.
3013 (collect_non_operand_hard_regs): Check for clobber high.
3014 (lra_set_insn_recog_data): Likewise.
3015 (add_regs_to_insn_regno_info): Likewise.
3016 (lra_update_insn_regno_info): Likewise.
3018 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3020 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
3021 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
3023 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3025 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
3026 (copy_insn_1): Likewise.
3027 (gen_hard_reg_clobber_high): New gen function.
3028 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
3029 * genemit.c (gen_exp): Likewise.
3030 (gen_emit_seq): Pass through info.
3031 (gen_insn): Check for CLOBBER_HIGH.
3032 (gen_expand): Pass through info.
3033 (gen_split): Likewise.
3034 (output_add_clobbers): Likewise.
3035 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
3036 (remove_clobbers): Likewise.
3037 * rtl.h (gen_hard_reg_clobber_high): New declaration.
3039 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3041 * doc/rtl.texi (clobber_high): Add.
3042 (parallel): Add in clobber high
3043 * rtl.c (rtl_check_failed_code3): Add function.
3044 * rtl.def (CLOBBER_HIGH): Add expression.
3045 * rtl.h (RTL_CHECKC3): Add macro.
3046 (rtl_check_failed_code3): Add declaration.
3047 (XC3EXP): Add macro.
3049 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
3052 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
3053 cfun->machine->max_used_stack_alignment if needed.
3055 2018-08-04 Martin Sebor <msebor@redhat.com>
3057 PR tree-optimization/86571
3058 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
3061 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
3064 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3067 2018-08-03 Jeff Law <law@redhat.com>
3070 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3071 Define to speculation_safe_value_not_needed.
3073 2018-08-03 David Malcolm <dmalcolm@redhat.com>
3075 * doc/gcov.texi (-x): Remove duplicate "to".
3076 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
3077 (-Wif-not-aligned): Remove duplicate "is".
3078 (-flto): Remove duplicate "the".
3079 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
3080 duplicate "v5.00.b".
3081 (MSP430 Options): Remove duplicate "and" from the description
3082 of "-mgprel-sec=regexp".
3083 (x86 Options): Remove duplicate copies of "vmldLog102" and
3084 vmlsLog104 from description of "-mveclibabi=type".
3086 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
3088 * internal-fn.h (first_commutative_argument): Declare.
3089 * internal-fn.c (first_commutative_argument): New function.
3090 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
3091 restrictions for pattern statements. Use first_commutative_argument
3092 to look for commutative operands in calls to internal functions.
3094 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
3096 * Makefile.in (wide-int-range.o): New.
3097 * tree-vrp.c: Move all the wide_int_* functions to...
3098 * wide-int-range.cc: ...here.
3099 * tree-vrp.h: Move all the wide_int_* prototypes to...
3100 * wide-int-range.h: ...here.
3102 2018-08-03 Tom de Vries <tdevries@suse.de>
3104 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
3106 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
3107 * except.c (output_function_exception_table): Do early exit if
3108 targetm_common.except_unwind_info (&global_options) == UI_NONE.
3110 2018-08-03 Martin Liska <mliska@suse.cz>
3112 * predict.c (dump_prediction): Change to 2 digits
3115 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
3117 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
3119 (falkor_am_1_gtov_gtov): ... a new insn reservation.
3121 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
3123 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
3124 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
3125 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
3126 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
3127 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
3129 2018-08-02 David Malcolm <dmalcolm@redhat.com>
3131 * diagnostic-show-locus.c (diagnostic_show_locus): Use
3132 pp_take_prefix when saving the existing prefix.
3133 * diagnostic.c (diagnostic_append_note): Likewise.
3134 * langhooks.c (lhd_print_error_function): Likewise.
3135 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
3136 param's type. Free the existing prefix.
3137 (pp_take_prefix): New function.
3138 (pretty_printer::pretty_printer): Drop the prefix parameter.
3139 Rename the length parameter to match the comment.
3140 (pretty_printer::~pretty_printer): Free the prefix.
3141 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
3143 (struct pretty_printer): Drop the "const" from "prefix" field's
3144 type and clarify memory management.
3145 (pp_set_prefix): Drop the "const" from the 2nd param.
3146 (pp_take_prefix): New decl.
3148 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
3150 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
3151 for word_mode_ok here instead of passing as argument.
3152 (expand_block_compare): Change select_block_compare_mode() call.
3153 (expand_strncmp_gpr_sequence): New function.
3154 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
3156 2018-08-02 Jeff Law <law@redhat.com>
3159 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3160 Define to speculation_safe_value_not_needed.
3163 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3164 Define to speculation_safe_value_not_needed.
3166 2018-08-02 Tom de Vries <tdevries@suse.de>
3169 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
3170 function. Return UI_TARGET unconditionally.
3171 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
3172 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
3174 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3176 * genemit.c (print_overload_test): Fix typo.
3178 2018-08-02 Richard Biener <rguenther@suse.de>
3180 PR tree-optimization/86816
3181 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
3182 which checks for value availability before querying it.
3183 (gvn_uses_equal): Use it.
3184 (same_succ_hash): Likewise.
3185 (gimple_equal_p): Likewise.
3187 2018-08-02 Nick Clifton <nickc@redhat.com>
3190 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3191 Define to speculation_safe_value_not_needed.
3194 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3195 Define to speculation_safe_value_not_needed.
3198 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3199 Define to speculation_safe_value_not_needed.
3202 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3203 Define to speculation_safe_value_not_needed.
3206 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3207 Define to speculation_safe_value_not_needed.
3210 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3211 Define to speculation_safe_value_not_needed.
3214 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3215 Define to speculation_safe_value_not_needed.
3218 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3219 Define to speculation_safe_value_not_needed.
3222 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
3223 speculation_safe_value_not_needed.
3226 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
3227 to speculation_safe_value_not_needed.
3229 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3231 * doc/md.texi: Expand the documentation of instruction names
3232 to mention port-local uses. Document '@' in pattern names.
3233 * read-md.h (overloaded_instance, overloaded_name): New structs.
3235 (md_reader::handle_overloaded_name): New member function.
3236 (md_reader::get_overloads): Likewise.
3237 (md_reader::m_first_overload): New member variable.
3238 (md_reader::m_next_overload_ptr): Likewise.
3239 (md_reader::m_overloads_htab): Likewise.
3240 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
3241 m_next_overload_ptr and m_overloads_htab.
3242 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
3243 (get_mode_token, get_code_token, get_int_token): New functions.
3244 (map_attr_string): Add an optional argument that passes back
3245 the associated iterator.
3246 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
3247 (md_reader::handle_overloaded_name, add_overload_instance): New
3249 (apply_iterators): Handle '@' names. Report an error if '@'
3250 is used without iterators.
3251 (initialize_iterators): Initialize the new iterator_group fields.
3252 * genopinit.c (handle_overloaded_code_for)
3253 (handle_overloaded_gen): New functions.
3254 (main): Use them to print declarations of maybe_code_for_* and
3255 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
3256 * genemit.c (print_overload_arguments, print_overload_test)
3257 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
3258 (main): Use it to print definitions of maybe_code_for_* and
3259 maybe_gen_* functions.
3260 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
3261 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
3262 instead of explicit mode checks.
3263 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
3264 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
3265 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
3266 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
3267 (aarch64_expand_compare_and_swap): Likewise
3268 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
3269 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
3270 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
3271 (aarch64_constant_pool_reload_icode): Delete.
3272 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
3273 instead of aarch64_constant_pool_reload_icode. Use
3274 code_for_aarch64_reload_mov instead of explicit mode checks.
3275 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
3276 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
3277 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
3278 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
3279 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
3280 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
3281 (aarch64_atomic_load_op_code): Delete.
3282 (aarch64_emit_atomic_load_op): Likewise.
3283 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
3284 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
3285 instead of aarch64_emit_atomic_load_op.
3286 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
3287 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
3288 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
3289 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
3290 character before the pattern name.
3291 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
3292 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
3293 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
3294 (aarch64_frecps<mode>): Likewise.
3295 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
3296 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
3297 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
3298 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
3299 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
3301 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3303 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
3304 Allow HFmode constants if TARGET_FP_F16INST.
3306 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
3309 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
3310 No longer check last store for clobber of address register.
3312 2018-08-02 Martin Liska <mliska@suse.cz>
3314 PR gcov-profile/86817
3315 * gcov.c (process_all_functions): New function.
3317 (process_file): Move functions processing to
3318 process_all_functions.
3320 2018-08-02 David Malcolm <dmalcolm@redhat.com>
3322 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
3323 "const" to the "gimple *" and "rtx_insn *" parameters.
3324 * dumpfile.h (dump_user_location_t::dump_user_location_t):
3326 (dump_location_t::dump_location_t): Likewise.
3328 2018-08-01 Martin Sebor <msebor@redhat.com>
3330 PR tree-optimization/86650
3331 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
3332 rather than a "gcall *". Directly pass the data of interest
3333 to percent_K_format, rather than building a temporary CALL_EXPR
3335 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
3336 (gimple_fold_builtin_strncat): Adjust.
3337 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
3338 gcall* argument with gimple*.
3339 * gimple-ssa-warn-restrict.c (check_call): Same.
3340 (wrestrict_dom_walker::before_dom_children): Same.
3341 (builtin_access::builtin_access): Same.
3342 (check_bounds_or_overlap): Same
3343 (maybe_diag_overlap): Same.
3344 (maybe_diag_offset_bounds): Same.
3345 * tree-diagnostic.c (default_tree_printer): Move usage of
3346 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
3348 * tree-pretty-print.c (percent_K_format): Add argument.
3349 * tree-pretty-print.h: Add argument.
3350 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
3351 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
3352 (maybe_diag_stxncpy_trunc): Same.
3353 (handle_builtin_stxncpy): Same.
3354 (handle_builtin_strcat): Same.
3356 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3358 * match.pd: Optimise pointer range checks.
3360 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3362 PR tree-optimization/86758
3363 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
3364 to remove pattern statements.
3366 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3368 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
3369 result of dfs_enumerate_from when constructing stmt_vec_infos,
3370 instead of additionally calling get_loop_body.
3372 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3374 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
3376 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
3377 When creating an iv, assert that the step is not known to be zero.
3378 (vect_setup_realignment): Update call accordingly.
3379 * tree-vect-stmts.c (vectorizable_store): Likewise.
3380 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
3382 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3384 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
3385 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
3386 (vectorizable_reduction): Likewise.
3387 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
3388 (vect_detect_hybrid_slp_stmts): Likewise.
3389 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
3391 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
3393 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
3394 (wide_int_set_zero_nonzero_bits): ...this.
3395 (zero_nonzero_bits_from_vr): Rename to...
3396 (vrp_set_zero_nonzero_bits): ...this.
3397 (extract_range_from_multiplicative_op_1): Abstract wide int
3399 (wide_int_range_multiplicative_op): ...here.
3400 (extract_range_from_binary_expr_1): Extract wide int binary
3401 operations into their own functions.
3402 (wide_int_range_lshift): New.
3403 (wide_int_range_can_optimize_bit_op): New.
3404 (wide_int_range_shift_undefined_p): New.
3405 (wide_int_range_bit_xor): New.
3406 (wide_int_range_bit_ior): New.
3407 (wide_int_range_bit_and): New.
3408 (wide_int_range_trunc_mod): New.
3409 (extract_range_into_wide_ints): New.
3410 (vrp_shift_undefined_p): New.
3411 (extract_range_from_multiplicative_op): New.
3412 (vrp_can_optimize_bit_op): New.
3413 * tree-vrp.h (value_range::dump): New.
3414 (wide_int_range_multiplicative_op): New.
3415 (wide_int_range_lshift):New.
3416 (wide_int_range_shift_undefined_p): New.
3417 (wide_int_range_bit_xor): New.
3418 (wide_int_range_bit_ior): New.
3419 (wide_int_range_bit_and): New.
3420 (wide_int_range_trunc_mod): New.
3421 (zero_nonzero_bits_from_bounds): Rename to...
3422 (wide_int_set_zero_nonzero_bits): ...this.
3423 (zero_nonzero_bits_from_vr): Rename to...
3424 (vrp_set_zero_nonzero_bits): ...this.
3425 (range_easy_mask_min_max): Rename to...
3426 (wide_int_range_can_optimize_bit_op): this.
3427 * vr-values.c (simplify_bit_ops_using_ranges): Rename
3428 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
3430 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3432 * tree-vectorizer.h (vect_orig_stmt): New function.
3433 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
3434 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
3435 (vect_create_epilog_for_reduction): Likewise.
3436 (vectorizable_live_operation): Likewise.
3437 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
3438 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
3439 * tree-vect-stmts.c (vectorizable_call): Likewise.
3440 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
3442 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3444 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
3446 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
3447 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
3448 (vect_transform_loop): Likewise.
3449 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3451 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3453 * tree-vectorizer.h (vect_schedule_slp): Return void.
3454 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3455 (vect_schedule_slp): Likewise.
3457 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3459 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
3461 (vect_transform_loop): Update calls accordingly. Schedule SLP
3462 instances before the main loop, if any exist.
3464 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3466 PR tree-optimization/86749
3467 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
3468 If the lhs is used in a COND_EXPR, check that it is being used
3469 as the "then" or "else" value.
3471 2018-08-01 Tom de Vries <tdevries@suse.de>
3474 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
3475 speculation_safe_value_not_needed.
3477 2018-08-01 Richard Biener <rguenther@suse.de>
3479 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
3482 2018-08-01 Martin Liska <mliska@suse.cz>
3484 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
3485 format how successful transformation is dumped.
3486 (gimple_mod_pow2_value_transform): Likewise.
3487 (gimple_mod_subtract_transform): Likewise.
3488 (gimple_stringops_transform): Likewise.
3490 2018-08-01 Martin Liska <mliska@suse.cz>
3493 * value-prof.c (interesting_stringop_to_profile_p):
3494 Simplify the code and add BUILT_IN_MEMMOVE.
3495 (gimple_stringops_transform): Likewise.
3497 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
3499 * config/aarch64/aarch64-simd.md
3500 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
3501 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
3502 use GPI iterator instead of SI mode.
3504 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
3506 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
3507 rs6000_speculation_barrier.
3508 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
3509 new barrier pattern name.
3511 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
3513 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
3514 (speculation_barrier): New insn.
3516 2018-08-01 Richard Biener <rguenther@suse.de>
3519 * graphite.h: Include isl/id.h and isl/space.h to allow build
3522 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
3525 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
3526 mode for COFF targets.
3527 * defaults.h (TARGET_COFF): Define.
3528 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
3529 TARGET_COFF): Define.
3530 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
3531 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
3532 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
3534 2018-07-31 Alexandre Oliva <oliva@adacore.com>
3536 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
3537 * gimple-streamer-out.c (output_bb): Save it.
3538 * lto-streamer-in.c (input_struct_function_base): Restore
3539 instance discriminator if available. Create map on demand.
3540 * lto-streamer-out.c (output_struct_function_base): Save it if
3542 * final.c (decl_to_instance_map): Document LTO strategy.
3544 2018-07-31 Alexandre Oliva <oliva@adacore.com>
3545 Olivier Hainque <hainque@adacore.com>
3547 * debug.h (decl_to_instance_map_t): New type.
3548 (decl_to_instance_map): Declare.
3549 (maybe_create_decl_to_instance_map): New inline function.
3550 * final.c (bb_discriminator, last_bb_discriminator): New statics,
3551 to track basic block discriminators.
3552 (final_start_function_1): Initialize them.
3553 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
3555 (decl_to_instance_map): New variable.
3556 (map_decl_to_instance, maybe_set_discriminator): New functions.
3557 (notice_source_line): Set discriminator.
3559 2018-07-31 Ian Lance Taylor <iant@golang.org>
3561 * targhooks.c (default_have_speculation_safe_value): Add
3564 2018-07-31 David Malcolm <dmalcolm@redhat.com>
3566 * dump-context.h: Include "pretty-print.h".
3567 (dump_context::refresh_dumps_are_enabled): New decl.
3568 (dump_context::emit_item): New decl.
3569 (class dump_context): Add fields "m_test_pp" and
3571 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
3572 (temp_dump_context::get_dumped_text): New decl.
3573 (class temp_dump_context): Add field "m_pp".
3574 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
3575 (dump_context::refresh_dumps_are_enabled): ...and add a test for
3577 (set_dump_file): Update for above change.
3578 (set_alt_dump_file): Likewise.
3579 (dump_loc): New overload, taking a pretty_printer *.
3580 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
3581 to any test pretty-printer.
3582 (make_item_for_dump_gimple_stmt): New function, adapted from
3583 optinfo::add_gimple_stmt.
3584 (dump_context::dump_gimple_stmt): Call it, and use the result,
3585 eliminating the direct usage of dump_file and alt_dump_file in
3586 favor of indirectly using them via emit_item.
3587 (make_item_for_dump_gimple_expr): New function, adapted from
3588 optinfo::add_gimple_expr.
3589 (dump_context::dump_gimple_expr): Call it, and use the result,
3590 eliminating the direct usage of dump_file and alt_dump_file in
3591 favor of indirectly using them via emit_item.
3592 (make_item_for_dump_generic_expr): New function, adapted from
3594 (dump_context::dump_generic_expr): Call it, and use the result,
3595 eliminating the direct usage of dump_file and alt_dump_file in
3596 favor of indirectly using them via emit_item.
3597 (make_item_for_dump_printf_va): New function, adapted from
3598 optinfo::add_printf_va.
3599 (make_item_for_dump_printf): New function.
3600 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
3601 and use the result, eliminating the direct usage of dump_file and
3602 alt_dump_file in favor of indirectly using them via emit_item.
3603 (make_item_for_dump_dec): New function.
3604 (dump_context::dump_dec): Call it, and use the result,
3605 eliminating the direct usage of dump_file and alt_dump_file in
3606 favor of indirectly using them via emit_item.
3607 (make_item_for_dump_symtab_node): New function, adapted from
3608 optinfo::add_symtab_node.
3609 (dump_context::dump_symtab_node): Call it, and use the result,
3610 eliminating the direct usage of dump_file and alt_dump_file in
3611 favor of indirectly using them via emit_item.
3612 (dump_context::begin_scope): Reimplement, avoiding direct usage
3613 of dump_file and alt_dump_file in favor of indirectly using them
3615 (dump_context::emit_item): New member function.
3616 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
3617 Set up test pretty-printer on the underlying context. Call
3618 refresh_dumps_are_enabled.
3619 (temp_dump_context::~temp_dump_context): Call
3620 refresh_dumps_are_enabled.
3621 (temp_dump_context::get_dumped_text): New member function.
3622 (selftest::verify_dumped_text): New function.
3623 (ASSERT_DUMPED_TEXT_EQ): New macro.
3624 (selftest::test_capture_of_dump_calls): Run all tests twice, with
3625 and then without optinfo enabled. Add uses of
3626 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
3627 * dumpfile.h: Update comment for the dump_* API.
3628 * optinfo-emit-json.cc
3629 (selftest::test_building_json_from_dump_calls): Update for new
3630 param for temp_dump_context ctor.
3631 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
3632 and "m_owned" field.
3633 (optinfo_item::~optinfo_item): Likewise.
3634 (optinfo::add_item): New member function.
3635 (optinfo::emit): Update comment.
3636 (optinfo::add_string): Delete.
3637 (optinfo::add_printf): Delete.
3638 (optinfo::add_printf_va): Delete.
3639 (optinfo::add_gimple_stmt): Delete.
3640 (optinfo::add_gimple_expr): Delete.
3641 (optinfo::add_tree): Delete.
3642 (optinfo::add_symtab_node): Delete.
3643 (optinfo::add_dec): Delete.
3644 * optinfo.h (class dump_context): New forward decl.
3645 (optinfo::add_item): New decl.
3646 (optinfo::add_string): Delete.
3647 (optinfo::add_printf): Delete.
3648 (optinfo::add_printf_va): Delete.
3649 (optinfo::add_gimple_stmt): Delete.
3650 (optinfo::add_gimple_expr): Delete.
3651 (optinfo::add_tree): Delete.
3652 (optinfo::add_symtab_node): Delete.
3653 (optinfo::add_dec): Delete.
3654 (optinfo::add_poly_int): Delete.
3655 (optinfo_item::optinfo_item): Remove "owned" param.
3656 (class optinfo_item): Remove field "m_owned".
3658 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3661 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
3662 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
3663 requested variable alignment.
3664 (expand_one_ssa_partition): Likewise.
3665 (expand_one_var): Likewise.
3667 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3669 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
3670 to speculation_safe_value_not_needed.
3672 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3674 * targhooks.h (speculation_safe_value_not_needed): New prototype.
3675 * targhooks.c (speculation_safe_value_not_needed): New function.
3676 * target.def (have_speculation_safe_value): Update documentation.
3677 * doc/tm.texi: Regenerated.
3679 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3681 * config/aarch64/iterators.md (ALLI_TI): New iterator.
3682 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
3684 (despeculate_copy<ALLI:mode>_insn): New insn.
3685 (despeculate_copyti_insn): New insn.
3686 (despeculate_simple<ALLI:mode>): New insn
3687 (despeculate_simpleti): New insn.
3688 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
3690 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
3691 aarch64_speculation_safe_value.
3692 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
3694 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3696 * config/aarch64/aarch64-speculation.cc: New file.
3697 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
3698 before pass_reorder_blocks.
3699 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
3701 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
3702 X14 and X15 when tracking speculation.
3703 * config/aarch64/aarch64.md (register name constants): Add
3704 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
3705 (unspec): Add UNSPEC_SPECULATION_TRACKER.
3706 (speculation_barrier): New insn attribute.
3707 (cmp<mode>): Allow SP in comparisons.
3708 (speculation_tracker): New insn.
3709 (speculation_barrier): Add speculation_barrier attribute.
3710 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
3711 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
3712 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
3714 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3716 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
3717 aarch64_track_speculation is true.
3718 (tb<optab><mode>1): Likewise.
3719 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
3720 generate CB[N]Z when tracking speculation.
3721 (aarch64_split_compare_and_swap): Likewise.
3722 (aarch64_split_atomic_op): Likewise.
3724 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3726 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
3728 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3730 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
3731 (speculation_barrier): New insn.
3733 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3735 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
3736 * config/arm/arm.md (speculation_barrier): New expand.
3737 (speculation_barrier_insn): New pattern.
3739 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3741 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
3742 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
3743 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
3744 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
3746 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
3747 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
3748 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
3749 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
3750 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
3751 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
3752 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
3753 * builtins.c (expand_speculation_safe_value): New function.
3754 (expand_builtin): Call it.
3755 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
3756 * doc/extend.texi: Document __builtin_speculation_safe_value.
3757 * doc/md.texi: Document "speculation_barrier" pattern.
3758 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
3759 TARGET_HAVE_SPECULATION_SAFE_VALUE.
3760 * doc/tm.texi: Regenerated.
3761 * target.def (have_speculation_safe_value, speculation_safe_value): New
3763 * targhooks.c (default_have_speculation_safe_value): New function.
3764 (default_speculation_safe_value): New function.
3765 * targhooks.h (default_have_speculation_safe_value): Add prototype.
3766 (default_speculation_safe_value): Add prototype.
3768 2018-07-31 David Malcolm <dmalcolm@redhat.com>
3770 * dump-context.h (dump_context::dump_loc): New decl.
3771 * dumpfile.c (dump_context::dump_loc): New member function.
3772 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
3773 and dump_gimple_stmt.
3774 (dump_context::dump_gimple_expr_loc): Likewise, using
3776 (dump_context::dump_generic_expr_loc): Likewise, using
3778 (dump_context::dump_printf_loc_va): Likewise, using
3780 (dump_context::begin_scope): Explicitly using the global function
3781 "dump_loc", rather than the member function.
3783 2018-07-31 Martin Sebor <msebor@redhat.com>
3785 PR tree-optimization/86741
3786 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
3788 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
3790 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
3791 depend on whether prefetch instructions will be emitted or not.
3792 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
3793 will be emitted or not.
3794 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
3795 (TARGET_SETMEM_PFD): New macros.
3797 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3799 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
3800 (NULL_STMT_VEC_INFO): Delete.
3801 (stmt_vec_info::operator*): Likewise.
3802 (stmt_vec_info::operator gimple *): Likewise.
3803 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
3804 of NULL_STMT_VEC_INFO.
3805 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
3806 (vect_reassociating_reduction_p): Likewise.
3807 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
3808 (vectorizable_store): Likewise.
3809 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
3810 (vec_info::free_stmt_vec_infos): Likewise.
3812 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3814 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
3815 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
3817 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3819 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
3820 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
3821 (vec_info::free_stmt_vec_info): New private member functions.
3822 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
3823 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
3824 * tree-parloops.c (gather_scalar_reductions): Remove calls to
3825 set_stmt_vec_info_vec and free_stmt_vec_infos.
3826 * tree-vect-loop.c (_loop_vec_info): Remove call to
3827 set_stmt_vec_info_vec.
3828 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
3829 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
3830 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
3831 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
3832 (vec_info::free_stmt_vec_info): ...these new functions. Remove
3833 assignments in {vec_info::,}new_stmt_vec_info that are redundant
3834 with the clearing in the xcalloc.
3835 (stmt_vec_info_vec): Delete.
3836 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
3837 (vectorize_loops): Likewise.
3838 (vec_info::~vec_info): Remove argument from call to
3839 free_stmt_vec_infos.
3840 (vec_info::add_stmt): Remove vinfo argument from call to
3843 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3845 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
3846 rather than a gimple stmt.
3847 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
3848 information for pattern statements when passed the original
3849 statement; instead wait to be passed the pattern statement itself.
3850 Don't call set_vinfo_for_stmt here.
3851 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
3852 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
3853 stmt_vec_infos here.
3854 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
3855 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
3856 stmt_vec_infos entry.
3858 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3860 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
3861 * tree-vectorizer.c (vec_info::replace_stmt): New function.
3862 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
3863 * tree-vect-stmts.c (vectorizable_call): Likewise.
3864 (vectorizable_simd_clone_call): Likewise.
3866 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3868 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
3869 * tree-vectorizer.c (vec_info::remove_stmt): New function.
3870 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
3871 * tree-vect-loop.c (vect_transform_loop): Likewise.
3872 * tree-vect-slp.c (vect_schedule_slp): Likewise.
3873 * tree-vect-stmts.c (vect_remove_stores): Likewise.
3875 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3877 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
3878 (vect_dr_stmt): Delete.
3879 * tree-vectorizer.c (vec_info::lookup_dr): New function.
3880 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
3882 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
3883 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
3884 (vect_verify_datarefs_alignment, vect_peeling_supportable)
3885 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
3886 (vect_analyze_data_refs): Likewise.
3887 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
3889 (vect_find_same_alignment_drs): Likewise.
3890 (vect_slp_analyze_node_dependences): Update calls accordingly.
3891 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
3892 instead of DR_VECT_AUX.
3893 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
3894 of a vector data references. Use vec_info::lookup_dr instead of
3896 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
3897 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
3898 instead of DR_VECT_AUX.
3900 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3902 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
3904 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
3906 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
3907 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
3908 (vect_gen_prolog_loop_niters): Likewise.
3910 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3912 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
3913 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
3914 (vect_known_alignment_in_bytes, vect_dr_behavior)
3915 (vect_get_scalar_dr_size): Take references as dr_vec_infos
3916 instead of data_references. Update calls to other routines for
3917 which the same change has been made.
3918 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
3919 dr_vec_infos instead of stmt_vec_infos.
3920 (vect_analyze_data_ref_dependence): Update call accordingly.
3921 (vect_slp_analyze_data_ref_dependence)
3922 (vect_record_base_alignments): Use DR_VECT_AUX.
3923 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
3924 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
3925 (vector_alignment_reachable_p, vect_get_data_access_cost)
3926 (vect_peeling_supportable, vect_analyze_group_access_1)
3927 (vect_analyze_group_access, vect_analyze_data_ref_access)
3928 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
3929 (vect_compile_time_alias, vect_small_gap_p)
3930 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
3931 (vect_supportable_dr_alignment): Take references as dr_vec_infos
3932 instead of data_references. Update calls to other routines for
3933 which the same change has been made.
3934 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
3935 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
3936 (vect_slp_analyze_and_verify_node_alignment)
3937 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
3938 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
3939 (vect_setup_realignment): Use dr_vec_infos. Update calls after
3941 (_vect_peel_info::dr): Replace with...
3942 (_vect_peel_info::dr_info): ...this new field.
3943 (vect_peeling_hash_get_most_frequent)
3944 (vect_peeling_hash_choose_best_peeling): Update accordingly.
3945 (vect_peeling_hash_get_lowest_cost):
3946 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
3947 routines for which the same change has been made.
3948 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
3950 * tree-vect-loop-manip.c (get_misalign_in_elems)
3951 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
3953 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
3954 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
3955 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
3956 (get_group_load_store_type, get_negative_load_store_type)
3957 (vect_get_data_ptr_increment, vectorizable_store)
3958 (vectorizable_load): Likewise.
3959 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
3960 Update calls to other routines for which the same change has been made.
3962 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3964 * tree-vectorizer.h (vec_info::move_dr): New member function.
3965 (dataref_aux): Rename to...
3966 (dr_vec_info): ...this and add "dr" and "stmt" fields.
3967 (_stmt_vec_info::dr_aux): Update accordingly.
3968 (_stmt_vec_info::data_ref_info): Delete.
3969 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
3970 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
3971 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
3972 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
3974 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
3975 (STMT_VINFO_DR_INFO): New macro.
3976 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
3977 (set_dr_misalignment): Update after rename of dataref_aux.
3978 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
3979 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
3980 initialization of STMT_VINFO_DATA_REF.
3981 * tree-vectorizer.c (vec_info::move_dr): New function.
3982 * tree-vect-patterns.c (vect_recog_bool_pattern)
3983 (vect_recog_mask_conversion_pattern)
3984 (vect_recog_gather_scatter_pattern): Use it.
3985 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
3986 the "dr" and "stmt" fields of dr_vec_info instead of
3987 STMT_VINFO_DATA_REF.
3989 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3991 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
3992 (is_pattern_stmt_p): Use it.
3993 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
3994 on pattern statements.
3996 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3998 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
3999 original stmt as a stmt_vec_info rather than a gimple stmt.
4000 (vect_pattern_recog_1): Take the statement directly as a
4001 stmt_vec_info, rather than via a gimple_stmt_iterator.
4002 Update call to vect_mark_pattern_stmts.
4003 (vect_pattern_recog): Update calls accordingly.
4005 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4007 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
4008 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
4009 a vect_def_type for the first argument.
4010 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
4011 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
4012 operand if it isn't defined by a vectorized statement.
4013 (vect_build_gather_load_calls): Remove the mask_dt argument and
4014 update calls to vect_get_vec_def_for_stmt_copy.
4015 (vectorizable_bswap): Likewise the dt argument.
4016 (vectorizable_call): Update calls to vectorizable_bswap and
4017 vect_get_vec_def_for_stmt_copy.
4018 (vectorizable_simd_clone_call, vectorizable_assignment)
4019 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
4020 (vectorizable_comparison): Update calls to
4021 vect_get_vec_def_for_stmt_copy.
4022 (vectorizable_store): Likewise. Remove now-unnecessary calls to
4024 (vect_get_loop_based_defs): Remove dt argument and update call
4025 to vect_get_vec_def_for_stmt_copy.
4026 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
4027 and vect_get_vec_def_for_stmt_copy.
4028 (vectorizable_load): Update calls to vect_build_gather_load_calls
4029 and vect_get_vec_def_for_stmt_copy.
4030 * tree-vect-loop.c (vect_create_epilog_for_reduction)
4031 (vectorizable_reduction, vectorizable_live_operation): Update calls
4032 to vect_get_vec_def_for_stmt_copy.
4034 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4036 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
4037 and gimple stmt arguments with a stmt_vec_info.
4038 (vect_record_base_alignments): Update calls accordingly.
4039 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
4040 and gimple stmt arguments with a stmt_vec_info.
4041 (vect_build_slp_tree_1): Remove vinfo argument and update call
4042 to vect_record_max_nunits.
4043 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
4044 and vect_record_max_nunits.
4046 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4048 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
4049 file and take a stmt_vec_info instead of a gimple stmt.
4050 (supportable_widening_operation, vect_finish_replace_stmt)
4051 (vect_finish_stmt_generation, vect_get_store_rhs)
4052 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
4053 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
4054 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
4055 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
4056 (vect_create_data_ref_ptr, bump_vector_ptr)
4057 (vect_permute_store_chain, vect_setup_realignment)
4058 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
4059 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
4060 (vectorizable_reduction, vectorizable_induction)
4061 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
4062 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
4063 than gimple stmts as arguments.
4064 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
4065 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
4066 (can_group_stmts_p, vect_check_gather_scatter)
4067 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
4068 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
4069 (vect_permute_load_chain, vect_shift_permute_load_chain)
4070 (vect_transform_grouped_load)
4071 (vect_record_grouped_load_vectors): Likewise.
4072 * tree-vect-loop.c (vect_fixup_reduc_chain)
4073 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
4074 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
4075 (vectorizable_reduction, vectorizable_induction)
4076 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
4077 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
4078 (vect_get_load_store_mask): Likewise.
4079 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
4080 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
4081 * tree-vect-stmts.c (vect_mark_relevant)
4082 (is_simple_and_all_uses_invariant)
4083 (exist_non_indexing_operands_for_use_p, process_use)
4084 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
4085 (vect_get_vec_def_for_operand, vect_get_vec_defs)
4086 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
4087 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
4088 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
4089 (get_negative_load_store_type, get_load_store_type)
4090 (vect_check_load_store_mask, vect_check_store_rhs)
4091 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
4092 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4093 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4094 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4095 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
4096 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
4097 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
4098 (supportable_widening_operation): Likewise.
4100 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4102 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
4103 a stmt_vec_info instead of a gcall.
4104 (vect_check_gather_scatter): Update call accordingly.
4105 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
4107 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
4108 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
4109 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
4110 instead of a gimple stmt.
4111 (vect_transform_loop): Update calls accordingly.
4112 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
4113 stmt_vec_infos instead of gimple stmts.
4114 (vect_analyze_slp_instance): Update use accordingly.
4115 * tree-vect-stmts.c (read_vector_array, write_vector_array)
4116 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
4117 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
4118 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
4119 (vect_gen_widened_results_half, vect_get_loop_based_defs)
4120 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
4121 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
4122 down to subroutines.
4124 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4126 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
4127 of the worklist from a vector of gimple stmts to a vector of
4129 * tree-vect-stmts.c (vect_mark_relevant, process_use)
4130 (vect_mark_stmts_to_be_vectorized): Likewise
4132 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4134 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
4135 statement before passing it to vect_analyze_stmt.
4136 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
4137 the chain of phi vector definitions. Track the exit phi via its
4139 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
4140 STMT_VINFO_REDUC_DEF.
4141 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
4142 stmt_vec_infos to handle the statement chains.
4143 (vect_get_slp_defs): Record the first statement in the node
4144 using a stmt_vec_info.
4145 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
4146 statements here and pass their stmt_vec_info down to subroutines.
4147 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
4148 down to vect_finish_stmt_generation.
4149 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
4150 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
4151 stmt_vec_infos to subroutines.
4152 (vect_remove_stores): Use stmt_vec_infos to handle the statement
4155 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4157 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
4158 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
4159 (vect_permute_store_chain, vect_setup_realignment)
4160 (vect_permute_load_chain, vect_shift_permute_load_chain)
4161 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
4162 stmts internally, and when passing values to other vectorizer routines.
4163 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
4164 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
4165 (vect_analyze_loop_operations, get_initial_def_for_reduction)
4166 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
4167 (vectorizable_reduction, vectorizable_induction)
4168 (vectorizable_live_operation, vect_transform_loop_stmt)
4169 (vect_transform_loop): Likewise.
4170 * tree-vect-patterns.c (vect_reassociating_reduction_p)
4171 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
4172 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
4173 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4174 (vect_slp_analyze_node_operations_1): Likewise.
4175 * tree-vect-stmts.c (vect_mark_relevant, process_use)
4176 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
4177 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
4178 (vect_finish_stmt_generation_1, get_group_load_store_type)
4179 (get_load_store_type, vect_build_gather_load_calls)
4180 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4181 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4182 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4183 (vectorizable_store, vectorizable_load, vectorizable_condition)
4184 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
4185 (supportable_widening_operation): Likewise.
4186 (vect_get_vector_types_for_stmt): Likewise.
4187 * tree-vectorizer.h (vect_dr_behavior): Likewise.
4189 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4191 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
4192 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
4193 (vect_permute_store_chain, vect_permute_load_chain)
4194 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
4195 repeated stmt_vec_info lookups.
4196 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
4197 (vect_update_ivs_after_vectorizer): Likewise.
4198 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
4199 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
4200 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
4201 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4202 (vect_bb_slp_scalar_cost): Likewise.
4203 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
4205 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4207 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
4208 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
4209 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
4210 to get gassigns and gcalls, rather than operating on generc gimple
4212 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
4213 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
4214 (vectorizable_load, vect_analyze_stmt): Likewise.
4215 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
4217 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4219 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
4220 return stmt_vec_infos rather than gimple stmts. Do not accept
4222 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
4224 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
4225 Update use of get_later_stmt.
4226 (vect_get_constant_vectors): Update call accordingly.
4227 (vect_schedule_slp_instance): Likewise
4228 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
4229 (vect_slp_analyze_instance_dependence): Likewise.
4230 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
4232 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4234 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
4235 (stmt_info_for_cost::stmt_info): ...this new field.
4236 (add_stmt_costs): Update accordingly.
4237 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
4238 (vect_get_known_peeling_cost): Likewise.
4239 (vect_estimate_min_profitable_iters): Likewise.
4240 * tree-vect-stmts.c (record_stmt_cost): Likewise.
4242 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4244 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
4245 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4246 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
4248 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
4250 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4252 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
4253 a gimple stmt to a stmt_vec_info.
4254 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
4256 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4258 * tree-vectorizer.h (vec_info::grouped_stores): Change from
4259 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4260 (_loop_vec_info::reduction_chains): Likewise.
4261 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
4263 * tree-vect-slp.c (vect_analyze_slp): Likewise.
4265 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4267 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
4268 a gimple stmt to a stmt_vec_info.
4269 (_stmt_vec_info::next_element): Likewise.
4270 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
4271 (vect_slp_analyze_and_verify_node_alignment)
4272 (vect_analyze_group_access_1, vect_analyze_group_access)
4273 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
4274 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
4275 (vect_supportable_dr_alignment): Update accordingly.
4276 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
4277 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
4278 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
4279 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
4280 * tree-vect-slp.c (vect_build_slp_tree_1)
4281 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
4282 (vect_split_slp_store_group, vect_analyze_slp_instance)
4283 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
4284 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
4285 (get_group_load_store_type, get_load_store_type)
4286 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
4287 (vect_transform_stmt, vect_remove_stores): Likewise.
4289 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4291 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
4293 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
4294 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
4295 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
4296 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
4297 (vector_alignment_reachable_p, vect_get_data_access_cost)
4298 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
4299 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
4300 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
4301 (vect_analyze_group_access_1, vect_analyze_group_access)
4302 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
4303 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
4304 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
4305 result of vect_dr_stmt and use the stmt_vec_info instead of
4306 the associated gimple stmt.
4307 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
4308 (vect_gen_prolog_loop_niters): Likewise.
4309 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
4311 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4313 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
4314 to a vec<stmt_vec_info>.
4315 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
4316 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
4318 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
4319 to a vec<stmt_vec_info>.
4320 (bst_traits::value_type, bst_traits::value_type): Likewise.
4321 (bst_traits::hash): Update accordingly.
4322 (vect_get_and_check_slp_defs): Change the stmts parameter from
4323 a vec<gimple *> to a vec<stmt_vec_info>.
4324 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
4325 (vect_build_slp_tree): Likewise.
4326 (vect_build_slp_tree_2): Likewise. Update uses of
4327 SLP_TREE_SCALAR_STMTS.
4328 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
4329 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
4330 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
4331 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
4332 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
4333 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
4334 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
4335 (vect_get_constant_vectors, vect_get_slp_defs)
4336 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
4337 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
4338 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
4339 instead of gimple stmts.
4340 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
4341 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
4342 (vect_slp_analyze_instance_dependence): Update uses of
4343 SLP_TREE_SCALAR_STMTS.
4344 (vect_slp_analyze_and_verify_node_alignment): Likewise.
4345 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
4346 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
4347 (get_initial_defs_for_reduction): Likewise.
4348 (vect_create_epilog_for_reduction): Likewise.
4349 (vectorize_fold_left_reduction): Likewise.
4350 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
4351 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
4352 (can_vectorize_live_stmts): Likewise.
4354 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4356 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
4357 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4358 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
4360 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
4362 (gather_scalar_reductions): Update after above interface changes.
4363 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
4364 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
4366 (vect_force_simple_reduction): Likewise.
4367 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
4368 LOOP_VINFO_REDUCTIONS.
4369 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4371 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4373 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
4374 a gimple stmt to a stmt_vec_info.
4375 * tree-vect-loop.c (vect_active_double_reduction_p)
4376 (vect_force_simple_reduction, vectorizable_reduction): Update
4379 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4381 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
4382 vec<gimple *> to a vec<stmt_vec_info>.
4383 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
4384 the reduction_phis argument from a vec<gimple *> to a
4386 (vectorizable_reduction): Likewise the phis local variable that
4387 is passed to vect_create_epilog_for_reduction. Update for new type
4388 of SLP_TREE_VEC_STMTS.
4389 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
4390 (vectorizable_live_operation): Likewise.
4391 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
4392 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
4394 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4396 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
4397 a gimple stmt to a stmt_vec_info.
4398 (vectorizable_condition, vectorizable_live_operation)
4399 (vectorizable_reduction, vectorizable_induction): Pass back the
4400 vectorized statement as a stmt_vec_info.
4401 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
4402 use of STMT_VINFO_VEC_STMT.
4403 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
4404 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
4405 as stmt_vec_infos rather than gimple stmts.
4406 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
4408 (vectorizable_live_operation): Likewise.
4409 (vectorizable_reduction, vectorizable_induction): Likewise,
4410 updating use of STMT_VINFO_VEC_STMT.
4411 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
4412 of STMT_VINFO_VEC_STMT.
4413 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
4414 (vectorizable_simd_clone_call, vectorizable_conversion)
4415 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4416 (vectorizable_store, vectorizable_load, vectorizable_condition)
4417 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
4418 from a gimple stmt to a stmt_vec_info.
4419 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
4420 pointer to a stmt_vec_info to the vectorizable_* routines.
4422 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4424 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
4425 a gimple stmt to a stmt_vec_info.
4426 (is_pattern_stmt_p): Update accordingly.
4427 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
4428 (vect_record_grouped_load_vectors): Likewise.
4429 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
4430 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
4431 (vect_model_reduction_cost): Likewise.
4432 (vect_create_epilog_for_reduction): Likewise.
4433 (vectorizable_reduction, vectorizable_induction): Likewise.
4434 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4435 Return the stmt_vec_info for the pattern statement.
4436 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
4437 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
4438 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
4439 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
4440 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
4441 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
4442 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
4443 (free_stmt_vec_info, vect_is_simple_use): Likewise.
4445 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4447 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
4448 (vect_finish_stmt_generation): Likewise.
4449 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
4450 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
4451 (vect_build_gather_load_calls): Use the return value of the above
4452 functions instead of a separate call to vinfo_for_stmt. Use narrow
4453 scopes for the input gimple stmt and wider scopes for the associated
4454 stmt_vec_info. Use vec_info::lookup_def when setting these
4455 stmt_vec_infos from an SSA_NAME definition.
4456 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4457 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4458 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4459 (vectorizable_store, vectorizable_load, vectorizable_condition)
4460 (vectorizable_comparison): Likewise.
4461 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
4462 (vectorizable_reduction): Likewise.
4464 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4466 * tree-vectorizer.h (vect_is_simple_use): Add an optional
4467 stmt_vec_info * parameter before the optional gimple **.
4468 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
4469 (process_use, vect_get_vec_def_for_operand_1): Update callers.
4470 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
4471 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4472 (vectorizable_live_operation): Likewise.
4473 * tree-vect-patterns.c (type_conversion_p): Likewise.
4474 (vect_look_through_possible_promotion): Likewise.
4475 (vect_recog_rotate_pattern): Likewise.
4476 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
4478 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4480 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
4481 a typedef to a wrapper class.
4482 (NULL_STMT_VEC_INFO): New macro.
4483 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
4484 (stmt_vec_info::operator*): New function.
4485 (stmt_vec_info::operator gimple *): Likewise.
4486 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
4487 (add_stmt_costs): Likewise.
4488 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
4489 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
4490 (vect_get_known_peeling_cost): Likewise.
4491 (vect_estimate_min_profitable_iters): Likewise.
4492 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4493 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
4494 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
4495 (vectorizable_store, free_stmt_vec_infos): Likewise.
4496 (new_stmt_vec_info): Change return type of xcalloc to
4499 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4501 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
4502 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
4503 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
4504 a single_imm_use-based sequence.
4505 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
4507 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4509 * tree-vectorizer.h (vec_info::lookup_def): Declare.
4510 * tree-vectorizer.c (vec_info::lookup_def): New function.
4511 * tree-vect-patterns.c (vect_get_internal_def): Use it.
4512 (vect_widened_op_tree): Likewise.
4513 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
4514 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
4515 (vectorizable_reduction): Likewise.
4516 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
4518 (vect_is_slp_reduction): Update calls accordingly. Use
4519 vec_info::lookup_def.
4520 (vect_is_simple_reduction): Likewise
4521 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
4523 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4525 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
4526 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
4527 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
4529 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
4530 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
4531 (vect_update_vf_for_slp, vect_analyze_loop_operations)
4532 (vect_is_slp_reduction, vectorizable_induction)
4533 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
4534 * tree-vect-patterns.c (vect_init_pattern_stmt):
4535 (vect_determine_min_output_precision_1, vect_determine_precisions)
4536 (vect_pattern_recog): Likewise.
4537 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
4538 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
4539 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
4540 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
4541 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
4542 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
4543 info field from a loop to a loop_vec_info.
4545 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4547 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
4548 (vec_info::add_stmt): Declare.
4549 * tree-vectorizer.c (vec_info::add_stmt): New function.
4550 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
4551 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
4552 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
4553 (vectorizable_induction): Likewise.
4554 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
4555 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
4556 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
4557 (vectorizable_load): Likewise.
4558 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4559 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
4560 (vect_recog_gather_scatter_pattern): Likewise.
4561 (append_pattern_def_seq): Likewise. Remove a check that is
4562 performed by add_stmt itself.
4564 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4566 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
4567 which make_ssa_name was called with new_stmt before new_stmt
4570 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4572 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
4574 (vect_is_slp_reduction): ...here...
4575 (vect_is_simple_reduction): ...and here. Remove repetition of tests
4576 that are already known to be false.
4578 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4580 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
4581 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
4582 STMT_VINFO_NUM_SLP_USES when it's true.
4583 (vect_free_slp_instance): Add a final_p parameter and pass it to
4585 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
4586 (vect_analyze_slp_instance): Likewise.
4587 (vect_slp_analyze_operations): Likewise.
4588 (vect_slp_analyze_bb_1): Likewise.
4589 * tree-vectorizer.c (vec_info): Likewise.
4590 * tree-vect-loop.c (vect_transform_loop): Likewise.
4592 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4594 * tree-vect-loop.c (vectorizable_reduction): Assert that the
4595 function is not called for second and subsequent members of
4598 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4600 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
4601 cases for nested loops from here to ...
4602 (vect_create_epilog_for_reduction): ...here. Only call
4603 vect_is_simple_use for inner-loop reductions.
4605 2018-07-31 Martin Liska <mliska@suse.cz>
4607 PR gcov-profile/85338
4608 PR gcov-profile/85350
4609 PR gcov-profile/85372
4610 * profile.c (struct location_triplet): New.
4611 (struct location_triplet_hash): Likewise.
4612 (output_location): Do not output a BB that
4613 is already recorded for a line.
4614 (branch_prob): Use streamed_locations.
4616 2018-07-31 Martin Liska <mliska@suse.cz>
4618 PR gcov-profile/85370
4619 * coverage.c (coverage_begin_function): Do not mark target
4620 clones as artificial functions.
4622 2018-07-31 Martin Liska <mliska@suse.cz>
4624 PR gcov-profile/83813
4625 PR gcov-profile/84758
4626 PR gcov-profile/85217
4627 PR gcov-profile/85332
4628 * profile.c (branch_prob): Do not record GOTO expressions
4629 for GIMPLE statements which locations are already streamed.
4631 2018-07-31 Olivier Hainque <hainque@adacore.com>
4633 * gcc.c (handle_spec_function): Accept a soft_matched_part
4634 argument, as do_spec_1. Pass it down to ...
4635 (eval_spec_function): Accept a soft_matched_part argument,
4636 and pass it down to ...
4637 (do_spec_2): Accept a soft_matched_part argument, and pass
4638 it down to do_spec_1.
4639 (do_spec_1): Pass soft_matched_part to handle_spec_function.
4640 (handle_braces): Update call to handle_spec_function.
4641 (driver::set_up_specs): Update calls to do_spec_2.
4642 (compare_debug_dump_opt_spec_function): Likewise.
4643 (compare_debug_self_opt_spec_function): Likewise.
4645 2018-07-31 Olivier Hainque <hainque@adacore.com>
4647 * common.opt (nolibc): New option.
4648 * doc/invoke.texi (Link Options): Document it.
4649 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
4650 * config/alpha/linux.h: Likewise.
4651 * config/arc/elf.h: Likewise.
4652 * config/arm/uclinux-elf.h: Likewise.
4653 * config/arm/unknown-elf.h: Likewise.
4654 * config/avr/avrlibc.h: Likewise.
4655 * config/bfin/bfin.h: Likewise.
4656 * config/bfin/linux.h: Likewise.
4657 * config/bfin/uclinux.h: Likewise.
4658 * config/darwin.h: Likewise.
4659 * config/darwin10.h: Likewise.
4660 * config/darwin12.h: Likewise.
4661 * config/gnu-user.h: Likewise.
4662 * config/lm32/uclinux-elf.h: Likewise.
4663 * config/pa/pa-hpux11.h: Likewise.
4664 * config/pa/pa64-hpux.h: Likewise.
4665 * config/sparc/sparc.h: Likewise.
4667 2018-07-31 Olivier Hainque <hainque@adacore.com>
4669 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
4670 undefined variables.
4672 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
4675 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
4678 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
4680 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
4681 terminated string literal.
4683 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
4685 PR rtl-optimization/85160
4686 * combine.c (is_just_move): New function.
4687 (try_combine): Allow combining two instructions into two if neither of
4688 the original instructions was a move.
4690 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
4693 * doc/extend.texi (Global Register Variables): Discourage use of type
4695 (Local Register Variables): Likewise.
4697 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
4699 PR tree-optimization/86506
4700 * hwint.h (ceil_log2): Resync with hwint.c implementation.
4702 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
4705 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
4706 hard_regno, make sure no insn between `from` and `to` clobbers it.
4708 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
4709 Tom de Vries <tdevries@suse.de>
4711 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
4712 (PTX_DEFAULT_RUNTIME_DIM): ... this.
4713 (nvptx_goacc_validate_dims): Set default worker and gang dims to
4714 PTX_DEFAULT_RUNTIME_DIM.
4715 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
4717 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
4719 * config/pa/pa.c (pa_output_addr_vec): Align address table.
4720 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
4721 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
4723 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
4725 * config/rs6000/constraints.md (wG constraint): Delete, no longer
4727 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
4728 predicate to reflect toc fusion has been deleted.
4729 (toc_fusion_mem_raw): Delete, no longer used.
4730 (toc_fusion_mem_wrapped): Likewise.
4731 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
4733 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
4734 Delete, no longer used.
4735 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
4736 meant to be used for toc fusion.
4737 (rs6000_debug_print_mode): Delete toc fusion debugging.
4738 (rs6000_debug_reg_global): Likewise.
4739 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
4740 fusion and secondary reload support that were never used.
4741 (rs6000_option_override_internal): Delete TOC fusion, that was only
4742 partially defined, and it did not work unless you also used the
4744 (rs6000_legitimate_address_p): Delete TOC fusion support.
4745 (rs6000_opt_masks): Likewise.
4746 (fusion_wrap_memory_address): Delete function, no longer used.
4747 (fusion_split_address); Delete TOC fusion support.
4748 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
4749 longer used with toc fusion being deleted.
4750 (TARGET_TOC_FUSION_FP): Likewise.
4751 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
4753 (toc fusion spliter): Delete TOC fusion support.
4754 (toc_fusionload_<mode>): Likewise.
4755 (toc_fusionload_di): Likewise.
4756 (fusion_gpr_load_<mode>): Delete generator function, this insn no
4757 longer needs to be named. Rename predicate to delete TOC fusion.
4758 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
4759 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
4760 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
4761 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
4762 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
4764 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
4766 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
4767 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
4768 __int128 in built-in function prototypes.
4769 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
4770 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
4772 2018-07-27 Martin Sebor <msebor@redhat.com>
4774 PR tree-optimization/86696
4775 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
4776 types, including enums.
4777 (handle_char_store): Be prepared for the above function to fail.
4779 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
4781 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
4782 when optimization level is lower than 2 or optimize for size.
4784 2018-07-26 Martin Sebor <msebor@redhat.com>
4786 PR tree-optimization/86043
4787 PR tree-optimization/86042
4788 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
4789 (get_string_cst_length): Rename...
4790 (get_min_string_length): ...to this. Add argument.
4791 (handle_char_store): Extend to handle multi-character stores by
4793 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
4794 * tree.h (initializer_zerop): Add argument.
4796 2018-07-26 Jakub Jelinek <jakub@redhat.com>
4799 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
4800 declare target to variables if they have always,{to,from,tofrom} map
4803 2018-07-26 Martin Liska <mliska@suse.cz>
4806 * lto-wrapper.c: Add linker_output as prefix
4807 for ltrans_output_file.
4809 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
4811 PR rtl-optimization/85805
4812 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
4813 value for hard registers if that was written in the same mode.
4815 2018-07-26 Martin Liska <mliska@suse.cz>
4817 PR gcov-profile/86536
4818 * gcov.c (format_gcov): Use printf format %.*f directly
4819 and do not handle special values.
4821 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4823 * common/config/arc/arc-common.c (arc_option_optimization_table):
4824 Update default optimizations for size.
4826 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4828 * config/arc/arc.md (movsf_insn): Add short instruction selection.
4829 * config/arc/constraints.md (CfZ): New constraint.
4830 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
4831 (subsf3_fpu): Likewise.
4832 (cmpsf_fpu): Likewise.
4833 (cmpsf_fpu_uneq): Likewise.
4835 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4837 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
4839 (arc_is_uncached_mem_p): uncached applies to both the variable and
4842 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4844 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
4847 2018-07-25 David Malcolm <dmalcolm@redhat.com>
4849 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
4850 field "m_scopes" from vec to auto_vec.
4852 2018-07-25 Martin Liska <mliska@suse.cz>
4854 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
4857 2018-07-25 Richard Biener <rguenther@suse.de>
4860 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
4861 special wrt context_die late.
4862 (gen_subprogram_die): Re-use DIEs in local scope.
4864 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
4866 PR tree-optimization/86644
4867 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
4869 2018-07-25 Martin Liska <mliska@suse.cz>
4872 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
4873 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
4875 2018-07-25 Martin Liska <mliska@suse.cz>
4878 * params.def: Explain ASan abbreviation and provide
4879 a documentation link.
4881 2018-07-24 Martin Sebor <msebor@redhat.com>
4883 PR tree-optimization/86622
4884 PR tree-optimization/86532
4885 * builtins.h (string_length): Declare.
4886 * builtins.c (c_strlen): Correct handling of non-constant offsets.
4887 (check_access): Be prepared for non-constant length ranges.
4888 (string_length): Make extern.
4889 * expr.c (string_constant): Only handle the minor non-constant
4890 array index. Use string_constant to compute the length of
4891 a generic string constant.
4893 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
4895 PR tree-optimization/86618
4896 * tree-vect-stmts.c (vectorizable_call): Don't take the address
4897 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
4899 2018-07-24 David Malcolm <dmalcolm@redhat.com>
4901 PR tree-optimization/86636
4902 * json.cc (json::object::set): Fix comment. Add assertions.
4903 (json::array::append): Move here from json.h. Add comment and an
4905 (json::string::string): Likewise.
4906 * json.h (json::array::append): Move to json.cc.
4907 (json::string::string): Likewise.
4908 * optinfo-emit-json.cc
4909 (optrecord_json_writer::impl_location_to_json): Assert that we
4910 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
4911 wrapper around it. Expand the location once, rather than three
4913 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
4914 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
4916 (optrecord_json_writer::optinfo_to_json): Likewise, in four
4917 places. Fix some overlong lines.
4919 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
4921 * config/aarch64/aarch64-simd.md
4922 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
4923 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
4924 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
4925 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
4926 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
4927 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
4928 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
4929 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
4930 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
4932 2018-07-24 Jakub Jelinek <jakub@redhat.com>
4935 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
4936 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
4937 and abs_d == d, do the power of two handling if profitable.
4939 2018-07-24 Richard Biener <rguenther@suse.de>
4941 * match.pd: Add BIT_FIELD_REF canonicalizations.
4943 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4946 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
4948 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4950 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
4951 terminated STRING_CST object.
4953 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4955 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
4957 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
4959 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
4960 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
4961 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
4962 the elements into a register.
4963 (rs6000_split_v4si_init_di_reg): Delete.
4964 (rs6000_split_v4si_init): Delete.
4965 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
4966 (vsx_init_v4si): Rewrite as a define_expand.
4968 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
4970 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
4971 zero_extend argument from memory): New.
4973 2018-07-22 Martin Sebor <msebor@redhat.com>
4976 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
4977 diagnosing calls with unknown arguments unless -Walloca-larger-than
4978 is restricted to less than PTRDIFF_MAX bytes.
4980 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
4982 * doc/gcov.texi (Invoking Gcov): Editorial changes.
4984 2018-07-20 David Malcolm <dmalcolm@redhat.com>
4986 * pretty-print.c (text_info::set_location): Remove redundant
4987 "line_table" parameter from call to rich_location::set_range.
4989 2018-07-20 Martin Sebor <msebor@redhat.com>
4992 * builtins.c (expand_builtin_alloca): Adjust.
4993 * calls.c (alloc_max_size): Simplify.
4994 * cgraphunit.c (cgraph_node::expand): Adjust.
4995 * common.opt (larger_than_size, warn_frame_larger_than): Remove
4997 (frame_larger_than_size): Same.
4998 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
4999 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
5001 * doc/invoke.texi (GCC Command Options): Document option arguments.
5002 Explain byte-size arguments and suffixes.
5003 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
5004 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
5005 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
5006 * doc/options.texi (UInteger): Expand.
5007 (Host_Wide_Int, ByteSize): Document new properties.
5008 * final.c (final_start_function_1): Include sizes in an error message.
5009 * function.c (frame_offset_overflow): Same.
5010 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
5011 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
5012 Diagnose unbounded alloca calls only for limits of less than
5014 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
5015 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
5017 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
5018 only for limits of less than PTRDIFF_MAX.
5019 * langhooks-def.h (lhd_handle_option): Change function argument
5021 * langhooks.c (lhd_handle_option): Same.
5022 * langhooks.h (handle_option): Same.
5023 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
5025 (var_type, var_type_struct): Same.
5026 (var_set): Handle ByteSize flag.
5027 * optc-gen.awk: Add comments to output to ease debugging. Make
5028 use of HOST_WIDE_INT where appropriate.
5029 * opts-gen-save.awk: Use %lx to format unsigned long.
5030 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
5031 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
5032 arguments. Parse bytes-size suffixes.
5033 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
5034 (enum_value_to_arg): Same.
5035 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
5036 (handle_option): Adjust.
5037 (generate_option): Change function argument to HOST_WIDE_INT.
5038 (cmdline_handle_error): Adjust.
5039 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
5040 (set_option): Change function argument to HOST_WIDE_INT.
5041 (option_enabled): Handle cl_host_wide_int.
5042 (get_option_state): Handle CLVC_SIZE.
5043 (control_warning_option): Same.
5044 * opts.c (common_handle_option): Change function argument to
5045 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
5046 OPT_Wvla_larger_than_.
5047 * opts.h (enum cl_var_type): Add an enumerator.
5048 * stor-layout.c (layout_decl): Print a more meaningful warning.
5049 * toplev.c (output_stack_usage): Adjust.
5051 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
5053 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
5054 call to inline_expand_builtin_string_cmp.
5055 (expand_builtin_strcmp): Likewise.
5056 (expand_builtin_strncmp): Likewise.
5057 (inline_string_cmp): Delete the last parameter, change char_type_node
5058 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
5060 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
5061 the inlining expansion on target where the type of the call has same or
5062 narrower precision than unsigned char.
5064 2018-07-20 David Malcolm <dmalcolm@redhat.com>
5066 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
5067 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
5068 * common.opt (fsave-optimization-record): New option.
5069 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
5070 * doc/invoke.texi (-fsave-optimization-record): New option.
5071 * dumpfile.c: Include "optinfo-emit-json.h".
5072 (struct kv_pair): Move to coretypes.h.
5073 (optgroup_options): Make non-static.
5074 (dump_context::end_scope): Call
5075 optimization_records_maybe_pop_dump_scope.
5076 * dumpfile.h (optgroup_options): New decl.
5077 * json.cc: New file.
5079 * optinfo-emit-json.cc: New file.
5080 * optinfo-emit-json.h: New file.
5081 * optinfo.cc: Include "optinfo-emit-json.h".
5082 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
5083 (optinfo_enabled_p): Check optimization_records_enabled_p.
5084 (optinfo_wants_inlining_info_p): Likewise.
5085 * optinfo.h: Update comment.
5086 * profile-count.c (profile_quality_as_string): New function.
5087 * profile-count.h (profile_quality_as_string): New decl.
5088 (profile_count::quality): New accessor.
5089 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
5090 and optinfo_emit_json_cc_tests.
5091 * selftest.h (selftest::json_cc_tests): New decl.
5092 (selftest::optinfo_emit_json_cc_tests): New decl.
5093 * toplev.c: Include "optinfo-emit-json.h".
5094 (compile_file): Call optimization_records_finish.
5095 (do_compile): Call optimization_records_start.
5096 * tree-ssa-live.c: Include optinfo.h.
5097 (remove_unused_scope_block_p): Retain inlining information if
5098 optinfo_wants_inlining_info_p returns true.
5100 2018-07-20 Richard Biener <rguenther@suse.de>
5103 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
5104 to cover -flto-partition=none.
5106 2018-07-20 Martin Liska <mliska@suse.cz>
5108 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
5109 (get_decl_source_range): Remove unused function.
5111 2018-07-20 Richard Biener <rguenther@suse.de>
5113 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
5114 (struct vn_phi_s): Likewise.
5115 (struct vn_reference_s): Likewise.
5116 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
5117 for searching the slot of an entry known to be in the hash itself.
5118 (vn_phi_hasher::equal): Likewise.
5119 (vn_reference_hasher::equal): Likewise.
5120 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
5122 (optimistic_info, current_info): Remove, keeping only valid_info.
5123 (vn_reference_lookup_1): Remove fallback lookup.
5124 (vn_reference_lookup_2): Likewise.
5125 (vn_nary_op_lookup_1): Likewise.
5126 (vn_phi_lookup): Likewise.
5127 (vn_nary_build_or_lookup_1): Make sure to not chain the built
5129 (vn_reference_insert): Adjust, chain the inserted hash element
5130 at last_inserted_ref.
5131 (vn_reference_insert_pieces): Likewise.
5132 (visit_reference_op_call): Likewise.
5133 (vn_nary_op_insert_into): Chain the inserted hash element at
5135 (vn_nary_op_insert_pieces): Adjust.
5136 (vn_nary_op_insert): Likewise.
5137 (vn_nary_op_insert_stmt): Likewise.
5138 (vn_phi_insert): Adjust, chain the inserted hash element at
5140 (process_scc): Remove clearing and copying the optimistic
5141 table. Instead remove elements inserted during an optimistic
5142 iteration from the single table we maintain.
5143 (init_scc_vn): Adjust.
5144 (free_scc_vn): Likewise.
5145 (sccvn_dom_walker::record_cond): Likewise.
5146 (sccvn_dom_walker::after_dom_children): Likewise.
5148 2018-07-19 Martin Sebor <msebor@redhat.com>
5150 PR tree-optimization/84047
5151 PR tree-optimization/83776
5152 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
5153 (check_array_bounds): Call it.
5155 2018-07-19 Martin Sebor <msebor@redhat.com>
5157 * align.h (align_flags): Use member initialization.
5159 2018-07-19 David Malcolm <dmalcolm@redhat.com>
5161 * Makefile.in (OBJS): Add optinfo.o.
5162 * coretypes.h (class symtab_node): New forward decl.
5163 (struct cgraph_node): New forward decl.
5164 (class varpool_node): New forward decl.
5165 * dump-context.h: New file.
5166 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
5168 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
5169 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
5170 (set_alt_dump_file): Likewise.
5171 (dump_context::~dump_context): New dtor.
5172 (dump_gimple_stmt): Move implementation to...
5173 (dump_context::dump_gimple_stmt): ...this new member function.
5174 Add the stmt to any pending optinfo, creating one if need be.
5175 (dump_gimple_stmt_loc): Move implementation to...
5176 (dump_context::dump_gimple_stmt_loc): ...this new member function.
5177 Start a new optinfo and add the stmt to it.
5178 (dump_gimple_expr): Move implementation to...
5179 (dump_context::dump_gimple_expr): ...this new member function.
5180 Add the stmt to any pending optinfo, creating one if need be.
5181 (dump_gimple_expr_loc): Move implementation to...
5182 (dump_context::dump_gimple_expr_loc): ...this new member function.
5183 Start a new optinfo and add the stmt to it.
5184 (dump_generic_expr): Move implementation to...
5185 (dump_context::dump_generic_expr): ...this new member function.
5186 Add the tree to any pending optinfo, creating one if need be.
5187 (dump_generic_expr_loc): Move implementation to...
5188 (dump_context::dump_generic_expr_loc): ...this new member
5189 function. Add the tree to any pending optinfo, creating one if
5191 (dump_printf): Move implementation to...
5192 (dump_context::dump_printf_va): ...this new member function. Add
5193 the text to any pending optinfo, creating one if need be.
5194 (dump_printf_loc): Move implementation to...
5195 (dump_context::dump_printf_loc_va): ...this new member function.
5196 Start a new optinfo and add the stmt to it.
5197 (dump_dec): Move implementation to...
5198 (dump_context::dump_dec): ...this new member function. Add the
5199 value to any pending optinfo, creating one if need be.
5200 (dump_context::dump_symtab_node): New member function.
5201 (dump_context::get_scope_depth): New member function.
5202 (dump_context::begin_scope): New member function.
5203 (dump_context::end_scope): New member function.
5204 (dump_context::ensure_pending_optinfo): New member function.
5205 (dump_context::begin_next_optinfo): New member function.
5206 (dump_context::end_any_optinfo): New member function.
5207 (dump_context::s_current): New global.
5208 (dump_context::s_default): New global.
5209 (dump_scope_depth): Delete global.
5210 (dumpfile_ensure_any_optinfo_are_flushed): New function.
5211 (dump_symtab_node): New function.
5212 (get_dump_scope_depth): Reimplement in terms of dump_context.
5213 (dump_begin_scope): Likewise.
5214 (dump_end_scope): Likewise.
5215 (selftest::temp_dump_context::temp_dump_context): New ctor.
5216 (selftest::temp_dump_context::~temp_dump_context): New dtor.
5217 (selftest::verify_item): New function.
5218 (ASSERT_IS_TEXT): New macro.
5219 (ASSERT_IS_TREE): New macro.
5220 (ASSERT_IS_GIMPLE): New macro.
5221 (selftest::test_capture_of_dump_calls): New test.
5222 (selftest::dumpfile_c_tests): Call it.
5223 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
5224 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
5225 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
5226 descriptive comment.
5227 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
5228 (dump_node, dump_bb): Move these unrelated decls.
5229 (class dump_manager): Add leading comment.
5230 * optinfo.cc: New file.
5231 * optinfo.h: New file.
5233 2018-07-19 Michael Collison <michael.collison@arm.com>
5234 Richard Henderson <rth@redhat.com>
5236 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
5237 (subti): Handle op1 zero.
5238 (subvti4, usub4ti4): New.
5239 (*sub<GPI>3_compare1_imm): New.
5240 (sub<GPI>3_carryinCV): New.
5241 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
5242 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
5244 2018-07-19 Michael Collison <michael.collison@arm.com>
5245 Richard Henderson <rth@redhat.com>
5247 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
5248 (addti3): Create simpler code if low part is already known to be 0.
5249 (addvti4, uaddvti4): New.
5250 (*add<GPI>3_compareC_cconly_imm): New.
5251 (*add<GPI>3_compareC_cconly): New.
5252 (*add<GPI>3_compareC_imm): New.
5253 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
5254 handle constants within this pattern..
5255 (*add<GPI>3_compareV_cconly_imm): New.
5256 (*add<GPI>3_compareV_cconly): New.
5257 (*add<GPI>3_compareV_imm): New.
5258 (add<GPI>3_compareV): New.
5259 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
5260 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
5261 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
5262 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
5263 with 'comparison' operator.
5264 (*add<GPI>3_compareV_cconly_imm): Ditto.
5265 (*add<GPI>3_compareV_cconly): Ditto.
5266 (*add<GPI>3_compareV_imm): Ditto.
5267 (add<GPI>3_compareV): Ditto.
5268 (add<mode>3_carryinC): Ditto.
5269 (*add<mode>3_carryinC_zero): Ditto.
5270 (*add<mode>3_carryinC): Ditto.
5271 (add<mode>3_carryinV): Ditto.
5272 (*add<mode>3_carryinV_zero): Ditto.
5273 (*add<mode>3_carryinV): Ditto.
5275 2018-07-19 Michael Collison <michael.collison@arm.com>
5276 Richard Henderson <rth@redhat.com>
5278 * config/aarch64/aarch64-modes.def (CC_V): New.
5279 * config/aarch64/aarch64-protos.h
5280 (aarch64_addti_scratch_regs): Declare
5281 (aarch64_subvti_scratch_regs): Declare.
5282 (aarch64_expand_subvti): Declare.
5283 (aarch64_gen_unlikely_cbranch): Declare
5284 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
5285 for signed overflow using CC_Vmode.
5286 (aarch64_get_condition_code_1): Handle CC_Vmode.
5287 (aarch64_gen_unlikely_cbranch): New function.
5288 (aarch64_addti_scratch_regs): New function.
5289 (aarch64_subvti_scratch_regs): New function.
5290 (aarch64_expand_subvti): New function.
5292 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5294 * config/aarch64/aarch64-option-extensions.def: New entry for profile
5296 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
5297 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
5300 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5303 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
5304 address check not strict.
5306 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5308 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
5310 (store_pair_lanes<mode>): Likewise.
5311 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
5312 enum value 'ADDR_QUERY_LDP_STP_N'.
5313 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
5314 (aarch64_print_address_internal): Add declaration.
5315 (aarch64_print_ldpstp_address): Remove.
5316 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
5317 (aarch64_print_operand): Change printing of 'y'.
5318 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
5319 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
5320 'true' rather than '1'.
5321 * config/aarch64/constraints.md (Uml): Likewise.
5322 (Uml): Rename to Umn.
5325 2018-07-19 Richard Biener <rguenther@suse.de>
5327 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
5329 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
5330 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
5331 (vn_reference_hasher): Likewise.
5332 (struct vn_tables_s): Remove obstack and alloc-pool members.
5333 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
5334 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
5335 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
5336 (vn_reference_insert_pieces): Likewise.
5337 (alloc_vn_nary_op_noinit): Adjust.
5338 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
5339 (vn_phi_eq): Adjust.
5340 (shared_lookup_phiargs): Remove.
5341 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
5342 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
5343 (visit_reference_op_call): Likewise.
5344 (copy_nary, copy_phi, copy_reference): Remove.
5345 (process_scc): Rewind the obstack when iterating. Do not
5346 copy the elements to valid_info but just move them from one
5347 hashtable to the other.
5348 (allocate_vn_table): Adjust.
5349 (free_vn_table): Likewise.
5350 (init_scc_vn): Likewise.
5351 (free_scc_vn): Likewise.
5353 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
5356 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
5357 indirect_return as function type attribute.
5358 (ix86_attribute_table): Change indirect_return to function
5360 * doc/extend.texi: Update indirect_return attribute.
5362 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
5364 * wide-int.h (widest2_int): New.
5365 * gimple-fold.c (arith_overflowed_p): Use it.
5366 * tree.h (widest2_int_cst): New.
5367 * tree-vrp.c (wide_int_binop_overflow): Rename from
5368 vrp_int_const_binop.
5369 Rewrite to work on trees.
5370 (extract_range_from_multiplicative_op_1): Abstract code to...
5371 (wide_int_range_min_max): ...here.
5372 (wide_int_range_cross_product): ...and here.
5373 (extract_range_from_binary_expr_1): Abstract overflow code to...
5374 (wide_int_range_mult_wrapping): ...here.
5375 * tree-vrp.h (wide_int_range_cross_product): New.
5376 (wide_int_range_mult_wrapping): New.
5378 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
5379 Julia Koval <julia.koval@intel.com>
5381 * config/i386/x86-tune-costs.h (skylake_memcpy,
5382 skylake_memset): Replace rep_prefix with unrolling for size 512.
5384 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
5387 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
5388 comparision with EQ_EXPR in last stmt.
5390 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
5392 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
5393 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
5394 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
5395 previously known as "PowerPC AltiVec Built-in Functions". Move
5396 some material to new subsubsections "PowerPC AltiVec Built-in
5397 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
5399 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
5400 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
5401 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
5402 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
5404 2018-07-18 Richard Biener <rguenther@suse.de>
5406 PR tree-optimization/86557
5407 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
5410 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
5412 * config/s390/s390.c (s390_function_profiler): Generate CFI.
5414 2018-07-17 Jeff Law <law@redhat.com>
5416 * config/arm/arm.c (get_label_padding): Update for recent
5417 changes to label_to_alignment.
5419 PR tree-optimization/86010
5420 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
5422 * config/mips/mips.c (vr4130_align_insns): Update for recent
5423 changes to label_to_alignment.
5425 * config/frv/frv.c (frv_label_align): Update for recent changes
5426 to label_to_alignment.
5428 * config/nios2/nios2.c (nios2_label_align): Update for recent
5429 changes which dropped ALIGN_LABELS_LOG.
5431 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
5433 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
5434 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
5436 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5438 * config/arc/arc.c (arc_label_align): Use align_labels instead of
5439 deprecated align_labels_log.
5441 2018-07-17 Richard Biener <rguenther@suse.de>
5444 * dwarf2out.c (init_sections_and_labels): Always generate
5445 a debug_line_str_section for early LTO debug.
5446 (dwarf2out_finish): Reset debug_line_str_hash output early.
5447 Bump counter for extra dwarf5 .debug_loc labels to not conflict
5449 (dwarf2out_early_finish): Output debug_line_str.
5451 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
5453 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
5454 index register on z196 or later.
5456 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
5458 * config/s390/s390.c (s390_default_align): Set default function
5460 (s390_override_options_after_change): Call s390_default align.
5461 (s390_option_override_internal): Call s390_default_align.
5462 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
5464 2018-07-17 Jakub Jelinek <jakub@redhat.com>
5467 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
5468 to _looptemp_ clauses, other than the first two.
5470 2018-07-17 Martin Liska <mliska@suse.cz>
5472 * opts.c: Do not enable OPT_falign_* for -Os.
5474 2018-07-17 Martin Liska <mliska@suse.cz>
5476 * align.h (MAX_CODE_ALIGN): New.
5477 (MAX_CODE_ALIGN_VALUE): New.
5478 * common/config/i386/i386-common.c (ix86_handle_option):
5479 (MAX_CODE_ALIGN): Moved to align.h.
5480 * final.c (MAX_CODE_ALIGN): Likewise.
5481 * opts.c (parse_and_check_align_values):
5482 (MAX_CODE_ALIGN): Likewise.
5483 (MAX_CODE_ALIGN_VALUE): Likewise.
5485 2018-07-17 Martin Liska <mliska@suse.cz>
5487 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
5488 in order to fulfil coding style.
5489 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
5490 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
5491 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
5492 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
5493 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
5494 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
5495 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
5496 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
5497 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
5499 2018-07-17 Martin Liska <mliska@suse.cz>
5501 * align.h: New file.
5502 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
5504 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
5505 align_flags of label_to_alignment.
5506 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
5508 * config/m68k/m68k.c: Do not use removed align_labels_value and
5510 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
5511 (LOOP_ALIGN): Likewise.
5512 (LABEL_ALIGN): Likewise.
5513 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
5514 Remove not used macro.
5515 (rs6000_loop_align): Change return type to align_flags.
5516 (rs6000_loop_align_max_skip): Remove.
5517 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
5518 Change return type to align_flags.
5519 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
5520 Remove not used macro.
5521 (rs6000_loop_align): Change return type to align_flags.
5522 (rs6000_loop_align_max_skip): Remove.
5523 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
5524 * config/rx/rx-protos.h (rx_align_for_label): Make it
5526 * config/rx/rx.c (rx_align_for_label): Change return type
5528 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
5530 into align_flags class.
5531 (LABEL_ALIGN): Likewise.
5532 (LOOP_ALIGN): Likewise.
5533 * config/s390/s390.c (s390_label_align): Use align_flags
5535 (s390_asm_output_function_label): Likewise.
5536 * config/sh/sh.c (sh_override_options_after_change):
5537 Use align_flags class directly without macros.
5538 (find_barrier): Likewise.
5539 (barrier_align): Likewise.
5540 (sh_loop_align): Likewise.
5541 * config/spu/spu.c (spu_option_override):
5542 Use align_flags_tuple::get_value instead of removed macros.
5543 (spu_sched_init): Likewise.
5544 * config/spu/spu.h (GTY): Likewise.
5545 * config/visium/visium.c (visium_option_override):
5546 Set "8" as default secondary alignment.
5547 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
5548 in order to guarantee secondary alignment of 8.
5549 * coretypes.h: Include align.h header file.
5550 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
5551 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
5552 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
5553 * doc/tm.texi.in: Likewise.
5554 * final.c (struct label_alignment): Remove not used structure.
5555 (LABEL_ALIGN): Change type to align_flags.
5556 (LOOP_ALIGN): Likewise.
5557 (JUMP_ALIGN): Likewise.
5558 (default_loop_align_max_skip): Remove.
5559 (default_label_align_max_skip): Likewise.
5560 (default_jump_align_max_skip): Likewise.
5561 (default_label_align_after_barrier_max_skip):
5562 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
5563 (LABEL_TO_MAX_SKIP): Remove.
5564 (label_to_alignment): Return align_flags type instead of integer.
5565 (label_to_max_skip): Remove.
5566 (align_fuzz): Use align_flags type.
5567 (compute_alignments): Use align_flags type and use align_flags::max
5568 to combine multiple alignments.
5569 (grow_label_align): Grow vec instead of C array.
5570 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
5571 (shorten_branches): Use align_flags type and use align_flags::max
5572 to combine multiple alignments.
5573 (final_scan_insn_1): Remove usage of secondary alignment that comes
5574 from label alignment, but instead use proper secondary alignment
5575 which is computed in grow_label_align.
5576 * flags.h (struct align_flags_tuple): Move to align.h.
5577 (struct align_flags): Likewise.
5578 (state_align_loops): Rename to align_loops.
5579 (state_align_jumps): Rename to align_jumps.
5580 (state_align_labels): Rename to align_labels.
5581 (state_align_functions): Rename to align_functions.
5582 (align_loops_log): Remove.
5583 (align_jumps_log): Remove.
5584 (align_labels_log): Remove.
5585 (align_functions_log): Remove.
5586 (align_loops_max_skip): Remove.
5587 (align_jumps_max_skip): Remove.
5588 (align_labels_max_skip): Remove.
5589 (align_functions_max_skip): Remove.
5590 (align_loops_value): Remove.
5591 (align_jumps_value): Remove.
5592 (align_labels_value): Remove.
5593 (align_functions_value): Remove.
5594 * output.h (label_to_alignment): Change return type to align_flags.
5595 (label_to_max_skip): Remove.
5596 * target.def: Remove loop_align_max_skip, label_align_max_skip,
5597 jump_align_max_skip macros.
5598 * targhooks.h (default_loop_align_max_skip): Remove.
5599 (default_label_align_max_skip): Likewise.
5600 (default_jump_align_max_skip): Likewise.
5601 (default_label_align_after_barrier_max_skip): Remove.
5602 * toplev.c (read_log_maxskip): Use ::normalize function.
5603 (parse_N_M): Remove not used argument and also call ::normalize.
5604 (parse_alignment_opts): Do not pass unused arguments.
5605 * varasm.c (assemble_start_function): Use directly align_functions
5606 instead of removed macros.
5607 * system.h: Do not poison removed macros.
5609 2018-07-17 Jakub Jelinek <jakub@redhat.com>
5612 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
5613 and cond temporaries don't have reference type if iterator has
5614 pointer type. For init use &for_pre_body instead of pre_p if
5615 for_pre_body is non-empty.
5617 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
5619 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
5620 double-double modes to SFmode directly directly.
5621 (trunc<mode>sf2_fprs): Delete.
5623 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
5625 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
5626 for conversions between IFmode and the decimal floating point modes.
5627 (init_float128_ieee): Use the correct names for conversions between
5628 KFmode and the decimal floating point modes.
5630 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
5632 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
5633 for the conversions between TDmode and IFmode.
5634 (init_float128_ieee): Use more correct names for the conversions
5635 between TDmode and KFmode.
5637 2018-07-16 Jakub Jelinek <jakub@redhat.com>
5639 PR tree-optimization/86526
5640 * builtins.c (expand_builtin_memcmp): Formatting fixes.
5641 (inline_expand_builtin_string_cmp): Likewise.
5642 (inline_string_cmp): Likewise. Use c_readstr instead of
5643 builtin_memcpy_read_str. Add unit_mode temporary.
5645 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
5648 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
5649 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
5651 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
5653 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
5654 Alphabetize prototypes of built-in functions, separating out
5655 built-in functions that are listed in this section but should be
5656 described elsewhere.
5658 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
5661 * expmed.c (emit_store_flag): Do not emit setcc followed by a
5662 conditional move when trapping comparison was split to a
5663 non-trapping one (and vice versa).
5665 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
5667 * config/s390/s390.c (s390_function_profiler): Generate nops
5668 instead of profiler call sequences.
5669 * config/s390/s390.opt: Add the new option.
5671 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
5673 * config/s390/s390.c (s390_function_profiler): Generate
5674 __mcount_loc section.
5675 * config/s390/s390.opt: Add the new option.
5677 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
5679 * common.opt: Add the new warning.
5680 * config/s390/s390.c (s390_function_profiler): Emit "brasl
5681 %r0,__fentry__" when -mfentry is specified.
5682 (s390_option_override_internal): Disallow -mfentry for 31-bit
5684 * config/s390/s390.opt: Add the new option.
5686 2018-07-16 Richard Biener <rguenther@suse.de>
5689 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
5690 for function-local FUNCTION_DECL and RESULT_DECL immediately.
5692 2018-07-16 Martin Liska <mliska@suse.cz>
5695 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
5698 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
5700 * config/arc/arcHS.md: Update ARCHS scheduling rules.
5702 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
5704 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
5706 * config/arc/arc-cpus.def (hs4x): New cpu.
5708 * config/arc/arc-tables.opt: Regenerate.
5709 * config/arc/arc.c (arc_sched_issue_rate): New function.
5710 (TARGET_SCHED_ISSUE_RATE): Define.
5711 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
5712 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
5714 (attr tune): Add ARCHS4x tune values.
5715 (attr tune_dspmpy): Define.
5716 (*tst): Correct instruction type.
5717 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
5718 * config/arc/arcHS4x.md: New file.
5719 * config/arc/fpu.md: Update instruction type attributes.
5720 * config/arc/t-multilib: Regenerate.
5722 2018-07-16 Tom de Vries <tdevries@suse.de>
5725 * var-tracking.c (vt_initialize): Fix pre_dec handling.
5727 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5729 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
5732 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
5734 PR tree-optimization/86514
5735 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
5736 conversion to a boolean type from a type with greater precision.
5738 2018-07-16 Tom de Vries <tdevries@suse.de>
5740 * var-tracking.c (vt_initialize): Print adjusted insn slim if
5741 dump_flags request TDF_SLIM.
5743 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
5745 * fold-const.c (int_const_binop_1): Abstract...
5746 (wide_int_binop): ...wide int code here.
5747 (poly_int_binop): ...poly int code here.
5748 Abstract the rest of int_const_binop_1 into int_const_binop.
5749 * fold-const.h (wide_int_binop): New.
5750 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
5751 Remove useless PLUS/MINUS_EXPR case.
5752 (zero_nonzero_bits_from_vr): Move wide int code...
5753 (zero_nonzero_bits_from_bounds): ...here.
5754 (extract_range_from_binary_expr_1): Move mask optimization code...
5755 (range_easy_mask_min_max): ...here.
5756 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
5757 (range_easy_mask_min_max): New.
5759 2018-07-15 Jeff Law <law@redhat.com>
5762 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
5765 2018-07-14 Jim Wilson <jimw@sifive.com>
5767 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
5769 2018-07-14 Paul Koning <ni1d@arrl.net>
5771 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
5773 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
5775 * lto-streamer-out.c (copy_function_or_variable): Dump info about
5778 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
5779 Steve Munroe <munroesj52@gmail.com>
5781 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
5782 (_mm_andnot_si128): Likewise.
5783 (_mm_or_si128): Likewise.
5784 (_mm_xor_si128): Likewise.
5786 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
5789 * builtins.c (expand_builtin_memcmp): Inline the calls first
5790 when result_eq is false.
5791 (expand_builtin_strcmp): Inline the calls first.
5792 (expand_builtin_strncmp): Likewise.
5793 (inline_string_cmp): New routine. Expand a string compare
5794 call by using a sequence of char comparison.
5795 (inline_expand_builtin_string_cmp): New routine. Inline expansion
5796 a call to str(n)cmp/memcmp.
5797 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
5799 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
5801 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
5803 * config/arm/driver-arm.c: Include arm-native.h.
5804 (host_detect_local_cpu): Use auto-generated data tables.
5805 (vendors, arm_cpu_table): Delete. Move part information to ...
5806 * config/arm/arm-cpus.in: ... here.
5807 * config/arm/parsecpu.awk (gen_native): New function.
5808 (vendor, part): New CPU fields.
5809 (END): Add support for building the native CPU detection tables.
5810 * config/arm/t-arm (arm-native.h): Add build rule.
5811 (driver-arm.o): Add dependency on arm-native.h.
5813 2018-07-13 Richard Biener <rguenther@suse.de>
5816 * match.pd (addr1 - addr2): Allow either of the operand to
5819 2018-07-13 Tom de Vries <tdevries@suse.de>
5821 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
5824 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
5826 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
5827 arrays instead of numbered variables.
5829 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
5831 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
5832 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
5834 2018-07-13 Richard Biener <rguenther@suse.de>
5837 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
5838 instead of get_context_die.
5840 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
5841 Richard Biener <rguenther@suse.de>
5844 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
5845 that the loop latch destination where phi is defined.
5847 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
5849 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
5850 (riscv_expand_epilogue): Add assertion to check interrupt mode.
5851 (riscv_set_current_function): Extract getting interrupt type to new
5853 (riscv_get_interrupt_type): New function.
5854 (riscv_merge_decl_attributes): New function, checking interrupt type is
5856 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
5858 2018-07-12 Paul Koning <ni1d@arrl.net>
5860 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
5863 2018-07-12 Paul Koning <ni1d@arrl.net>
5865 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
5866 zero reference, add doloop_end instead.
5867 * doc/md.texi (decrement_and_branch_until_zero): Remove.
5868 (Looping patterns): Remove decrement_and_branch_until_zero. Add
5869 detail for doloop_end.
5871 2018-07-12 Martin Sebor <msebor@redhat.com>
5874 * attribs.c (decl_attributes): Reject conflicting attributes before
5875 calling attribute handlers.
5877 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
5879 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
5881 (gcc::dump_manager::get_dump_file_name): likewise.
5882 (dump_begin): Likewise.
5883 * dumpfile.h (dump_begin): Update prototype.
5884 (gcc::dump_manager::get_dump_file_name,
5885 gcc::dump_manager::get_dump_file_name): Update prototype.
5887 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
5889 * internal-fn.h (vectorizable_internal_fn_p): New function.
5890 * tree-vect-slp.c (compatible_calls_p): Likewise.
5891 (vect_build_slp_tree_1): Remove nops argument. Handle calls
5892 to internal functions.
5893 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
5895 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
5897 * fold-const.h (inverse_conditions_p): Declare.
5898 * fold-const.c (inverse_conditions_p): New function.
5899 * match.pd: Use inverse_conditions_p. Add folds of view_converts
5900 that test the inverse condition of a conditional internal function.
5901 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
5902 * internal-fn.c (internal_fn_mask_index): Handle conditional
5904 (vectorized_internal_fn_supported_p): New function.
5905 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
5906 (any_pred_load_store): Replace with...
5907 (need_to_predicate): ...this new variable.
5908 (redundant_ssa_names): New variable.
5909 (ifcvt_can_use_mask_load_store): Move initial checks to...
5910 (ifcvt_can_predicate): ...this new function. Handle tree codes
5911 for which a conditional internal function exists.
5912 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
5913 instead of ifcvt_can_use_mask_load_store. Update after variable
5915 (predicate_load_or_store): New function, split out from
5916 predicate_mem_writes.
5917 (check_redundant_cond_expr): New function.
5918 (value_available_p): Likewise.
5919 (predicate_rhs_code): Likewise.
5920 (predicate_mem_writes): Rename to...
5921 (predicate_statements): ...this. Use predicate_load_or_store
5922 and predicate_rhs_code.
5923 (combine_blocks, tree_if_conversion): Update after above name changes.
5924 (ifcvt_local_dce): Handle redundant_ssa_names.
5925 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
5926 general conditional functions.
5927 * tree-vect-stmts.c (vectorizable_call): Likewise.
5929 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
5930 Alan Hayward <alan.hayward@arm.com>
5931 David Sherwood <david.sherwood@arm.com>
5933 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
5934 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
5935 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
5936 plus and minus and convert them into IFN_COND_FMA-based sequences.
5937 (convert_mult_to_fma): Handle conditional plus and minus.
5939 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
5941 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
5942 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
5943 (cond_fnms_optab): New optabs.
5944 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
5946 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
5947 * internal-fn.h (get_conditional_internal_fn): Declare.
5948 (get_unconditional_internal_fn): Likewise.
5949 * internal-fn.c (cond_ternary_direct): New macro.
5950 (expand_cond_ternary_optab_fn): Likewise.
5951 (direct_cond_ternary_optab_supported_p): Likewise.
5952 (FOR_EACH_COND_FN_PAIR): Likewise.
5953 (get_conditional_internal_fn): New function.
5954 (get_unconditional_internal_fn): Likewise.
5955 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
5956 (gimple_match_op::gimple_match_op): Add a new overload for 5
5958 (gimple_match_op::set_op): Likewise.
5959 (gimple_resimplify5): Declare.
5960 * genmatch.c (decision_tree::gen): Generate simplifications for
5962 * gimple-match-head.c (gimple_simplify): Define an overload for
5963 5 operands. Handle calls with 5 arguments in the top-level overload.
5964 (convert_conditional_op): Handle conversions from unconditional
5965 internal functions to conditional ones.
5966 (gimple_resimplify5): New function.
5967 (build_call_internal): Pass a fifth operand.
5968 (maybe_push_res_to_seq): Likewise.
5969 (try_conditional_simplification): Try converting conditional
5970 internal functions to unconditional internal functions.
5971 Handle 3-operand unconditional forms.
5972 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
5973 Define ternary equivalents of the current rules for binary conditional
5975 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
5977 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
5978 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
5979 (optab): Handle them.
5980 (SVE_COND_FP_TERNARY): New int iterator.
5981 (sve_fmla_op, sve_fmad_op): New int attributes.
5982 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
5983 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
5984 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
5986 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
5988 * target.def (preferred_else_value): New target hook.
5989 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
5990 * doc/tm.texi: Regenerate.
5991 * targhooks.h (default_preferred_else_value): Declare.
5992 * targhooks.c (default_preferred_else_value): New function.
5993 * internal-fn.h (conditional_internal_fn_code): Declare.
5994 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
5995 (get_conditional_internal_fn): Use it.
5996 (conditional_internal_fn_code): New function.
5997 * gimple-match.h (gimple_match_cond): New struct.
5998 (gimple_match_op): Add a cond member function.
5999 (gimple_match_op::gimple_match_op): Update all forms to take a
6001 * genmatch.c (expr::gen_transform): Use the same condition as res_op
6002 for the suboperation, but don't specify a particular else_value.
6003 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
6004 (visit_nary_op, visit_reference_op_load): Pass
6005 gimple_match_cond::UNCOND to the gimple_match_op constructor.
6006 * gimple-match-head.c: Include tree-eh.h
6007 (convert_conditional_op): New function.
6008 (maybe_resimplify_conditional_op): Likewise.
6009 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
6010 (gimple_resimplify2): Likewise.
6011 (gimple_resimplify3): Likewise.
6012 (gimple_resimplify4): Likewise.
6013 (maybe_push_res_to_seq): Return null for conditional operations.
6014 (try_conditional_simplification): New function.
6015 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
6017 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
6019 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
6021 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
6023 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
6025 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
6027 (hash_tree): Do not hash DECL_FCONTEXT
6028 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
6029 Do not stream DECL_FCONTEXT.
6030 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
6031 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
6033 2018-07-12 Richard Biener <rguenther@suse.de>
6036 * dwarf2out.c (gen_block_die): Only output blocks when they have
6037 at least one !DECL_IGNORED_P variable.
6039 2018-07-12 Richard Biener <rguenther@suse.de>
6042 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
6043 Remove -mieee-fp handling.
6045 2018-07-12 Richard Biener <rguenther@suse.de>
6047 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
6048 left-over from last patch.
6050 2018-07-12 Jakub Jelinek <jakub@redhat.com>
6052 PR tree-optimization/86492
6053 * gimple-ssa-store-merging.c
6054 (imm_store_chain_info::coalesce_immediate_stores): Call
6055 check_no_overlap even for the merge_overlapping case. Formatting fix.
6057 2018-07-12 Richard Biener <rguenther@suse.de>
6060 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
6061 move possibly trapping operations into the conditional.
6063 2018-07-12 Richard Biener <rguenther@suse.de>
6065 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
6066 (vn_lookup_simplify_result): Remove recursion limit applied
6068 (vn_nary_build_or_lookup_1): Adjust.
6069 (try_to_simplify): Likewise.
6070 * gimple-match-head.c (gimple_resimplify1): Instead apply one
6072 (gimple_resimplify2): Likewise.
6073 (gimple_resimplify3): Likewise.
6074 (gimple_resimplify4): Likewise.
6076 2018-07-11 Jakub Jelinek <jakub@redhat.com>
6078 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
6079 Use __mmask64 type instead of __mmask8 for __M argument.
6080 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
6081 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
6083 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
6084 __mmask16 for __M argument.
6085 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
6086 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
6087 to __mmask16 instead of __mmask8.
6088 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
6089 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
6090 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
6091 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
6092 instead of __mmask16 for __U argument.
6093 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
6094 __mmask16 instead of __mmask8 for __U argument.
6095 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
6097 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
6099 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
6101 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
6103 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
6105 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
6107 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
6109 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
6111 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
6113 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
6115 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
6116 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
6117 return type as well as __M argument type and all casts from __mmask8
6119 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
6120 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
6121 return type as well as __M argument type and all casts from __mmask8
6123 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
6124 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
6125 return type as well as __M argument type and all casts from __mmask8
6127 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
6128 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
6129 return type as well as __M argument type and all casts from __mmask8
6131 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
6132 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
6135 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
6137 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
6138 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
6141 2018-07-11 Paul Koning <ni1d@arrl.net>
6143 * doc/md.texi (define_subst): Document how multiple occurrences of
6144 the same argument in the replacement pattern are handled.
6146 2018-07-11 Paul Koning <ni1d@arrl.net>
6148 * doc/extend.texi (Common Variable Attributes): Move "mode" into
6150 (Common Type Attributes): Add "mode" attribute.
6152 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
6154 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
6155 stream DECL_ORIGINAL_TYPE.
6156 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
6157 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
6158 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
6159 Do not walk original type.
6160 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
6162 (write_ts_decl_non_common_tree_pointers): Do not stream
6164 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
6165 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
6167 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
6169 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
6170 thread twice from the same starting edge.
6172 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
6174 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
6175 * gimple.c (gimple_call_nonnull_result_p): ...here...
6176 (gimple_call_nonnull_arg): ...and here.
6177 * gimple.h (gimple_call_nonnull_result_p): New.
6178 (gimple_call_nonnull_arg): New.
6180 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
6182 * config/arm/arm-cpus.in: Move information from fpu field of each
6183 cpu definition to the isa field.
6184 * config/arm/parsecpu.awk (fpu): Delete match rule.
6185 (gen_comm_data): Don't add bits from the CPU's FPU entry.
6187 2018-07-11 Richard Biener <rguenther@suse.de>
6190 * dwarf2out.c (init_sections_and_labels): Use
6191 output_asm_line_debug_info consistently.
6192 (dwarf2out_early_finish): Likewise.
6193 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
6196 2018-07-11 Richard Biener <rguenther@suse.de>
6198 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
6199 Rework father_bb setting in a way to avoid propagating constants
6200 multiple times on a loop body.
6202 2018-07-10 Mark Wielaard <mark@klomp.org>
6205 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
6207 2018-07-10 Richard Biener <rguenther@suse.de>
6209 * hash-map.h (hash_map::iterator::operator*): Return
6210 references to key and value.
6212 2018-07-10 Jakub Jelinek <jakub@redhat.com>
6215 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
6216 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
6217 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
6218 (gimplify_omp_for): For composite loops, move outer
6219 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
6220 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
6221 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
6222 TREE_LIST for both the original class iterator and the "last" helper
6223 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
6224 loop, remember has_decl_expr from outer composite loops for the
6225 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
6227 2018-07-09 Martin Sebor <msebor@redhat.com>
6231 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
6232 accessing implicitly initialized array elements.
6233 * expr.c (string_constant): Handle string initializers of
6234 character arrays within aggregates.
6235 * gimple-fold.c (fold_array_ctor_reference): Add argument.
6236 Store element offset. As a special case, handle zero size.
6237 (fold_nonarray_ctor_reference): Same.
6238 (fold_ctor_reference): Add argument. Store subobject offset.
6239 * gimple-fold.h (fold_ctor_reference): Add argument.
6241 2018-07-09 Paul Koning <ni1d@arrl.net>
6243 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
6244 (pdp11_insn_cost): New function.
6245 (pdp11_md_asm_adjust): New function.
6246 (TARGET_INVALID_WITHIN_DOLOOP): Define.
6247 (pdp11_rtx_costs): Update to match machine better.
6248 (output_addr_const_pdp11): Correct format mismatch warnings.
6249 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
6250 * config/pdp11/pdp11.md: General change to add base_cost and/or
6251 length attributes for use by new pdp11_insn_cost function.
6252 (MIN_BRANCH): Correct definition.
6254 (doloop_end): Use standard pattern name for looping pattern.
6255 (doloop_end_nocc): New.
6256 (movsf): Add another constraint alternative.
6257 (zero_extendqihi2): Add constraint alternatives for not in place
6259 (zero_extendhisi2): Remove.
6260 (shift patterns): Add CC handling variants.
6264 (define_peephole2): New peephole to recognize mov that sets CC for
6267 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6269 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
6270 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
6273 2018-07-09 Jakub Jelinek <jakub@redhat.com>
6276 * real.c (real_nextafter): Return true if result is denormal.
6278 2018-07-09 Martin Liska <mliska@suse.cz>
6280 * common.opt: Add back wrongly removed attribute.
6282 2018-07-09 Richard Biener <rguenther@suse.de>
6285 * dwarf2out.c (gen_block_die): For an early generated DIE
6286 always output high/low PC attributes.
6288 2018-07-09 Tom de Vries <tdevries@suse.de>
6290 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
6291 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
6292 onto VAR_DECL with abstract origin.
6294 2018-07-07 Jim Wilson <jimw@sifive.com>
6296 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
6298 2018-07-07 Tom de Vries <tdevries@suse.de>
6300 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
6302 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
6304 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
6306 (combine_bound): Use wide-int overflow calculation instead of
6308 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
6310 * fold-const.c (int_const_binop_2): Same.
6311 (extract_muldiv_1): Same.
6312 (fold_div_compare): Same.
6313 (fold_abs_const): Same.
6315 * poly-int.h (add): Same.
6319 * predict.c (predict_iv_comparison): Same.
6320 * profile-count.c (slow_safe_scale_64bit): Same.
6321 * simplify-rtx.c (simplify_const_binary_operation): Same.
6322 * tree-chrec.c (tree_fold_binomial): Same.
6323 * tree-data-ref.c (split_constant_offset_1): Same.
6324 * tree-if-conv.c (idx_within_array_bound): Same.
6325 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
6326 * tree-ssa-phiopt.c (minmax_replacement): Same.
6327 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
6328 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
6329 * vr-values.c (vr_values::adjust_range_with_scev): Same.
6330 * wide-int.cc (wi::add_large): Same.
6331 (wi::mul_internal): Same.
6332 (wi::sub_large): Same.
6333 (wi::divmod_internal): Same.
6334 * wide-int.h: Change overflow type to overflow_type for neg, add,
6335 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
6336 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
6337 mul_internal, divmod_internal.
6338 (overflow_type): New enum.
6339 (accumulate_overflow): New.
6341 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
6343 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
6344 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
6346 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
6348 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
6349 argument is checked for zero before entering loop, avoid checking again.
6351 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
6353 * gimplify.h (generic_expr_could_trap_p): Set as global function.
6354 * gimplify.h (generic_expr_could_trap_p): Likwise.
6355 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
6357 2018-07-06 Jakub Jelinek <jakub@redhat.com>
6359 PR tree-optimization/86401
6360 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
6361 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
6362 (fold_bit_and_mask): ... here. New helper function for match.pd.
6363 * fold-const.h (fold_bit_and_mask): Declare.
6364 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
6366 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
6369 * target.def (translate_mode_attribute): New hook.
6370 * targhooks.h (default_translate_mode_attribute): Declare.
6371 * targhooks.c (default_translate_mode_attribute): New function.
6372 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
6373 * doc/tm.texi: Regenerate.
6374 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
6375 (rs6000_translate_mode_attribute): New function.
6377 2018-07-06 Paul Koning <ni1d@arrl.net>
6379 * doc/md.texi (define_split): Document DONE and FAIL.
6380 (define_peephole2): Ditto.
6382 2018-07-05 Jeff Law <law@redhat.com>
6384 PR tree-optimization/86010
6385 * tree-ssa-dse.c (compute_trims): More aggressively trim at
6386 both the head and tail of mem* and str* calls.
6388 2018-07-05 Jim Wilson <jimw@sifive.com>
6390 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
6392 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
6394 * config/aarch64/aarch64-simd.md: correct flags text for
6395 MIN_EXPR replacement.
6397 2018-07-05 James Clarke <jrtc27@jrtc27.com>
6399 * configure: Regenerated.
6401 2018-07-05 Carl Love <cel@us.ibm.com>
6403 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
6404 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
6405 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
6406 VSX_BUILTIN_DOUBLEL_V4SF.
6408 2018-07-05 Martin Sebor <msebor@redhat.com>
6411 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
6412 than its domain to compute its the upper bound of a char array.
6414 2018-07-05 Nathan Sidwell <nathan@acm.org>
6416 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
6417 * doc/cpp.texi: Update comment.
6418 * doc/tm.texi: Rebuilt.
6419 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
6420 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
6421 * doc/extend.texi (Backwards Compatibility): Clarify it is system
6422 headers affected by extern "C".
6423 * system.h: Poison NO_IMPLICIT_EXTERN_C.
6424 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
6425 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
6426 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
6427 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
6428 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
6429 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
6430 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
6431 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
6432 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
6433 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
6434 config/sparc/sp64-elf.h, config/spu/spu.h,
6435 config/stormy16/stormy16.h, config/v850/v850.h,
6436 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
6437 define NO_IMPLICIT_EXTERN_C.
6438 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
6440 2018-07-05 Tamar Christina <tamar.christina@arm.com>
6443 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
6444 instead of GET_MODE_SIZE when comparing Units.
6446 2018-07-05 Tamar Christina <tamar.christina@arm.com>
6449 * rtlanal.c (set_noop_p): Constrain on mode change,
6450 include hard-reg-set.h
6452 2018-07-05 Tamar Christina <tamar.christina@arm.com>
6454 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
6456 2018-07-05 Jakub Jelinek <jakub@redhat.com>
6459 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
6462 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
6464 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
6466 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
6469 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
6471 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
6473 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
6476 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
6478 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
6480 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
6481 neg_*_op* variables.
6483 2018-07-04 Martin Liska <mliska@suse.cz>
6485 * tree-switch-conversion.c: Define
6486 max_ratio_for_speed and max_ratio_for_size constants.
6488 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
6489 Martin Liska <mliska@suse.cz>
6494 * common.opt: Rename align options with 'str_' prefix.
6495 * common/config/i386/i386-common.c (set_malign_value): New
6497 (ix86_handle_option): Use it to set -falign-* options/
6498 * config/aarch64/aarch64-protos.h (struct tune_params): Change
6499 type from int to string.
6500 * config/aarch64/aarch64.c: Update default values from int
6502 * config/alpha/alpha.c (alpha_override_options_after_change):
6504 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
6505 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6506 max skip conditionally.
6507 * config/i386/freebsd.h (SUBALIGN_LOG): New.
6508 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6509 max skip conditionally.
6510 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6511 max skip conditionally.
6512 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
6513 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6514 max skip conditionally.
6515 * config/i386/i386.c (struct ptt): Change type from int to
6517 (ix86_default_align): Set default values.
6518 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
6519 max skip conditionally.
6520 * config/i386/iamcu.h (SUBALIGN_LOG): New.
6521 (ASM_OUTPUT_MAX_SKIP_ALIGN):
6522 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
6523 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6524 max skip conditionally.
6525 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
6526 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
6527 * config/i386/x86-64.h (SUBALIGN_LOG): New.
6528 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6529 max skip conditionally.
6530 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
6531 * config/ia64/ia64.c (ia64_option_override): Set default values
6532 for alignment options.
6533 * config/m68k/m68k.c: Handle new str_align_* options.
6534 * config/mips/mips.c (mips_set_compression_mode): Change
6536 (mips_option_override): Set default values for options.
6537 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
6539 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6541 * config/rx/rx.c (rx_option_override): Likewise.
6542 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
6543 (LABEL_ALIGN): Use align_labels_log.
6544 (LOOP_ALIGN): Use align_loops_align.
6545 * config/s390/s390.c (s390_asm_output_function_label): Use new
6547 * config/sh/sh.c (sh_override_options_after_change):
6548 Change type of constants.
6549 * config/spu/spu.c (spu_sched_init): Likewise.
6550 * config/sparc/sparc.c (sparc_option_override): Set default
6552 * config/visium/visium.c (visium_option_override): Likewise.
6553 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
6554 emit p2align format with last argument if it's not needed.
6555 * doc/invoke.texi: Document extended format of -falign-*.
6556 * final.c: Use align_labels alignment.
6557 * flags.h (struct target_flag_state): Change type to use
6559 (struct align_flags_tuple): New.
6560 (struct align_flags): Likewise.
6561 (align_loops_log): Redefine macro to use new types.
6562 (align_loops_max_skip): Redefine macro to use new types.
6563 (align_jumps_log): Redefine macro to use new types.
6564 (align_jumps_max_skip): Redefine macro to use new types.
6565 (align_labels_log): Redefine macro to use new types.
6566 (align_labels_max_skip): Redefine macro to use new types.
6567 (align_functions_log): Redefine macro to use new types.
6568 (align_loops): Redefine macro to use new types.
6569 (align_jumps): Redefine macro to use new types.
6570 (align_labels): Redefine macro to use new types.
6571 (align_functions): Redefine macro to use new types.
6572 (align_functions_max_skip): Redefine macro to use new types.
6573 (align_loops_value): New macro.
6574 (align_jumps_value): New macro.
6575 (align_labels_value): New macro.
6576 (align_functions_value): New macro.
6577 * function.c (invoke_set_current_function_hook): Propagate
6578 alignment values from flags to global variables default in
6580 * ipa-icf.c (sem_function::equals_wpa): Use
6581 cl_optimization_option_eq instead of memcmp.
6582 * lto-streamer.h (cl_optimization_stream_out): Support streaming
6584 (cl_optimization_stream_in): Likewise.
6585 * optc-save-gen.awk: Support strings in cl_optimization.
6586 * opth-gen.awk: Likewise.
6587 * opts.c (finish_options): Remove error checking of invalid
6589 (MAX_CODE_ALIGN): Remove.
6590 (MAX_CODE_ALIGN_VALUE): Likewise.
6591 (parse_and_check_align_values): New function.
6592 (check_alignment_argument): Likewise.
6593 (common_handle_option): Use check_alignment_argument.
6594 * opts.h (parse_and_check_align_values): Declare.
6595 * toplev.c (init_alignments): Remove.
6596 (read_log_maxskip): New.
6597 (parse_N_M): Likewise.
6598 (parse_alignment_opts): Likewise.
6599 (backend_init_target): Remove usage of init_alignments.
6600 * toplev.h (parse_alignment_opts): Declare.
6601 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
6603 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
6604 * tree.c (cl_option_hasher::equal): New.
6605 * varasm.c: Use new global macros.
6607 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
6609 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
6610 Use a simpler align directive also if MAXSKIP = ALIGN-1.
6611 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6612 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6613 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6614 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
6615 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
6616 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
6617 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
6618 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6619 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6620 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6621 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6623 2018-07-04 Martin Liska <mliska@suse.cz>
6624 Jonathan Wakely <jwakely@redhat.com>
6626 * coverage.c: Use correct type.
6627 * doc/invoke.texi: Language correction.
6629 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
6632 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
6633 ENDBRANCH for non-tail call which may return via indirect branch.
6634 * doc/extend.texi: Document indirect_return attribute.
6636 2018-07-03 Martin Sebor <msebor@redhat.com>
6638 PR tree-optimization/86274
6639 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
6641 (format_floating): Correct handling of infinities and NaNs.
6643 2018-07-03 Martin Sebor <msebor@redhat.com>
6645 * print-tree.c (print_real_cst): New function.
6646 (print_node_brief): Call it.
6647 (print_node): Ditto.
6649 2018-07-03 Jeff Law <law@redhat.com>
6651 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
6652 into a single pattern.
6654 * config/h8300/h8300.md (ors code_iterator): New.
6655 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
6656 a single pattern and single splitter.
6657 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
6658 (iorqi3_1, xorqi3_1): Likewise.
6659 (iorqi3, xorqi3 expanders): Similarly.
6661 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
6662 (movmd_internal) into a single pattern using the P mode iterator.
6663 (movmd splitters): Similarly.
6664 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
6665 (movsd splitters): Similarly.
6667 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
6668 ADDB, ADDW and ADDL into a single ADD attribute which selects the
6669 right table based on the size of the operand.
6670 * config/h8300/h8300.md (length_table): Corresponding changes. All
6671 references to "addb", "addw" and "addl" changed to "add".
6672 (btst patterns): Merge two variants into a single pattern.
6673 (tstqi, tsthi): Likewise.
6674 (addhi3_incdec, addsi3_incdec): Likewise.
6675 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
6676 (mulhi3, mulsi3): Likewise.
6677 (udivhi3, udivsi3): Likewise.
6678 (divhi3, divsi3): Likewise.
6679 (andorqi3, andorhi3, andorsi3): Likewise.
6681 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
6684 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
6685 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
6687 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6689 PR tree-optimization/85694
6690 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
6691 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
6693 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
6694 (<u>avg<mode>3_ceil): New patterns.
6696 2018-07-03 David Malcolm <dmalcolm@redhat.com>
6698 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
6699 scan-tree-dump directive.
6700 * gcc.dg/vect/slp-perm-2.c: Likewise.
6701 * gcc.dg/vect/slp-perm-3.c: Likewise.
6702 * gcc.dg/vect/slp-perm-5.c: Likewise.
6703 * gcc.dg/vect/slp-perm-6.c: Likewise.
6704 * gcc.dg/vect/slp-perm-7.c: Likewise.
6705 * gcc.dg/vect/slp-perm-8.c: Likewise.
6707 2018-07-03 Marek Polacek <polacek@redhat.com>
6710 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
6712 2018-07-03 Richard Biener <rguenther@suse.de>
6715 * tree-ssa-structalias.c (find_func_clobbers): Properly
6716 handle indirect calls.
6718 2018-07-03 Jeff Law <law@redhat.com>
6720 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
6721 (shifts): New code iterator.
6722 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
6723 expander. Fix HImode handling on H8/SX.
6724 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
6725 (subqi3, subhi3, subsi3 expanders): Likewise.
6726 (andqi3, andhi3, andsi3 expanders): Likewise.
6727 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
6728 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
6729 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
6730 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
6731 (zero_extendqihi2, zero_extendqisi2): Likewise.
6732 (extendqihi2, extendqisi2): Likewise.
6733 (rotlqi3, rotlhi3, rotlsi3): Likewise.
6734 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
6735 (rotlqi3_1, rotlhi3_1): Likewise.
6736 (logicalhi3_sn, logicalsi3_sn): Likewise.
6737 (logicalhi3, logicalsi3): Likewise.
6739 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6741 * tree-vect-patterns.c (vect_recog_rotate_pattern)
6742 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
6743 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
6744 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
6745 type to append_pattern_def_seq instead of creating a stmt_vec_info
6747 (build_mask_conversion): Likewise. Remove vinfo argument.
6748 (vect_add_conversion_to_patterm): Likewise, renaming to...
6749 (vect_add_conversion_to_pattern): ...this.
6750 (vect_recog_mask_conversion_pattern): Update call to
6751 build_mask_conversion. Pass the vector type to
6752 append_pattern_def_seq here too.
6753 (vect_recog_gather_scatter_pattern): Update call to
6754 vect_add_conversion_to_pattern.
6756 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6758 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
6759 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
6760 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
6761 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
6762 STMT_VINFO_PATTERN_DEF_SEQ to null here.
6763 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
6764 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
6765 append_pattern_def_seq instead of new_pattern_def_seq.
6766 (vect_recog_divmod_pattern): Do both of the above.
6767 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
6770 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6772 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
6773 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
6774 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
6775 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
6776 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
6777 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
6778 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
6779 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
6780 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
6781 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
6782 parameter with a single stmt_vec_info.
6783 (vect_recog_func_ptr): Likewise.
6784 (vect_recog_gather_scatter_pattern): Likewise, folding in...
6785 (vect_try_gather_scatter_pattern): ...this.
6786 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
6787 the stmt_vec_info of the statement to be matched. Don't clear
6788 STMT_VINFO_RELATED_STMT.
6789 (vect_pattern_recog): Update call accordingly.
6791 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6793 PR tree-optimization/85694
6794 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
6795 (uavgM3_ceil): Document new optabs.
6796 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
6797 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
6799 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
6800 (savg_ceil_optab): New optabs.
6801 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
6802 (vect_vect_recog_func_ptrs): Add it.
6803 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
6804 constant directly from the associated lhs.
6806 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6808 * tree-vect-patterns.c (vect_split_statement): New function.
6809 (vect_convert_input): Use it to try to split an existing cast.
6811 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6813 * poly-int.h (print_hex): New function.
6814 * dumpfile.h (dump_dec, dump_hex): Declare.
6815 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
6816 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
6817 min_input_precision, operation_precision and operation_sign.
6818 * tree-vect-patterns.c (vect_get_range_info): New function.
6819 (vect_same_loop_or_bb_p, vect_single_imm_use)
6820 (vect_operation_fits_smaller_type): Delete.
6821 (vect_look_through_possible_promotion): Add an optional
6822 single_use_p parameter.
6823 (vect_recog_over_widening_pattern): Rewrite to use new
6824 stmt_vec_info infomration. Handle one operation at a time.
6825 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
6826 (vect_truncatable_operation_p, vect_set_operation_type)
6827 (vect_set_min_input_precision): New functions.
6828 (vect_determine_min_output_precision_1): Likewise.
6829 (vect_determine_min_output_precision): Likewise.
6830 (vect_determine_precisions_from_range): Likewise.
6831 (vect_determine_precisions_from_users): Likewise.
6832 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
6833 (vect_vect_recog_func_ptrs): Put over_widening first.
6835 (vect_pattern_recog): Call vect_determine_precisions.
6837 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6839 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
6840 statements that have been replaced by further pattern statements.
6841 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
6843 2018-07-03 Richard Biener <rguenther@suse.de>
6845 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
6846 always set *dt. Dump vectype in vectype overload.
6847 * dumpfile.h (dump_gimple_expr): New function.
6848 (dump_gimple_expr_loc): Likewise.
6849 * dumpfile.c (dump_gimple_expr): New function.
6850 (dump_gimple_expr_loc): Likewise.
6852 2018-07-02 Jeff Law <law@redhat.com>
6854 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
6855 the H8/300, H8/300H and H8/S variants into a single pattern.
6856 (movhi_h8300, movqi_h8300hs): Similarly.
6857 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
6858 (QHI mode iterator): New.
6860 * config/h8300/h8300.md: Remove trailing whitespace.
6862 2018-07-02 Jim Wilson <jimw@sifive.com>
6864 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
6865 instead of emit_insn for interrupt returns.
6866 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
6867 (riscv_sret, riscv_uret): Likewise.
6869 2018-07-02 David Malcolm <dmalcolm@redhat.com>
6871 * pretty-print.c (selftest::test_pp_format): Move save and restore
6872 of quotes to class auto_fix_quotes, and add an instance.
6873 * selftest.c: Include "intl.h".
6874 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
6875 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
6876 * selftest.h (selftest::auto_fix_quotes): New class.
6878 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
6880 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
6881 (aarch64_sve_prepare_conditional_op): Remove.
6882 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
6883 Allow aarch64_simd_reg_or_zero as select operand; remove
6884 the aarch64_sve_prepare_conditional_op call.
6885 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
6886 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
6887 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
6888 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
6889 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
6890 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
6891 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
6892 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
6893 and a splitters to match all of the *_any patterns.
6894 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
6896 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
6897 (SVE_COND_FP_BINARY_REV): Remove.
6898 (sve_int_op_rev, sve_fp_op_rev): New.
6899 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
6900 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
6901 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
6902 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
6903 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
6904 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
6905 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
6906 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
6907 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
6909 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
6910 Remove match_dup 1 from the inner unspec.
6911 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
6913 * config/aarch64/aarch64.md (movprfx): New attr.
6914 (length): Default movprfx to 8.
6915 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
6916 (*madd<SVE_I>, *msub<SVE_I): Likewise.
6917 (*<su>mul<SVE_I>3_highpart): Likewise.
6918 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
6919 (*v<ASHIFT><SVE_I>3): Likewise.
6920 (*<su><MAXMIN><SVE_I>3): Likewise.
6921 (*<su><MAXMIN><SVE_F>3): Likewise.
6922 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
6923 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
6924 (*div<SVE_F>4): Likewise.
6926 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
6928 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
6931 2018-07-02 Richard Biener <rguenther@suse.de>
6933 PR tree-optimization/86363
6934 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
6935 memset argument refers to a non-variable address.
6937 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
6939 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
6940 {PLUS,MINUS}_EXPR code to...
6941 (adjust_symbolic_bound): ...here,
6942 (combine_bound): ...here,
6943 (set_value_range_with_overflow): ...and here.
6945 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
6947 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
6949 (extract_range_from_abs_expr): ...here.
6951 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
6953 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
6954 -fno-omit-frame-pointer when not optimizing.
6956 2018-07-02 Martin Liska <mliska@suse.cz>
6959 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
6960 (propagate_nothrow): Likewise.
6962 2018-07-02 Martin Liska <mliska@suse.cz>
6965 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
6967 2018-07-02 David Malcolm <dmalcolm@redhat.com>
6969 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
6970 function in r262149, changing "loc" param from source_location to
6971 const dump_location_t &.
6972 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
6973 declaration, as above.
6975 2018-07-01 Paul Koning <ni1d@arrl.net>
6977 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
6978 -munit-asm, -mgnu-asm, -mdec-asm.
6979 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
6980 (pdp11_output_labelref): New.
6981 (pdp11_output_def): New.
6982 (pdp11_output_addr_vec_elt): New.
6983 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
6984 %# and %@ format codes.
6985 (pdp11_option_override): New.
6986 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
6987 (pdp11_output_ident): New.
6988 (pdp11_asm_named_section): New.
6989 (pdp11_asm_init_sections): New.
6990 (pdp11_file_start): New.
6991 (pdp11_file_end): New.
6992 (output_ascii): Use .ascii/.asciz for -mdec-asm.
6993 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
6994 %o, like %c but octal.
6995 (pdp11_option_override): New.
6996 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
6998 (DATA_SECTION_ASM_OP): Ditto.
6999 (READONLY_DATA_SECTION_ASM_OP): New.
7000 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
7001 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
7002 (ASM_OUTPUT_LABELREF): Ditto.
7003 (ASM_OUTPUT_DEF): Ditto.
7004 (ASM_OUTPUT_EXTERNAL): New.
7005 (ASM_OUTPUT_SOURCE_FILENAME): New.
7006 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
7007 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
7008 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
7009 %# and %@ format codes.
7010 * config/pdp11/pdp11.opt (mgnu-asm): New.
7011 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
7012 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
7013 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
7015 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
7017 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
7018 dereferencing path[] beyond its length.
7020 (debug_all_paths): New.
7021 (rewire_first_differing_edge): New.
7022 (adjust_paths_after_duplication): New.
7023 (duplicate_thread_path): Call adjust_paths_after_duplication.
7025 (thread_through_all_blocks): Add new argument to
7026 duplicate_thread_path.
7028 2018-06-30 Jim Wilson <jimw@sifive.com>
7030 * config/riscv/predicates.md (p2m1_shift_operand): New.
7031 (high_mask_shift_operand): New.
7032 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
7033 pattern using p2m1_shift_operand.
7034 (lshsi3_zero_extend_3+2): New combiner pattern using
7035 high_mask_shift_operand.
7037 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7039 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
7041 (vect_recog_rotate_pattern): ...here.
7042 (vect_convert_input): Try to insert casts of invariants in the
7044 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
7045 preheader to be empty.
7047 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7049 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
7050 vector type. If given, install it in the new statement's
7052 (vect_element_precision): New function.
7053 (vect_unpromoted_value): New struct.
7054 (vect_unpromoted_value::vect_unpromoted_value): New function.
7055 (vect_unpromoted_value::set_op): Likewise.
7056 (vect_look_through_possible_promotion): Likewise.
7057 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
7058 (vect_widened_op_tree, vect_convert_input): Likewise.
7059 (vect_convert_inputs, vect_convert_output): Likewise.
7060 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
7061 to handle the optional cast of the multiplication result and
7062 vect_widened_op_tree to detect the widened multiplication itself.
7063 Do not require the input and output of promotion casts to have
7064 the same sign, but base the signedness of the operation on the
7065 input rather than the result. If the pattern includes two
7066 promotions, check that those promotions have the same sign.
7067 Do not restrict the MULT_EXPR handling to a double-width result;
7068 handle quadruple-width results and wider. Use vect_convert_inputs
7069 to convert the inputs to the common type.
7070 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
7071 to handle the optional cast of the ABS result. Also allow a sign
7072 change or a sign extension between the ABS and MINUS.
7073 Use vect_widened_op_tree to detect the widened subtraction and use
7074 vect_convert_inputs to convert the inputs to the common type.
7075 (vect_handle_widen_op_by_const): Delete.
7076 (vect_recog_widen_op_pattern): New function.
7077 (vect_recog_widen_mult_pattern): Use it.
7078 (vect_recog_widen_shift_pattern): Likewise.
7079 (vect_recog_widen_sum_pattern): Use
7080 vect_look_through_possible_promotion to handle the promoted
7083 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7085 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
7086 the containing gimple_seq *.
7087 * gimple-iterator.h (gsi_for_stmt): Declare it.
7088 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
7089 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
7090 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
7091 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
7092 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
7094 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
7096 (vect_mark_pattern_stmts): ...here. Handle cases in which the
7097 statement being replaced is part of an existing pattern
7098 definition sequence, inserting the new pattern statements before
7100 (vect_pattern_recog_1): Don't return a bool. If the statement
7101 is already part of a pattern, instead apply pattern matching
7102 to the pattern definition statements. Don't clear the
7103 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
7104 (vect_pattern_recog): Don't break after the first match;
7105 continue processing the pattern definition statements instead.
7106 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
7108 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7110 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
7111 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
7112 (vect_recog_widen_sum_pattern): Use it.
7114 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7116 * tree-vect-loop.c (vectorizable_reduction): Assert that the
7117 phi is not a pattern statement and has not been replaced by
7118 a pattern statement.
7119 * tree-vect-patterns.c (type_conversion_p): Don't check
7120 STMT_VINFO_IN_PATTERN_P.
7121 (vect_recog_vector_vector_shift_pattern): Likewise.
7122 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
7123 the pattern statement rather than the original statement; check
7124 directly for a WIDEN_MULT_EXPR here.
7125 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
7126 vect_is_simple_use to return the pattern statement rather
7127 than the original statement; use is_pattern_stmt_p to check
7128 for such a pattern statement.
7129 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
7130 to return the pattern statement rather than the original statement;
7131 don't do the same transformation here.
7132 (vect_is_simple_use): If the defining statement has been replaced
7133 by a pattern statement, return the pattern statement instead.
7134 Remove the corresponding (local) transformation from the vectype
7137 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7139 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
7140 end and default to null.
7141 * tree-vect-loop.c (vect_create_epilog_for_reduction)
7142 (vectorizable_reduction): Update calls accordingly, dropping the
7143 gimple ** argument if the passed-back statement isn't needed.
7144 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
7145 (vect_recog_rotate_pattern): Likewise.
7146 (vect_recog_mask_conversion_pattern): Likewise.
7147 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
7148 (vect_mask_constant_operand_p): Likewise.
7149 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
7150 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
7151 (get_group_load_store_type, get_load_store_type): Likewise.
7152 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
7153 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
7154 (vectorizable_conversion, vectorizable_assignment): Likewise.
7155 (vectorizable_shift, vectorizable_operation): Likewise.
7156 (vectorizable_store, vect_is_simple_cond): Likewise.
7157 (vectorizable_condition, vectorizable_comparison): Likewise.
7158 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
7159 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
7160 and move it to the end. Cope with null def_stmt_outs.
7162 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
7164 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
7166 2018-06-29 Jeff Law <law@redhat.com>
7168 * config/v850/v850.c (v850_legitimate_address_p): Handle large
7169 displacements for TARGET_V850E2V3 and newer.
7170 (TARGET_LRA_P): Remove. Defaults to LRA now.
7171 * config/v850/v850.md (sign23byte_load): Remove.
7172 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
7173 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
7175 2018-06-29 Martin Liska <mliska@suse.cz>
7178 * coverage.c (coverage_init): Mangle full path name.
7179 * doc/invoke.texi: Document the change.
7180 * gcov-io.c (mangle_path): New.
7181 * gcov-io.h (mangle_path): Likewise.
7182 * gcov.c (mangle_name): Use mangle_path for path mangling.
7184 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7186 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
7187 if starting source register is not even.
7189 2018-06-29 Martin Liska <mliska@suse.cz>
7191 PR tree-optimization/86263
7192 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
7193 Make edge redirection.
7195 2018-06-29 David Malcolm <dmalcolm@redhat.com>
7197 * dumpfile.c (dump_loc): Add indentation based on scope depth.
7198 (dump_scope_depth): New variable.
7199 (get_dump_scope_depth): New function.
7200 (dump_begin_scope): New function.
7201 (dump_end_scope): New function.
7202 * dumpfile.h (get_dump_scope_depth): New declaration.
7203 (dump_begin_scope): New declaration.
7204 (dump_end_scope): New declaration.
7205 (class auto_dump_scope): New class.
7206 (AUTO_DUMP_SCOPE): New macro.
7207 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
7210 2018-06-29 Richard Biener <rguenther@suse.de>
7212 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
7213 compute_all_dependences succeeds.
7214 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
7215 exceed --param loop-max-datarefs-for-datadeps.
7217 2018-06-29 Jakub Jelinek <jakub@redhat.com>
7219 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
7221 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
7224 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
7225 alternative 0 in preferred_for_speed attribute.
7227 2018-06-28 Paul Koning <ni1d@arrl.net>
7229 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
7230 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
7231 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
7232 * config/pdp11/pdp11.md: Correct "length" attribute calculation
7233 for shift insn patterns.
7235 2018-06-28 David Malcolm <dmalcolm@redhat.com>
7237 * cgraph.c (cgraph_node::get_body): Replace assignments to
7238 "dump_file" with calls to set_dump_file.
7239 * dumpfile.c (alt_dump_file): Make static, and group with...
7240 (alt_flags): ...this definition.
7241 (dumps_are_enabled): New variable.
7242 (refresh_dumps_are_enabled): New function.
7243 (set_dump_file): New function.
7244 (set_alt_dump_file): New function.
7245 (gcc::dump_manager::dump_start): Replace assignments to
7246 "dump_file" and "alt_dump_file" with calls to set_dump_file and
7248 (gcc::dump_manager::dump_finish): Likewise.
7249 * dumpfile.h (alt_dump_file): Delete decl.
7250 (dumps_are_enabled): New variable decl.
7251 (set_dump_file): New function decl.
7252 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
7254 * tree-nested.c (lower_nested_functions): Replace assignments to
7255 "dump_file" with calls to set_dump_file.
7257 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
7259 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
7260 goto_locus of each outgoing edge of each basic block.
7262 2018-06-28 Richard Biener <rguenther@suse.de>
7264 * dwarf2out.c (decl_scope_table): Remove.
7265 (push_decl_scope): Likewise.
7266 (pop_decl_scope): Likewise.
7267 (gen_type_die_for_member): Do not call push/pop_decl_scope.
7268 (gen_struct_or_union_type_die): Likewise.
7269 (gen_tagged_type_die): Likewise.
7270 (dwarf2out_init): Do not initialize decl_scope_table.
7271 (dwarf2out_c_finalize): Do not free it.
7273 2018-06-28 Richard Biener <rguenther@suse.de>
7275 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
7276 deciding whether to not re-use a DIE.
7278 2018-06-28 Richard Biener <rguenther@suse.de>
7280 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
7281 DW_AT_abstract_origin attribute.
7283 2018-06-28 Martin Liska <mliska@suse.cz>
7285 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
7286 Use newly introduced constants.
7287 * tree-switch-conversion.h (struct jump_table_cluster):
7288 Define max_ratio_for_size and max_ratio_for_speed.
7290 2018-06-28 Martin Liska <mliska@suse.cz>
7292 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
7293 Add new checking assert to catch invalid state.
7294 (jump_table_cluster::can_be_handled): Handle single case
7296 (jump_table_cluster::is_beneficial): Bail out for such case.
7297 (bit_test_cluster::find_bit_tests):
7298 Add new checking assert to catch invalid state.
7299 (bit_test_cluster::can_be_handled): Handle single case
7301 (bit_test_cluster::is_beneficial): Bail out for such case.
7302 (switch_decision_tree::analyze_switch_statement):
7305 2018-06-28 Martin Liska <mliska@suse.cz>
7307 * common.opt: Introduce -completion option.
7308 * gcc.c (driver_handle_option): Handle it.
7309 (driver::main): Print completions if completion
7311 * opt-suggestions.c (option_proposer::get_completions):
7313 (option_proposer::suggest_completion): Likewise.
7314 (option_proposer::find_param_completions): Likewise.
7315 (verify_autocompletions): Likewise.
7316 (test_completion_valid_options): Likewise.
7317 (test_completion_valid_params): Likewise.
7318 (in_completion_p): Likewise.
7319 (empty_completion_p): Likewise.
7320 (test_completion_partial_match): Likewise.
7321 (test_completion_garbage): Likewise.
7322 (opt_proposer_c_tests): Likewise.
7323 * opt-suggestions.h: Declare new functions.
7324 * opts.c (common_handle_option): Handle OPT__completion_.
7325 * selftest-run-tests.c (selftest::run_tests): Add
7326 opt_proposer_c_tests.
7327 * selftest.c (assert_str_startswith): New.
7328 * selftest.h (assert_str_startswith): Likewise.
7329 (opt_proposer_c_tests): New.
7330 (ASSERT_STR_STARTSWITH): Likewise.
7332 2018-06-28 Martin Liska <mliska@suse.cz>
7334 * Makefile.in: Add opt-suggestions.o.
7335 * gcc-main.c: Include opt-suggestions.h.
7336 * gcc.c (driver::driver): Likewise.
7337 (driver::~driver): Remove m_option_suggestions.
7338 (driver::build_option_suggestions): Moved to option_proposer.
7339 (driver::suggest_option): Likewise.
7340 (driver::handle_unrecognized_options): Use option_proposer.
7341 * gcc.h (class driver): Add new memver m_option_proposer.
7342 * opt-suggestions.c: New file.
7343 * opt-suggestions.h: New file.
7345 2018-06-28 Martin Liska <mliska@suse.cz>
7347 * vec.h (class auto_string_vec): New (moved from auto_argvec).
7348 (auto_string_vec::~auto_string_vec): Likewise.
7350 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
7352 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
7353 prevent_decl_creation_for_types fields up and add reset_location field.
7354 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
7355 statement if id->reset_location is true.
7356 (copy_edges_for_bb): Do not set goto_locus on the new edges if
7357 id->reset_location is true.
7358 (copy_phis_for_bb): Force input_location on the arguments if
7359 id->reset_location is true.
7360 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
7361 is set on the function to be inlined.
7363 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
7365 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
7367 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
7369 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
7370 registers for Pmode.
7371 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
7372 hard registers for the clobbered pseudo.
7374 2018-06-27 Paul Koning <ni1d@arrl.net>
7376 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
7377 mutually exclusive options.
7378 * config/pdp11/constraints.md (h): New constraint.
7379 (O): Update definition to match shift code generation.
7380 (D): New constraint.
7381 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
7383 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
7385 (output_jump): Change arguments.
7386 (pdp11_fixed_cc_regs): New function.
7387 (pdp11_cc_mode): Ditto.
7388 (pdp11_expand_shift): Ditto.
7389 (pdp11_assemble_shift): Ditto.
7390 (pdp11_small_shift): Ditto.
7391 (pdp11_branch_cost): Remove.
7392 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
7394 (pdp11_register_move_cost): Update for CC registers.
7395 (pdp11_rtx_costs): Add case for LSHIFTRT.
7396 (pdp11_output_jump): Add CCNZ mode conditional branches.
7397 (notice_update_cc_on_set): Remove.
7398 (pdp11_cc_mode): New function.
7399 (simple_memory_operand): Correct pre/post decrement case.
7400 (no_side_effect_operand): New function.
7401 (pdp11_regno_reg_class): Add CC_REGS class.
7402 (pdp11_fixed_cc_regs): New function.
7403 (pdp11_small_shift): New function.
7404 (pdp11_expand_shift): New function to expand shift insns.
7405 (pdp11_assemble_shift): New function to output shifts.
7406 (pdp11_branch_cost): Remove.
7407 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
7408 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
7409 (WCHAR_TYPE): Ditto.
7410 (PTRDIFF_TYPE): Ditto.
7411 (ADJUST_INSN_LENGTH): New macro.
7412 (FIXED_REGISTERS): Add CC registers.
7413 (CALL_USED_REGISTERS): Ditto.
7415 (REG_CLASS_NAMES): Ditto.
7416 (REG_CLASS_CONTENTS): Ditto.
7417 (SELECT_CC_MODE): Use new function.
7418 (TARGET_FLAGS_REGNUM): New macro.
7419 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
7420 (cc0_reg_rtx): Remove.
7421 (CC_STATUS_MDEP): Remove.
7422 (CC_STATUS_MDEFP_INIT): Remove.
7423 (CC_IN_FPU): Remove.
7424 (NOTICE_UPDATE_CC): Remove.
7425 (REGISTER_NAMES): Add CC registers.
7426 (BRANCH_COST): Change to constant 1.
7427 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
7429 * config/pdp11/pdp11.opt (mbcopy): Remove.
7430 (mbcopy-builtin): Remove.
7431 (mbranch-cheap): Remove.
7432 (mbranch-expensive): Remove.
7433 * config/pdp11/predicates.md (expand_shift_operand): Update to
7434 match shift code generation.
7435 (ccnz_operator): New predicate.
7436 * doc/invoke.texi (PDP-11 Options): Remove deleted options
7437 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
7438 Remove non-existent option -mabshi, -mno-abshi. Document mutually
7440 * doc/md.texi (PDP-11): Document new D and h constraints. Update
7441 description of O constraint.
7443 2018-06-27 Jeff Law <law@redhat.com>
7444 Austin Law <austinklaw@gmail.com>
7446 * config/v850/v850.md (addsi3_set_flags): New pattern.
7447 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
7448 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
7449 (zero_extendhisi2_v850_set_flags): Likewise.
7450 (zero_extendqisi2_v850_set_flags): Likewise.
7451 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
7452 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
7453 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
7455 * config/v850/v850-protos.h (notice_update_cc): Remove.
7456 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
7457 (v850_print_operand): Handle 'D' and "d".
7458 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
7459 Add handling of arithmetic/logical operations compared against zero.
7460 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
7461 Do not look at v850_compare_op, instead get mode from last argument.
7462 (v850_gen_compare): Remove
7463 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
7464 after reload for prologue insns.
7465 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
7467 (construct_save_jarl): Likewise.
7468 (TARGET_FLAGS_REGNUM): Define.
7469 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
7470 (NOTICE_UPDATE_CC): Remove.
7471 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
7472 than cc0. Conditionalize on reload_completed.
7473 (cmpsi_insn, setfcc_insn): Likewise.
7474 (tst1 splitter): Turn into define_and_split which sets the flags
7476 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
7477 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
7478 (cstoresf4, cstoredf4): Clobber the flags.
7479 (cmpsi, cmpsf, cmpdf): Remove expanders.
7480 (setf_insn): Remove pattern.
7481 (addsi3): Turn into define_and_split which clobbers the flags after
7482 reload and a suitable pattern (addsi3_clobber_flags) for use after
7484 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
7485 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
7486 (ashrsi3, ashrsi3_v850e2): Likewise.
7487 (bins): Clobber the flags.
7488 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
7489 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
7490 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
7491 (call_value_internal_short, call_value_internal_long): Likewise.
7492 (callt_save_interrupt, callt_return_interrupt): Likewise.
7493 (save_interrupt, return_interrupt): Likewise.
7494 (callt_save_all_interrupt, save_all_interrupt): Likewise.
7495 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
7496 (restore_all_interrupt, _restore_all_interrupt): Likewise.
7497 (All FP comparisons): Only allow after reload has completed.
7499 (divh, divhu): Tweak output template.
7500 (branch_z_normal, branch_z_invert): Remove
7501 (branch_nz_normal, branch_nz_invert): Likewise.
7502 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
7504 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
7505 * config/v850/v850.c (notice_update_cc): Remove.
7506 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
7507 (CC_NO_CARRY): Likewise.
7508 (NOTICE_UPDATE_CC): Define to nothing.
7509 * config/v850/v850.md: Remove block comment on cc0 handling
7510 Remove "cc" attribute from all patterns. Remove cc_status handling
7511 from all patterns. Minor formatting fixes.
7513 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7515 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
7516 (cortex-a76.cortex-a55): Likewise.
7517 * config/aarch64/aarch64-tune.md: Regenerate.
7518 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
7519 cortex-a76.cortex-a55.
7521 2018-06-27 Jeff Law <law@redhat.com>
7523 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
7524 (MULTILIB_DIRNAMES): Similarly.
7526 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
7528 * gimple.h (gimple_return_retbnd): Delete.
7529 (gimple_return_set_retbnd): Likewise.
7530 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
7531 gimple_return_set_retbnd.
7532 * gimple-pretty-print.c (dump_gimple_return): Remove call to
7533 gimple_return_retbnd and adjust.
7534 * tree-inline.h (struct copy_body_data): Remove retbnd field.
7535 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
7536 Explicitly return NULL in a couple more cases. Move assertion
7537 on debug statements and remove unreachable code.
7538 (reset_debug_binding): Do not test id->retbnd.
7539 (expand_call_inline): Do not set it.
7541 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
7543 * configure.ac: Add --disable-gcov option.
7544 * configure: Regenerate.
7545 * Makefile.in: Honour @enable_gcov@.
7546 * doc/install.texi: Document --disable-gcov.
7548 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7550 * config/arm/arm-cpus.in (cortex-a76): New entry.
7551 (cortex-a76.cortex-a55): Likewise.
7552 * config/arm/arm-tables.opt: Regenerate.
7553 * config/arm/arm-tune.md: Likewise.
7554 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
7555 * doc/invoke.texi (ARM Options): Document cortex-a76 and
7556 cortex-a76.cortex-a55.
7558 2018-06-27 Tamar Christina <tamar.christina@arm.com>
7561 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
7563 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
7565 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
7567 (EPILOGUE_USES): Likewise.
7569 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
7571 * tree-inline.c (remap_location): New function extracted from...
7572 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
7573 (copy_phis_for_bb): ...here. Call remap_location.
7574 (copy_cfg_body): Adjust call to copy_edges_for_bb.
7576 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
7578 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
7579 unaligned vsx for 16B memset.
7581 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
7584 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
7585 ieee128_float_type_node to long_double_type_node unless
7586 TARGET_LONG_DOUBLE_128 is set.
7588 2018-06-26 David Malcolm <dmalcolm@redhat.com>
7590 * cfgloop.c (get_loop_location): Convert return type from
7591 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
7592 by implicit construction from rtx_insn *, and using
7593 dump_user_location_t::from_function_decl for the fallback case.
7594 * cfgloop.h (get_loop_location): Convert return type from
7595 location_t to dump_user_location_t.
7596 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
7597 dump_printf_loc to pass in a dump_location_t rather than a
7598 location_t, via the gimple stmt.
7599 * coverage.c (get_coverage_counts): Update calls to
7600 dump_printf_loc to pass in dump_location_t rather than a
7602 * doc/optinfo.texi (Dump types): Convert example of
7603 dump_printf_loc from taking "locus" to taking "insn". Update
7604 description of the "_loc" calls to cover dump_location_t.
7605 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
7607 (dump_user_location_t::dump_user_location_t): New constructors,
7608 from gimple *stmt and rtx_insn *.
7609 (dump_user_location_t::from_function_decl): New function.
7610 (dump_loc): Make static.
7611 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
7612 const dump_location_t &.
7613 (dump_generic_expr_loc): Delete.
7614 (dump_printf_loc): Convert param "loc" from location_t to
7615 const dump_location_t &.
7616 (selftest::test_impl_location): New function.
7617 (selftest::dumpfile_c_tests): New function.
7618 * dumpfile.h: Include "profile-count.h".
7619 (class dump_user_location_t): New class.
7620 (struct dump_impl_location_t): New struct.
7621 (class dump_location_t): New class.
7622 (dump_printf_loc): Convert 2nd param from source_location to
7623 const dump_location_t &.
7624 (dump_generic_expr_loc): Delete.
7625 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
7626 const dump_location_t &.
7627 * gimple-fold.c (fold_gimple_assign): Update call to
7628 dump_printf_loc to pass in a dump_location_t rather than a
7629 location_t, via the gimple stmt.
7630 (gimple_fold_call): Likewise.
7631 * gimple-loop-interchange.cc
7632 (loop_cand::analyze_iloop_reduction_var): Update for change to
7633 check_reduction_path.
7634 (tree_loop_interchange::interchange): Update for change to
7636 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
7637 change in return-type of find_loop_location.
7638 (graphite_regenerate_ast_isl): Likewise.
7639 * graphite-optimize-isl.c (optimize_isl): Likewise.
7640 * graphite.c (graphite_transform_loops): Likewise.
7641 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
7642 pass in a dump_location_t rather than a location_t, via the
7644 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
7645 * ipa.c (walk_polymorphic_call_targets): Likewise.
7646 * loop-unroll.c (report_unroll): Convert "locus" param from
7647 location_t to dump_location_t.
7648 (decide_unrolling): Update for change to get_loop_location's
7650 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
7651 location_t to dump_user_location_t.
7652 (grid_find_single_omp_among_assignments_1): Updates calls to
7653 dump_printf_loc to pass in a dump_location_t rather than a
7654 location_t, via the gimple stmt.
7655 (grid_parallel_clauses_gridifiable): Convert "tloc" from
7656 location_t to dump_location_t. Updates calls to dump_printf_loc
7657 to pass in a dump_location_t rather than a location_t, via the
7659 (grid_inner_loop_gridifiable_p): Likewise.
7660 (grid_dist_follows_simple_pattern): Likewise.
7661 (grid_gfor_follows_tiling_pattern): Likewise.
7662 (grid_target_follows_gridifiable_pattern): Likewise.
7663 (grid_attempt_target_gridification): Convert initialization
7664 of local "grid" from memset to zero-initialization; FIXME: does
7665 this require C++11? Update call to dump_printf_loc to pass in a
7666 optinfo_location rather than a location_t, via the gimple stmt.
7667 * profile.c (read_profile_edge_counts): Updates call to
7668 dump_printf_loc to pass in a dump_location_t rather than a
7670 (compute_branch_probabilities): Likewise.
7671 * selftest-run-tests.c (selftest::run_tests): Call
7673 * selftest.h (dumpfile_c_tests): New decl.
7674 * tree-loop-distribution.c (pass_loop_distribution::execute):
7675 Update for change in return type of find_loop_location.
7676 * tree-parloops.c (parallelize_loops): Likewise.
7677 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
7678 "locus" from location_t to dump_user_location_t.
7679 (canonicalize_loop_induction_variables): Likewise.
7680 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
7681 for change in return type of find_loop_location.
7682 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
7683 to dump_printf_loc to pass in a dump_location_t rather than a
7684 location_t, via the stmt.
7685 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
7687 * tree-vect-loop-manip.c (find_loop_location): Convert return
7688 type from source_location to dump_user_location_t.
7689 (vect_do_peeling): Update for above change.
7690 (vect_loop_versioning): Update for change in type of
7692 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
7693 from location_t to dump_user_location_t.
7694 (vect_estimate_min_profitable_iters): Update for change in type
7696 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
7697 location_t to dump_location_t.
7698 (vect_slp_bb): Update for change in type of vect_location.
7699 * tree-vectorizer.c (vect_location): Convert from source_location
7700 to dump_user_location_t.
7701 (try_vectorize_loop_1): Update for change in vect_location's type.
7702 (vectorize_loops): Likewise.
7703 (increase_alignment): Likewise.
7704 * tree-vectorizer.h (vect_location): Convert from source_location
7705 to dump_user_location_t.
7706 (find_loop_location): Convert return type from source_location to
7707 dump_user_location_t.
7708 (check_reduction_path): Convert 1st param from location_t to
7709 dump_user_location_t.
7710 * value-prof.c (check_counter): Update call to dump_printf_loc to
7711 pass in a dump_user_location_t rather than a location_t; update
7712 call to error_at for change in type of "locus".
7713 (check_ic_target): Update call to dump_printf_loc to
7714 pass in a dump_user_location_t rather than a location_t, via the
7717 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
7719 * config/s390/s390.h (enum processor_flags): Do not use
7720 default tune parameter when -march was specified.
7722 2018-06-26 Jakub Jelinek <jakub@redhat.com>
7725 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
7726 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
7729 2018-06-26 Richard Biener <rguenther@suse.de>
7731 PR tree-optimization/86287
7733 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
7734 (vect_analyze_loop): Initialize n_stmts.
7736 2018-06-26 Richard Biener <rguenther@suse.de>
7739 * fold-const.c (fold_convertible_p): Pointer extension
7742 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
7745 * dwarf2out.c (loc_list_has_views): Adjust comments.
7746 (dw_loc_list): Split single cross-partition range with
7749 2018-06-25 Jeff Law <law@redhat.com>
7751 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
7752 on -mbig-switch by default.
7754 * config/v850/predicates.md (const_float_1_operand): Fix match_code
7756 (const_float_0_operand): Remove unused predicate.
7757 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
7758 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
7759 (recipsf2): New expander. Original pattern now called
7761 (recipdf2, recipdf2_insn): Similarly.
7762 (rsqrtsf2, rsqrtsf2_insn): Similarly
7763 (rsqrtdf2, rsqrtdf2_insn): Similarly
7765 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
7767 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
7768 Simplify logic for FreeBSD (twice).
7770 2018-06-25 Martin Sebor <msebor@redhat.com>
7772 PR tree-optimization/86204
7773 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
7774 a strnlen result if it's less than the length of the string.
7776 2018-06-25 Martin Sebor <msebor@redhat.com>
7778 PR tree-optimization/85700
7779 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
7780 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
7781 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
7783 2018-06-25 Martin Sebor <msebor@redhat.com>
7785 * doc/extend.texi (Zero-length arrays): Update and clarify.
7787 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
7789 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
7790 added IEEE/IBM long double multilib support on PowerPC little
7791 endian Linux systems.
7792 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
7793 (MULTILIB_DEFAULTS): Likewise.
7794 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7796 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
7797 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
7798 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
7800 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
7803 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
7804 (REORDER_45): Likewise.
7806 2018-06-25 Jeff Law <law@redhat.com>
7808 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
7809 dividend to 32 bits. Adjust length.
7810 (udivmodhi4): Cleanup output template. Fix length.
7812 2018-06-25 Carl Love <cel@us.ibm.com>
7814 * config/rs6000/vsx.md: Change word selector to prefered location.
7816 2018-06-25 Richard Biener <rguenther@suse.de>
7818 PR tree-optimization/86304
7819 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
7820 epilogue-if-converted loops as well.
7822 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
7824 * lto-section-out.c (lto_begin_section): Do not print section
7825 name for noaddr and unnumbered dumps.
7827 2018-06-25 Richard Biener <rguenther@suse.de>
7829 * tree-vectorizer.h (struct vec_info_shared): New structure
7830 with parts split out from struct vec_info and loop_nest from
7831 struct _loop_vec_info.
7832 (struct vec_info): Adjust accordingly.
7833 (struct _loop_vec_info): Likewise.
7834 (LOOP_VINFO_LOOP_NEST): Adjust.
7835 (LOOP_VINFO_DATAREFS): Likewise.
7836 (LOOP_VINFO_DDRS): Likewise.
7837 (struct _bb_vec_info): Likewise.
7838 (BB_VINFO_DATAREFS): Likewise.
7839 (BB_VINFO_DDRS): Likewise.
7840 (struct _stmt_vec_info): Add dr_aux member.
7841 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
7842 (DR_MISALIGNMENT_UNINITIALIZED): New.
7843 (set_dr_misalignment): Adjust.
7844 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
7845 (vect_analyze_loop): Adjust prototype.
7846 (vect_analyze_loop_form): Likewise.
7847 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
7848 Compute dependences lazily.
7849 (vect_record_base_alignments): Use shared datarefs/ddrs.
7850 (vect_verify_datarefs_alignment): Likewise.
7851 (vect_analyze_data_refs_alignment): Likewise.
7852 (vect_analyze_data_ref_accesses): Likewise.
7853 (vect_analyze_data_refs): Likewise.
7854 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
7855 constructor parameter for shared part.
7856 (vect_analyze_loop_form): Pass in shared part and adjust.
7857 (vect_analyze_loop_2): Pass in storage for the number of
7858 stmts. Move loop nest finding to the caller. Compute
7860 (vect_analyze_loop): Pass in shared part.
7861 (vect_transform_loop): Verify shared datarefs are unchanged.
7862 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
7863 constructor parameter for shared part.
7864 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
7865 (vect_slp_bb): Verify shared datarefs are unchanged before
7867 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
7869 (new_stmt_vec_info): Initialize DR_AUX misalignment to
7870 DR_MISALIGNMENT_UNINITIALIZED.
7871 * tree-vectorizer.c (vec_info::vec_info): Add constructor
7872 parameter for shared part.
7873 (vec_info::~vec_info): Adjust.
7874 (vec_info_shared::vec_info_shared): New.
7875 (vec_info_shared::~vec_info_shared): Likewise.
7876 (vec_info_shared::save_datarefs): Likewise.
7877 (vec_info_shared::check_datarefs): Likewise.
7878 (try_vectorize_loop_1): Construct shared part live for analyses
7879 of a single loop for multiple vector sizes.
7880 * tree-parloops.c (gather_scalar_reductions): Adjust.
7882 2018-06-25 Richard Biener <rguenther@suse.de>
7884 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
7885 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
7886 (vect_analyze_data_refs): Remove similar code from here and
7887 simplify accordingly.
7889 2018-06-25 Richard Biener <rguenther@suse.de>
7891 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
7892 for reverse storage order accesses rather than asserting
7893 they cannot happen here.
7895 2018-06-25 Tom de Vries <tdevries@suse.de>
7898 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
7899 Use data16 instead of .byte for insn prefix.
7901 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
7904 * parser.c (make_char_string_pack): Pass this literal chars
7905 through cpp_interpret_string.
7906 (cp_parser_userdef_numeric_literal): Check the result of
7907 make_char_string_pack.
7909 2018-06-24 Maya Rashish <coypu@sdf.org>
7911 * ginclude/stddef.h: Simplify conditions around avoiding
7912 re-definition of __size_t.
7914 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
7916 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
7917 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
7919 2018-06-22 Maya Rashish <coypu@sdf.org>
7921 * doc/invoke.texi (mno-fancy-math-387): Update for changes
7922 made to OpenBSD and NetBSD through the years.
7924 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
7926 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
7927 behavior of vec_pack (vector double, vector double) to match
7928 behavior of vec_float2 (vector double, vector double).
7930 2018-06-22 Olivier Hainque <hainque@adacore.com>
7932 * gimplify.c (gimplify_function_tree): Prevent creation
7933 of a trampoline for the address of the current function
7934 passed to entry/exit instrumentation hooks.
7936 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
7939 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
7942 2018-06-22 Martin Liska <mliska@suse.cz>
7944 PR tree-optimization/86263
7945 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
7946 Bail out if is_enabled is false.
7947 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
7949 (jump_table_cluster::is_enabled): New function.
7951 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
7953 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
7954 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
7955 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
7956 (lto_input_ts_binfo_tree_pointers): Likewise.
7957 * tree-streamer-out.c (streamer_write_tree_bitfields,
7958 write_ts_binfo_tree_pointers): Likewise.
7959 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
7961 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
7963 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
7965 2018-06-22 Martin Liska <mliska@suse.cz>
7967 * symbol-summary.h (get): Make it pure and inline move
7968 functionality from ::get function.
7969 (get): Remove and inline into ::get and ::get_create.
7970 (get_create): Move code from ::get function.
7972 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7975 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
7976 -x assembler-with-cpp.
7978 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7980 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
7981 _FILE_OFFSET_BITS=64 for C++.
7983 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
7985 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
7986 conversion insn that shows up when pr85657-3.c is compiled using
7987 IEEE 128-bit long double.
7988 (neg<mode>2_internal): Use the correct mode to check whether the
7989 mode is IBM extended.
7990 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
7991 multiply and divide external functions from being created more
7994 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
7996 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
7998 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
7999 the edge can be forwarded.
8000 (cfg_layout_merge_blocks): Likewise.
8002 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
8004 * except.c (finish_eh_generation): Commit edge insertions only after
8005 the EH edges have been redirected from post-landing to landing pads.
8007 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
8009 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
8010 create_tmp_var_for to create the FRAME decl.
8011 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
8013 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
8015 * tree-inline.c (copy_edges_for_bb): Minor tweak.
8016 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
8017 debug statement when resetting its value.
8018 (expand_call_inline): Copy the locus of the call onto the assignment
8019 of the return value, if any. Use local variable in more cases.
8021 2018-06-21 Martin Liska <mliska@suse.cz>
8023 * ipa-pure-const.c (propagate_nothrow): Use
8024 funct_state_summaries->get.
8025 (dump_malloc_lattice): Likewise.
8026 (propagate_malloc): Likewise.
8028 2018-06-21 Richard Biener <rguenther@suse.de>
8030 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
8031 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
8032 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
8034 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
8035 BLOCK_ABSTRACT_ORIGIN unconditionally.
8037 2018-06-21 David Malcolm <dmalcolm@redhat.com>
8039 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
8041 * ipa-reference.c (ipa_reference_c_finalize): Delete
8042 ipa_ref_opt_sum_summaries and set it to NULL.
8044 2018-06-21 Tom de Vries <tdevries@suse.de>
8046 PR tree-optimization/85859
8047 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
8048 test with comment from bb_no_side_effects_p.
8050 2018-06-21 Richard Biener <rguenther@suse.de>
8052 PR tree-optimization/86232
8053 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
8054 max for constant niter.
8056 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
8058 * config/aarch64/aarch64-simd.md
8059 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
8061 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
8063 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
8064 Make opernads of the unspec commutative.
8066 2018-06-21 Richard Biener <rguenther@suse.de>
8068 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
8069 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
8070 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
8071 (vect_analyze_data_ref_dependence): Re-order checks to deal with
8073 (vect_record_base_alignments): Do not record base alignment
8074 for gathers or scatters.
8075 (vect_compute_data_ref_alignment): Drop return value that is always
8076 true. Bail out early for gathers or scatters.
8077 (vect_enhance_data_refs_alignment): Bail out early for gathers
8079 (vect_find_same_alignment_drs): Likewise.
8080 (vect_analyze_data_refs_alignment): Remove dead code.
8081 (vect_slp_analyze_and_verify_node_alignment): Likewise.
8082 (vect_analyze_data_refs): For possible gathers or scatters do
8083 not create an alternate DR, just check their possible validity
8084 and mark them. Adjust DECL_NONALIASED handling to not rely
8086 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
8087 update inits of gathers or scatters.
8088 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
8089 Also copy gather/scatter flag to pattern vinfo.
8091 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
8093 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
8094 behavior of vec_packsu (vector unsigned long long, vector unsigned
8095 long long) to match behavior of vec_packs with same signature.
8097 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
8098 Thomas Schwinge <thomas@codesourcery.com>
8099 Cesar Philippidis <cesar@codesourcery.com>
8101 * gimplify.c (gimplify_scan_omp_clauses): Add support for
8102 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
8103 (gimplify_adjust_omp_clauses): Likewise.
8104 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
8105 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
8106 (gimplify_omp_target_update): Update handling of acc update and
8108 * omp-low.c (install_var_field): Remove unused parameter
8109 base_pointers_restrict.
8110 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
8111 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
8113 (omp_target_base_pointers_restrict_p): Delete.
8114 (scan_omp_target): Update call to scan_sharing_clauses.
8115 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
8117 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
8118 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
8119 (convert_local_omp_clauses): Likewise.
8120 * tree-pretty-print.c (dump_omp_clause): Likewise.
8121 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
8123 (omp_clause_code_name): Likewise.
8125 2018-06-20 Jakub Jelinek <jakub@redhat.com>
8128 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
8131 PR tree-optimization/86231
8132 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
8133 anti-range don't overwrite *vr0min before using it to compute *vr0max.
8135 2018-06-20 Tom de Vries <tdevries@suse.de>
8137 PR tree-optimization/86097
8138 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
8139 iv type if signedness of iv type is not the same as that of *nit.
8141 2018-06-20 Jakub Jelinek <jakub@redhat.com>
8143 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
8144 EDGE_EH edges, verify they are all EDGE_EH.
8146 2018-06-20 Maya Rashish <coypu@sdf.org>
8148 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
8150 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8152 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
8153 * config/aarch64/aarch64.c (xgene1_tunings): Add
8154 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
8155 (aarch64_mode_valid_for_sched_fusion_p):
8156 Allow 16-byte modes.
8157 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
8158 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
8160 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
8162 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
8163 * config/aarch64/iterators.md (VQ2): New mode iterator.
8165 2018-06-20 Martin Liska <mliska@suse.cz>
8167 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
8168 Change default ratio from 10 to 8.
8170 2018-06-20 Martin Liska <mliska@suse.cz>
8172 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
8174 (bit_test_cluster::find_bit_tests): Likewise.
8175 (switch_decision_tree::analyze_switch_statement): Find clusters.
8176 * tree-switch-conversion.h (struct jump_table_cluster): Document
8179 2018-06-20 Martin Liska <mliska@suse.cz>
8181 * tree-switch-conversion.c (switch_conversion::collect):
8182 Record m_uniq property.
8183 (switch_conversion::expand): Bail out for special conditions.
8184 (group_cluster::~group_cluster): New.
8185 (group_cluster::group_cluster): Likewise.
8186 (group_cluster::dump): Likewise.
8187 (jump_table_cluster::emit): New.
8188 (switch_decision_tree::fix_phi_operands_for_edges): New.
8189 (struct case_node): Remove struct.
8190 (jump_table_cluster::can_be_handled): New.
8191 (case_values_threshold): Moved to header.
8192 (reset_out_edges_aux): Likewise.
8193 (jump_table_cluster::is_beneficial): New.
8194 (bit_test_cluster::can_be_handled): Likewise.
8195 (add_case_node): Remove.
8196 (bit_test_cluster::is_beneficial): New.
8197 (case_bit_test::cmp): New.
8198 (bit_test_cluster::emit): New.
8199 (expand_switch_as_decision_tree_p): Remove.
8200 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
8201 (fix_phi_operands_for_edge): Likewise.
8202 (switch_decision_tree::analyze_switch_statement): New.
8203 (compute_cases_per_edge): Move ...
8204 (switch_decision_tree::compute_cases_per_edge): ... here.
8205 (try_switch_expansion): Likewise.
8206 (switch_decision_tree::try_switch_expansion): Likewise.
8207 (record_phi_operand_mapping): Likewise.
8208 (switch_decision_tree::record_phi_operand_mapping): Likewise.
8209 (emit_case_decision_tree): Likewise.
8210 (switch_decision_tree::emit): Likewise.
8211 (balance_case_nodes): Likewise.
8212 (switch_decision_tree::balance_case_nodes): Likewise.
8213 (dump_case_nodes): Likewise.
8214 (switch_decision_tree::dump_case_nodes): Likewise.
8215 (emit_jump): Likewise.
8216 (switch_decision_tree::emit_jump): Likewise.
8217 (emit_cmp_and_jump_insns): Likewise.
8218 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
8219 (emit_case_nodes): Likewise.
8220 (switch_decision_tree::emit_case_nodes): Likewise.
8221 (conditional_probability): Remove.
8222 * tree-switch-conversion.h (enum cluster_type): New.
8224 (struct cluster): Likewise.
8225 (cluster::cluster): Likewise.
8226 (struct simple_cluster): Likewise.
8227 (simple_cluster::simple_cluster): Likewise.
8228 (struct group_cluster): Likewise.
8229 (struct jump_table_cluster): Likewise.
8230 (struct bit_test_cluster): Likewise.
8231 (struct min_cluster_item): Likewise.
8232 (struct case_tree_node): Likewise.
8233 (case_tree_node::case_tree_node): Likewise.
8234 (jump_table_cluster::case_values_threshold): Likewise.
8235 (struct case_bit_test): Likewise.
8236 (struct switch_decision_tree): Likewise.
8237 (struct switch_conversion): Likewise.
8238 (switch_decision_tree::reset_out_edges_aux): Likewise.
8240 2018-06-20 Martin Liska <mliska@suse.cz>
8242 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
8243 (hoist_edge_and_branch_if_true): Likewise.
8244 (expand_switch_using_bit_tests_p): Likewise.
8245 (struct case_bit_test): Likewise.
8246 (case_bit_test_cmp): Likewise.
8247 (emit_case_bit_tests): Likewise.
8248 (switch_conversion::switch_conversion): New class.
8249 (struct switch_conv_info): Remove old struct.
8250 (collect_switch_conv_info): More to ...
8251 (switch_conversion::collect): ... this.
8252 (check_range): Likewise.
8253 (switch_conversion::check_range): Likewise.
8254 (check_all_empty_except_final): Likewise.
8255 (switch_conversion::check_all_empty_except_final): Likewise.
8256 (check_final_bb): Likewise.
8257 (switch_conversion::check_final_bb): Likewise.
8258 (create_temp_arrays): Likewise.
8259 (switch_conversion::create_temp_arrays): Likewise.
8260 (free_temp_arrays): Likewise.
8261 (gather_default_values): Likewise.
8262 (switch_conversion::gather_default_values): Likewise.
8263 (build_constructors): Likewise.
8264 (switch_conversion::build_constructors): Likewise.
8265 (constructor_contains_same_values_p): Likewise.
8266 (switch_conversion::contains_same_values_p): Likewise.
8267 (array_value_type): Likewise.
8268 (switch_conversion::array_value_type): Likewise.
8269 (build_one_array): Likewise.
8270 (switch_conversion::build_one_array): Likewise.
8271 (build_arrays): Likewise.
8272 (switch_conversion::build_arrays): Likewise.
8273 (gen_def_assigns): Likewise.
8274 (switch_conversion::gen_def_assigns): Likewise.
8275 (prune_bbs): Likewise.
8276 (switch_conversion::prune_bbs): Likewise.
8277 (fix_phi_nodes): Likewise.
8278 (switch_conversion::fix_phi_nodes): Likewise.
8279 (gen_inbound_check): Likewise.
8280 (switch_conversion::gen_inbound_check): Likewise.
8281 (process_switch): Use the newly created class.
8282 (switch_conversion::expand): New.
8283 (switch_conversion::~switch_conversion): New.
8284 * tree-switch-conversion.h: New file.
8286 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8288 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
8289 tree-vect-patterns.c.
8290 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
8291 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
8292 (vect_recog_sad_pattern): Likewise.
8293 (vect_recog_widen_sum_pattern): Likewise.
8294 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
8295 (vect_recog_widen_shift_pattern): Remove the type_in argument.
8296 (vect_recog_rotate_pattern): Likewise.
8297 (vect_recog_mult_pattern): Likewise.
8298 (vect_recog_vector_vector_shift_pattern): Likewise.
8299 (vect_recog_divmod_pattern): Likewise.
8300 (vect_recog_mixed_size_cond_pattern): Likewise.
8301 (vect_recog_bool_pattern): Likewise.
8302 (vect_recog_mask_conversion_pattern): Likewise.
8303 (vect_try_gather_scatter_pattern): Likewise.
8304 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
8305 (vect_recog_over_widening_pattern): Likewise.
8306 (vect_recog_gather_scatter_pattern): Likewise.
8307 (vect_recog_func_ptr): Move from tree-vectorizer.h
8308 (vect_vect_recog_func_ptrs): Move further down the file.
8309 (vect_recog_func): Likewise. Remove the third argument.
8310 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
8311 (vect_pattern_recog_1): Expect the pattern function to do any
8312 necessary target tests. Also expect it to provide a vector type.
8313 Remove the type_in handling.
8315 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8317 * tree-vect-patterns.c (vect_pattern_detected): New function.
8318 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
8319 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
8320 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
8321 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
8322 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
8323 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
8324 (vect_recog_mask_conversion_pattern)
8325 (vect_try_gather_scatter_pattern): Likewise.
8327 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8329 * tree-vect-patterns.c (vect_get_internal_def): New function.
8330 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
8331 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
8332 (search_type_for_mask_1): Use it.
8334 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8336 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
8337 redundant WIDEN_SUM_EXPR handling.
8338 (vect_recog_sad_pattern): Likewise.
8340 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8342 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
8343 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
8344 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
8345 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
8346 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
8348 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8350 * tree-vect-stmts.c (vectorizable_call): Make sure that we
8351 use the stmt_vec_info of the original bb statement for the
8352 new zero assignment, even if the call is part of a pattern.
8354 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8356 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
8357 that the sequence is attached to the original statement rather
8358 than the pattern statement.
8359 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
8360 PATTERN_DEF_SEQ from the original statement rather than
8361 the main pattern statement.
8362 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
8363 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
8364 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
8366 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8368 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
8369 definition statements before the early exit for statements that aren't
8371 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
8373 (vect_transform_loop): ...here. Process pattern definition
8374 statements without first checking whether the main pattern
8375 statement is live or relevant.
8377 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
8379 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
8380 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
8382 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
8384 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
8385 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
8386 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
8387 (expand_block_compare): Change select_block_compare_mode call.
8388 (expand_strncmp_align_check): Use new functions, fix comment.
8389 (emit_final_str_compare_gpr): New function.
8390 (expand_strn_compare): Refactor and clean up code.
8391 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
8393 2018-06-19 Tony Reix <tony.reix@atos.com>
8394 Damien Bergamini <damien.bergamini@atos.com>
8395 David Edelsohn <dje.gcc@gmail.com>
8397 * collect2.c (static_obj): New variable.
8398 (static_libs): New variable.
8399 (is_in_list): Uncomment declaration.
8400 (main): Track AIX libraries linked statically.
8401 (is_in_list): Uncomment definition.
8402 (scan_prog_file): Don't add AIX shared libraries initializer
8403 to constructor list if linking statically.
8405 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
8407 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
8409 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
8411 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
8413 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
8416 2018-06-19 Martin Liska <mliska@suse.cz>
8418 * config/i386/i386.c (ix86_can_inline_p): Do not use
8419 ipa_fn_summaries::get_create.
8420 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
8422 (devirtualization_time_bonus): Likewise.
8423 (ipcp_propagate_stage): Likewise.
8424 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
8425 (edge_set_predicate): Likewise.
8426 (evaluate_conditions_for_known_args): Likewise.
8427 (evaluate_properties_for_edge): Likewise.
8428 (ipa_call_summary::reset): Tranform to ...
8429 (ipa_call_summary::~ipa_call_summary): ... this.
8430 (ipa_fn_summary::reset): Transform to ...
8431 (ipa_fn_summary::~ipa_fn_summary): ... this.
8432 (ipa_fn_summary_t::remove): Rename to ...
8433 (ipa_fn_summary_t::remove_callees): ... this.
8434 (ipa_fn_summary_t::duplicate): Use placement new
8435 instead of memory copy.
8436 (ipa_call_summary_t::duplicate): Likewise.
8437 (ipa_call_summary_t::remove): Remove.
8438 (dump_ipa_call_summary): Change get_create to get.
8439 (ipa_dump_fn_summary): Dump only when summary exists.
8440 (analyze_function_body): Use symbol_summary::get instead
8442 (compute_fn_summary): Likewise.
8443 (estimate_edge_devirt_benefit): Likewise.
8444 (estimate_edge_size_and_time): Likewise.
8445 (inline_update_callee_summaries): Likewise.
8446 (remap_edge_change_prob): Likewise.
8447 (remap_edge_summaries): Likewise.
8448 (ipa_merge_fn_summary_after_inlining): Likewise.
8449 (write_ipa_call_summary): Likewise.
8450 (ipa_fn_summary_write): Likewise.
8451 (ipa_free_fn_summary): Likewise.
8452 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
8453 (struct ipa_call_summary): Likewise.
8454 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
8456 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
8457 (estimate_size_after_inlining): Likewise.
8458 (estimate_growth): Likewise.
8459 (growth_likely_positive): Likewise.
8460 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
8461 (inline_call): Likewise.
8462 * ipa-inline.c (caller_growth_limits): Likewise.
8463 (can_inline_edge_p): Likewise.
8464 (can_inline_edge_by_limits_p): Likewise.
8465 (compute_uninlined_call_time): Likewise.
8466 (compute_inlined_call_time): Likewise.
8467 (want_inline_small_function_p): Likewise.
8468 (edge_badness): Likewise.
8469 (update_caller_keys): Likewise.
8470 (update_callee_keys): Likewise.
8471 (inline_small_functions): Likewise.
8472 (inline_to_all_callers_1): Likewise.
8473 (dump_overall_stats): Likewise.
8474 (early_inline_small_functions): Likewise.
8475 (early_inliner): Likewise.
8476 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
8477 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
8478 * ipa-pure-const.c (malloc_candidate_p): Likewise.
8479 * ipa-split.c (execute_split_functions): Likewise.
8480 * symbol-summary.h: Likewise.
8481 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
8483 2018-06-19 Richard Biener <rguenther@suse.de>
8485 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
8486 (vectorize_loops): ... here. Fix dbgcnt handling.
8487 (try_vectorize_loop): Wrap try_vectorize_loop_1.
8489 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
8492 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
8493 ieee128 argument takes up only one (vector) register, not two (floating
8496 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
8498 * gimplify.c (gimplify_init_constructor): Really never clear for an
8499 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
8501 2018-06-19 Richard Biener <rguenther@suse.de>
8503 PR tree-optimization/86179
8504 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
8505 after failed recognition.
8507 2018-06-18 Martin Sebor <msebor@redhat.com>
8510 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
8511 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
8512 Handle integer subtraction.
8513 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
8514 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
8516 2018-06-18 David Malcolm <dmalcolm@redhat.com>
8518 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
8519 param from rtx to rtx_insn *.
8520 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
8522 (frv_ifcvt_modify_insn): Likwise.
8523 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
8524 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
8525 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
8526 as_a <rtx_insn *> cast to local "unprotected_region" once
8527 it's been established that it's not NULL or pc_rtx.
8528 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
8529 param "sethi" from rtx to rtx_insn *.
8530 (nds32_group_float_insns): Likewise for param "insn".
8531 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
8533 (vax_output_int_subtract): Likewise.
8534 * config/vax/vax.c (vax_output_int_add): Likewise for param
8536 (vax_output_int_subtract): Likewise.
8537 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
8538 (emit_pattern_after): Likewise for param "after".
8539 (emit_insn_after): Likewise.
8540 (emit_jump_insn_after): Likewise.
8541 (emit_call_insn_after): Likewise.
8542 (emit_debug_insn_after): Likewise.
8543 (emit_pattern_before): Likewise for param "before".
8544 (emit_insn_before): Likewise.
8545 (emit_jump_insn_before): Likewise.
8546 * final.c (get_insn_template): Likewise for param "insn", removing
8548 * output.h (get_insn_template): Likewise for 2nd param.
8549 * rtl.h (emit_insn_before): Likewise.
8550 (emit_jump_insn_before): Likewise.
8551 (emit_debug_insn_before_noloc): Likewise.
8552 (emit_insn_after): Likewise.
8553 (emit_jump_insn_after): Likewise.
8554 (emit_call_insn_after): Likewise.
8555 (emit_debug_insn_after): Likewise.
8556 (set_insn_deleted): Likewise for param.
8558 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
8561 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
8562 floating point modes, so that IFmode is numerically greater than
8563 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
8564 to declare the ordering. This prevents IFmode from being
8565 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
8566 machine. Include rs6000-modes.h to share the fractional values
8567 between genmodes* and the rest of the compiler.
8571 * config/rs6000/rs6000-modes.h: New file.
8572 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
8573 meaning of rs6000_long_double_size so that 126..128 selects an
8574 appropriate 128-bit floating point type.
8575 (rs6000_option_override_internal): Likewise.
8576 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
8577 (TARGET_LONG_DOUBLE_128): Change the meaning of
8578 rs6000_long_double_size so that 126..128 selects an appropriate
8579 128-bit floating point type.
8580 (LONG_DOUBLE_TYPE_SIZE): Update comment.
8581 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
8582 source and destination to match the standard usage.
8583 (truncifkf2): Likewise.
8584 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
8585 ISA 2.07 to use an explicit clobber, instead of passing in a
8587 (copysign<mode>3_soft): Likewise.
8589 2018-06-18 David Malcolm <dmalcolm@redhat.com>
8591 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
8592 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
8593 (vect_slp_analyze_instance_dependence): Likewise.
8594 (vect_enhance_data_refs_alignment): Likewise.
8595 (vect_analyze_data_refs_alignment): Likewise.
8596 (vect_slp_analyze_and_verify_instance_alignment
8597 (vect_analyze_data_ref_accesses): Likewise.
8598 (vect_prune_runtime_alias_test_list): Likewise.
8599 (vect_analyze_data_refs): Likewise.
8600 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
8601 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
8602 (vect_analyze_scalar_cycles_1): Likewise.
8603 (vect_get_loop_niters): Likewise.
8604 (vect_analyze_loop_form_1): Likewise.
8605 (vect_update_vf_for_slp): Likewise.
8606 (vect_analyze_loop_operations): Likewise.
8607 (vect_analyze_loop): Likewise.
8608 (vectorizable_induction): Likewise.
8609 (vect_transform_loop): Likewise.
8610 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
8611 * tree-vect-slp.c (vect_analyze_slp): Likewise.
8612 (vect_make_slp_decision): Likewise.
8613 (vect_detect_hybrid_slp): Likewise.
8614 (vect_slp_analyze_operations): Likewise.
8615 (vect_slp_bb): Likewise.
8616 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
8617 (vectorizable_bswap): Likewise.
8618 (vectorizable_call): Likewise.
8619 (vectorizable_simd_clone_call): Likewise.
8620 (vectorizable_conversion): Likewise.
8621 (vectorizable_assignment): Likewise.
8622 (vectorizable_shift): Likewise.
8623 (vectorizable_operation): Likewise.
8624 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
8626 2018-06-18 Martin Sebor <msebor@redhat.com>
8628 PR tree-optimization/81384
8629 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
8630 * builtins.c (expand_builtin_strnlen): New function.
8631 (expand_builtin): Call it.
8632 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
8633 * builtins.def (BUILT_IN_STRNLEN): New.
8634 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
8635 Warn for bounds in excess of maximum object size.
8636 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
8637 single-value ranges. Handle strnlen.
8638 (handle_builtin_strlen): Handle strnlen.
8639 (strlen_check_and_optimize_stmt): Same.
8640 * doc/extend.texi (Other Builtins): Document strnlen.
8642 2018-06-18 Maya Rashish <coypu@sdf.org>
8644 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
8645 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
8646 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
8648 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
8650 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
8652 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8654 * tree.c (escaped_string::escape): Replace cast to char * by
8655 const_cast<char *> (unescaped).
8657 2018-06-18 Nick Clifton <nickc@redhat.com>
8660 * tree.c (escaped_string): New class. Converts an unescaped
8661 string into its escaped equivalent.
8662 (warn_deprecated_use): Use the new class to convert the
8663 deprecation message, if present.
8664 (test_escaped_strings): New self test.
8665 (test_c_tests): Add test_escaped_strings.
8666 * doc/extend.texi (deprecated): Add a note that the
8667 deprecation message is affected by the -fmessage-length
8668 option, and that control characters will be escaped.
8669 (#pragma GCC error): Document this pragma.
8670 (#pragma GCC warning): Likewise.
8671 * doc/invoke.texi (-fmessage-length): Document this option's
8672 effect on the #warning and #error preprocessor directives and
8673 the deprecated attribute.
8675 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
8677 * tree.c (decl_value_expr_lookup): Revert latest change.
8678 (decl_value_expr_insert): Likewise.
8680 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
8682 * gimplify.c (nonlocal_vlas): Delete.
8683 (nonlocal_vla_vars): Likewise.
8684 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
8686 (gimplify_body): Do not create and destroy nonlocal_vlas.
8687 * tree-nested.c: Include diagnostic.h.
8688 (use_pointer_in_frame): Tweak.
8689 (lookup_field_for_decl): Add assertion and declare the transformation.
8690 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
8691 internal error when the reference is in a wrong context. Do not
8692 create a debug decl by default.
8693 (note_nonlocal_block_vlas): Delete.
8694 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
8695 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
8696 create a debug decl by default.
8697 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
8698 call is in a wrong context.
8699 (fixup_vla_decls): New function.
8700 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
8701 debug variables were created.
8702 * tree.c (decl_value_expr_lookup): Add checking assertion.
8703 (decl_value_expr_insert): Likewise.
8705 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
8708 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
8709 * tree-scalar-evolution.c (interpret_expr): Likewise.
8710 (expression_expensive_p): Likewise.
8711 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
8712 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
8713 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
8714 (ssa_defined_by_minus_one_stmt_p): New.
8716 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
8719 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
8720 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
8721 * dojump.c (do_jump): Likewise.
8722 * expr.c (expand_expr_real_2): Check operand type's sign.
8723 * fold-const.c (const_unop): Handle ABSU_EXPR.
8724 (fold_abs_const): Likewise.
8725 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
8726 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
8727 (strip_sign_op_1): Likesise.
8728 * match.pd: Add new pattern to generate ABSU_EXPR.
8729 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
8730 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
8731 * tree-eh.c (operation_could_trap_helper_p): Likewise.
8732 * tree-inline.c (estimate_operator_cost): Likewise.
8733 * tree-pretty-print.c (dump_generic_node): Likewise.
8734 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
8735 * tree.def (ABSU_EXPR): New.
8737 2018-06-16 Jakub Jelinek <jakub@redhat.com>
8740 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
8741 documented as preserved for backward compatibility only.
8742 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
8744 PR rtl-optimization/86108
8745 * bb-reorder.c (create_forwarder_block): Renamed to ...
8746 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
8747 jump from new landing pad to the second part.
8748 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
8751 2018-06-15 Jakub Jelinek <jakub@redhat.com>
8754 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
8755 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
8756 Only call store_expr for halves if the mode is the same.
8759 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
8760 Fix up comment formatting.
8762 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8764 * typed-splay-tree.h (typed_splay_tree::remove): New function.
8765 (typed_splay_tree::closure,
8766 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
8767 (typed_splay_tree::typed_splay_tree,
8768 typed_splay_tree::operator =): Declared private.
8769 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
8770 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
8771 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
8772 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
8773 typed_splay_tree::splay_tree_splay,
8774 typed_splay_tree::splay_tree_foreach_helper,
8775 typed_splay_tree::splay_tree_insert,
8776 typed_splay_tree::splay_tree_remove,
8777 typed_splay_tree::splay_tree_lookup,
8778 typed_splay_tree::splay_tree_predecessor,
8779 typed_splay_tree::splay_tree_successor,
8780 typed_splay_tree::splay_tree_min,
8781 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
8782 (typed_splay_tree::root, typed_splay_tree::comp,
8783 typed_splay_tree::delete_key,
8784 typed_splay_tree::delete_value): New data members.
8785 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
8786 typed_splay_tree::remove.
8788 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
8790 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
8791 -mginv and -mno-ginv to the assembler.
8792 * config/mips/mips.opt (-mcrc): New option.
8794 * doc/invoke.text (-mcrc): Document.
8797 2018-06-15 Nick Clifton <nickc@redhat.com>
8800 * tree.c (escaped_string): New class. Converts an unescaped
8801 string into its escaped equivalent.
8802 (warn_deprecated_use): Use the new class to convert the
8803 deprecation message, if present.
8804 (test_escaped_strings): New self test.
8805 (test_c_tests): Add test_escaped_strings.
8806 * doc/extend.texi (deprecated): Add a note that the
8807 deprecation message is affected by the -fmessage-length
8808 option, and that control characters will be escaped.
8809 (#pragma GCC error): Document this pragma.
8810 (#pragma GCC warning): Likewise.
8811 * doc/invoke.texi (-fmessage-length): Document this option's
8812 effect on the #warning and #error preprocessor directives and
8813 the deprecated attribute.
8815 2018-06-15 Richard Biener <rguenther@suse.de>
8817 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
8818 here, also noting vector size used.
8819 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
8820 size used in MSG_OPTIMIZED_LOCATIONS dump.
8821 (pass_slp_vectorize::execute): Adjust.
8823 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
8826 * config/arc/arc.c (arc_return_address_register): Fix
8829 2018-06-15 Richard Biener <rguenther@suse.de>
8832 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
8833 leave useless conversion stripping to force_gimple_operand_gsi.
8834 (gimplify_build2): Likewise.
8835 (gimplify_build1): Likewise.
8837 2018-06-15 Richard Biener <rguenther@suse.de>
8840 * tree-cfg.c (move_stmt_op): unshare invariant addresses
8841 before adjusting their block.
8843 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
8845 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
8846 multilibs for *-*-rtems*.
8847 * config/riscv/t-rtems: New file.
8849 2018-06-14 Jakub Jelinek <jakub@redhat.com>
8852 * match.pd ((A +- CST1) +- CST2): Punt if last resort
8853 unsigned_type_for returns NULL.
8856 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
8857 subregs of multi-word pseudos unless the float mode has word size.
8859 2018-06-14 Richard Biener <rguenther@suse.de>
8862 * tree-vect-generic.c (build_word_mode_vector_type): Remove
8863 duplicate and harmful type_hash_canon.
8864 * tree.c (type_hash_canon): Assert we didn't find ourselves.
8866 2018-06-14 Richard Biener <rguenther@suse.de>
8869 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
8872 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
8874 * config/rtems.h (STDINT_LONG32): Define.
8876 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
8877 Prachi Godbole <prachi.godbole@imgtec.com>
8879 * config/mips/mips-cpus.def: Define P6600.
8880 * config/mips/mips-tables.opt: Regenerate.
8881 * config/mips/mips.c (mips_ucbranch_type): New enum.
8882 (mips_rtx_cost_data): Add support for P6600.
8883 (mips_issue_rate): Likewise.
8884 (mips_multipass_dfa_lookahead): Likewise.
8885 (mips_avoid_hazard): Likewise.
8886 (mips_reorg_process_insns): Likewise.
8887 (mips_classify_branch_p6600): New function.
8888 * config/mips/mips.h (TUNE_P6600): New define.
8889 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
8890 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
8891 * config/mips/mips.md: Include p6600.md.
8892 (processor): Add p6600.
8893 * config/mips/p6600.md: New file.
8894 * doc/invoke.texi: Add p6600 to supported architectures.
8896 2018-06-13 Martin Sebor <msebor@redhat.com>
8898 PR tree-optimization/86114
8899 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
8901 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
8903 2018-06-13 Richard Biener <rguenther@suse.de>
8905 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
8906 Properly set vector type of the intermediate stmt.
8907 * tree-vect-stmts.c (vectorizable_operation): The destination
8908 var always has vectype_out type.
8910 2018-06-13 Jeff Law <law@redhat.com>
8912 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
8913 integer 0 for argument to print_rtl_with_bb.
8914 (rl78_reorg): Likewise.
8916 2018-06-13 David Malcolm <dmalcolm@redhat.com>
8918 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
8919 from rtx to rtx_insn *.
8920 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
8922 (add_sched_insns_for_speculation): Likewise for local "target",
8923 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
8924 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
8925 from rtx_insn ** to rtx_code_label **.
8926 (reorg_emit_nops): Likewise.
8927 (c6x_reorg): Likewise for local "call_labels".
8928 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
8930 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
8931 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
8932 the loops over LABEL_REFS.
8933 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
8935 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
8936 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
8937 (split_branches): Strengthen local "olabel" from rtx to
8938 rtx_insn *, adding a safe_as_a cast.
8939 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
8941 (add_insn_after): Likewise for first two params.
8942 (add_insn_before): Likewise.
8943 (remove_insn): Likewise for param.
8944 (emit_pattern_before_noloc): Likewise for second and third params.
8945 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
8946 (emit_call_insn_before_noloc): Likewise.
8947 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
8949 (emit_barrier_before): Likewise.
8950 (emit_label_before): Strengthen "label" param from "rtx" to
8951 "rtx_code_label *". Strengthen "before" param from "rtx" to
8953 (emit_insn_after_1): Strengthen "after" param from "rtx" to
8955 (emit_pattern_after_noloc): Likewise.
8956 (emit_insn_after_noloc): Likewise.
8957 (emit_jump_insn_after_noloc): Likewise.
8958 (emit_call_insn_after_noloc): Likewise.
8959 (emit_debug_insn_after_noloc): Likewise.
8960 (emit_barrier_after): Likewise.
8961 (emit_label_after): Likewise for both params.
8962 (emit_pattern_after_setloc): Likewise for "after" param. Convert
8963 "loc" param from "int" to "location_t".
8964 (emit_insn_after_setloc): Likewise.
8965 (emit_jump_insn_after_setloc): Likewise.
8966 (emit_call_insn_after_setloc): Likewise.
8967 (emit_debug_insn_after_setloc): Likewise.
8968 (emit_pattern_before_setloc): Likewise for "before" param. Convert
8969 "loc" param from "int" to "location_t".
8970 (emit_pattern_before): Convert NULL_RTX to NULL.
8971 (emit_insn_before_setloc): Convert "loc" param from "int" to
8973 (emit_jump_insn_before_setloc): Likewise.
8974 (emit_call_insn_before_setloc): Likewise.
8975 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
8976 rtx_insn *. Convert "loc" param from "int" to "location_t".
8977 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
8978 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
8979 Convert 3rd param from "int" to "location_t".
8980 (emit_barrier_before, emit_barrier_after, next_real_insn):
8981 Strengthen param from rtx to rtx_insn *.
8982 (emit_label_before): Strengthen 1st param from "rtx" to
8983 "rtx_code_label *". Strengthen 2nd param from "rtx" to
8985 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
8986 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
8987 Strengthen 2nd param from "rtx" to "rtx_insn *".
8988 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
8989 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
8990 Likewise. Convert 3rd param from "int" to "location_t".
8991 (emit_label_after): Strengthen 1st param from "rtx" to
8993 (next_real_insn, remove_insn): Strengthen param from "rtx" to
8995 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
8996 from "rtx" to "rtx_insn *".
8998 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
9000 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
9001 bodies streamed in with -Q.
9002 * dumpfile.c (dump_files): Add lto-stream-out dump file.
9003 * dumpfile.h (tree_dump_index): Add lto_stream_out.
9004 * gimple-streamer-out.c: Include gimple-pretty-print.h
9005 (output_bb): Dump stmts streamed.
9006 * lto-section-out.c: Include print-tree.h
9007 (lto_begin_section): Dump sections created.
9008 (lto_output_decl_index): Dump decl encoded.
9009 * lto-streamer-out.c: Include print-tree.h
9010 (create_output_block): Dump output block created.
9011 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
9012 (output_function): Dump function output.
9013 (output_constructor): Dump constructor streamed.
9014 (write_global_stream): Output indexes encoded.
9015 (produce_asm_for_decls): Dump streams encoded.
9016 * lto-streamer.c (streamer_dump_file): New global var.
9017 * lto-streamer.h (streamer_dump_file): Declare.
9018 * passes.c (ipa_write_summaries): Initialize streamer dump.
9019 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
9022 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
9025 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
9026 offsets for register save directives. Emit a second batch of save
9027 directives, if need be, when the function accesses prior frames.
9029 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9031 * config/arc/fpu.md (fmasf4): Force operand to register.
9032 (fnmasf4): Likewise.
9034 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9036 * config/arc/arc-protos.h (arc_pad_return): Remove.
9037 * config/arc/arc.c (machine_function): Remove force_short_suffix
9039 (arc_print_operand): Adjust printing of '&'.
9040 (arc_verify_short): Remove conditional printing of short suffix.
9041 (arc_final_prescan_insn): Remove reference to size_reason.
9042 (pad_return): New function.
9043 (arc_reorg): Call pad_return.
9044 (arc_pad_return): Remove.
9045 (arc_init_machine_status): Remove reference to force_short_suffix.
9046 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
9047 (attr length): When attribute iscompact is true force to 2
9048 regardless; in the case of maybe check if we want to force the
9049 instruction to have 4 bytes length.
9050 (nopv): Change it to generate 4 byte long nop as well.
9051 (blockage): New pattern.
9052 (simple_return): Remove call to arc_pad_return.
9053 (p_return_i): Likewise.
9055 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9057 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
9059 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9061 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
9064 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9066 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
9067 for ARC700 and ARCv2.
9069 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
9072 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
9073 operands[2] instead of operands[1].
9076 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9078 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
9079 case, check whether the outer register overlaps an unallocatable
9080 register, not just whether it fits the required class.
9082 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9084 * poly-int.h (can_div_trunc_p): Add new overload in which all values
9086 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
9087 (memrefs_conflict_p): Likewise.
9088 (init_alias_analysis): Likewise.
9089 * cfgexpand.c (expand_debug_expr): Likewise.
9090 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
9091 * cse.c (fold_rtx): Likewise.
9092 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
9093 * expr.c (emit_block_move_hints): Likewise.
9094 (clear_storage_hints, push_block, emit_push_insn): Likewise.
9095 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
9096 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
9097 (emit_group_store): Likewise.
9098 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
9099 to read the PRE/POST_MODIFY increment.
9100 * calls.c (store_one_arg): Use strip_offset.
9101 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
9103 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
9105 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
9106 (simplify_binary_operation_1): Extend CONST_INT handling to
9108 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
9109 than a HOST_WIDE_INT.
9110 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
9112 (adjust_mems, add_stores): Update accodingly.
9113 (vt_canonicalize_addr): Track polynomial offsets.
9114 (emit_note_insn_var_location): Likewise.
9115 (vt_add_function_parameter): Likewise.
9116 (vt_initialize): Likewise.
9118 2018-06-12 Jeff Law <law@redhat.com>
9120 * config.gcc (alpha*-*-freebsd*): Remove.
9121 * config/alpha/freebsd.h: Remove.
9123 2018-06-12 David Malcolm <dmalcolm@redhat.com>
9126 * spellcheck-tree.c (levenshtein_distance): Rename to...
9127 (get_edit_distance): ...this, and update for underlying renaming.
9128 * spellcheck-tree.h (levenshtein_distance): Rename to...
9129 (get_edit_distance): ...this.
9130 * spellcheck.c (levenshtein_distance): Rename to...
9131 (get_edit_distance): ...this. Convert from Levenshtein distance
9132 to Damerau-Levenshtein distance by supporting transpositions of
9133 adjacent characters. Rename "v1" to "v_next" and "v0" to
9135 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
9136 (selftest::test_edit_distance_unit_test_oneway): ...this, and
9137 update for underlying renaming.
9138 (selftest::levenshtein_distance_unit_test): Rename to...
9139 (selftest::test_get_edit_distance_unit): ...this, and update for
9140 underlying renaming.
9141 (selftest::test_find_closest_string): Add example from PR 69968
9142 where transposition helps
9143 (selftest::test_metric_conditions): Update for renaming.
9144 (selftest::test_metric_conditions): Likewise.
9145 (selftest::spellcheck_c_tests): Likewise.
9146 * spellcheck.h (levenshtein_distance): Rename both overloads to...
9147 (get_edit_distance): ...this.
9148 (best_match::consider): Update for renaming.
9150 2018-06-12 Martin Sebor <msebor@redhat.com>
9152 PR tree-optimization/85259
9153 * builtins.c (compute_objsize): Handle constant offsets.
9154 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
9155 true iff a warning has been issued.
9156 * gimple.h (gimple_nonartificial_location): New function.
9157 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
9158 gimple_nonartificial_location and handle -Wno-system-headers.
9159 (handle_builtin_stxncpy): Same.
9161 2018-06-12 Martin Sebor <msebor@redhat.com>
9164 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
9166 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9168 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9169 BUILTIN_VEC_XST entries for pointer to double and long long.
9171 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
9174 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
9176 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
9179 2018-06-12 Martin Liska <mliska@suse.cz>
9181 * doc/options.texi: Document IntegerRange.
9183 2018-06-12 Martin Liska <mliska@suse.cz>
9185 * config/i386/i386.opt: Make MPX-related options as Deprecated.
9186 * opt-functions.awk: Handle Deprecated flag.
9187 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
9189 (read_cmdline_option): Report warning for a deprecated option.
9190 * opts.h (struct cl_option): Add new field cl_deprecated.
9191 (CL_ERR_DEPRECATED): New.
9193 2018-06-12 Martin Liska <mliska@suse.cz>
9195 * doc/options.texi: Document Deprecated option flag.
9197 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9199 * config/arc/arc-arch.h (arc_extras): New enum.
9200 (arc_cpu_t):Add field extra.
9201 (arc_cpu_types): Consider the extras.
9202 * config/arc/arc-cpus.def: Add extras info.
9203 * config/arc/arc-opts.h (processor_type): Consider extra field.
9204 * config/arc/arc.c (arc_override_options): Handle extra field.
9206 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9208 * config/arc/arc-arch.h: Update ARC_OPTX macro.
9209 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
9211 * config/arc/arc.c (arc_init): Update pic warning.
9212 (irq_range): Update irq range parsing warnings.
9213 (arc_override_options): Update various warning messages.
9214 (arc_handle_aux_attribute): Likewise.
9216 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
9218 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
9220 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
9222 * doc/sourcebuild.texi: Document usage of line number 0 in verify
9223 compiler messages directives.
9225 2018-06-12 Matthew Fortune <mfortune@gmail.com>
9227 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
9228 * config/mips/mips-tables.opt: Regenerate.
9229 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
9231 * doc/invoke.texi: Document -march=i6500.
9233 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
9235 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
9236 (i6400_gpmul): Add cpu_unit.
9237 (i6400_gpdiv): Likewise.
9238 (i6400_msa_add_d): Update reservations.
9239 (i6400_msa_int_add) Likewise.
9240 (i6400_msa_short_logic3) Likewise.
9241 (i6400_msa_short_logic2) Likewise.
9242 (i6400_msa_short_logic) Likewise.
9243 (i6400_msa_move) Likewise.
9244 (i6400_msa_cmp) Likewise.
9245 (i6400_msa_short_float2) Likewise.
9246 (i6400_msa_div_d) Likewise.
9247 (i6400_msa_long_logic1) Likewise.
9248 (i6400_msa_long_logic2) Likewise.
9249 (i6400_msa_mult) Likewise.
9250 (i6400_msa_long_float2) Likewise.
9251 (i6400_msa_long_float4) Likewise.
9252 (i6400_msa_long_float5) Likewise.
9253 (i6400_msa_long_float8) Likewise.
9254 (i6400_fpu_fadd): Include frint type.
9255 (i6400_fpu_store): New define_insn_reservation.
9256 (i6400_fpu_load): Likewise.
9257 (i6400_fpu_move): Likewise.
9258 (i6400_fpu_fcmp): Likewise.
9259 (i6400_fpu_fmadd): Likewise.
9260 (i6400_int_mult): Include imul3nc type and update reservation.
9261 (i6400_int_div): Include idiv3 type and update reservation.
9262 (i6400_int_load): Update to check type not move_type.
9263 (i6400_int_store): Likewise.
9264 (i6400_int_prefetch): Set zero latency.
9266 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
9268 * gcc.c: Document new %@{...} sequence.
9269 (LINK_COMMAND_SPEC): Use it for the -L switches.
9270 (cpp_unique_options): Use it for the -I switches.
9271 (at_file_argbuf): New global variable.
9272 (in_at_file): Likewise.
9273 (alloc_args): Create at_file_argbuf.
9274 (clear_args): Truncate at_file_argbuf.
9275 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
9276 (open_at_file): New function.
9277 (close_at_file): Likewise.
9278 (create_at_file): Delete.
9279 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
9282 (validate_switches_from_spec): Deal with %@{...} sequence.
9283 (validate_switches): Likewise.
9284 (driver::finalize): Call clear_args.
9286 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
9288 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
9290 2018-06-11 Martin Sebor <msebor@redhat.com>
9292 * doc/invoke.texi (-Wall): List -Wc++17-compat.
9293 (Wno-class-memaccess): Add @opindex.
9294 (Wno-templates, Wno-multiple-inheritance): Same.
9295 (Wno-virtual-inheritance, Wno-namespaces): Same.
9296 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
9297 (Wno-format-overflow, Wno-format-truncation): Same.
9298 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
9299 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
9300 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
9301 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
9302 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
9303 (Wno-misspelled-isr): Same.
9305 2018-06-11 Martin Sebor <msebor@redhat.com>
9307 * PR tree-optimization/86083
9308 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
9310 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
9312 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
9314 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
9317 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
9318 on the correct operand.
9319 (*movdi_internal64): Ditto.
9321 2018-06-11 Martin Liska <mliska@suse.cz>
9323 PR tree-optimization/86089
9324 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
9326 2018-06-11 Julia Koval <julia.koval@intel.com>
9328 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
9329 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
9330 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
9332 2018-06-11 Olivier Hainque <hainque@adacore.com>
9334 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
9335 for Ada with strict dwarf2.
9337 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
9340 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
9343 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
9345 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
9347 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
9349 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
9352 2018-06-08 Martin Liska <mliska@suse.cz>
9354 * tree-cfg.h (debug_function): Fix argument type to match
9357 2018-06-08 Martin Liska <mliska@suse.cz>
9359 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
9360 Remove usage of MPX-related (and removed) fields.
9361 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
9363 2018-06-08 David Malcolm <dmalcolm@redhat.com>
9365 * cfg.c (debug): Use TDF_NONE rather than 0.
9366 * cfghooks.c (debug): Likewise.
9367 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
9368 (struct dump_option_value_info): Convert to...
9369 (struct kv_pair): ...this template type.
9370 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
9372 (optinfo_verbosity_options): Likewise.
9373 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
9375 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
9376 than int for "optgroup_flags" param.
9377 (dump_generic_expr_loc): Use dump_flags_t rather than int for
9379 (dump_dec): Likewise.
9380 (dump_finish): Use TDF_NONE rather than 0.
9381 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
9382 rather than int for "optgroup_flags" param. Use TDF_NONE rather
9383 than 0. Update for change to option_ptr.
9384 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
9385 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
9386 0. Update for changes to optinfo_verbosity_options and
9388 (opt_info_switch_p): Convert optgroup_flags from int to
9390 (dump_basic_block): Use dump_flags_t rather than int
9391 for "dump_kind" param.
9392 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
9393 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
9394 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
9395 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
9396 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
9397 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
9398 TDF_NONE): Convert from macros to...
9399 (enum dump_flag): ...this new enum.
9400 (dump_flags_t): Update to use enum.
9401 (operator|, operator&, operator~, operator|=, operator&=):
9402 Implement for dump_flags_t.
9403 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
9404 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
9405 Convert from macros to...
9406 (enum optgroup_flag): ...this new enum.
9407 (optgroup_flags_t): New typedef.
9408 (operator|, operator|=): Implement for optgroup_flags_t.
9409 (struct dump_file_info): Convert field "alt_flags" to
9410 dump_flags_t. Convert field "optgroup_flags" to
9412 (dump_basic_block): Use dump_flags_t rather than int for param.
9413 (dump_generic_expr_loc): Likewise.
9414 (dump_dec): Likewise.
9415 (dump_register): Convert param "optgroup_flags" to
9417 (opt_info_enable_passes): Likewise.
9418 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
9420 * gimple-pretty-print.c (debug): Likewise.
9421 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
9422 (merged_store_group::apply_stores): Likewise.
9423 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
9424 * gimple.c (verify_gimple_pp): Likewise.
9425 * graphite-poly.c (print_pbb_body): Likewise.
9426 * passes.c (pass_manager::register_one_dump_file): Convert
9427 local "optgroup_flags" to optgroup_flags_t.
9428 * print-tree.c (print_node): Use TDF_NONE rather than 0.
9430 (debug_body): Likewise.
9431 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
9432 to optgroup_flags_t.
9433 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
9435 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
9436 (convert_mult_to_fma): Likewise.
9437 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
9438 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
9439 * tree-vect-data-refs.c (dump_lower_bound): Convert param
9440 "dump_kind" to dump_flags_t.
9442 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
9444 * config/rs6000/rs6000.c (min, max): Delete.
9446 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
9448 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
9449 -mabi=spe and -mabi=no-spe.
9451 2018-06-08 Martin Liska <mliska@suse.cz>
9453 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
9454 where we expect an existing summary.
9456 2018-06-08 Martin Liska <mliska@suse.cz>
9458 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
9459 * ipa-inline.h (estimate_edge_growth): Likewise.
9461 2018-06-08 Martin Liska <mliska@suse.cz>
9463 * cgraph.c (function_version_hasher::hash): Use
9464 cgraph_node::get_uid ().
9465 (function_version_hasher::equal):
9466 * cgraph.h (cgraph_node::get_uid): New method.
9467 * ipa-inline.c (update_caller_keys): Use
9468 cgraph_node::get_uid ().
9469 (update_callee_keys): Likewise.
9470 * ipa-utils.c (searchc): Likewise.
9471 (ipa_reduced_postorder): Likewise.
9472 * lto-cgraph.c (input_node): Likewise.
9473 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
9474 * symbol-summary.h (symtab_insertion): Likewise.
9475 (symtab_removal): Likewise.
9476 (symtab_duplication): Likewise.
9477 * tree-pretty-print.c (dump_function_header): Likewise.
9478 * tree-sra.c (convert_callers_for_node): Likewise.
9480 2018-06-08 Martin Liska <mliska@suse.cz>
9482 * cgraph.c (symbol_table::create_edge): Always assign a new
9484 (symbol_table::free_edge): Do not recycle numbers.
9485 * cgraph.h (cgraph_edge::get): New method.
9486 * symbol-summary.h (symtab_removal): Use it.
9487 (symtab_duplication): Likewise.
9488 (call_summary::hashable_uid): Remove.
9490 2018-06-08 Martin Liska <mliska@suse.cz>
9492 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
9493 (initialize_growth_caches): Remove.
9494 (free_growth_caches): Likewise.
9495 (do_estimate_edge_time): Use edge_growth_cache.
9496 (do_estimate_edge_size): Likewise.
9497 (do_estimate_edge_hints): Likewise.
9498 * ipa-inline.c (reset_edge_caches): Likewise.
9499 (recursive_inlining): Likewise.
9500 (inline_small_functions): Likewise.
9501 * ipa-inline.h (initialize_growth_caches): Remove.
9502 (estimate_edge_size): Likewise.
9503 (estimate_edge_time): Likewise.
9504 (estimate_edge_hints): Likewise.
9505 (reset_edge_growth_cache): Likewise.
9506 * symbol-summary.h (call_summary::remove): New method.
9508 2018-06-08 Martin Liska <mliska@suse.cz>
9510 * ipa-cp.c (class edge_clone_summary): New summary.
9511 (grow_edge_clone_vectors): Remove.
9512 (ipcp_edge_duplication_hook): Remove.
9513 (class edge_clone_summary_t): New call_summary class.
9514 (ipcp_edge_removal_hook): Remove.
9515 (edge_clone_summary_t::duplicate): New function.
9516 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
9517 (create_specialized_node): Likewise.
9518 (ipcp_driver): Initialize edge_clone_summaries and do not
9521 2018-06-08 Martin Liska <mliska@suse.cz>
9523 * symbol-summary.h (get): New function.
9524 (call_summary::m_initialize_when_cloning): New class member.
9526 2018-06-08 Martin Liska <mliska@suse.cz>
9528 * cgraph.c (cgraph_node::remove): Do not recycle uid.
9529 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
9530 (symbol_table::allocate_cgraph_symbol): Do not set uid.
9531 * passes.c (uid_hash_t): Record removed_nodes by their uids.
9532 (remove_cgraph_node_from_order): Use the removed_nodes set.
9533 (do_per_function_toporder): Likwise.
9534 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
9535 instead of summary_uid.
9536 (symtab_removal): Likewise.
9537 (symtab_duplication): Likewise.
9539 2018-06-08 Martin Liska <mliska@suse.cz>
9541 * ipa-cp.c (ipcp_store_bits_results): Use
9542 ipcp_transformation_sum.
9543 (ipcp_store_vr_results): Likewise.
9544 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
9546 (ipcp_transformation_initialize): ... this.
9547 (ipa_set_node_agg_value_chain):
9548 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
9549 (write_ipcp_transformation_info): Likewise.
9550 (read_ipcp_transformation_info): Likewise.
9551 (ipcp_update_bits): Likewise.
9552 (ipcp_update_vr): Likewise.
9553 (ipcp_transform_function): Likewise.
9554 * ipa-prop.h: Rename ipcp_transformation_summary to
9555 ipcp_transformation.
9556 (class ipcp_transformation_t): New function summary.
9557 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
9558 (ipa_get_agg_replacements_for_node): Likewise.
9560 2018-06-08 Martin Liska <mliska@suse.cz>
9562 * ipa-pure-const.c (struct funct_state_d): Do it class instead
9564 (class funct_state_summary_t): New function_summary class.
9565 (has_function_state): Remove.
9566 (get_function_state): Likewise.
9567 (set_function_state): Likewise.
9568 (add_new_function): Likewise.
9569 (funct_state_summary_t::insert): New function.
9570 (duplicate_node_data): Remove.
9571 (remove_node_data): Remove.
9572 (funct_state_summary_t::duplicate): New function.
9573 (register_hooks): Create new funct_state_summaries.
9574 (pure_const_generate_summary): Use it.
9575 (pure_const_write_summary): Likewise.
9576 (pure_const_read_summary): Likewise.
9577 (propagate_pure_const): Likewise.
9578 (propagate_nothrow): Likewise.
9579 (dump_malloc_lattice): Likewise.
9580 (propagate_malloc): Likewise.
9581 (execute): Do not register hooks, just remove summary
9583 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
9586 2018-06-08 Martin Liska <mliska@suse.cz>
9588 * ipa-reference.c (remove_node_data): Remove.
9589 (duplicate_node_data): Likewise.
9590 (class ipa_ref_var_info_summary_t): New class.
9591 (class ipa_ref_opt_summary_t): Likewise.
9592 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
9593 (get_reference_optimization_summary): Use
9594 ipa_ref_opt_sum_summaries.
9595 (set_reference_vars_info): Remove.
9596 (set_reference_optimization_summary): Likewise.
9597 (ipa_init): Create summaries.
9598 (init_function_info): Use function summary.
9599 (ipa_ref_opt_summary_t::duplicate): New function.
9600 (ipa_ref_opt_summary_t::remove): New function.
9601 (get_read_write_all_from_node): Fix GNU coding style.
9602 (propagate): Use function summary.
9603 (write_node_summary_p): Fix GNU coding style.
9604 (stream_out_bitmap): Likewise.
9605 (ipa_reference_read_optimization_summary): Use function summary.
9606 (ipa_reference_c_finalize): Do not release hooks.
9608 2018-06-08 Martin Liska <mliska@suse.cz>
9610 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
9611 (analyze_function_body): Extract multiple calls of get_create.
9612 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
9613 * ipa-inline.c (recursive_inlining): Use ::get method.
9614 * ipa-inline.h (estimate_edge_growth): Likewise.
9616 2018-06-08 Martin Liska <mliska@suse.cz>
9618 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
9620 (hsa_function_summary::hsa_function_summary): Use the new enum
9622 (hsa_gpu_implementation_p): Use hsa_summaries::get.
9623 * hsa-gen.c (hsa_get_host_function): Likewise.
9624 (get_brig_function_name): Likewise.
9625 * ipa-hsa.c (process_hsa_functions): Likewise.
9626 (ipa_hsa_write_summary): Likewise.
9627 * symbol-summary.h (symtab_duplication): Use ::get function/
9628 (get): New function.
9630 2018-06-08 Martin Liska <mliska@suse.cz>
9632 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
9634 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
9635 (hsa_register_kernel): Likewise.
9636 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
9637 * hsa-gen.c (hsa_get_host_function): Likewise.
9638 (get_brig_function_name): Likewise.
9639 (generate_hsa): Likewise.
9640 (pass_gen_hsail::execute): Likewise.
9641 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
9642 (devirtualization_time_bonus): Likewise.
9643 (ipcp_propagate_stage): Likewise.
9644 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
9645 (edge_set_predicate): Likewise.
9646 (evaluate_conditions_for_known_args): Likewise.
9647 (evaluate_properties_for_edge): Likewise.
9648 (ipa_fn_summary::reset): Likewise.
9649 (ipa_fn_summary_t::duplicate): Likewise.
9650 (dump_ipa_call_summary): Likewise.
9651 (ipa_dump_fn_summary): Likewise.
9652 (analyze_function_body): Likewise.
9653 (compute_fn_summary): Likewise.
9654 (estimate_edge_devirt_benefit): Likewise.
9655 (estimate_edge_size_and_time): Likewise.
9656 (estimate_calls_size_and_time): Likewise.
9657 (estimate_node_size_and_time): Likewise.
9658 (inline_update_callee_summaries): Likewise.
9659 (remap_edge_change_prob): Likewise.
9660 (remap_edge_summaries): Likewise.
9661 (ipa_merge_fn_summary_after_inlining): Likewise.
9662 (ipa_update_overall_fn_summary): Likewise.
9663 (read_ipa_call_summary): Likewise.
9664 (inline_read_section): Likewise.
9665 (write_ipa_call_summary): Likewise.
9666 (ipa_fn_summary_write): Likewise.
9667 (ipa_free_fn_summary): Likewise.
9668 * ipa-hsa.c (process_hsa_functions): Likewise.
9669 (ipa_hsa_write_summary): Likewise.
9670 (ipa_hsa_read_section): Likewise.
9671 * ipa-icf.c (sem_function::merge): Likewise.
9672 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
9673 (do_estimate_edge_time): Likewise.
9674 (estimate_size_after_inlining): Likewise.
9675 (estimate_growth): Likewise.
9676 (growth_likely_positive): Likewise.
9677 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
9678 (inline_call): Likewise.
9679 * ipa-inline.c (caller_growth_limits): Likewise.
9680 (can_inline_edge_p): Likewise.
9681 (can_inline_edge_by_limits_p): Likewise.
9682 (compute_uninlined_call_time): Likewise.
9683 (compute_inlined_call_time): Likewise.
9684 (want_inline_small_function_p): Likewise.
9685 (edge_badness): Likewise.
9686 (update_caller_keys): Likewise.
9687 (update_callee_keys): Likewise.
9688 (recursive_inlining): Likewise.
9689 (inline_small_functions): Likewise.
9690 (inline_to_all_callers_1): Likewise.
9691 (dump_overall_stats): Likewise.
9692 (early_inline_small_functions): Likewise.
9693 (early_inliner): Likewise.
9694 * ipa-inline.h (estimate_edge_growth): Likewise.
9695 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
9696 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
9697 * ipa-prop.h (IPA_NODE_REF): Likewise.
9698 (IPA_EDGE_REF): Likewise.
9699 * ipa-pure-const.c (malloc_candidate_p): Likewise.
9700 (propagate_malloc): Likewise.
9701 * ipa-split.c (execute_split_functions): Likewise.
9702 * symbol-summary.h: Rename get to get_create.
9704 (get_create): Likewise.
9705 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
9707 2018-06-08 Martin Liska <mliska@suse.cz>
9709 * symbol-summary.h (release): Move definition out of class
9711 (symtab_removal): Likewise.
9712 (symtab_duplication): Likewise.
9714 2018-06-08 Martin Liska <mliska@suse.cz>
9716 * symbol-summary.h (function_summary): Move constructor
9717 implementation out of class declaration.
9718 (release): Likewise.
9719 (symtab_insertion): Likewise.
9720 (symtab_removal): Likewise.
9721 (symtab_duplication): Likewise.
9724 2018-06-08 Martin Liska <mliska@suse.cz>
9726 * Makefile.in: Remove support for MPX (macros, related functions,
9727 fields in cgraph_node, ...).
9728 * builtin-types.def (BT_BND): Likewise.
9729 (BT_FN_BND_CONST_PTR): Likewise.
9730 (BT_FN_CONST_PTR_BND): Likewise.
9731 (BT_FN_VOID_PTR_BND): Likewise.
9732 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
9733 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
9734 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
9735 (expand_builtin_mempcpy_with_bounds): Likewise.
9736 (expand_builtin_memset_with_bounds): Likewise.
9737 (expand_builtin_memset_args): Likewise.
9738 (std_expand_builtin_va_start): Likewise.
9739 (expand_builtin): Likewise.
9740 (expand_builtin_with_bounds): Likewise.
9741 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
9742 (DEF_LIB_BUILTIN_CHKP): Likewise.
9743 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
9744 (DEF_CHKP_BUILTIN): Likewise.
9745 (BUILT_IN_MEMCPY): Likewise.
9746 (BUILT_IN_MEMMOVE): Likewise.
9747 (BUILT_IN_MEMPCPY): Likewise.
9748 (BUILT_IN_MEMSET): Likewise.
9749 (BUILT_IN_STPCPY): Likewise.
9750 (BUILT_IN_STRCAT): Likewise.
9751 (BUILT_IN_STRCHR): Likewise.
9752 (BUILT_IN_STRCPY): Likewise.
9753 (BUILT_IN_STRLEN): Likewise.
9754 (BUILT_IN_MEMCPY_CHK): Likewise.
9755 (BUILT_IN_MEMMOVE_CHK): Likewise.
9756 (BUILT_IN_MEMPCPY_CHK): Likewise.
9757 (BUILT_IN_MEMSET_CHK): Likewise.
9758 (BUILT_IN_STPCPY_CHK): Likewise.
9759 (BUILT_IN_STRCAT_CHK): Likewise.
9760 (BUILT_IN_STRCPY_CHK): Likewise.
9761 * calls.c (store_bounds): Likewise.
9762 (emit_call_1): Likewise.
9763 (special_function_p): Likewise.
9764 (maybe_warn_nonstring_arg): Likewise.
9765 (initialize_argument_information): Likewise.
9766 (finalize_must_preallocate): Likewise.
9767 (compute_argument_addresses): Likewise.
9768 (expand_call): Likewise.
9769 * cfgexpand.c (expand_call_stmt): Likewise.
9770 (expand_return): Likewise.
9771 (expand_gimple_stmt_1): Likewise.
9772 (pass_expand::execute): Likewise.
9773 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
9774 (cgraph_node::remove): Likewise.
9775 (cgraph_node::dump): Likewise.
9776 (cgraph_node::verify_node): Likewise.
9777 * cgraph.h (chkp_function_instrumented_p): Likewise.
9778 (symtab_node::get_alias_target): Likewise.
9779 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
9780 (cgraph_local_p): Likewise.
9781 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
9782 (cgraph_edge::rebuild_references): Likewise.
9783 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
9784 (walk_polymorphic_call_targets): Likewise.
9785 (cgraph_node::expand_thunk): Likewise.
9786 (symbol_table::output_weakrefs): Likewise.
9787 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
9788 (ix86_handle_option): Likewise.
9789 * config/i386/constraints.md: Likewise.
9790 * config/i386/i386-builtin-types.def (BND): Likewise.
9794 * config/i386/i386-builtin.def (BDESC_END): Likewise.
9795 (BDESC_FIRST): Likewise.
9797 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
9798 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
9799 * config/i386/i386.c (enum reg_class): Likewise.
9800 (ix86_target_string): Likewise.
9801 (ix86_option_override_internal): Likewise.
9802 (ix86_conditional_register_usage): Likewise.
9803 (ix86_valid_target_attribute_inner_p): Likewise.
9804 (ix86_set_indirect_branch_type): Likewise.
9805 (ix86_set_current_function): Likewise.
9806 (ix86_function_arg_regno_p): Likewise.
9807 (init_cumulative_args): Likewise.
9808 (ix86_function_arg_advance): Likewise.
9809 (ix86_function_arg): Likewise.
9810 (ix86_pass_by_reference): Likewise.
9811 (ix86_function_value_regno_p): Likewise.
9812 (ix86_function_value_1): Likewise.
9813 (ix86_function_value_bounds): Likewise.
9814 (ix86_return_in_memory): Likewise.
9815 (ix86_setup_incoming_vararg_bounds): Likewise.
9816 (ix86_va_start): Likewise.
9817 (indirect_thunk_need_prefix): Likewise.
9818 (print_reg): Likewise.
9819 (ix86_print_operand): Likewise.
9820 (ix86_expand_call): Likewise.
9821 (ix86_output_function_return): Likewise.
9822 (reg_encoded_number): Likewise.
9823 (BDESC_VERIFYS): Likewise.
9824 (ix86_init_mpx_builtins): Likewise.
9825 (ix86_init_builtins): Likewise.
9826 (ix86_emit_cmove): Likewise.
9827 (ix86_emit_move_max): Likewise.
9828 (ix86_expand_builtin): Likewise.
9829 (ix86_builtin_mpx_function): Likewise.
9830 (ix86_get_arg_address_for_bt): Likewise.
9831 (ix86_load_bounds): Likewise.
9832 (ix86_store_bounds): Likewise.
9833 (ix86_load_returned_bounds): Likewise.
9834 (ix86_store_returned_bounds): Likewise.
9835 (ix86_class_likely_spilled_p): Likewise.
9836 (ix86_hard_regno_mode_ok): Likewise.
9837 (x86_order_regs_for_local_alloc): Likewise.
9838 (ix86_mitigate_rop): Likewise.
9839 (ix86_bnd_prefixed_insn_p): Likewise.
9840 (ix86_mpx_bound_mode): Likewise.
9841 (ix86_make_bounds_constant): Likewise.
9842 (ix86_initialize_bounds): Likewise.
9843 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
9844 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
9845 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
9846 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
9847 (TARGET_CHKP_BOUND_MODE): Likewise.
9848 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
9849 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
9850 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
9851 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
9852 * config/i386/i386.h (TARGET_MPX): Likewise.
9853 (TARGET_MPX_P): Likewise.
9854 (VALID_BND_REG_MODE): Likewise.
9855 (FIRST_BND_REG): Likewise.
9856 (LAST_BND_REG): Likewise.
9857 (enum reg_class): Likewise.
9858 (BND_REG_P): Likewise.
9859 (BND_REGNO_P): Likewise.
9860 (BNDmode): Likewise.
9861 (ADJUST_INSN_LENGTH): Likewise.
9862 * config/i386/i386.md: Likewise.
9863 * config/i386/i386.opt: Likewise.
9864 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
9865 (defined): Likewise.
9866 (LINK_MPX): Likewise.
9867 (MPX_SPEC): Likewise.
9868 (LIBMPX_SPEC): Likewise.
9869 (LIBMPXWRAPPERS_SPEC): Likewise.
9870 (CHKP_SPEC): Likewise.
9871 * config/i386/predicates.md: Likewise.
9872 * dbxout.c (dbxout_type): Likewise.
9873 * doc/extend.texi: Likewise.
9874 * doc/invoke.texi: Likewise.
9875 * doc/md.texi: Likewise.
9876 * doc/tm.texi: Likewise.
9877 * doc/tm.texi.in: Likewise.
9878 * dwarf2out.c (is_base_type): Likewise.
9879 (gen_formal_types_die): Likewise.
9880 (gen_subprogram_die): Likewise.
9881 (gen_type_die_with_usage): Likewise.
9882 (gen_decl_die): Likewise.
9883 (dwarf2out_late_global_decl): Likewise.
9884 * expr.c (expand_assignment): Likewise.
9885 (emit_storent_insn): Likewise.
9886 (store_expr_with_bounds): Likewise.
9887 (store_expr): Likewise.
9888 (expand_expr_real_1): Likewise.
9889 * expr.h (store_expr_with_bounds): Likewise.
9890 * function.c (use_register_for_decl): Likewise.
9891 (struct bounds_parm_data): Likewise.
9892 (assign_parms_augmented_arg_list): Likewise.
9893 (assign_parm_find_entry_rtl): Likewise.
9894 (assign_parm_is_stack_parm): Likewise.
9895 (assign_parm_load_bounds): Likewise.
9896 (assign_bounds): Likewise.
9897 (assign_parms): Likewise.
9898 (expand_function_start): Likewise.
9899 * gcc.c (CHKP_SPEC): Likewise.
9900 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9901 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
9902 (wrestrict_dom_walker::check_call): Likewise.
9903 * gimple.c (gimple_build_call_from_tree): Likewise.
9904 * gimple.h (enum gf_mask): Likewise.
9905 (gimple_call_with_bounds_p): Likewise.
9906 (gimple_call_set_with_bounds): Likewise.
9907 * gimplify.c (gimplify_init_constructor): Likewise.
9908 * ipa-cp.c (initialize_node_lattices): Likewise.
9909 (propagate_constants_across_call): Likewise.
9910 (find_more_scalar_values_for_callers_subset): Likewise.
9911 * ipa-hsa.c (process_hsa_functions): Likewise.
9912 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
9913 * ipa-icf.c (sem_function::merge): Likewise.
9914 * ipa-inline.c (early_inliner): Likewise.
9915 * ipa-pure-const.c (warn_function_noreturn): Likewise.
9916 (warn_function_cold): Likewise.
9917 (propagate_pure_const): Likewise.
9918 * ipa-ref.h (enum GTY): Likewise.
9919 * ipa-split.c (find_retbnd): Likewise.
9920 (consider_split): Likewise.
9921 (split_function): Likewise.
9922 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
9923 * ipa.c (walk_polymorphic_call_targets): Likewise.
9924 (symbol_table::remove_unreachable_nodes): Likewise.
9925 (process_references): Likewise.
9926 (cgraph_build_static_cdtor_1): Likewise.
9927 * lto-cgraph.c (lto_output_node): Likewise.
9928 (output_refs): Likewise.
9929 (compute_ltrans_boundary): Likewise.
9930 (input_overwrite_node): Likewise.
9931 (input_node): Likewise.
9932 (input_cgraph_1): Likewise.
9933 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
9934 * passes.c (pass_manager::execute_early_local_passes): Likewise.
9935 (class pass_chkp_instrumentation_passes): Likewise.
9936 (make_pass_chkp_instrumentation_passes): Likewise.
9937 * passes.def: Likewise.
9938 * rtl.h (struct GTY): Likewise.
9939 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
9940 * stor-layout.c (layout_type): Likewise.
9941 * symtab.c: Likewise.
9942 * target.def: Likewise.
9943 * targhooks.c (default_chkp_bound_type): Likewise.
9944 (default_chkp_bound_mode): Likewise.
9945 (default_builtin_chkp_function): Likewise.
9946 (default_chkp_function_value_bounds): Likewise.
9947 (default_chkp_make_bounds_constant): Likewise.
9948 (default_chkp_initialize_bounds): Likewise.
9949 * targhooks.h (default_chkp_bound_type): Likewise.
9950 (default_chkp_bound_mode): Likewise.
9951 (default_builtin_chkp_function): Likewise.
9952 (default_chkp_function_value_bounds): Likewise.
9953 (default_chkp_make_bounds_constant): Likewise.
9954 (default_chkp_initialize_bounds): Likewise.
9955 * toplev.c (compile_file): Likewise.
9956 (process_options): Likewise.
9957 * tree-core.h (DEF_BUILTIN): Likewise.
9958 (DEF_BUILTIN_CHKP): Likewise.
9959 * tree-inline.c (declare_return_variable): Likewise.
9960 (remap_gimple_stmt): Likewise.
9961 (copy_bb): Likewise.
9962 (initialize_inlined_parameters): Likewise.
9963 (expand_call_inline): Likewise.
9964 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
9965 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
9966 (make_pass_ipa_chkp_produce_thunks): Likewise.
9967 (make_pass_chkp): Likewise.
9968 (make_pass_chkp_opt): Likewise.
9969 (make_pass_chkp_instrumentation_passes): Likewise.
9970 * tree-pretty-print.c (dump_generic_node): Likewise.
9971 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
9972 * tree-ssa-dce.c (propagate_necessity): Likewise.
9973 (eliminate_unnecessary_stmts): Likewise.
9974 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
9975 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
9976 * tree-ssa-sccvn.h: Likewise.
9977 * tree-ssa-strlen.c (get_string_length): Likewise.
9978 (valid_builtin_call): Likewise.
9979 (adjust_last_stmt): Likewise.
9980 (handle_builtin_strchr): Likewise.
9981 (handle_builtin_strcpy): Likewise.
9982 (handle_builtin_stxncpy): Likewise.
9983 (handle_builtin_memcpy): Likewise.
9984 (handle_builtin_strcat): Likewise.
9985 (strlen_check_and_optimize_stmt): Likewise.
9986 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
9987 * tree-streamer-in.c: Likewise.
9988 * tree-streamer.c (record_common_node): Likewise.
9989 * tree.c (tree_code_size): Likewise.
9990 (wide_int_to_tree_1): Likewise.
9991 (type_contains_placeholder_1): Likewise.
9992 (build_common_tree_nodes): Likewise.
9993 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
9994 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
9995 (POINTER_BOUNDS_P): Likewise.
9996 (BOUNDED_TYPE_P): Likewise.
9997 (BOUNDED_P): Likewise.
9998 (CALL_WITH_BOUNDS_P): Likewise.
9999 (pointer_bounds_type_node): Likewise.
10000 * value-prof.c (gimple_ic): Likewise.
10001 * var-tracking.c (vt_add_function_parameters): Likewise.
10002 * varasm.c (make_decl_rtl): Likewise.
10003 (assemble_start_function): Likewise.
10004 (output_constant): Likewise.
10005 (maybe_assemble_visibility): Likewise.
10006 * varpool.c (ctor_for_folding): Likewise.
10007 * chkp-builtins.def: Remove.
10008 * ipa-chkp.c: Remove.
10009 * ipa-chkp.h: Remove.
10010 * rtl-chkp.c: Remove.
10011 * rtl-chkp.h: Remove.
10012 * tree-chkp-opt.c: Remove.
10013 * tree-chkp.c: Remove.
10014 * tree-chkp.h: Remove.
10016 2018-06-07 Carl Love <cel@us.ibm.com>
10018 * config/rs6000/vsx.md (vextract_fp_from_shorth,
10019 vextract_fp_from_shortl): Add BE support.
10021 2018-06-07 Paul Koning <ni1d@arrl.net>
10023 * compare-elim.c (try_merge_compare): Don't merge compare if
10024 address contains a side effect.
10025 (try_eliminate_compare): Likewise.
10027 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
10029 * config.gcc: Support "tremont".
10030 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
10031 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
10033 * config/i386/i386.c (m_TREMONT): Define.
10034 (processor_target_table): Add "tremont".
10035 (PTA_TREMONT): Define.
10036 (ix86_lea_outperforms): Add TARGET_TREMONT.
10037 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
10038 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
10039 and M_INTEL_GOLDMONT_PLUS.
10040 (fold_builtin_cpu): Add "tremont".
10041 (ix86_add_stmt_cost): Add TARGET_TREMONT.
10042 (ix86_option_override_internal): Add "tremont".
10043 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
10044 (processor_type): Add PROCESSOR_TREMONT.
10045 * config/i386/x86-tune.def: Add m_TREMONT.
10046 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
10048 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10050 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
10051 symbol defined for msp430i* devices to be lower case.
10053 2018-06-07 Richard Biener <rguenther@suse.de>
10055 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
10056 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
10057 Properly wrap signed arithmetic if overflow wraps.
10059 2018-06-07 Jakub Jelinek <jakub@redhat.com>
10061 PR tree-optimization/69615
10062 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
10063 of a cast from a same precision integral SSA_NAME in a bb dominated
10064 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
10065 cast to utype if rhs2 has already a compatible type.
10067 2018-06-07 Richard Biener <rguenther@suse.de>
10069 PR tree-optimization/85935
10070 * graphite-scop-detection.c (find_params_in_bb): Analyze
10071 condition operands with respect to the correct loop. Assert
10072 the analysis doesn't fail.
10074 2018-06-04 Carl Love <cel@us.ibm.com>
10076 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
10077 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
10078 as it is slightly cheaper.
10079 (first_match_or_eos_index_<mode>):
10080 Calculate index using natural element order.
10081 (first_match_index_<mode>):
10082 Calculate index using natural element order.
10083 (first_match_or_eos_index_<mode>):
10084 Calculate index using natural order.
10085 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
10086 for BE and LE modes.
10087 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
10088 P9V_BUILTIN_VCLZLSBB_V16QI.
10089 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
10092 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
10094 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
10095 indentation and line wrap for many prototypes. Add missing
10096 @smallexample directives around block of prototypes for vec_xl and
10099 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
10101 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
10102 track if we pass or return IEEE 128-bit floating point.
10103 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
10104 C++ mangling that is compatible with GCC 8.1.
10105 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
10106 (init_cumulative_args): Note if we pass or return IEEE 128-bit
10107 floating point types.
10108 (rs6000_function_arg_advance_1): Likewise.
10109 (rs6000_mangle_type): Optionally generate mangled names that match
10110 what GCC 8.1 generated for IEEE 128-bit floating point types.
10111 (rs6000_globalize_decl_name): If we have an external function that
10112 passes or returns IEEE 128-bit types, generate a weak reference
10113 from the mangled name used in GCC 8.1 to the current mangled
10115 (rs6000_init_builtins): Make __ibm128 use the long double type if
10116 long double is IBM extended double. Make __float128 use the long
10117 double type if long double is IEEE 128-bit.
10120 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
10121 macro for __ibm128 built-in functions.
10122 (PACK_IF): Add __ibm128 pack/unpack functions.
10123 (UNPACK_IF): Likewise.
10124 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
10125 enable long double built-in functions if long double is IEEE
10126 128-bit floating point.
10127 (rs6000_invalid_builtin): Update long double built-in function
10129 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
10130 functions, adjust the built-in function to use the long double
10131 built-in function if __ibm128 and long double are the same type.
10132 * doc/extend.texi (PowerPC builtins): Update documention for
10133 __builtin_{,un}pack_longdouble. Add documentation for
10134 __builtin_{,un}pack_ibm128.
10136 2018-06-06 Jim Wilson <jimw@sifive.com>
10138 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
10139 (struct machine_function): New field interrupt_mode.
10140 (riscv_handle_type_attribute): New function. Add forward declaration.
10141 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
10142 (riscv_expand_epilogue): Check interrupt_mode field.
10143 (riscv_set_current_function): Check interrupt attribute args and
10144 set interrupt_mode field.
10145 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
10146 (riscv_sret, riscv_uret): New.
10147 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
10148 new arguments to interrupt attribute.
10150 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
10153 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
10154 Don't handle -mcpu=power8 if -mpower9-vector is also used.
10156 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
10158 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
10159 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
10160 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
10161 several redundant entries.
10163 2018-06-06 David Malcolm <dmalcolm@redhat.com>
10165 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
10166 type from "rtx" to "rtx_insn *".
10167 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
10168 for local "call_insn", removing cast.
10169 (ix86_expand_call): Likewise, introducing a "call_insn" local.
10171 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
10173 PR tree-optimization/86066
10174 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
10175 for BIT_INSERT_EXPR stores.
10177 2018-06-06 Richard Biener <rguenther@suse.de>
10179 PR tree-optimization/86062
10180 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
10181 component refs ontop
10182 of to be offsetted base.
10184 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10186 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
10187 to be static and remove check on interrupt attribute name.
10189 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
10191 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
10192 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
10194 2018-06-05 Steve Ellcey <sellcey@cavium.com>
10197 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
10199 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
10200 Remove second argument, change how error is called.
10201 (aarch64_layout_arg): Remove second argument from
10202 aarch64_err_no_fpadvsimd call.
10203 (aarch64_init_cumulative_args): Ditto.
10204 (aarch64_gimplify_va_arg_expr): Ditto.
10205 * config/aarch64/aarch64.md (mov<mode>): Ditto.
10207 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
10209 * config/i386/i386.md (simple_return_indirect_internal): New expander.
10210 (*simple_return_indirect_internal<mode>): Rename from
10211 simple_return_indirect_internal. Use W mode iterator.
10212 (rstorssp): New expander.
10213 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
10214 (clrssbsy): New expander.
10215 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
10217 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
10219 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
10221 (cmse_check_pointed_object): Likewise.
10223 2018-06-05 Martin Liska <mliska@suse.cz>
10225 PR gcov-profile/47618
10226 * doc/invoke.texi: Document how -fprofile-dir format
10229 2018-06-05 Richard Biener <rguenther@suse.de>
10231 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
10232 removal pretend DOM info isn't available so we do not update
10233 it and only remove edges, not dominated blocks. Actually free
10234 DOM info in case we removed something. Remove unreachable blocks.
10235 (mfb_keep_latches): Work with either DOM info or marked backedges.
10236 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
10237 first. Mark backedges if DOM info isn't available.
10238 (Re-)compute DOM info after cleanup_control_flow_pre.
10240 2018-06-05 Richard Biener <rguenther@suse.de>
10242 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
10243 (locus_discrim_hasher::hash): Adjust.
10244 (locus_discrim_hasher::equal): Likewise.
10245 (next_discriminator_for_locus): Work on line directly.
10246 (same_line_p): Pass in expanded locus1 as well.
10247 (assign_discriminators): Avoid redundant location expansions.
10249 2018-06-05 Richard Biener <rguenther@suse.de>
10251 PR tree-optimization/86046
10252 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
10253 if required after clearing TREE_ADDRESSABLE.
10255 2018-06-05 Richard Biener <rguenther@suse.de>
10257 PR tree-optimization/86047
10258 * tree-ssa-loop.c (for_each_index): Glob handling of all
10259 decls and constants and really handle all of them.
10261 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10264 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
10265 qualifier_void_pointer and qualifier_const_void_pointer.
10266 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
10267 (arm_init_builtins): Handle the above.
10268 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
10269 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
10270 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
10273 2018-06-05 Martin Liska <mliska@suse.cz>
10275 * auto-profile.c (read_autofdo_file): Do not use
10276 gcov_ctr_summary struct.
10277 (afdo_callsite_hot_enough_for_early_inline): Likewise.
10278 * coverage.c (struct counts_entry): Likewise.
10279 (read_counts_file): Read just single summary entry.
10280 (get_coverage_counts): Use gcov_summary struct.
10281 * coverage.h (get_coverage_counts): Likewise.
10282 * gcov-dump.c (dump_working_sets): Likewise.
10283 (tag_summary): Dump just single summary.
10284 * gcov-io.c (gcov_write_summary): Write just histogram
10286 (gcov_read_summary): Read just single summary.
10287 (compute_working_sets): Use gcov_summary struct.
10288 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
10289 of GCOV_COUNTERS_SUMMABLE.
10290 (GCOV_COUNTERS_SUMMABLE): Remove.
10291 (GCOV_FIRST_VALUE_COUNTER): Replace with
10292 GCOV_COUNTER_V_INTERVAL.
10293 (struct gcov_ctr_summary): Remove.
10294 (struct gcov_summary): Directly use fields of former
10296 (compute_working_sets): Use gcov_summary struct.
10297 * gcov.c (read_count_file): Do not use ctrs fields.
10298 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
10300 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
10302 * profile.c: Likewise.
10303 * profile.h: Likewise.
10305 2018-06-05 Martin Liska <mliska@suse.cz>
10307 PR gcov-profile/84846
10308 * gcov.c (output_lines): Print working directory only
10309 in intermediate format.
10311 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
10313 * config/s390/s390-builtin-types.def: Add void function type.
10314 * config/s390/s390-builtins.def: Use the function type for the
10317 2018-06-04 Jim Wilson <jimw@sifive.com>
10319 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
10321 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
10322 and maybe_eh_return. Change regno to unsigned int. Use new args to
10323 handle EH_RETURN_DATA_REGNO registers properly.
10324 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
10325 (riscv_expand_epilogue): Update comment. Change argument name and
10326 type. Update code to use new name and type. Pass new args to
10327 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
10329 * config/riscv/riscv.md (NORMAL_RETURN): New.
10330 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
10331 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
10332 (eh_return): Call gen_eh_return_internal and emit barrier.
10333 (eh_return_internal): Call riscv_expand_epilogue.
10335 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
10337 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
10338 bit_insertion field and declare can_be_merged_into method.
10339 (merged_store_group::can_be_merged_into): New method.
10340 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
10341 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
10342 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
10344 2018-06-04 Richard Biener <rguenther@suse.de>
10346 PR tree-optimization/85955
10347 * builtins.c (fold_builtin_sincos): Convert pointers to
10348 destination to appropriate type before dereferencing.
10350 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
10352 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
10354 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
10356 * expr.c (expand_expr_real_1): Force the operand into memory if
10357 its TYPE_MODE is BLKmode and if there is no integer mode for
10358 the number of bits being extracted.
10360 2018-06-04 Jakub Jelinek <jakub@redhat.com>
10364 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
10365 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
10367 2018-06-04 Richard Biener <rguenther@suse.de>
10369 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
10370 (cleanup_tree_cfg_noloop): ... single caller. Do
10371 start_recording_case_labels later.
10373 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
10375 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
10376 to _IMMINTRIN_H_INCLUDED.
10377 * config/i386/pconfigintrin.h: Ditto.
10378 * config/i386/waitpkgintrin.h: Ditto.
10379 * config/i386/immintrin.h: Add includes for sgxintrin.h,
10380 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
10381 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
10382 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
10383 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
10384 waitpkgintrin.h and cldemoteintrin.h.
10386 2018-06-04 Richard Biener <rguenther@suse.de>
10388 PR tree-optimization/86038
10389 * tracer.c (find_best_successor): Check probability for
10390 being initialized, bail out if not.
10392 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
10395 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
10396 of bits to ignore when comparing architectures.
10398 2018-06-04 Jakub Jelinek <jakub@redhat.com>
10400 PR tree-optimization/69615
10401 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
10402 maximum or minimum of the type, try to merge it also as if
10403 range1 is + [-, x - 1] or + [x + 1, -].
10406 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
10408 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
10410 PR tree-optimization/86034
10411 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
10412 the unsigned bitfield type in a bit insertion sequence if it does not
10413 have a larger precision than the bitfield size.
10414 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
10416 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
10418 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
10420 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
10422 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
10423 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
10424 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
10425 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
10427 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
10429 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
10430 Disable -fdelete-null-pointer-checks for ELF toolchain.
10432 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
10433 Kito Cheng <kito.cheng@gmail.com>
10435 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
10436 (nds32le-*-*, nds32be-*-*): Integrate checking process.
10437 (nds32*-*-*): Add glibc and uclibc conditions.
10438 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
10439 (TARGET_EXCEPT_UNWIND_INFO): Define.
10440 * config/nds32/elf.h: New file.
10441 * config/nds32/linux.h: New file.
10442 * config/nds32/nds32-elf.opt: New file.
10443 * config/nds32/nds32-linux.opt: New file.
10444 * config/nds32/nds32-fp-as-gp.c
10445 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
10446 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
10448 (nds32_asm_file_end): Ditto.
10449 (nds32_print_operand): Ditto.
10450 (nds32_insert_attributes): Ditto.
10451 (nds32_init_libfuncs): New function.
10452 (TARGET_HAVE_TLS): Define.
10453 (TARGET_INIT_LIBFUNCS): Define.
10454 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
10456 (TARGET_ELF): Apply different mcmodel setting.
10457 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
10458 been migrated into elf.h and linux.h files.
10459 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
10460 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
10461 (mcmodel): The content has been migrated into nds32-elf.opt and
10462 nds32-linux.opt files.
10463 * config/nds32/t-elf: New file.
10464 * config/nds32/t-linux: New file.
10466 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
10467 Shiva Chen <shiva0217@gmail.com>
10469 * config/nds32/constants.md (unspec_volatile_element): Add
10470 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
10471 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
10473 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
10474 (make_pass_nds32_fp_as_gp): Declare.
10475 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
10477 (nds32_asm_function_end_prologue): Remove unused asm output.
10478 (nds32_asm_function_begin_epilogue): Remove unused asm output.
10479 (nds32_asm_file_start): Output necessary fp_as_gp information.
10480 (nds32_option_override): Adjust register usage.
10481 (nds32_expand_prologue): Consider fp_as_gp situation.
10482 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
10483 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
10486 (simple_return): Ditto.
10487 (omit_fp_begin): Output special directive for fp_as_gp.
10488 (omit_fp_end): Output special directive for fp_as_gp.
10489 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
10490 mforbid-fp-as-gp): New options.
10492 2018-06-01 Mark Wielaard <mark@klomp.org>
10494 * dwarf2out.c (dwarf2out_finish): Remove generation of
10495 DW_AT_loclists_base.
10497 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
10499 * gimple-ssa-store-merging.c: Include gimple-fold.h.
10500 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
10501 (struct merged_store_group): Add bit_insertion field.
10502 (dump_char_array): Use standard hexadecimal format.
10503 (merged_store_group::merged_store_group): Set bit_insertion to false.
10504 (merged_store_group::apply_stores): Use optimal buffer size. Deal
10505 with BIT_INSERT_EXPR stores. Move up code updating the mask and
10506 also print the mask in the dump file.
10507 (pass_store_merging::gate): Minor tweak.
10508 (imm_store_chain_info::coalesce_immediate): Fix wrong association
10509 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
10510 stores with INTEGER_CST stores.
10511 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
10512 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
10513 and use it throughout. Generate bit insertion sequences if need be.
10514 (pass_store_merging::process_store): Remove redundant condition.
10515 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
10517 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
10519 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
10520 the 128-bit floating point types. Fix function comment.
10522 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10524 * config/aarch64/aarch64-simd.md
10525 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
10527 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
10530 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
10532 PR tree-optimization/85989
10533 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
10535 (backprop::intersect_uses): Check it when deciding whether this
10536 is a backedge reference.
10537 (backprop::process_block): Add each phi to m_visited_phis
10538 after visiting it, then clear it at the end.
10540 2018-06-01 Richard Biener <rguenther@suse.de>
10542 * tree-vectorizer.h (vect_dr_stmt): New function.
10543 (vect_get_load_cost): Adjust.
10544 (vect_get_store_cost): Likewise.
10545 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
10546 Use vect_dr_stmt instead of DR_SMTT.
10547 (vect_record_base_alignments): Likewise.
10548 (vect_calculate_target_alignment): Likewise.
10549 (vect_compute_data_ref_alignment): Likewise and make static.
10550 (vect_update_misalignment_for_peel): Likewise.
10551 (vect_verify_datarefs_alignment): Likewise.
10552 (vector_alignment_reachable_p): Likewise.
10553 (vect_get_data_access_cost): Likewise. Pass down
10554 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
10555 (vect_get_peeling_costs_all_drs): Likewise.
10556 (vect_peeling_hash_get_lowest_cost): Likewise.
10557 (vect_enhance_data_refs_alignment): Likewise.
10558 (vect_find_same_alignment_drs): Likewise.
10559 (vect_analyze_data_refs_alignment): Likewise.
10560 (vect_analyze_group_access_1): Likewise.
10561 (vect_analyze_group_access): Likewise.
10562 (vect_analyze_data_ref_access): Likewise.
10563 (vect_analyze_data_ref_accesses): Likewise.
10564 (vect_vfa_segment_size): Likewise.
10565 (vect_small_gap_p): Likewise.
10566 (vectorizable_with_step_bound_p): Likewise.
10567 (vect_prune_runtime_alias_test_list): Likewise.
10568 (vect_analyze_data_refs): Likewise.
10569 (vect_supportable_dr_alignment): Likewise.
10570 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
10571 (vect_gen_prolog_loop_niters): Likewise.
10572 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
10573 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
10575 (vect_recog_mask_conversion_pattern): Likewise.
10576 (vect_try_gather_scatter_pattern): Likewise.
10577 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
10578 to vect_get_store_cost.
10579 (vect_get_store_cost): Get stmt_info instead of DR.
10580 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
10581 (vect_get_load_cost): Get stmt_info instead of DR.
10583 2018-06-01 Richard Biener <rguenther@suse.de>
10585 PR middle-end/86017
10586 * gimple-fold.c (var_decl_component_p): Also allow offsetted
10587 vars wrapped in MEM_REFs.
10589 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
10591 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10592 Fix subreg tests so that we only return a choice between
10593 GENERAL_REGS and FP_REGS if the original classes included both.
10595 2018-06-01 Richard Biener <rguenther@suse.de>
10598 * tree-ssa-structalias.c (get_function_part_constraint):
10599 Handle NULL fi->decl.
10600 (find_func_aliases_for_call): Properly handle indirect
10601 fi from direct call.
10602 (find_func_clobbers): Likewise.
10603 (ipa_pta_execute): Likewise.
10604 (create_variable_info_for): For functions that are ifunc_resolver
10605 resolve to a varinfo that contains the result of the resolver call.
10606 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
10609 2018-05-31 Michael Collison <michael.collison@arm.com>
10611 * config/aarch64/aarch64.md:
10612 (*fix_to_zero_extenddfdi2): New pattern.
10613 * gcc.target/aarch64/fix_extend1.c: New testcase.
10615 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
10617 PR middle-end/78809
10618 PR middle-end/83026
10619 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
10620 and BUILT_IN_STRNCMP_EQ.
10621 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
10622 BUILT_IN_STRNCMP_EQ.
10623 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
10624 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
10625 (gimple_fold_builtin): Likewise.
10626 * tree-ssa-strlen.c (compute_string_length): New function.
10627 (determine_min_obsize): New function.
10628 (handle_builtin_string_cmp): New function to handle calls to
10629 string compare functions.
10630 (strlen_optimize_stmt): Add handling to builtin string compare
10632 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
10633 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
10634 * tree.c (build_common_builtin_nodes): Add new defines of
10635 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
10637 2018-05-31 Jakub Jelinek <jakub@redhat.com>
10640 * bb-reorder.c (pass_partition_blocks::gate): Return false for
10641 functions with naked attribute.
10643 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
10645 * config/i386/sse.md (avx_vec_concat<mode>):
10646 Substitute concat_tg_mode mode attribute with xtg_mode.
10647 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
10648 (concat_tg_mode): Remove mode attribute.
10650 2018-05-31 Martin Sebor <msebor@redhat.com>
10653 * calls.c (alloc_max_size): Correct a logic error/typo.
10654 Treat excessive arguments as infinite. Warn for invalid arguments.
10655 * doc/invoke.texi (-Walloc-size-larger-than): Update.
10657 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
10660 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
10661 and movx for Haswell.
10663 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
10664 Cesar Philippidis <cesar@codesourcery.com>
10666 PR middle-end/85879
10667 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
10668 when emitting error on private/firstprivate reductions.
10669 * omp-low.c (lower_omp_target): Avoid reference-type processing
10670 on pointers for firstprivate clause.
10672 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
10674 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
10677 * config/aarch64/aarch64-simd.md
10678 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
10679 (aarch64_ld1_x3_<mode>): Likewise
10680 (aarch64_st1x2<VALLDIF:mode>): Likewise
10681 (aarch64_st1_x2_<mode>): Likewise
10682 (aarch64_st1x3<VALLDIF:mode>): Likewise
10683 (aarch64_st1_x3_<mode>): Likewise
10684 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
10685 (vld1_s8_x3): Likewise.
10686 (vld1_u16_x3): Likewise.
10687 (vld1_s16_x3): Likewise.
10688 (vld1_u32_x3): Likewise.
10689 (vld1_s32_x3): Likewise.
10690 (vld1_u64_x3): Likewise.
10691 (vld1_s64_x3): Likewise.
10692 (vld1_f16_x3): Likewise.
10693 (vld1_f32_x3): Likewise.
10694 (vld1_f64_x3): Likewise.
10695 (vld1_p8_x3): Likewise.
10696 (vld1_p16_x3): Likewise.
10697 (vld1_p64_x3): Likewise.
10698 (vld1q_u8_x3): Likewise.
10699 (vld1q_s8_x3): Likewise.
10700 (vld1q_u16_x3): Likewise.
10701 (vld1q_s16_x3): Likewise.
10702 (vld1q_u32_x3): Likewise.
10703 (vld1q_s32_x3): Likewise.
10704 (vld1q_u64_x3): Likewise.
10705 (vld1q_s64_x3): Likewise.
10706 (vld1q_f16_x3): Likewise.
10707 (vld1q_f32_x3): Likewise.
10708 (vld1q_f64_x3): Likewise.
10709 (vld1q_p8_x3): Likewise.
10710 (vld1q_p16_x3): Likewise.
10711 (vld1q_p64_x3): Likewise.
10712 (vst1_s64_x2): Likewise.
10713 (vst1_u64_x2): Likewise.
10714 (vst1_f64_x2): Likewise.
10715 (vst1_s8_x2): Likewise.
10716 (vst1_p8_x2): Likewise.
10717 (vst1_s16_x2): Likewise.
10718 (vst1_p16_x2): Likewise.
10719 (vst1_s32_x2): Likewise.
10720 (vst1_u8_x2): Likewise.
10721 (vst1_u16_x2): Likewise.
10722 (vst1_u32_x2): Likewise.
10723 (vst1_f16_x2): Likewise.
10724 (vst1_f32_x2): Likewise.
10725 (vst1_p64_x2): Likewise.
10726 (vst1q_s8_x2): Likewise.
10727 (vst1q_p8_x2): Likewise.
10728 (vst1q_s16_x2): Likewise.
10729 (vst1q_p16_x2): Likewise.
10730 (vst1q_s32_x2): Likewise.
10731 (vst1q_s64_x2): Likewise.
10732 (vst1q_u8_x2): Likewise.
10733 (vst1q_u16_x2): Likewise.
10734 (vst1q_u32_x2): Likewise.
10735 (vst1q_u64_x2): Likewise.
10736 (vst1q_f16_x2): Likewise.
10737 (vst1q_f32_x2): Likewise.
10738 (vst1q_f64_x2): Likewise.
10739 (vst1q_p64_x2): Likewise.
10740 (vst1_s64_x3): Likewise.
10741 (vst1_u64_x3): Likewise.
10742 (vst1_f64_x3): Likewise.
10743 (vst1_s8_x3): Likewise.
10744 (vst1_p8_x3): Likewise.
10745 (vst1_s16_x3): Likewise.
10746 (vst1_p16_x3): Likewise.
10747 (vst1_s32_x3): Likewise.
10748 (vst1_u8_x3): Likewise.
10749 (vst1_u16_x3): Likewise.
10750 (vst1_u32_x3): Likewise.
10751 (vst1_f16_x3): Likewise.
10752 (vst1_f32_x3): Likewise.
10753 (vst1_p64_x3): Likewise.
10754 (vst1q_s8_x3): Likewise.
10755 (vst1q_p8_x3): Likewise.
10756 (vst1q_s16_x3): Likewise.
10757 (vst1q_p16_x3): Likewise.
10758 (vst1q_s32_x3): Likewise.
10759 (vst1q_s64_x3): Likewise.
10760 (vst1q_u8_x3): Likewise.
10761 (vst1q_u16_x3): Likewise.
10762 (vst1q_u32_x3): Likewise.
10763 (vst1q_u64_x3): Likewise.
10764 (vst1q_f16_x3): Likewise.
10765 (vst1q_f32_x3): Likewise.
10766 (vst1q_f64_x3): Likewise.
10767 (vst1q_p64_x3): Likewise.
10769 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10771 * config/msp430/msp430.c (msp430_output_labelref): Prepend
10772 user_label_prefix to name.
10774 * tree-core.h: Update comment about the format of NAME string
10775 passed to handler in attribute_spec.
10777 * config/msp430/msp430.md: Remove erroneous subreg expression from
10778 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
10779 zero_extend{q,h}isi2.
10781 2018-05-30 Borislav Petkov <bp@suse.de>
10783 * doc/extend.texi: Document some architecture specific
10784 constraints and sort entries.
10786 2018-05-30 Martin Sebor <msebor@redhat.com>
10788 PR middle-end/85369
10789 * builtins.c (expand_builtin_stpcpy_1): New function.
10790 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
10791 only if the former succeeds.
10793 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
10795 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
10798 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10800 * doc/invoke.texi (-flinker-output): Document
10802 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10804 * passes.c (ipa_write_summaries): Only modify statements if body
10806 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
10807 incrementally linking.
10808 (ipa_passes): Likewise.
10809 * lto-cgraph.c (lto_output_node): When incrementally linking do not
10810 pass down resolution info.
10811 * common.opt (flag_incremental_link): Update info.
10812 * gcc.c (plugin specs): Turn flinker-output=* to
10813 -plugin-opt=-linker-output-known
10814 * toplev.c (compile_file): Also cut compilation when doing incremental
10816 * flag-types. (enum lto_partition_model): Add
10817 LTO_LINKER_OUTPUT_NOLTOREL.
10818 (invoke.texi): Add -flinker-output docs.
10819 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
10820 link same way as WPA; do not stream in dead initializers.
10822 * dwarf2out.c (dwarf2out_die_ref_for_decl,
10823 darf2out_register_external_decl): Support incremental link.
10825 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10827 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
10829 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10831 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
10832 it down to simple_object_copy_lto_debug_sections.
10833 (run_gcc): Determine incremental LTO link time and configure
10834 lto1 into non-wpa mode, disable renaming of debug sections.
10836 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
10838 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
10839 descriptions of various incorrectly documented functions.
10841 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
10844 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
10845 address check not strict.
10847 2018-05-30 Richard Biener <rguenther@suse.de>
10849 PR tree-optimization/85964
10850 * tracer.c (better_p): Drop initialized count check, we only
10851 call the function with initialized counts now.
10852 (find_best_successor): Do find a best edge if one
10853 has uninitialized count.
10854 (find_best_predecessor): Likewise. Do BB frequency check only
10855 if count is initialized.
10857 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
10859 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
10860 (aarch64_ldrstr_offset_compare): New.
10861 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
10862 load/store orderings.
10863 (aarch64_gen_adjusted_ldpstp): Likewise.
10865 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
10867 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10868 Check for subset of GENERAL_REGS and FP_REGS.
10869 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
10872 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
10874 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
10875 and wi::to_poly_offset. Add the current offset and then check
10876 whether the sum fits, rather than using an unchecked addition of
10877 a checked term. Check for a shwi rather than a uhwi.
10878 * expr.c (get_bit_range): Use tree_to_poly_uint64.
10879 (store_constructor): Use poly_int_tree_p.
10880 (expand_expr_real_1): Likewise.
10881 * function.c (assign_temp): Likewise.
10882 * fold-const.c (const_binop): Use poly_int_tree_p and
10883 wi::to_poly_offset.
10884 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
10886 * ipa-icf-gimple.c (func_checker::compare_operand): Use
10887 to_poly_offset for MEM offsets.
10888 * ipa-icf.c (sem_variable::equals): Likewise.
10889 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
10890 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
10891 wi::to_poly_offset for BIT_FIELD_REF offsets.
10892 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
10893 wi::to_poly_offset.
10894 * var-tracking.c (emit_note_insn_var_location): Use
10895 tree_to_poly_uint64.
10897 2018-05-29 Jim Wilson <jimw@sifive.com>
10899 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
10901 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
10904 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
10905 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
10907 (sse4_1_round<mode>2): Use nonimmediate_operand
10908 for operand 1 predicate.
10910 2018-05-29 Martin Sebor <msebor@redhat.com>
10911 Richard Biener <rguenther@suse.de>
10914 * calls.c (get_size_range): Call determine_value_range instead
10915 of get_value_range..
10916 * tree-vrp.h (determine_value_range): Declared new function.
10917 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
10919 2018-05-29 Richard Biener <rguenther@suse.de>
10921 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
10922 sure to use non-pattern stmts for get_earlier_stmt arguments.
10923 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
10924 called on pattern stmts.
10925 (get_later_stmt): Likewise.
10927 2018-05-29 Martin Liska <mliska@suse.cz>
10929 PR gcov-profile/85759
10930 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
10933 2018-05-29 Jakub Jelinek <jakub@redhat.com>
10935 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
10937 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
10941 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
10942 VEC_PACK_FLOAT_EXPR): New tree codes.
10943 * tree-pretty-print.c (op_code_prio): Handle
10944 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
10945 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
10946 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
10947 * tree-inline.c (estimate_operator_cost): Likewise.
10948 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
10949 * fold-const.c (const_binop): Likewise.
10950 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
10951 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
10952 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
10953 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
10954 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
10955 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
10956 * expr.c (expand_expr_real_2): Likewise.
10957 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
10958 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
10959 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
10961 * optabs.c (expand_widen_pattern_expr): For
10962 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
10963 sign from result type rather than operand's type.
10964 (expand_binop_directly): For vec_packu_float_optab and
10965 vec_packs_float_optab allow result type to be different from operand's
10967 * optabs-tree.c (optab_for_tree_code): Handle
10968 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
10969 VEC_PACK_FLOAT_EXPR. Formatting fixes.
10970 * tree-vect-generic.c (expand_vector_operations_1): Handle
10971 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
10972 VEC_PACK_FLOAT_EXPR.
10973 * tree-vect-stmts.c (supportable_widening_operation): Handle
10975 (supportable_narrowing_operation): Handle FLOAT_EXPR.
10976 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
10977 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
10978 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
10979 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
10981 (vec_pack<floatprefix>_float_<mode>): New expander.
10982 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
10984 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
10985 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
10986 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
10987 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
10988 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
10990 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
10991 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
10992 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
10993 VEC_PACK_FLOAT_EXPR): Document.
10995 2018-05-29 Richard Biener <rguenther@suse.de>
10997 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
10999 (stmt_vec_info_vec): Make pointer.
11000 (init_stmt_vec_info_vec): Remove.
11001 (free_stmt_vec_info_vec): Likewise.
11002 (set_stmt_vec_info_vec): New function.
11003 (free_stmt_vec_infos): Likewise.
11004 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
11005 (set_vinfo_for_stmt): Likewise.
11006 (get_earlier_stmt): Likewise.
11007 (get_later_stmt): Likewise.
11008 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
11009 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
11010 (vec_info::~vec_info): Free stmt_vec_infos.
11011 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
11012 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
11013 (pass_slp_vectorize::execute): Likewise.
11014 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
11015 (free_stmt_vec_info_vec): Likewise.
11016 (set_stmt_vec_info_vec): New function.
11017 (free_stmt_vec_infos): Likewise.
11018 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
11019 the global stmt_vec_info_vec.
11020 * tree-parloops.c (gather_scalar_reductions): Use
11021 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
11024 2018-05-29 Richard Biener <rguenther@suse.de>
11026 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
11028 2018-05-29 Martin Liska <mliska@suse.cz>
11029 David Malcolm <dmalcolm@redhat.com>
11031 * vec.c (test_reverse): New.
11032 (vec_c_tests): Add new test.
11033 * vec.h (vl_ptr>::reverse): New function.
11035 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
11037 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
11039 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
11042 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
11044 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
11046 2018-05-28 Richard Biener <rguenther@suse.de>
11048 PR tree-optimization/85933
11049 * tree-vect-data-refs.c (vect_record_base_alignments): Only
11050 look at stmts marked as vectorizable.
11052 2018-05-28 Richard Biener <rguenther@suse.de>
11054 PR tree-optimization/85934
11055 * tree-vect-generic.c (expand_vector_operations_1): Hoist
11056 vector boolean check before scalar optimization.
11058 2018-05-28 Jakub Jelinek <jakub@redhat.com>
11060 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
11063 2018-05-28 Mark Wielaard <mark@klomp.org>
11065 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
11066 if it is an expression containing a minus sign.
11068 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
11070 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
11072 2018-05-27 Paul Koning <ni1d@arrl.net>
11074 * config/pdp11/pdp11.md (truncsihi2): Remove.
11076 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
11077 Chung-Ju Wu <jasonwucj@gmail.com>
11079 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
11081 (unaligned_store_dw): Ditto.
11082 * config/nds32/nds32-memory-manipulation.c
11083 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
11084 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
11085 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
11086 (emit_setmem_word_loop): Rename to ...
11087 (emit_setmem_doubleword_loop): ... this.
11088 (nds32_gen_dup_4_byte_to_word_value): New function.
11089 (nds32_gen_dup_8_byte_to_double_word_value): New function.
11090 (nds32_expand_setmem_loop): Refine implementation.
11091 (nds32_expand_setmem_loop_v3m): Ditto.
11092 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
11095 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
11097 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
11099 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
11101 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
11102 (nds32_init_machine_status): Initialize machine->attr_naked_p and
11103 machine->attr_no_prologue_p.
11104 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
11105 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
11106 (nds32_expand_epilogue): Consider attr_naked_p.
11107 (nds32_expand_epilogue_v3pop): Likewise.
11108 (nds32_can_use_return_insn): Likewise.
11109 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
11110 attr_no_prologue_p fields.
11111 * config/nds32/nds32.opt (mret-in-naked-func): New option.
11113 2018-05-27 Jakub Jelinek <jakub@redhat.com>
11116 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
11118 * config/i386/sse.md
11119 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
11121 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
11123 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
11125 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
11127 (*<floatsuffix>floatv2div2sf2): Rename to ...
11128 (*float<floatunssuffix>v2div2sf2): ... this.
11129 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
11130 (float<floatunssuffix>v2div2sf2_mask): ... this.
11131 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
11132 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
11133 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
11135 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
11137 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
11139 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
11141 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
11143 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
11145 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
11146 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
11147 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
11148 gen_ufix_truncv8dfv8si2.
11149 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
11150 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
11151 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
11152 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
11153 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
11154 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
11155 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
11156 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
11158 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
11162 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
11164 2018-05-25 Jim Wilson <jimw@sifive.com>
11166 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
11167 * config/riscv/riscv.c (struct machine_function): Add
11168 interrupt_handler_p and attribute_checked_p fields.
11169 (riscv_attribute_table): Add interrupt.
11170 (riscv_interrupt_type_p): New.
11171 (riscv_save_reg_p): Save extra regs for interrupt handler.
11172 (riscv_use_save_libcall): Return false for interrupt handler.
11173 (riscv_first_stack_step): Add forward declaration.
11174 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
11175 for interrupt handler with large frame. Use it for saved reg list.
11176 (riscv_expand_prologue): Move flag_stack_usage_info support to
11177 eliminate duplication.
11178 (riscv_expand_epilogue): Generate mret for interrupt handler.
11179 (riscv_epilogue_uses): New.
11180 (riscv_can_use_return_insn): Return false for interrupt handler.
11181 (riscv_function_ok_for_sibcall): Likewise.
11182 (riscv_set_current_function): Add interrupt handler support.
11183 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
11184 * config/riscv/riscv.md (UNSPECV_MRET): New.
11186 (riscv_frflags, riscv_fsflags): Use tab after opcode.
11188 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
11190 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
11192 PR tree-optimization/85712
11193 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
11194 this candidate has already been replaced in-situ by a copy.
11196 2018-05-25 Jason Merrill <jason@redhat.com>
11198 PR c++/80485 - inline function non-zero address.
11199 * symtab.c (nonzero_address): Check DECL_COMDAT.
11201 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
11204 * config/alpha/alpha.md (ashlsi3): New insn pattern.
11205 (*ashlsi_se): Rename from *ashldi_se. Define as sign
11206 extension of SImode operation. Use const123_operand predicate.
11207 (*saddsi_1): Remove.
11208 (*saddl_se_1): Ditto.
11209 (*ssubsi_1): Ditto.
11210 (*ssubl_se_1): Ditto.
11211 * config/alpha/predicates.md (const123_operand): New predicate.
11212 * config/alpha/constraints.md (P): Use IN_RANGE.
11214 2018-05-25 Richard Biener <rguenther@suse.de>
11216 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
11218 (ref_maybe_used_by_stmt_p): Likewise.
11219 (stmt_may_clobber_ref_p): Likewise.
11220 (stmt_may_clobber_ref_p_1): Likewise.
11221 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
11223 (ref_maybe_used_by_stmt_p): Likewise.
11224 (stmt_may_clobber_ref_p): Likewise.
11225 (stmt_may_clobber_ref_p_1): Likewise.
11226 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
11227 the alias oracle to disambiguate DRs with stmts DR analysis
11229 (vect_analyze_data_refs): Do not give up on not analyzable
11230 DRs for BB vectorization. Remove code truncating the dataref
11233 2018-05-25 Jakub Jelinek <jakub@redhat.com>
11236 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
11237 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
11238 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
11240 2018-05-25 Richard Biener <rguenther@suse.de>
11242 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
11243 function, combining stmt data ref gathering and fatal analysis
11245 (vect_analyze_data_refs): Remove now redudnant code and simplify.
11246 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
11247 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
11248 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
11249 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
11251 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11253 PR tree-optimization/85720
11254 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
11255 SCC if all partitions are builtins.
11256 (version_loop_by_alias_check): New parameter. Generate cancelable
11257 runtime alias check if all partitions are builtins.
11258 (distribute_loop): Update call to above function.
11260 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11262 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
11263 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
11264 (parm_default_def_partition_arg): Ditto.
11265 (set_parm_default_def_partition): Ditto.
11266 (get_parm_default_def_partitions): Ditto and make it static.
11267 (get_undefined_value_partitions): Ditto and make it static.
11268 (remove_ssa_form): Refactor call to init_var_map here.
11269 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
11270 computation for loop region.
11271 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
11272 (register_default_def): Delete.
11273 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
11274 (parm_default_def_partition_arg): Ditto.
11275 (set_parm_default_def_partition): Ditto.
11276 (get_parm_default_def_partitions): Ditto and make it static.
11277 (get_undefined_value_partitions): Ditto and make it static.
11278 (coalesce_with_default, coalesce_with_default): Update comment.
11279 (create_coalesce_list_for_region): New func factored out from
11280 create_outofssa_var_map.
11281 (populate_coalesce_list_for_outofssa): New func factored out from
11282 create_outofssa_var_map and coalesce_ssa_name.
11283 (create_outofssa_var_map): Delete.
11284 (coalesce_ssa_name): Refactor to support live range computation.
11285 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
11286 (get_parm_default_def_partitions): Delete.
11287 (get_undefined_value_partitions): Ditto.
11288 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
11289 computation for loop region.
11290 (new_tree_live_info, loe_visit_block): Ditto.
11291 (live_worklist, set_var_live_on_entry): Ditto.
11292 (calculate_live_on_exit, verify_live_on_entry): Ditto.
11293 * tree-ssa-live.h (struct _var_map): New fields.
11294 (init_var_map): Change decl.
11295 (region_contains_p): New.
11297 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11299 * tree-ssa-live.h (live_merge_and_clear): Delete.
11301 2018-05-25 Richard Biener <rguenther@suse.de>
11304 * tree-dump.c (dequeue_and_dump): Remove access to removed
11305 operand 2 of a SWITCH_EXPR.
11307 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11309 * doc/sourcebuild.texi (vect_double_cond_arith): Include
11310 multiplication and division.
11311 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
11312 (cond_udiv@var{m}, cond_umod@var{m}): Document.
11313 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
11314 (cond_udiv_optab, cond_umod_optab): New optabs.
11315 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
11316 (IFN_COND_RDIV): New internal functions.
11317 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
11318 TRUNC_MOD_EXPR and RDIV_EXPR.
11319 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
11320 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
11322 (SVE_INT_BINARY): Include mult.
11323 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
11324 (optab, sve_int_op): Handle mult.
11325 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
11327 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
11328 for SVE_INT_BINARY_SD.
11330 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11332 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
11333 (optab, sve_int_op): Handle div and udiv.
11334 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
11335 for SVE_INT_BINARY_SD.
11336 (*<optab><mode>3): New insn for the same.
11338 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11340 * tree-vect-patterns.c: Include predict.h.
11341 (vect_recog_divmod_pattern): Restrict check for division support
11342 to when optimizing for size.
11344 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11346 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
11347 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
11348 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
11349 (gimple_match_op::set_op): Likewise.
11350 (gimple_resimplify4): Declare.
11351 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
11352 (expr::gen_transform): Likewise.
11353 (decision_tree::gen): Generate a simplification routine for 4 operands.
11354 * gimple-match-head.c (gimple_simplify): Add an overload for
11355 4 operands. In the top-level function, handle up to 4 call
11356 arguments and call gimple_resimplify4.
11357 (gimple_resimplify4): New function.
11358 (build_call_internal): Pass a fourth operand.
11359 (maybe_push_to_seq): Likewise.
11360 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
11361 Fold VEC_COND_EXPRs of an operation and a default value into
11362 an IFN_COND_* function if possible.
11363 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
11365 (SVE_COND_FP_BINARY): Include them.
11366 (optab, sve_fp_op): Handle them.
11367 (SVE_INT_BINARY_REV): New code iterator.
11368 (SVE_COND_FP_BINARY_REV): New int iterator.
11369 (commutative): New int attribute.
11370 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
11372 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
11374 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
11375 (*cond_<optab><mode>): New patterns for reversed operands.
11377 2018-05-25 Richard Biener <rguenther@suse.de>
11379 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
11380 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
11381 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
11382 (STMT_VINFO_GROUPED_ACCESS): Adjust.
11383 * tree-vect-data-refs.c (everywhere): Adjust users.
11384 * tree-vect-loop.c (everywhere): Likewise.
11385 * tree-vect-slp.c (everywhere): Likewise.
11386 * tree-vect-stmts.c (everywhere): Likewise.
11387 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
11389 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11391 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
11393 (gcc_cv_as_section_exclude): ... this.
11394 Try Solaris as #exclude syntax.
11395 * configure: Regenerate.
11396 * config.in: Regenerate.
11397 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
11399 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
11400 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
11402 * varasm.c (default_elf_asm_named_section): Don't check if
11403 HAVE_GAS_SECTION_EXCLUDE is defined.
11405 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11407 * doc/md.texi: Update the documentation of the cond_* optabs
11408 to mention the new final operand. Fix GET_MODE_NUNITS call.
11409 Describe the scalar case too.
11410 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
11411 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
11413 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
11414 (get_conditional_internal_fn): Update comment.
11415 * tree-vect-loop.c (vectorizable_reduction): Pass the original
11416 accumulator value as a final argument to conditional functions.
11417 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
11418 a define_expand and add an "else" operand. Assert for now that
11419 the else operand is equal to operand 2. Use SVE_INT_BINARY and
11420 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
11421 (*cond_<optab><mode>): New patterns.
11422 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
11423 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
11424 (UNSPEC_COND_EOR): Delete.
11425 (optab): Remove associated mappings.
11426 (SVE_INT_BINARY): New code iterator.
11427 (sve_int_op): Remove int attribute and add "minus" to the code
11429 (SVE_COND_INT_OP): Delete.
11430 (SVE_COND_FP_OP): Rename to...
11431 (SVE_COND_FP_BINARY): ...this.
11433 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11435 * optabs.c (can_reuse_operands_p): New function.
11436 (maybe_legitimize_operands): Try to reuse the results for
11439 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
11441 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
11442 Add {q} suffix to insn mnemonic.
11444 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
11446 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
11447 (msp430_warn_func_return): New.
11449 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
11451 * fold-const.c (tree_nonzero_bits): New function.
11452 * fold-const.h (tree_nonzero_bits): Likewise.
11453 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
11454 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
11456 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
11460 * varasm.c (assemble_alias): Check ifunc_resolver only on
11463 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
11466 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
11467 when memory input operand is handled.
11469 2018-05-24 Luis Machado <luis.machado@linaro.org>
11471 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
11473 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
11475 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
11477 * match.pd: Delay FMA folds until after vectorization.
11479 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
11482 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
11483 address check not strict.
11485 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
11487 * gimple-match.h (gimple_match_op): New class.
11488 (mprts_hook): Replace parameters with a gimple_match_op *.
11489 (maybe_build_generic_op): Likewise.
11490 (gimple_simplified_result_is_gimple_val): Replace parameters with
11491 a const gimple_match_op *.
11492 (gimple_simplify): Replace code_helper * and tree * parameters with
11493 a gimple_match_op * parameter.
11494 (gimple_resimplify1): Replace code_helper *, tree and tree *
11495 parameters with a gimple_match_op * parameter.
11496 (gimple_resimplify2): Likewise.
11497 (gimple_resimplify3): Likewise.
11498 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
11499 parameters with a gimple_match_op * parameter.
11500 * gimple-match-head.c (gimple_simplify): Change prototypes of
11501 auto-generated functions to take a gimple_match_op * instead of
11502 separate code_helper * and tree * parameters. Make the same
11503 change in the top-level overload and update calls to the
11504 gimple_resimplify routines. Update calls to the auto-generated
11505 functions and to maybe_push_res_to_seq in the publicly-facing
11506 operation-specific gimple_simplify overloads.
11507 (gimple_match_op::MAX_NUM_OPS): Define.
11508 (gimple_resimplify1): Replace rcode and ops with a single res_op
11509 parameter. Update call to gimple_simplify.
11510 (gimple_resimplify2): Likewise.
11511 (gimple_resimplify3): Likewise.
11512 (mprts_hook): Replace parameters with a gimple_match_op *.
11513 (maybe_build_generic_op): Likewise.
11514 (build_call_internal): Replace type, nargs and ops with
11515 a gimple_match_op *.
11516 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
11517 with a single gimple_match_op *. Update calls to mprts_hook,
11518 build_call_internal and gimple_simplified_result_is_gimple_val.
11519 Factor out code that is common to the tree_code and combined_fn cases.
11520 * genmatch.c (expr::gen_transform): Replace tem_code and
11521 tem_ops with a gimple_match_op called tem_op. Update calls
11522 to the gimple_resimplify functions and maybe_push_res_to_seq.
11523 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
11524 res_ops. Update call to the gimple_resimplify functions.
11525 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
11526 (decision_tree::gen): Make the functions take a gimple_match_op *
11527 called res_op instead of separate res_code and res_ops parameters.
11528 Update call accordingly.
11529 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
11530 and ops with a single res_op parameter. Update calls to
11531 maybe_build_generic_op and maybe_push_res_to_seq.
11532 (fold_stmt_1): Update calls to gimple_simplify and
11533 replace_stmt_with_simplification.
11534 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
11535 and gimple_simplified_result_is_gimple_val.
11536 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
11538 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
11539 with a gimple_match_op *.
11540 (vn_nary_build_or_lookup): Likewise. Update call to
11541 vn_nary_build_or_lookup_1.
11542 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
11543 gimple_match_op *. Update calls to the gimple_resimplify routines
11544 and to gimple_simplified_result_is_gimple_val.
11545 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
11546 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
11547 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
11548 (visit_nary_op): Likewise.
11549 (visit_reference_op_load): Likewise.
11551 2018-05-23 Luis Machado <luis.machado@linaro.org>
11553 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
11554 modifier for printing the step amount.
11556 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11559 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
11562 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
11564 * doc/sourcebuild.texi (Endianness): New subsubsection.
11566 2018-05-23 Luis Machado <luis.machado@linaro.org>
11568 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
11569 <prefetch_dynamic_strides>: New const bool field.
11570 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
11571 prefetch_dynamic_strides.
11572 (exynosm1_prefetch_tune): Likewise.
11573 (thunderxt88_prefetch_tune): Likewise.
11574 (thunderx_prefetch_tune): Likewise.
11575 (thunderx2t99_prefetch_tune): Likewise.
11576 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
11578 (aarch64_override_options_internal): Update to set
11579 PARAM_PREFETCH_DYNAMIC_STRIDES.
11580 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
11581 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
11582 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
11583 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
11584 prefetch-dynamic-strides setting.
11586 2018-05-23 Luis Machado <luis.machado@linaro.org>
11588 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
11589 <minimum_stride>: New const int field.
11590 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
11591 minimum_stride field defaulting to -1.
11592 (exynosm1_prefetch_tune): Likewise.
11593 (thunderxt88_prefetch_tune): Likewise.
11594 (thunderx_prefetch_tune): Likewise.
11595 (thunderx2t99_prefetch_tune): Likewise.
11596 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
11597 <default_opt_level>: Set to 3.
11598 (aarch64_override_options_internal): Update to set
11599 PARAM_PREFETCH_MINIMUM_STRIDE.
11600 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
11601 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
11602 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
11603 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
11604 stride is constant and is below the minimum stride threshold.
11606 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11608 * config/arm/arm-cpus.in (mode26): Delete.
11609 (armv4): Delete mode26 reference.
11610 * config/arm/arm.c (arm_configure_build_target): Delete use of
11613 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
11615 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
11617 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
11618 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
11620 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
11621 (floatunsdidf2): Ditto.
11623 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
11625 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
11626 (fixuns_trunc<mode>si2_avx512f): Ditto.
11627 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
11628 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
11629 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
11631 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
11633 PR rtl-optimization/79985
11634 * df-scan.c (df_insn_refs_collect): Remove special case for
11635 global registers and asm statements.
11637 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
11639 * extend.texi (Global Register Variables): Rewrite the bullet list.
11640 Note that the register is available for allocation. Note that access
11641 via inline asm must use constraints. Add note about async-signal
11642 handlers. Remove paragraph about automagic register selection.
11644 2018-05-23 Richard Biener <rguenther@suse.de>
11646 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
11647 of fixed offset from memset VN.
11649 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
11651 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
11652 first_interp field.
11653 (alloc_cand_and_find_basis): Initialize first_interp field.
11654 (slsr_process_mul): Modify first_interp field.
11655 (slsr_process_add): Likewise.
11656 (slsr_process_cast): Modify first_interp field for each new
11658 (slsr_process_copy): Likewise.
11659 (dump_candidate): Dump first_interp field.
11660 (replace_mult_candidate): Process all interpretations, not just
11662 (replace_rhs_if_not_dup): Likewise.
11663 (replace_one_candidate): Likewise.
11665 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
11667 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
11669 (aarch64_needs_frame_chain): New function.
11670 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
11672 2018-05-23 Sudakshina Das <sudi.das@arm.com>
11675 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
11676 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
11677 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
11678 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
11679 as true for strict-align.
11680 (aarch64_can_inline_p): Perform checks even when callee has no
11681 attributes to check for strict alignment.
11682 * doc/extend.texi (AArch64 Function Attributes): Document
11684 * doc/invoke.texi: (AArch64 Options): Likewise.
11686 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
11688 PR tree-optimization/85853
11689 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
11690 the handling of the root of the node to...
11691 (vect_slp_analyze_node_operations_1): ...this new function,
11692 and run the whole thing with the child nodes' def types
11693 set according to their SLP node's def type.
11695 2018-05-23 Richard Biener <rguenther@suse.de>
11697 PR middle-end/85874
11698 * tree-data-ref.c (create_runtime_alias_checks): Defer
11699 and ignore overflow warnings.
11701 2018-05-23 Yury Gribov <tetra2005@gmail.com>
11703 PR tree-optimization/85822
11704 * tree-vrp.c (is_masked_range_test): Fix handling of negative
11707 2018-05-23 Richard Biener <rguenther@suse.de>
11709 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
11710 memset constants via native_interpret_expr.
11712 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
11715 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
11717 (cgraph_node::create_alias): Likewise.
11718 (cgraph_node::get_availability): Check ifunc_resolver instead
11719 of looking up ifunc attribute.
11720 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
11721 * varasm.c (do_assemble_alias): Likewise.
11722 (assemble_alias): Likewise.
11723 (default_binds_local_p_3): Likewise.
11724 * cgraph.h (cgraph_node): Add ifunc_resolver.
11725 (cgraph_node::only_called_directly_or_aliased_p): Return false
11726 for IFUNC resolver.
11727 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
11729 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
11730 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
11731 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
11732 instead of looking up ifunc attribute.
11734 2018-05-22 Luis Machado <luis.machado@linaro.org>
11736 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
11738 2018-05-22 Martin Sebor <msebor@redhat.com>
11740 PR middle-end/85359
11741 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
11742 only when expasion succeeds.
11743 (expand_builtin_strcmp): Same.
11744 (expand_builtin_strncmp): Same.
11746 2018-05-22 Martin Sebor <msebor@redhat.com>
11748 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
11750 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
11751 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11753 * config/aarch64/aarch64-ldpstp.md: Replace uses of
11754 aarch64_mem_pair_operand with memory_operand and delete operand swapping
11756 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
11757 Add check for legitimate_address.
11758 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
11759 (aarch64_swap_ldrstr_operands): New.
11760 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
11763 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
11764 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11766 * config/aarch64/aarch64.md: New patterns to generate stp
11768 (store_pair_sw, store_pair_dw): New patterns to generate stp for
11769 single words and double words.
11770 (load_pair_sw, load_pair_dw): Likewise.
11771 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
11773 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
11775 * config/aarch64/aarch64-ldpstp.md: Modify peephole
11776 for different mode ldpstp and add peephole for merged zero stores.
11777 Likewise for loads.
11778 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
11780 (aarch64_gen_store_pair): Rename calls to match new patterns.
11781 (aarch64_gen_load_pair): Rename calls to match new patterns.
11782 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
11783 (load_pair<DREG:mode><DREG2:mode>): ... This.
11784 (store_pair<mode>): Rename to...
11785 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
11786 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
11787 New mode iterators.
11788 (V_INT_EQUIV): Handle SImode.
11789 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
11792 2018-05-22 Martin Sebor <msebor@redhat.com>
11795 * calls.c (maybe_warn_nonstring_arg): Use string length to set
11796 or ajust the presumed bound on an operation to avoid unnecessary
11799 2018-05-22 Martin Sebor <msebor@redhat.com>
11801 PR tree-optimization/85826
11802 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
11803 assuming that a DECL necesarily has a constant size.
11805 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
11807 PR middle-end/85862
11808 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
11810 2018-05-22 Richard Biener <rguenther@suse.de>
11812 PR tree-optimization/85834
11813 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
11814 non-constant and non-zero memset arguments.
11816 2018-05-22 Martin Liska <mliska@suse.cz>
11819 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
11821 2018-05-22 Richard Biener <rguenther@suse.de>
11823 PR tree-optimization/85863
11824 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
11825 comparisons when vectype is specified.
11826 (vectorizable_condition): Do not specify vectype for
11827 vect_is_simple_cond when SLP vectorizing.
11829 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
11832 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
11833 define __ibm128 as long double.
11834 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
11835 as a distinct type when IEEE 128-bit support is enabled.
11836 (init_float128_ieee): Fix up conversions between IFmode and IEEE
11837 128-bit types to use the correct functions.
11838 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
11839 convert between 128-bit floating point types that have different
11840 modes but the same representation, instead of using gen_lowpart to
11842 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
11844 (IFKF_reg): New attributes to give the register constraints for
11846 (extend<mode>tf2_internal): New insns to mark an explicit
11847 conversion between 128-bit floating point types that have a
11848 different mode but share the same representation.
11850 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
11852 PR tree-optimization/85814
11853 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
11854 a null return from get_strinfo when unsharing the next
11855 strinfo in the chain.
11857 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
11860 * varasm.c (weak_finish): Clean up weak_decls.
11862 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11864 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
11865 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
11866 UNSPEC_UADALP values.
11867 * config/aarch64/iterators.md (ABAL): New int iterator.
11870 (sur): Add mappings for the above.
11871 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
11873 (aarch64_<sur>abal<mode>_4): Likewise.
11874 (aarch64_<sur>adalp<mode>_3): Likewise.
11875 (<sur>sadv16qi): New define_expand.
11877 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
11879 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
11880 (*movdf_internal): Ditto.
11881 (*rcpsf2_sse): Ditto.
11882 (*rsqrtsf2_sse): Ditto.
11883 (*sqrt<mode>2_sse): Ditto.
11885 2018-05-21 Tamar Christina <tamar.christina@arm.com>
11887 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
11889 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
11890 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
11891 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
11892 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
11894 * config/aarch64/arm_neon.h: Likewise.
11895 * config/aarch64/iterators.md (VQ_I): New.
11897 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
11899 * config.gcc: Add arc/t-multilib-linux to tmake_file for
11901 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
11904 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
11906 * config/nds32/constraints.md (S): New constraint.
11907 * config/nds32/nds32.md (call_internal): Use constraint S.
11908 (call_value_internal): Likewise.
11909 (sibcall_internal): Likewise.
11910 (sibcall_value_internal): Likewise.
11912 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
11913 Chung-Ju Wu <jasonwucj@gmail.com>
11915 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
11916 into consideration.
11918 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
11919 Chung-Ju Wu <jasonwucj@gmail.com>
11921 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
11922 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
11923 (nds32_rtx_costs_impl): Simplify.
11924 (nds32_address_cost_impl): Simplify.
11925 (nds32_init_rtx_costs): New function.
11926 (nds32_rtx_costs_speed_prefer): Likewise.
11927 (nds32_rtx_costs_size_prefer): Likewise.
11928 (nds32_address_cost_speed_prefer): Likewise.
11929 (nds32_address_cost_speed_fwprop): Likewise.
11930 (nds32_address_cost_size_prefer): Likewise.
11931 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
11932 * config/nds32/nds32.c (nds32_option_override): Use
11933 nds32_init_rtx_costs function.
11935 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
11937 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
11938 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
11939 (TARGET_PIPELINE_N8): Likewise.
11940 (TARGET_PIPELINE_N10): Likewise.
11941 (TARGET_PIPELINE_N13): Likewise.
11942 (TARGET_PIPELINE_GRAYWOLF): Likewise.
11944 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
11946 * config/nds32/nds32-fpu.md: Update copyright year.
11948 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
11950 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
11952 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
11954 * config/nds32/nds32.c
11955 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
11956 * config/nds32/nds32.opt (minline-asm-r15): New option.
11958 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
11960 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
11962 * config/nds32/nds32.md (abssi2): New pattern.
11964 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
11966 * config/i386/i386.md (rex64namesuffix): New mode attribute.
11967 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
11968 Merge insn pattern from sse_cvtsi2ss<round_name> and
11969 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
11970 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
11971 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
11972 using SWI48 mode iterator.
11973 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
11974 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
11975 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
11976 pattern from sse_cvttss2si<round_saeonly_name>
11977 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
11978 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
11979 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
11980 using SWI48 mode iterator.
11981 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
11982 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
11983 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
11984 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
11985 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
11986 using SWI48 mode iterator.
11987 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
11988 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
11989 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
11990 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
11991 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
11992 SWI48 mode iterator.
11993 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
11994 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
11995 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
11996 pattern from sse_cvttsd2si<round_saeonly_name>
11997 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
11999 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12001 * config/nds32/nds32-md-auxiliary.c
12002 (nds32_valid_smw_lwm_base_p): Refine.
12003 (nds32_output_smw_single_word): Refine.
12004 (nds32_output_smw_double_word): New.
12005 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
12007 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12009 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
12010 (nds32_output_stack_pop): Refine.
12011 (nds32_expand_unaligned_load): Refine.
12012 (nds32_expand_unaligned_store): Refine.
12014 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
12015 Chung-Ju Wu <jasonwucj@gmail.com>
12017 * config/nds32/constants.md: Add TP_REGNUM constant.
12018 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
12019 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
12021 * config/nds32/nds32-doubleword.md: Consider flag_pic.
12022 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
12023 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
12024 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
12025 and PIC code generation.
12026 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
12028 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
12030 * config/nds32/nds32.md: Support TLS and PIC.
12031 * config/nds32/nds32.c: Support TLS and PIC.
12032 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
12033 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
12036 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12038 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
12039 mode with E_ prefix.
12041 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
12042 Chung-Ju Wu <jasonwucj@gmail.com>
12044 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
12045 * config/nds32/nds32-md-auxiliary.c
12046 (symbolic_reference_mentioned_p): New.
12047 (nds32_legitimize_ict_address): New.
12048 (nds32_expand_ict_move): New.
12049 (nds32_indirect_call_referenced_p): New.
12050 (nds32_symbol_binds_local_p): Delete.
12051 (nds32_long_call_p): Modify.
12052 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
12053 * config/nds32/nds32-protos.h
12054 (symbolic_reference_mentioned_p): Declare.
12055 (nds32_legitimize_ict_address): Declare.
12056 (nds32_expand_ict_move): Declare.
12057 (nds32_indirect_call_referenced_p): Declare.
12058 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
12059 (nds32_relax_group): Use nds32_ict_const_p as condition.
12060 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
12061 (nds32_asm_file_start): Output ict_model directive in asm code.
12062 (nds32_legitimate_address_p): Consider indirect call.
12063 (nds32_print_operand): Consider indirect call.
12064 (nds32_print_operand_address): Consider indirect call.
12065 (nds32_insert_attributes): Handle "indirect_call" attribute.
12066 (TARGET_LEGITIMATE_ADDRESS_P): Define.
12067 (TARGET_LEGITIMATE_CONSTANT_P): Define.
12068 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
12069 (TARGET_DELEGITIMIZE_ADDRESS): Define.
12070 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
12071 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
12072 (TARGET_ICT_MODEL_SMALL): Define.
12073 (TARGET_ICT_MODEL_LARGE): Define.
12074 * config/nds32/nds32.md (movsi): Consider ict model.
12075 (call, call_value): Consider ict model.
12076 (sibcall, sibcall_value): Consider ict model.
12077 * config/nds32/nds32.opt (mict-model): New option.
12078 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
12081 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
12082 Monk Chiang <sh.chiang04@gmail.com>
12083 Jim Wilson <jimw@sifive.com>
12085 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
12086 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
12087 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
12088 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
12089 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
12090 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
12091 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
12092 compute save_libcall_adjustment properly.
12093 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
12094 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
12095 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
12096 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
12097 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
12098 (ABI_SPEC): Handle mabi=ilp32e.
12099 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
12100 (RVE): Add RVE mask.
12101 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
12102 <-march>: Add rv32e as an example.
12104 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
12107 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
12108 (intra_create_variable_infos): Handle C++ constructors.
12110 2018-05-18 Martin Liska <mliska@suse.cz>
12112 * passes.def: Remove a redundant pass.
12114 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
12117 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
12119 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12121 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
12123 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
12125 (armv2, armv2a, armv3, armv3m): Delete architectures.
12126 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
12127 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
12128 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
12130 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
12131 (*mulsidi3adddi): Likewise.
12132 (mulsidi3): Likewise.
12133 (*mulsidi3_nov6): Likewise.
12134 (umulsidi3): Likewise.
12135 (umulsidi3_nov6): Likewise.
12136 (umaddsidi4): Likewise.
12137 (*umulsidi3adddi): Likewise.
12138 (smulsi3_highpart): Likewise.
12139 (*smulsi3_highpart_nov6): Likewise.
12140 (umulsi3_highpart): Likewise.
12141 (*umulsi3_highpart_nov6): Likewise.
12142 * config/arm/arm.h (arm_arch3m): Delete.
12143 * config/arm/arm.c (arm_arch3m): Delete.
12144 (arm_option_override_internal): Update armv3-related comment.
12145 (arm_configure_build_target): Delete use of isa_bit_mode32.
12146 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
12147 (arm_rtx_costs_internal): Delete check of arm_arch3m.
12148 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
12149 (mulsa3): Likewise.
12150 (mulusa3): Likewise.
12151 * config/arm/arm-protos.h (arm_arch3m): Delete.
12152 * config/arm/arm-tables.opt: Regenerate.
12153 * config/arm/arm-tune.md: Likewise.
12154 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
12155 deleted architectures.
12157 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12159 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
12160 (armv5t, armv5te): New features.
12161 (ARMv5, ARMv5e): Delete fgroups.
12162 (ARMv5t, ARMv5te): Adjust for above changes.
12163 (ARMv6m): Likewise.
12164 (armv5, armv5e): Delete arches.
12165 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
12167 (*call_reg_arm): Likewise.
12168 (*call_value_reg_armv5): Likewise.
12169 (*call_value_reg_arm): Likewise.
12170 (*call_symbol): Likewise.
12171 (*call_value_symbol): Likewise.
12172 (*sibcall_insn): Likewise.
12173 (*sibcall_value_insn): Likewise.
12174 (clzsi2): Likewise.
12175 (prefetch): Likewise.
12176 (define_split and define_peephole2 dependent on arm_arch5):
12178 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
12180 (TARGET_ARM_QBIT): Likewise.
12181 (TARGET_DSP_MULTIPLY): Likewise.
12182 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
12183 (arm_arch5, arm_arch5e): Delete.
12184 (arm_arch5t, arm_arch5te): Declare.
12185 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
12186 (arm_arch5t): Declare.
12187 (arm_option_reconfigure_globals): Update for the above.
12188 (arm_options_perform_arch_sanity_checks): Update comment, replace
12189 use of arm_arch5 with arm_arch5t.
12190 (use_return_insn): Likewise.
12191 (arm_emit_call_insn): Likewise.
12192 (output_return_instruction): Likewise.
12193 (arm_final_prescan_insn): Likewise.
12194 (arm_coproc_builtin_available): Likewise.
12195 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
12196 arm_arch5e with arm_arch5t and arm_arch5te.
12197 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
12198 (arm_arch5t, arm_arch5te): Declare.
12199 * config/arm/arm-tables.opt: Regenerate.
12200 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
12201 * config/arm/t-multilib: Likewise.
12202 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
12203 instead of arm_arch5.
12204 (*call_reg_thumb1): Likewise.
12205 (*call_value_reg_thumb1_v5): Likewise.
12206 (*call_value_reg_thumb1): Likewise.
12207 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
12209 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
12211 2018-05-18 Martin Liska <mliska@suse.cz>
12213 PR gcov-profile/84846
12214 * doc/gcov.texi: Document -t option of gcov tool.
12216 2018-05-18 Martin Liska <mliska@suse.cz>
12218 PR gcov-profile/84846
12219 * gcov.c (print_usage): Add new -t option.
12220 (process_args): Handle the option.
12221 (generate_results): Use stdout as output when requested by
12224 2018-05-18 Martin Liska <mliska@suse.cz>
12226 PR gcov-profile/84846
12227 * coverage.c (coverage_init): Write PWD to .gcno file.
12228 * doc/gcov.texi: Document how working directory is printed.
12229 * gcov-dump.c (dump_gcov_file): Print PWD.
12230 * gcov.c (output_intermediate_file): Likewise.
12231 (read_graph_file): Read PWD string.
12232 (output_lines): Print PWD.
12234 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12236 PR middle-end/85817
12237 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
12238 for retval and return false if all args to phi are zero.
12240 2018-05-18 Richard Biener <rguenther@suse.de>
12242 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
12244 (evrp_dom_walker::before_dom_children): Call it.
12246 2018-05-18 Richard Biener <rguenther@suse.de>
12248 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
12249 results when processing array refs with variable index.
12251 2018-05-18 Toon Moene <toon@moene.org>
12253 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
12254 directly after that of -floop-interchange. Indicate that both
12255 options are enabled by default when specifying -O3.
12257 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12259 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
12260 iterator. Delete separate integer-mode vec_set<mode> expander.
12261 (aarch64_simd_vec_setv2di): Delete.
12262 (vec_setv2di): Delete.
12263 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
12264 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
12265 the "w, r" alternative.
12267 2018-05-18 Martin Liska <mliska@suse.cz>
12269 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
12270 * tree-pass.h (make_pass_lower_switch_O0): New function.
12271 * tree-switch-conversion.c (node_has_low_bound): Remove.
12272 (node_has_high_bound): Likewise.
12273 (node_is_bounded): Likewise.
12274 (class pass_lower_switch): Make it a template type and create
12276 (pass_lower_switch::execute): Add template argument.
12277 (make_pass_lower_switch): New function.
12278 (make_pass_lower_switch_O0): New function.
12279 (do_jump_if_equal): Remove.
12280 (emit_case_nodes): Simplify to just handle all 3 cases and leave
12281 all the hard work to tree optimization passes.
12283 2018-05-18 Martin Liska <mliska@suse.cz>
12285 * dbgcnt.c (limit_low): Renamed from limit.
12286 (limit_high): New variable.
12287 (dbg_cnt_is_enabled): Check for upper limit.
12288 (dbg_cnt): Adjust dumping.
12289 (dbg_cnt_set_limit_by_index): Add new argument for high
12291 (dbg_cnt_set_limit_by_name): Likewise.
12292 (dbg_cnt_process_single_pair): Parse new format.
12293 (dbg_cnt_process_opt): Use strtok.
12294 (dbg_cnt_list_all_counters): Remove 'value' and add
12296 * doc/invoke.texi: Document changes.
12298 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
12300 * doc/sourcebuild.texi (scalar_all_fma): Document.
12301 * tree.def (FMA_EXPR): Delete.
12302 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
12303 * internal-fn.c (ternary_direct): New macro.
12304 (expand_ternary_optab_fn): Likewise.
12305 (direct_ternary_optab_supported_p): Likewise.
12306 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
12307 * builtins.c (fold_builtin_fma): Delete.
12308 (fold_builtin_3): Don't call it.
12309 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
12310 * expr.c (expand_expr_real_2): Likewise.
12311 * fold-const.c (operand_equal_p): Likewise.
12312 (fold_ternary_loc): Likewise.
12313 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
12314 * gimple.c (DEFTREECODE): Likewise.
12315 * gimplify.c (gimplify_expr): Likewise.
12316 * optabs-tree.c (optab_for_tree_code): Likewise.
12317 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12318 * tree-eh.c (operation_could_trap_p): Likewise.
12319 (stmt_could_throw_1_p): Likewise.
12320 * tree-inline.c (estimate_operator_cost): Likewise.
12321 * tree-pretty-print.c (dump_generic_node): Likewise.
12322 (op_code_prio): Likewise.
12323 * tree-ssa-loop-im.c (stmt_cost): Likewise.
12324 * tree-ssa-operands.c (get_expr_operands): Likewise.
12325 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
12326 * fold-const-call.h (fold_fma): Delete.
12327 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
12328 CFN_FNMA and CFN_FNMS.
12329 (fold_fma): Delete.
12330 * genmatch.c (combined_fn): New enum.
12331 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
12332 (commutative_op): New function.
12333 (commutate): Use it. Handle more than 2 operands.
12334 (dt_operand::gen_gimple_expr): Use commutative_op.
12335 (parser::parse_expr): Allow :c to be used with non-binary
12336 operators if the commutative operand is known.
12337 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
12338 CFN_FMS, CFN_FNMA and CFN_FNMS.
12339 (backprop::process_assign_use): Remove FMA_EXPR handling.
12340 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
12341 (gen_hsa_fma): New function.
12342 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
12343 IFN_FNMA and IFN_FNMS.
12344 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
12345 * gimple-fold.h (follow_all_ssa_edges): Declare.
12346 * gimple-fold.c (follow_all_ssa_edges): New function.
12347 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
12348 gimple_build interface and use follow_all_ssa_edges to fold the result.
12349 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
12350 instead of checking for optabs directly.
12351 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
12352 rather than FMA_EXPRs.
12353 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
12354 call to IFN_FMA instead of an FMA_EXPR.
12356 2018-05-17 Jim Wilson <jimw@sifive.com>
12358 * expr.c (do_tablejump): When converting index to Pmode, if we have a
12359 sign extended promoted subreg, and the range does not have the sign bit
12360 set, then do a sign extend.
12362 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
12363 test, check for sign extended subreg and/or constant operands, and
12364 do a sign extend in that case.
12366 2018-05-17 Steve Ellcey <sellcey@cavium.com>
12368 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
12369 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
12371 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
12372 Change logics_shift_reg to logics_shift_imm.
12373 (thunderx2t99_fp_loadpair_basic): Delete.
12374 (thunderx2t99_fp_storepair_basic): Delete.
12375 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
12376 (thunderx2t99_asimd_polynomial): Delete.
12377 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
12378 and neon_fp_mul_d_scalar_q.
12379 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
12380 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
12381 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
12382 (thunderx2t99_asimd_lut): Add missing tbl types.
12383 (thunderx2t99_asimd_ext): Delete.
12384 (thunderx2t99_asimd_load1_1_mult): Delete.
12385 (thunderx2t99_asimd_load1_2_mult): Delete.
12386 (thunderx2t99_asimd_load1_ldp): New.
12387 (thunderx2t99_asimd_load1): New.
12388 (thunderx2t99_asimd_load2): Add missing *load2* types.
12389 (thunderx2t99_asimd_load3): New.
12390 (thunderx2t99_asimd_load4): New.
12391 (thunderx2t99_asimd_store1_1_mult): Delete.
12392 (thunderx2t99_asimd_store1_2_mult): Delete.
12393 (thunderx2t99_asimd_store2_mult): Delete.
12394 (thunderx2t99_asimd_store2_onelane): Delete.
12395 (thunderx2t99_asimd_store_stp): New.
12396 (thunderx2t99_asimd_store1): New.
12397 (thunderx2t99_asimd_store2): New.
12398 (thunderx2t99_asimd_store3): New.
12399 (thunderx2t99_asimd_store4): New.
12401 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
12403 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
12404 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
12406 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
12407 Segher Boessenkool <segher@kernel.crashing.org>
12410 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
12413 2018-05-17 Richard Biener <rguenther@suse.de>
12415 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
12416 for pruning loop and prune defs feeding only already visited PHIs.
12418 2018-05-17 Richard Biener <rguenther@suse.de>
12420 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
12422 2018-05-17 Bin Cheng <bin.cheng@arm.com>
12423 Richard Biener <rguenther@suse.de>
12425 PR tree-optimization/85793
12426 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
12427 for VMAT_ELEMENTWISE.
12429 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
12431 * internal-fn.h (lookup_internal_fn): Declare
12432 * internal-fn.c (lookup_internal_fn): New function.
12433 * gimple.c (gimple_build_call_from_tree): Handle calls to
12434 internal functions.
12435 * gimple-pretty-print.c (dump_gimple_call): Print "." before
12436 internal function names.
12437 * tree-pretty-print.c (dump_generic_node): Likewise.
12438 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
12440 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
12442 * gimple-fold.h (gimple_build): Make the function forms take
12443 combined_fn rather than built_in_function.
12444 (gimple_simplify): Likewise.
12445 * gimple-match-head.c (gimple_simplify): Likewise.
12446 * gimple-fold.c (gimple_build): Likewise.
12447 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
12448 rather than gimple_build_call_internal.
12449 (get_initial_defs_for_reduction): Likewise.
12450 (vect_create_epilog_for_reduction): Likewise.
12451 (vectorizable_live_operation): Likewise.
12453 2018-05-17 Martin Liska <mliska@suse.cz>
12455 * gimple-ssa-sprintf.c (format_directive): Do not use
12456 space in between 'G_' and '('.
12458 2018-05-17 Jakub Jelinek <jakub@redhat.com>
12461 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
12462 even if the mask is not all ones.
12465 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
12467 (ix86_gimple_fold_builtin): Likewise.
12470 * config/i386/i386.c: Include tree-vector-builder.h.
12471 (ix86_vector_shift_count): New function.
12472 (ix86_fold_builtin): Fold shift builtins by scalar count.
12473 (ix86_gimple_fold_builtin): Likewise.
12475 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
12476 _mm512_setzero): New intrinsics.
12478 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
12479 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12481 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
12482 code generation for cases where splatting a value is not useful.
12483 * simplify-rtx.c (simplify_ternary_operation): Simplify
12484 vec_merge across a vec_duplicate and a paradoxical subreg forming
12485 a vector mode to a vec_concat.
12487 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
12489 * config.gcc: Support "goldmont-plus".
12490 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
12492 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12493 PROCESSOR_GOLDMONT_PLUS.
12494 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
12495 (processor_target_table): Add "goldmont-plus".
12496 (PTA_GOLDMONT_PLUS): Define.
12497 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
12498 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
12499 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
12500 (fold_builtin_cpu): Add "goldmont-plus".
12501 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
12502 (ix86_option_override_internal): Add "goldmont-plus".
12503 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
12504 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
12505 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
12506 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
12508 2018-05-17 Richard Biener <rguenther@suse.de>
12510 PR tree-optimization/85757
12511 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
12512 remove defs that only feed that PHI from further processing.
12514 2018-05-16 Jim Wilson <jimw@sifive.com>
12516 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
12518 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
12520 2018-05-16 Mark Wielaard <mark@klomp.org>
12522 * dwarf2out.c (count_index_strings): New function.
12523 (output_indirect_strings): Call count_index_strings and generate
12524 header for dwarf_version >= 5.
12526 2018-05-16 Mark Wielaard <mark@klomp.org>
12528 * dwarf2out.c (dwarf_FORM): New function.
12529 (set_indirect_string): Use dwarf_FORM.
12530 (reset_indirect_string): Likewise.
12531 (size_of_die): Likewise.
12532 (value_format): Likewise.
12533 (output_die): Likewise.
12534 (add_skeleton_AT_string): Likewise.
12535 (output_macinfo_op): Likewise.
12536 (index_string): Likewise.
12537 (output_index_string_offset): Likewise.
12538 (output_index_string): Likewise.
12539 (count_index_strings): Likewise.
12541 2018-05-16 Carl Love <cel@us.ibm.com>
12543 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
12544 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
12546 2018-05-16 Martin Jambor <mjambor@suse.cz>
12548 * ipa-prop.c (ipa_free_all_edge_args): Remove.
12549 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
12551 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
12553 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
12554 (fnma<mode>4): Likewise.
12555 (fms<mode>4): Likewise.
12556 (fnms<mode>4): Likewise.
12557 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
12558 (aarch64_fnma<mode>4): Likewise.
12559 (aarch64_fms<mode>4): Likewise.
12560 (aarch64_fnms<mode>4): Likewise.
12561 (aarch64_fnmadd<mode>4): Likewise.
12563 2018-05-16 Jason Merrill <jason@redhat.com>
12565 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
12567 2018-05-16 Richard Biener <rguenther@suse.de>
12569 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
12570 (dump_stmt_cost): Declare.
12571 (add_stmt_cost): Dump cost we add.
12572 (add_stmt_costs): New function.
12573 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
12574 No longer exported.
12575 (vect_analyze_stmt): Adjust prototype.
12576 (vectorizable_condition): Likewise.
12577 (vectorizable_live_operation): Likewise.
12578 (vectorizable_reduction): Likewise.
12579 (vectorizable_induction): Likewise.
12580 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
12581 cost vector to pass to vectorizable_ and record afterwards.
12582 (vect_model_reduction_cost): Take cost vector argument and adjust.
12583 (vect_model_induction_cost): Likewise.
12584 (vectorizable_reduction): Likewise.
12585 (vectorizable_induction): Likewise.
12586 (vectorizable_live_operation): Likewise.
12587 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
12588 SLP_TREE_NUMBER_OF_VEC_STMTS.
12589 (vect_analyze_slp_cost_1): Remove.
12590 (vect_analyze_slp_cost): Likewise.
12591 (vect_slp_analyze_node_operations): Take visited args and
12592 a target cost vector. Avoid processing already visited stmt sets.
12593 (vect_slp_analyze_operations): Use a local cost vector to gather
12594 costs and register those of non-discarded instances.
12595 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
12596 (vect_schedule_slp_instance): Remove copying of
12597 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
12599 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
12600 adding cost. Record cost entry location.
12601 (vect_prologue_cost_for_slp_op): Function to compute cost of
12602 a constant or invariant generated for SLP vect in the prologue,
12603 split out from vect_analyze_slp_cost_1.
12604 (vect_model_simple_cost): Make static. Adjust for SLP costing.
12605 (vect_model_promotion_demotion_cost): Likewise.
12606 (vect_model_store_cost): Likewise, make static.
12607 (vect_model_load_cost): Likewise.
12608 (vectorizable_bswap): Add cost vector arg and adjust.
12609 (vectorizable_call): Likewise.
12610 (vectorizable_simd_clone_call): Likewise.
12611 (vectorizable_conversion): Likewise.
12612 (vectorizable_assignment): Likewise.
12613 (vectorizable_shift): Likewise.
12614 (vectorizable_operation): Likewise.
12615 (vectorizable_store): Likewise.
12616 (vectorizable_load): Likewise.
12617 (vectorizable_condition): Likewise.
12618 (vectorizable_comparison): Likewise.
12619 (can_vectorize_live_stmts): Likewise.
12620 (vect_analyze_stmt): Likewise.
12621 (vect_transform_stmt): Adjust calls to vectorizable_*.
12622 * tree-vectorizer.c: Include gimple-pretty-print.h.
12623 (dump_stmt_cost): New function.
12625 2018-05-16 Richard Biener <rguenther@suse.de>
12627 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
12628 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
12629 * tree-ssa-dse.c: Include tree-ssa-loop.h.
12630 (check_name): New callback.
12631 (dse_classify_store): Track cycles via a visited bitmap of PHI
12632 defs and simplify handling of in-loop and across loop dead stores
12633 and properly fail for loop-variant refs. Handle byte-tracking with
12634 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
12637 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
12639 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
12640 (vect_get_mask_type_for_stmt): Likewise.
12641 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
12643 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
12644 to determine the statement's vector type and the vector type that
12645 should be used for calculating nunits. Deal with cases in which
12646 the type has to be deferred.
12647 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
12648 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
12649 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
12650 (vect_determine_vf_for_stmt): New functions, split out from...
12651 (vect_determine_vectorization_factor): ...here.
12652 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
12653 (vect_get_mask_type_for_stmt): New functions, split out from
12654 vect_determine_vectorization_factor.
12656 2018-05-16 Richard Biener <rguenther@suse.de>
12658 * tree-cfg.c (verify_gimple_assign_ternary): Properly
12659 verify the [VEC_]COND_EXPR embedded comparison.
12661 2018-05-15 Martin Sebor <msebor@redhat.com>
12663 PR tree-optimization/85753
12664 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
12665 RECORD_TYPE in addition to ARRAY_TYPE.
12667 2018-05-15 Martin Sebor <msebor@redhat.com>
12669 PR middle-end/85643
12670 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
12672 2018-05-15 Richard Biener <rguenther@suse.de>
12674 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
12675 add by_clobber_p one. Change algorithm to collect all defs
12676 representing uses we need to walk and try reducing them to
12677 a single one before failing.
12678 (dse_dom_walker::dse_optimize_stmt): Adjust.
12680 2018-05-13 Mark Wielaard <mark@klomp.org>
12682 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
12683 (size_of_loc_descr): Likewise.
12684 (output_loc_operands): Likewise.
12685 (output_loc_operands_raw): Likewise.
12686 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
12687 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
12688 (hash_loc_operands): Likewise.
12689 (compare_loc_operands): Likewise.
12691 2018-05-14 Mark Wielaard <mark@klomp.org>
12693 * dwarf2out.c (count_index_addrs): New function.
12694 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
12696 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12698 PR tree-optimization/83648
12699 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
12700 return value as malloc candidate.
12702 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12705 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
12706 param as true in call to suggest_attribute.
12708 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
12710 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
12711 -mreadonly-in-sdata.
12713 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12715 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
12717 (aarch64_crypto_aesd_fused): Likewise.
12719 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
12721 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
12722 (movsi_aarch64): Likewise.
12723 (load_pairsi): Likewise.
12724 (load_pairdi): Likewise.
12725 (store_pairsi): Likewise.
12726 (store_pairdi): Likewise.
12727 (load_pairsf): Likewise.
12728 (load_pairdf): Likewise.
12729 (store_pairsf): Likewise.
12730 (store_pairdf): Likewise.
12731 (zero_extend): Likewise.
12732 (trunc): Swap alternatives.
12733 (fcvt_target): Add '?' to prefer w over r.
12735 2018-05-14 Jakub Jelinek <jakub@redhat.com>
12738 * config/i386/i386.md: Disallow non-commutative arithmetics in
12739 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
12740 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
12741 in the peephole2 before it.
12743 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
12745 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
12746 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
12747 (ix86_handle_option): Handle -mcldemote.
12748 * config.gcc: New header.
12749 * config/i386/cldemoteintrin.h: New file.
12750 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
12751 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
12753 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12754 OPTION_MASK_ISA_CLDEMOTE.
12755 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
12756 (ix86_valid_target_attribute_inner_p): Ditto.
12757 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
12758 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
12759 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
12760 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
12761 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
12763 * config/i386/i386.opt: Add -mcldemote.
12764 * config/i386/x86intrin.h: New header.
12765 * doc/invoke.texi: Add -mcldemote.
12767 2018-05-14 Richard Biener <rguenther@suse.de>
12769 * doc/match-and-simplify.texi: Adjust :s documentation.
12771 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
12773 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
12774 intended memcpy size.
12775 (REORDER_45): Likewise.
12777 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
12779 * sort.cc: New file.
12780 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
12781 * vec.c (qsort_chk): Use gcc_qsort.
12782 * Makefile.in (OBJS-libcommon): Add sort.o.
12783 (build/sort.o): New target. Use it...
12784 (BUILD_RTL): ... here, and...
12785 (build/gencfn-macros): ... here, and...
12786 (build/genmatch): ... here.
12788 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
12789 Chung-Ju Wu <jasonwucj@gmail.com>
12791 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
12792 * config/nds32/nds32-graywolf.md: New file.
12793 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
12794 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
12796 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
12797 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
12798 * config/nds32/nds32.md (pipeline_model): Add graywolf.
12799 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
12800 * config/nds32/pipelines.md: Include n15 settings.
12802 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
12803 Chung-Ju Wu <jasonwucj@gmail.com>
12805 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
12806 * config/nds32/nds32-n13.md: New file.
12807 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
12808 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
12810 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
12811 * config/nds32/nds32.md (pipeline_model): Add n13.
12812 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
12813 * config/nds32/pipelines.md: Include n13 settings.
12815 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
12816 Chung-Ju Wu <jasonwucj@gmail.com>
12818 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
12819 * config/nds32/nds32-n10.md: New file.
12820 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
12821 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
12823 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
12824 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
12825 * config/nds32/nds32.md (pipeline_model): Add n10.
12826 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
12827 * config/nds32/pipelines.md: Include n10 settings.
12829 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
12830 Kito Cheng <kito.cheng@gmail.com>
12831 Chung-Ju Wu <jasonwucj@gmail.com>
12833 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
12834 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
12835 Add enum values for DSP extension instructions.
12836 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
12838 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
12839 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
12840 New code iterators.
12841 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
12842 * config/nds32/nds32-dspext.md: New file for DSP implementation.
12843 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
12844 * config/nds32/nds32-intrinsic.md: Likewise.
12845 * config/nds32/nds32_intrinsic.h: Likewise.
12846 * config/nds32/nds32-md-auxiliary.c: Likewise.
12847 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
12848 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
12849 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
12850 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
12851 * config/nds32/nds32-protos.h: New declarations for DSP extension.
12852 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
12853 TYPE_DMAC in switch statement.
12854 * config/nds32/nds32.c: New checking and implementation for DSP
12855 extension instructions.
12856 * config/nds32/nds32.h: Likewise.
12857 * config/nds32/nds32.md: Likewise.
12858 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
12859 * config/nds32/predicates.md: Implement new predicates for DSP
12862 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
12864 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
12865 Reformat alternatives and attributes so it is easier to identify
12866 which constraints/attributes go with which instruction.
12867 (mov<mode>_hardfloat32, FMOVE64): Likewise.
12868 (mov<mode>_softfloat32, FMOVE64): Likewise.
12869 (mov<mode>_hardfloat64, FMOVE64): Likewise.
12870 (mov<mode>_softfloat64, FMOVE64): Likewise.
12872 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
12874 * doc/extend.texi (PowerPC Built-in Functions): Rename this
12876 (Basic PowerPC Built-in Functions): The new name of the
12877 subsection previously known as "PowerPC Built-in Functions".
12878 (Basic PowerPC Built-in Functions Available on all Configurations):
12880 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
12881 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
12882 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
12883 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
12885 2018-05-11 Martin Jambor <mjambor@suse.cz>
12888 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
12891 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
12894 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
12896 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
12898 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
12899 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
12900 (ix86_handle_option): Handle -mwaitpkg.
12901 * config.gcc: New header.
12902 * config/i386/cpuid.h (bit_WAITPKG): New bit.
12903 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
12904 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
12906 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12907 OPTION_MASK_ISA_WAITPKG.
12908 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
12909 (ix86_option_override_internal): Add PTA_WAITPKG.
12910 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
12911 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
12912 IX86_BUILTIN_TPAUSE.
12913 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
12914 __builtin_ia32_umwait and __builtin_ia32_tpause.
12915 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
12916 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
12917 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
12918 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
12919 UNSPECV_TPAUSE): New.
12920 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
12921 * config/i386/i386.opt: Add -mwaitpkg.
12922 * config/i386/waitpkgintrin.h: New file.
12923 * config/i386/x86intrin.h: New header.
12924 * doc/invoke.texi: Add -mwaitpkg.
12926 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
12929 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
12931 (cortex-m0): Use armv6s-m isa.
12932 (cortex-m0plus): Likewise.
12933 (cortex-m1): Likewise.
12934 (cortex-m0.small-multiply): Likewise.
12935 (cortex-m0plus.small-multiply): Likewise.
12936 (cortex-m1.small-multiply): Likewise.
12938 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
12939 Jakub Jelinek <jakub@redhat.com>
12941 PR tree-optimization/85692
12942 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
12943 source permute as well.
12945 2018-05-11 Martin Liska <mliska@suse.cz>
12948 * doc/extend.texi: Document LLVM style format for no_sanitize
12951 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
12953 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
12954 mode_supports_vsx_dform_quad to mode_supports_dq_form.
12955 (mode_supports_vsx_dform_quad): Likewise.
12956 (mode_supports_vmx_dform): Move these functions to be next to the
12957 other mode_supports functions.
12958 (mode_supports_dq_form): Likewise.
12959 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
12960 mode_supports_dq_form.
12961 (reg_offset_addressing_ok_p): Likewise.
12962 (offsettable_ok_by_alignment): Likewise.
12963 (rs6000_legitimate_offset_address_p): Likewise.
12964 (legitimate_lo_sum_address_p): Likewise.
12965 (rs6000_legitimize_address): Likewise.
12966 (rs6000_legitimize_reload_address): Likewise.
12967 (rs6000_secondary_reload_inner): Likewise.
12968 (rs6000_preferred_reload_class): Likewise.
12969 (rs6000_output_move_128bit): Likewise.
12971 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
12973 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
12974 Generate SImode target register for null target.
12975 <case IX86_BUILTIN_XGETBV>: Ditto.
12976 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
12977 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
12979 2018-05-10 Carl Love <cel@us.ibm.com>
12981 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
12982 dcbtt and dcbtstt if operands[2] is 0.
12984 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
12987 * config/i386/sse.md (usadv64qi): New expander.
12989 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
12991 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
12992 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
12993 -maltivec=be support.
12994 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
12995 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
12996 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12997 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
12998 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
12999 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
13000 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
13001 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
13002 altivec_vsumsws): Adjust.
13003 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
13004 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
13005 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
13007 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
13008 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
13009 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
13010 (altivec_lve<VI_char>x): Delete expand.
13011 (*altivec_lve<VI_char>x_internal): Rename to...
13012 (altivec_lve<VI_char>x): ... this.
13013 (altivec_lvxl_<mode>): Delete expand.
13014 (*altivec_lvxl_<mode>_internal): Rename to ...
13015 (altivec_lvxl_<mode>): ... this.
13016 (altivec_stvxl_<mode>): Delete expand.
13017 (*altivec_stvxl_<mode>_internal): Rename to ...
13018 (altivec_stvxl_<mode>): ... this.
13019 (altivec_stve<VI_char>x): Delete expand.
13020 (*altivec_stve<VI_char>x_internal): Rename to ...
13021 (altivec_stve<VI_char>x): ... this.
13022 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
13023 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
13024 reduc_plus_scal_<mode>): Adjust.
13025 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
13027 (rs6000_cpu_cpp_builtins): Adjust.
13028 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
13029 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
13030 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
13031 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
13032 -maltivec=be support.
13033 (rs6000_split_vec_extract_var): Adjust.
13034 (rs6000_split_v4si_init): Adjust.
13035 (swap_selector_for_mode): Delete.
13036 (altivec_expand_lvx_be, altivec_expand_stvx_be,
13037 altivec_expand_stvex_be): Delete.
13038 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
13039 -maltivec=be support.
13040 (rs6000_gimple_fold_builtin): Ditto.
13041 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
13043 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
13044 (TARGET_DIRECT_MOVE_64BIT): Adjust.
13045 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
13046 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
13047 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
13048 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
13049 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
13050 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
13051 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
13052 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
13053 anonymous split): Adjust.
13054 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
13055 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
13057 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
13059 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
13060 when --with-gxx-include-dir is also specified.
13061 * configure: Regenerate.
13063 2018-05-09 Jim Wilson <jimw@sifive.com>
13066 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
13067 * config/riscv/t-withmultilib: New.
13068 * config/riscv/withmultilib.h: New.
13069 * doc/install.texi: Document RISC-V --with-multilib-list support.
13071 2018-05-09 Richard Biener <rguenther@suse.de>
13073 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
13075 (vect_bb_vectorization_profitable_p): Adjust. Compute
13076 actual scalar cost using the cost vector and the add_stmt_cost
13079 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13081 PR rtl-optimization/85645
13082 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
13083 in the REG_CFA_REGISTER note for LR, don't leave it empty.
13085 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13087 PR rtl-optimization/85645
13088 * shrink-wrap.c (spread_components): Return a boolean saying if
13089 anything was changed.
13090 (try_shrink_wrapping_separate): Iterate spread_components until
13091 nothing changes anymore.
13093 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13095 PR rtl-optimization/85645
13096 * regrename.c (build_def_use): Also kill the chains that include the
13097 destination of a REG_CFA_REGISTER note.
13099 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13101 PR rtl-optimization/85645
13102 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
13103 insn that has a REG_CFA_REGISTER note.
13105 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
13107 * cfgexpand.c (expand_clobber): New function.
13108 (expand_gimple_stmt_1): Use it.
13109 * tree-vect-stmts.c (vect_clobber_variable): New function,
13111 (vectorizable_simd_clone_call): ...here.
13112 (vectorizable_store): Emit a clobber either side of an
13113 IFN_STORE_LANES sequence.
13114 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
13116 2018-05-09 Tom de Vries <tom@codesourcery.com>
13119 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
13120 (define_insn "trap_if_false"): Add exit after trap.
13122 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
13124 PR rtl-optimization/85638
13125 * bb-reorder.c: Include common/common-target.h.
13126 (create_forwarder_block): New function extracted from...
13127 (fix_up_crossing_landing_pad): ...here. Rename into...
13128 (dw2_fix_up_crossing_landing_pad): ...this.
13129 (sjlj_fix_up_crossing_landing_pad): New function.
13130 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
13131 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
13132 from both partitions and exit the loop after one iteration.
13134 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13137 * doc/extend.texi (PowerPC Built-in Functions): Rename this
13139 (Basic PowerPC Built-in Functions): The new name of the
13140 subsection previously known as "PowerPC Built-in Functions".
13141 (Basic PowerPC Built-in Functions Available on all Configurations):
13143 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
13145 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
13147 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
13149 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
13152 2018-05-08 Jim Wilson <jimw@sifive.com>
13154 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
13155 (LD_EMUL_SUFFIX): New.
13156 (LINK_SPEC): Use it.
13158 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13160 * doc/extend.texi (PowerPC Built-in Functions): Rename this
13162 (Basic PowerPC Built-in Functions): The new name of the
13163 subsection previously known as "PowerPC Built-in Functions".
13164 (Basic PowerPC Built-in Functions Available on all Configurations):
13166 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
13168 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
13170 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
13172 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
13175 2018-05-08 Jakub Jelinek <jakub@redhat.com>
13178 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
13179 after cmpelim optimization.
13181 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
13183 * config.gcc: Support "goldmont".
13184 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
13185 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13186 PROCESSOR_GOLDMONT.
13187 * config/i386/i386.c (m_GOLDMONT): Define.
13188 (processor_target_table): Add "goldmont".
13189 (PTA_GOLDMONT): Define.
13190 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
13191 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
13192 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
13193 (fold_builtin_cpu): Add "goldmont".
13194 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
13195 (ix86_option_override_internal): Add "goldmont".
13196 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
13197 (processor_type): Add PROCESSOR_GOLDMONT.
13198 * config/i386/i386.md: Add CPU "glm".
13199 * config/i386/glm.md: New file.
13200 * config/i386/x86-tune.def: Add m_GOLDMONT.
13201 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
13203 2018-05-08 Jakub Jelinek <jakub@redhat.com>
13206 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
13208 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
13209 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
13210 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
13213 * config/i386/i386.c (ix86_fold_builtin): Handle
13214 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
13217 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
13218 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
13220 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
13223 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
13224 (check_arch): Likewise.
13225 (check_fpu): Return the result rather than printing it.
13226 (end arch): Fix operator precedence.
13227 (end cpu): Likewise.
13228 (END): Print the result from check_fpu.
13230 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
13231 Alan Hayward <alan.hayward@arm.com>
13232 David Sherwood <david.sherwood@arm.com>
13234 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
13235 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
13236 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
13237 (*fcmuo<mode>_and): New patterns.
13239 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
13241 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
13242 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
13243 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
13244 (cmp_op, sve_imm_con): New code attributes.
13245 (SVE_COND_INT_CMP, imm_con): Delete.
13246 (cmp_op): Remove above unspecs from int attribute.
13247 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
13249 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
13250 comparison-specific unspecs.
13251 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
13252 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
13253 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
13254 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
13255 (*vec_fcm<cmp_op><mode>): Rename to...
13256 (*fcm<cmp_op><mode>): ...this and adjust likewise.
13257 (*vec_fcmuo<mode>): Rename to...
13258 (*fcmuo<mode>): ...this and adjust likewise.
13259 (*pred_fcm<cmp_op><mode>): New pattern.
13260 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
13261 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
13263 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
13265 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
13266 (aarch64_emit_sve_predicated_cond): New function.
13267 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
13268 (aarch64_emit_unspec_cond_or): Replace with...
13269 (aarch64_emit_sve_or_conds): ...this new function. Use
13270 aarch64_emit_sve_ptrue_op for the individual comparisons and
13271 aarch64_emit_binop to OR them together.
13272 (aarch64_emit_inverted_unspec_cond): Replace with...
13273 (aarch64_emit_sve_inverted_cond): ...this new function. Use
13274 aarch64_emit_sve_ptrue_op for the comparison and
13275 aarch64_emit_unop to invert the result.
13276 (aarch64_expand_sve_vec_cmp_float): Update after the above
13277 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
13279 2018-05-07 Nathan Sidwell <nathan@acm.org>
13281 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
13282 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
13283 (Backwards Compatibility): Likewise.
13285 2018-05-07 Luis Machado <luis.machado@linaro.org>
13289 2018-05-07 Luis Machado <luis.machado@linaro.org>
13291 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13292 <prefetch_dynamic_strides>: New const bool field.
13293 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13294 prefetch_dynamic_strides.
13295 (exynosm1_prefetch_tune): Likewise.
13296 (thunderxt88_prefetch_tune): Likewise.
13297 (thunderx_prefetch_tune): Likewise.
13298 (thunderx2t99_prefetch_tune): Likewise.
13299 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
13301 (aarch64_override_options_internal): Update to set
13302 PARAM_PREFETCH_DYNAMIC_STRIDES.
13303 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
13304 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
13305 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
13306 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
13307 prefetch-dynamic-strides setting.
13309 2018-05-07 Luis Machado <luis.machado@linaro.org>
13311 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13312 <minimum_stride>: New const int field.
13313 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13314 minimum_stride field.
13315 (exynosm1_prefetch_tune): Likewise.
13316 (thunderxt88_prefetch_tune): Likewise.
13317 (thunderx_prefetch_tune): Likewise.
13318 (thunderx2t99_prefetch_tune): Likewise.
13319 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
13320 (aarch64_override_options_internal): Update to set
13321 PARAM_PREFETCH_MINIMUM_STRIDE.
13322 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
13323 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
13324 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
13325 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
13326 stride is constant and is below the minimum stride threshold.
13328 2018-05-07 Luis Machado <luis.machado@linaro.org>
13330 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
13333 2018-05-07 Luis Machado <luis.machado@linaro.org>
13335 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13336 <prefetch_dynamic_strides>: New const bool field.
13337 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13338 prefetch_dynamic_strides.
13339 (exynosm1_prefetch_tune): Likewise.
13340 (thunderxt88_prefetch_tune): Likewise.
13341 (thunderx_prefetch_tune): Likewise.
13342 (thunderx2t99_prefetch_tune): Likewise.
13343 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
13345 (aarch64_override_options_internal): Update to set
13346 PARAM_PREFETCH_DYNAMIC_STRIDES.
13347 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
13348 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
13349 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
13350 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
13351 prefetch-dynamic-strides setting.
13353 2018-05-07 Luis Machado <luis.machado@linaro.org>
13355 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13356 <minimum_stride>: New const int field.
13357 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13358 minimum_stride field.
13359 (exynosm1_prefetch_tune): Likewise.
13360 (thunderxt88_prefetch_tune): Likewise.
13361 (thunderx_prefetch_tune): Likewise.
13362 (thunderx2t99_prefetch_tune): Likewise.
13363 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
13364 (aarch64_override_options_internal): Update to set
13365 PARAM_PREFETCH_MINIMUM_STRIDE.
13366 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
13367 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
13368 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
13369 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
13370 stride is constant and is below the minimum stride threshold.
13372 2018-05-06 Jakub Jelinek <jakub@redhat.com>
13375 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
13376 the type is addressable. Don't force op into register if it has
13379 2018-05-05 Roland McGrath <mcgrathr@google.com>
13382 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
13383 any section for which we don't know a specific type it should have,
13384 regardless of name. Previously this was done only for the exact
13385 names ".init_array", ".fini_array", and ".preinit_array".
13386 (default_elf_asm_named_section): Add comment about
13387 relationship with default_section_type_flags and SECTION_NOTYPE.
13388 (get_section): Don't consider it a type conflict if one side has
13389 SECTION_NOTYPE and the other doesn't, as long as neither has the
13390 SECTION_BSS et al used in the default_section_type_flags logic.
13392 2018-05-05 Tom de Vries <tom@codesourcery.com>
13395 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
13396 (workaround_barsyncs): New function.
13397 (nvptx_reorg): Use workaround_barsyncs.
13398 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
13399 (define_expand "nvptx_membar_cta"): New define_expand.
13400 (define_insn "*nvptx_membar_cta"): New insn.
13402 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
13404 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
13405 To improve optimization opportunities.
13406 * builtin-types.def: The new needed builtin types for the above.
13408 2018-05-04 Richard Biener <rguenther@suse.de>
13410 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
13411 * gimple-ssa-store-merging.c
13412 (imm_store_chain_info::output_merged_store): Remove redundant create,
13413 release split_store vector contents on failure.
13414 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
13415 scalar stmt vector on cache hit.
13417 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
13419 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
13421 * config.gcc (powerpc-xilinx-eabi*): Remove.
13422 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
13424 (fusion_addis_mem_combo_load): Ditto.
13425 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
13427 (rs6000_cpu_cpp_builtins): Ditto.
13428 * config/rs6000/rs6000-linux.c
13429 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
13430 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
13431 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
13433 (rs6000_setup_reg_addr_masks): Ditto.
13434 (rs6000_init_hard_regno_mode_ok): Ditto.
13435 (rs6000_option_override_internal): Ditto.
13436 (legitimate_lo_sum_address_p): Ditto.
13437 (rs6000_legitimize_address): Ditto.
13438 (rs6000_legitimize_reload_address): Ditto.
13439 (rs6000_legitimate_address_p): Ditto.
13440 (abi_v4_pass_in_fpr): Ditto.
13441 (setup_incoming_varargs): Ditto.
13442 (rs6000_gimplify_va_arg): Ditto.
13443 (rs6000_split_multireg_move): Ditto.
13444 (rs6000_savres_strategy): Ditto.
13445 (rs6000_emit_prologue_components): Ditto.
13446 (rs6000_emit_epilogue_components): Ditto.
13447 (rs6000_emit_prologue): Ditto.
13448 (rs6000_emit_epilogue): Ditto.
13449 (rs6000_elf_file_end): Ditto.
13450 (rs6000_function_value): Ditto.
13451 (rs6000_libcall_value): Ditto.
13452 * config/rs6000/rs6000.h: Ditto.
13453 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
13454 (TARGET_MINMAX): ... this. New.
13455 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
13456 * config/rs6000/rs6000.md: Remove Xilinx FP support.
13457 (*movsi_internal1_single): Delete.
13458 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
13459 mfpu=, mxilinx-fpu): Delete.
13460 * config/rs6000/singlefp.h: Delete.
13461 * config/rs6000/sysv4.h: Remove Xilinx FP support.
13462 * config/rs6000/t-rs6000: Ditto.
13463 * config/rs6000/t-xilinx: Delete.
13464 * config/rs6000/titan.md: Adjust for fp_type removal.
13465 * config/rs6000/vsx.md: Remove Xilinx FP support.
13466 (VStype_simple): Delete.
13467 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
13468 * config/rs6000/xfpu.h: Delete.
13469 * config/rs6000/xfpu.md: Delete.
13470 * config/rs6000/xilinx.h: Delete.
13471 * config/rs6000/xilinx.opt: Delete.
13472 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
13473 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
13475 2018-05-04 Tom de Vries <tom@codesourcery.com>
13478 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
13481 2018-05-04 Richard Biener <rguenther@suse.de>
13483 PR middle-end/85627
13484 * tree-complex.c (update_complex_assignment): We are always in SSA form.
13485 (expand_complex_div_wide): Likewise.
13486 (expand_complex_operations_1): Likewise.
13487 (expand_complex_libcall): Preserve EH info of the original stmt.
13488 (tree_lower_complex): Handle removed blocks.
13489 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
13490 on complex multiplication and division libcall builtins.
13492 2018-05-04 Richard Biener <rguenther@suse.de>
13494 PR middle-end/85574
13495 * fold-const.c (negate_expr_p): Restrict negation of operand
13496 zero of a division to when we know that can happen without
13498 (fold_negate_expr_1): Likewise.
13500 2018-05-04 Jakub Jelinek <jakub@redhat.com>
13503 * real.h (real_nextafter): Declare.
13504 * real.c (real_nextafter): New function.
13505 * fold-const-call.c (fold_const_nextafter): New function.
13506 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
13507 CASE_CFN_NEXTTOWARD.
13508 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
13509 even when arg1_mode is different from arg0_mode.
13511 2018-05-03 Nathan Sidwell <nathan@acm.org>
13513 * doc/extend.texi (Deprecated Features): Remove
13514 -ffriend-injection.
13515 (Backwards Compatibility): Likewise.
13516 * doc/invoke.texi (C++ Language Options): Likewise.
13517 (C++ Dialect Options): Likewise.
13519 2018-05-03 Jakub Jelinek <jakub@redhat.com>
13522 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
13523 _mm512_mask_mullox_epi64): New intrinsics.
13525 2018-05-03 Tom de Vries <tom@codesourcery.com>
13528 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
13529 dump files): Add offload-tree.
13531 2018-05-03 Richard Biener <rguenther@suse.de>
13533 PR tree-optimization/85615
13534 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
13535 to loops not nested in BBs loop father to avoid creating multi-entry
13538 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13540 PR tree-optimization/70291
13541 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
13542 arguments. Change return type to tree. Emit libcall as a new
13543 statement rather than replacing existing one when inplace_p is true.
13544 (expand_complex_multiplication_components): New function.
13545 (expand_complex_multiplication): Expand floating-point complex
13546 multiplication using the above.
13547 (expand_complex_division): Rename inner_type parameter to type.
13548 Update expand_complex_libcall call-site.
13549 (expand_complex_operations_1): Update expand_complex_multiplication
13550 and expand_complex_division call-sites.
13552 2018-05-02 Jakub Jelinek <jakub@redhat.com>
13555 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
13556 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
13557 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
13558 the highest significant bit of the shift count mask is clear. In
13559 check whether and[sq]i3 is needed verify that all significant bits
13560 of the shift count other than the highest are set.
13562 2018-05-02 Tom de Vries <tom@codesourcery.com>
13565 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
13566 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
13567 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
13568 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
13569 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
13570 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
13571 __builtin_goacc_parlevel_size.
13573 2018-05-02 Richard Biener <rguenther@suse.de>
13575 PR tree-optimization/85597
13576 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
13577 do not use split vect_get_vec_defs call but call vect_get_slp_defs
13580 2018-05-02 Tom de Vries <tom@codesourcery.com>
13583 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
13584 dump files): Add ltrans-tree.
13586 2018-05-02 Tom de Vries <tom@codesourcery.com>
13589 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
13590 dump files): Add wpa-ipa.
13592 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
13594 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
13595 powerpc*-*-linux*paired* target.
13596 * config/rs6000/750cl.h: Delete.
13597 * config/rs6000/paired.h: Delete.
13598 * config/rs6000/paired.md: Delete.
13599 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
13601 * config/rs6000/rs6000-builtin.def: Remove paired float support.
13602 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13603 comment. Remove paired float support.
13604 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
13605 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
13607 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
13608 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
13610 * config/rs6000/rs6000.c: Remove paired float support.
13611 (paired_expand_vector_init, paired_expand_vector_move,
13612 paired_emit_vector_compare, paired_emit_vector_cond_expr,
13613 (paired_expand_lv_builtin, paired_expand_stv_builtin,
13614 paired_expand_builtin, paired_expand_predicate_builtin,
13615 paired_init_builtins): Delete.
13616 * config/rs6000/rs6000.h: Remove paired float support.
13617 * config/rs6000/rs6000.md: Remove paired float support.
13618 (move_from_CR_ov_bit): Delete.
13619 * config/rs6000/rs6000.opt (mpaired): Delete.
13620 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
13621 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
13623 2018-05-02 Richard Biener <rguenther@suse.de>
13625 PR middle-end/85567
13626 * gimplify.c (gimplify_save_expr): When in SSA form allow
13627 SAVE_EXPRs to compute to SSA vars.
13629 2018-05-02 Jakub Jelinek <jakub@redhat.com>
13632 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
13633 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
13634 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
13635 clobber operands[2], instead use a new pseudo. Formatting fixes.
13637 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
13639 PR tree-optimization/85586
13640 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
13641 exit early for statements in the same group if the accesses are
13644 2018-05-02 Tom de Vries <tom@codesourcery.com>
13647 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
13650 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
13652 PR tree-optimization/85143
13653 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
13655 2018-05-01 Tom de Vries <tom@codesourcery.com>
13658 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
13659 not found" error message.
13661 2018-05-01 Tom de Vries <tom@codesourcery.com>
13664 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
13665 * vec.c (test_ordered_remove_if): New function.
13666 (vec_c_tests): Call test_ordered_remove_if.
13667 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
13668 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
13669 * tree-vect-patterns.c (vect_pattern_recog_1): Use
13670 VEC_ORDERED_REMOVE_IF.
13672 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13674 PR tree-optimization/82665
13675 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
13676 pointer subtraction where arguments come from a memchr call.
13678 2018-05-01 Jakub Jelinek <jakub@redhat.com>
13680 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
13681 --push-state --as-needed and --pop-state instead of --as-needed and
13682 --no-as-needed if ld supports it.
13683 * configure: Regenerated.
13686 * doc/install.texi2html: Replace _002d with - and _002a with * in
13687 generated html files using sed.
13689 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13692 * gcc-rich-location.c (blank_line_before_p): New function.
13693 (use_new_line): New function.
13694 (gcc_rich_location::add_fixit_insert_formatted): New function.
13695 * gcc-rich-location.h
13696 (gcc_rich_location::add_fixit_insert_formatted): New function.
13698 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13700 * selftest.c (assert_streq): Rename "expected" and "actual" to
13701 "val1" and "val2". Extend NULL-handling to cover both inputs
13702 symmetrically, while still requiring both to be non-NULL for a pass.
13703 * selftest.h (assert_streq): Rename "expected" and "actual" to
13705 (ASSERT_EQ): Likewise.
13706 (ASSERT_EQ_AT): Likewise.
13707 (ASSERT_KNOWN_EQ): Likewise.
13708 (ASSERT_KNOWN_EQ_AT): Likewise.
13709 (ASSERT_NE): Likewise.
13710 (ASSERT_MAYBE_NE): Likewise.
13711 (ASSERT_MAYBE_NE_AT): Likewise.
13712 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
13713 the assertion to pass.
13714 (ASSERT_STREQ_AT): Likewise.
13716 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
13718 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
13719 interaction with -pie.
13721 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13723 * selftest.h: Fix alphabetization of per-source-file selftest
13726 2018-04-30 Jason Merrill <jason@redhat.com>
13728 PR c++/61982 - dead stores to destroyed objects.
13729 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
13732 2018-04-30 Jason Merrill <jason@redhat.com>
13734 * tree.c (build_clobber): New.
13735 * tree.h: Declare it.
13736 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
13738 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13740 * diagnostic-show-locus.c (layout::layout): Update for
13741 location_get_source_line returning a char_span.
13742 (struct char_span): Move to input.h.
13743 (struct correction): Update for fields in char_span becoming
13745 (struct source_line): Update for location_get_source_line
13746 returning a char_span.
13747 (layout::print_line): Likewise.
13748 * edit-context.c (edited_file::print_content): Likewise.
13749 (edited_file::print_diff_hunk): Likewise.
13750 (edited_file::print_run_of_changed_lines): Likewise.
13751 (edited_file::get_num_lines): Likewise.
13752 (edited_line::edited_line): Likewise.
13753 * final.c (asm_show_source): Likewise.
13754 * input.c (location_get_source_line): Convert return type
13755 from const char * to char_span, losing the final "line_len"
13757 (dump_location_info): Update for the above.
13758 (get_substring_ranges_for_loc): Likewise. Use a char_span
13759 when handling the literal within the line.
13760 (test_reading_source_line): Update for location_get_source_line
13761 returning a char_span.
13762 * input.h (class char_span): Move here from
13763 diagnostic-show-locus.c, converting from a struct to a class.
13764 Make data members private.
13765 (char_span::operator bool): New.
13766 (char_span::length): New.
13767 (char_span::get_buffer): New.
13768 (char_span::operator[]): New.
13769 (char_span::subspan): Make const.
13770 (char_span::xstrdup): New.
13771 (location_get_source_line): Convert return type from const char *
13772 to char_span, losing the final "line_size" param.
13774 2018-04-30 Jan Hubicka <jh@suse.cz>
13776 * lto-wrapper.c (ltrans_priorities): New static var.
13777 (cmp_priority): New.
13778 (run_gcc): Read priorities and if doing parallel build order
13779 the Makefile by them.
13781 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13783 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
13785 2018-04-30 Richard Biener <rguenther@suse.de>
13787 * tree-cfg.c (verify_address): Remove base argument, add
13788 flag whether to check TREE_ADDRESSABLE and do that.
13789 (verify_expr): Remove.
13790 (verify_types_in_gimple_reference): Add pieces from verify_expr.
13791 (verify_gimple_assign_single): Likewise.
13792 (verify_gimple_switch): Likewise.
13793 (verify_expr_location_1): Dereference tp once. Add (disabled)
13794 piece from verify_expr.
13795 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
13797 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
13799 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
13801 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
13803 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
13804 (small_data_pattern): Likewise.
13805 (arc_rewrite_small_data): Likewise.
13806 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
13807 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
13808 (get_symbol_alignment): New function.
13809 (legitimate_small_data_address_p): Likewise.
13810 (legitimate_scaled_address): Update, call
13811 legitimate_small_data_address_p.
13812 (output_sdata): New static variable.
13813 (arc_print_operand): Update how we handle small data operands.
13814 (arc_print_operand_address): Likewise.
13815 (arc_legitimate_address_p): Update, use
13816 legitimate_small_data_address_p.
13817 (arc_rewrite_small_data_p): Remove.
13818 (arc_rewrite_small_data_1): Likewise.
13819 (arc_rewrite_small_data): Likewise.
13820 (small_data_pattern): Likewise.
13821 (compact_sda_memory_operand): Update to use
13822 legitimate_small_data_address_p and get_symbol_alignment.
13823 (prepare_move_operands): Don't rewite sdata pattern.
13824 (prepare_extend_operands): Remove.
13825 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
13827 (zero_extendqisi2): Likewise.
13828 (zero_extendhisi2): Likewise.
13829 (extendqihi2): Likewise.
13830 (extendqisi2): Likewise.
13831 (extendhisi2): Likewise.
13832 (addsi3): Likewise.
13833 (subsi3): Likewise.
13834 (andsi3): Likewise.
13835 * config/arc/constraints.md (Usd): Change it to memory constraint.
13837 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
13839 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
13840 as source of std instructions.
13841 * config/arc/arc.md (movsi_insn): Update pattern predicate to
13842 allow 6-bit constants as source for store instructions.
13843 (movdi_insn): Update instruction pattern to allow 6-bit constants
13844 as source for store instructions.
13846 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
13848 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
13850 2018-04-30 Nathan Sidwell <nathan@acm.org>
13851 Sandra Loosemore <sandra@codesourcery.com>
13853 * dumpfile.c (dump_open): Allow '-' for stdout.
13854 * doc/invoke.texi (Developer Options): Document dump filename
13855 determination early. Document stdin/stdout selection.
13857 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
13859 Microblaze Target: PIC data text relative
13861 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
13862 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
13864 * config/microblaze/microblaze.h (microblaze_constant_address_p):
13865 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
13866 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
13867 New addressing mode for data-text relative position indepenedent code.
13868 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
13869 'ADDRESS_SYMBOLIC_TXT_REL'.
13870 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
13871 (microblaze_legitimate_pic_operand): Exclude function calls from
13872 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
13873 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
13875 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
13876 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
13877 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
13878 for 'address + offset'.
13879 (microblaze_expand_prologue): Add new function prologue call for
13881 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
13882 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
13883 table in case of TARGET_PIC_DATA_TEXT_REL.
13884 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
13885 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
13886 Add new macros 'UNSPEC_TEXT',
13887 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
13888 + exclude function calls from 'UNSPEC_PLT' in case of data text
13890 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
13891 new target hook for generating address diff vector tables in case of
13893 * doc/tm.texi : Regenerate.
13894 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
13895 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
13896 of addr diff vector generation.
13897 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
13898 target hook definition.
13899 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
13900 Add default function for generate_pic_addr_diff_vec -> flag_pic.
13901 * doc/invoke.texi (Add new pic option): Add new microblaze pic
13902 option for data text relative.
13904 2018-04-30 Richard Biener <rguenther@suse.de>
13906 * tree-chrec.h (evolution_function_is_constant_p): Remove
13908 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
13910 2018-04-30 Richard Biener <rguenther@suse.de>
13913 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
13915 2018-04-30 Richard Biener <rguenther@suse.de>
13917 PR tree-optimization/28364
13918 PR tree-optimization/85275
13919 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
13920 copying first exit test.
13922 2018-04-28 Mark Wielaard <mark@klomp.org>
13924 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
13925 dwarf_version >= 5.
13926 (dwarf_AT): Handle DW_AT_addr_base.
13927 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
13929 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
13932 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
13933 (*ashl<dwi>3_doubleword_mask_1): Ditto.
13934 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
13935 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
13937 2018-04-28 Richard Biener <rguenther@suse.de>
13939 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
13940 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
13941 to reflect use. Only add interesting stmts.
13943 2018-04-27 Martin Jambor <mjambor@suse.cz>
13946 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
13947 the jump function allows for passing through aggregate values.
13949 2018-04-27 David Malcolm <dmalcolm@redhat.com>
13951 * input.h (in_system_header_at): Convert from macro to inline
13953 (from_macro_expansion_at): Likewise.
13954 (from_macro_definition_at): Likewise.
13956 2018-04-27 Jeff Law <law@redhat.com>
13958 * config.gcc: Mark tile* targets as deprecated/obsolete.
13960 2018-04-27 Richard Biener <rguenther@suse.de>
13962 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
13965 2018-04-27 Richard Biener <rguenther@suse.de>
13967 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
13969 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
13971 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
13972 with Yd constraint. Set "preferred_for_speed" attribute from
13973 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
13974 with Yd constraint.
13975 (*movdi_internal): Ditto.
13976 (movti_interunit splitters): Remove
13977 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
13978 (movdi_interunit splitters): Ditto.
13979 * config/i386/constraints.md (Ye): Remove.
13980 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
13982 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13985 * config/aarch64/constraints.md (Usg): Limit to 31.
13986 (Usj): Limit to 63.
13988 2018-04-27 Jakub Jelinek <jakub@redhat.com>
13990 PR tree-optimization/85529
13991 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
13992 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
13993 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
13994 zero extension or masking of the MSB bit.
13995 (optimize_range_tests): Add FIRST_BB argument, pass it through
13996 to optimize_range_tests_var_bound.
13997 (maybe_optimize_range_tests, reassociate_bb): Adjust
13998 optimize_range_tests callers.
14000 2018-04-26 Richard Biener <rguenther@suse.de>
14001 Jakub Jelinek <jakub@redhat.com>
14003 * cgraph.h (symbol_table): Just declare debug method here.
14004 * symtab.c (symbol_table::debug): Define.
14006 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
14008 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
14010 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
14012 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
14013 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
14014 (*movdi_internal): Substitute Yi and Yj constraint with x
14015 and Ym and Yn constraint with y constraint. Update "isa"
14016 attribute and set "preferred_for_speed" attribute from
14017 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
14018 (*movsi_internal): Ditto.
14019 (*movdf_internal): Ditto.
14020 (*movsf_internal): Ditto.
14021 (*zero_extendsidi2): Ditto.
14022 * config/i386/sse.md (vec_set<mode>_0): Ditto.
14023 (sse2_loadld): Ditto.
14024 (*vec_extract<ssevecmodelower>_0): Ditto.
14025 (*vec_extractv4si_0_zext_sse4): Ditto.
14026 (vec_concatv2di): Ditto.
14027 (*vec_dup<mode>): Ditto.
14028 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
14029 * config/i386/constraints.md (Yi): Remove.
14034 2018-04-26 Nathan Sidwell <nathan@acm.org>
14036 * dumpfile.c (dump_open): New.
14037 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
14038 (dump_finish): Detect stdio/stderr by value not name.
14040 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
14042 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
14044 2018-04-26 Tom de Vries <tom@codesourcery.com>
14047 * config/nvptx/nvptx.c (verify_neutering_jumps)
14048 (verify_neutering_labels): New function
14049 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
14051 2018-04-26 Tom de Vries <tom@codesourcery.com>
14054 * config/nvptx/nvptx.c (needs_neutering_p): New function.
14055 (nvptx_single): Use needs_neutering_p to skip over insns that do not
14058 2018-04-26 Richard Biener <rguenther@suse.de>
14059 Tom de Vries <tom@codesourcery.com>
14062 * lto-streamer-out.c (output_function): Fixup loops if required to match
14063 discovery done in the reader.
14065 2018-04-26 Richard Biener <rguenther@suse.de>
14067 PR tree-optimization/85116
14068 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
14069 have a loop exit from the single latch predecessor. Remove
14070 case of header with just condition.
14071 (ch_base::copy_headers): Exclude infinite loops from any
14073 (pass_ch::execute): Record exits.
14075 2018-04-26 Richard Biener <rguenther@suse.de>
14077 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
14078 prologue cost vector and pass it to vect_get_load_cost.
14079 (vect_get_peeling_costs_all_drs): Likewise.
14080 (vect_peeling_hash_get_lowest_cost): Likewise.
14081 (vect_enhance_data_refs_alignment): Likewise.
14083 2018-04-26 Richard Biener <rguenther@suse.de>
14085 PR middle-end/85450
14086 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
14087 checking of integer<->pointer conversions.
14088 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
14089 sign-/zero-extending pointer types.
14090 (expand_omp_for_static_chunk): Likewise.
14092 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
14093 Jean Lee <xiaoyur347@gmail.com>
14095 * config/mips/mips.c (mips_asan_shadow_offset): New function.
14096 (TARGET_ASAN_SHADOW_OFFSET): Define.
14097 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
14098 true for -fsanitize=address.
14100 2018-04-25 Mark Wielaard <mark@klomp.org>
14102 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
14105 2018-04-25 Jakub Jelinek <jakub@redhat.com>
14107 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
14108 than "alu", remove explicit "memory" and "imm_disp" attributes.
14109 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
14111 PR middle-end/85414
14112 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
14113 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
14114 gen_lowpart_no_emit.
14116 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
14119 * config/i386/i386.c (ix86_expand_builtin): Change memory
14120 operand to XI, extend p0 to Pmode.
14121 * config/i386/i386.md: Change unspec volatile and operand
14122 1 mode to XI, change operand 0 mode to P.
14124 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14126 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
14127 GET_MODE_MASK before any checking.
14128 (nds32_can_use_bset_p): Likewise.
14129 (nds32_can_use_btgl_p): Likewise.
14131 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14133 * config/nds32/nds32-doubleword.md: New define_split pattern for
14134 illegal register number.
14136 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14138 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
14140 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14142 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
14144 2018-04-25 Richard Biener <rguenther@suse.de>
14146 * lto-streamer.h (LTO_major_version): Bump to 8.
14148 2018-04-25 Jakub Jelinek <jakub@redhat.com>
14150 * BASE-VER: Set to 9.0.0.
14152 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
14154 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
14155 in __abskf2 and __powikf2.
14157 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14160 * config/aarch64/constraints.md (Usg, Usj): New constraints.
14161 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
14162 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
14163 Use the above on operand 2. Reindent.
14164 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
14166 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
14169 * common/config/i386/i386-common.c (ix86_handle_option): Don't
14171 * config/i386/i386.opt (mcet): Removed.
14172 * doc/install.texi: Remove -mcet documentation.
14173 * doc/invoke.texi: Likewise.
14175 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
14178 * doc/install.texi: Remove -mcet from bootstrap-cet.
14180 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14183 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
14184 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
14188 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
14189 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
14190 containing a CONST_VECTOR.
14192 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
14194 * doc/install.texi: Update newlib dependency for nvptx.
14196 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14199 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
14200 instead of INTVAL when shifting x left.
14202 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
14204 PR tree-optimization/85478
14205 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
14206 vect_grouped_store_supported for single element vectors.
14208 2018-04-24 Richard Biener <rguenther@suse.de>
14211 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
14212 load cost increase to the case of non-constant step.
14214 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14217 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
14218 destination if any_malformed_asm.
14220 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
14222 PR middle-end/85496
14223 * expr.c (store_field): In the bitfield case, if the value comes from
14224 a function call and is returned in registers by means of a PARALLEL,
14225 do not change the mode of the temporary unless BLKmode and VOIDmode.
14227 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
14229 PR rtl-optimization/85423
14230 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
14231 dependencies to debug insns when the previous insn is non-debug.
14233 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
14235 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
14236 enums into a single definition.
14237 (fls): Fix predicates and printing.
14240 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
14242 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
14243 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
14244 and short u6 immediate.
14245 (check_if_valid_sleep_operand): Remove.
14246 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
14248 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14250 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
14251 flag_always_save_lp condition.
14252 * config/nds32/nds32.opt (malways-save-lp): New option.
14254 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14256 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
14257 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
14258 * config/nds32/nds32.h
14259 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
14260 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
14262 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14264 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
14265 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
14267 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14268 Chung-Ju Wu <jasonwucj@gmail.com>
14270 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
14272 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
14273 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
14275 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14277 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
14279 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14281 * config/nds32/nds32-protos.h (nds32_data_alignment,
14282 nds32_local_alignment): Declare.
14283 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
14284 nds32_local_alignment): New functions.
14285 (TARGET_CONSTANT_ALIGNMENT): Define.
14286 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
14288 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14290 * config/nds32/nds32.c
14291 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
14292 (TARGET_MODES_TIEABLE_P): Likewise.
14294 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14296 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
14297 level Ofast and Og.
14299 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
14300 Chung-Ju Wu <jasonwucj@gmail.com>
14302 * config/nds32/constants.md (unspec_volatile_element): Add enum values
14303 for unaligned access.
14304 * config/nds32/nds32-intrinsic.c: Implementation of expanding
14306 * config/nds32/nds32-intrinsic.md: Likewise.
14307 * config/nds32/nds32_intrinsic.h: Likewise.
14308 * config/nds32/nds32.h (nds32_builtins): Likewise.
14309 * config/nds32/nds32.opt (munaligned-access): New option.
14310 * config/nds32/nds32.c (nds32_asm_file_start): Display
14311 flag_unaligned_access status.
14313 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
14315 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
14316 -mno-relax is present.
14317 * config/riscv/linux.h (LINK_SPEC): Ditto.
14319 2018-04-20 Martin Sebor <msebor@redhat.com>
14322 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
14324 (gimple_fold_builtin_stxcpy_chk): Same.
14325 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
14327 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
14330 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
14331 __powikf2 when long double is IEEE 128-bit.
14333 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
14335 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
14336 step to make sure stack always aligned.
14338 2018-04-20 Carl Love <cel@us.ibm.com>
14341 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
14342 size check for arg0.
14344 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
14345 Tom de Vries <tom@codesourcery.com>
14348 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
14349 Emit insns for calls too.
14350 (nvptx_find_par): Always look for worker-level predecessor insn.
14351 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
14353 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
14354 (nvptx_process_pars): Propagate frames for calls.
14356 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
14359 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
14361 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
14362 (ix86_handle_option): Don't handle OPT_mibt.
14363 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
14365 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
14367 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
14368 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
14369 (ix86_target_macros): Define __CET__ with flag_cf_protection
14370 for -fcf-protection.
14371 * config/i386/i386.c (isa2_opts): Remove -mibt.
14372 * config/i386/i386.h (TARGET_IBT): Removed.
14373 (TARGET_IBT_P): Likewise.
14374 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
14375 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
14376 * config/i386/i386.opt (mcet): Update help message.
14377 (mshstk): Likewise.
14379 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
14380 -mcet as an alias for -mshstk.
14382 2018-04-20 Richard Biener <rguenther@suse.de>
14384 PR middle-end/85475
14385 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
14386 complexity by forcing a single use of the multiply operand.
14388 2018-04-20 Martin Jambor <mjambor@suse.cz>
14391 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
14392 recursion dependency to only apply to non-clones.
14394 2018-04-20 Martin Jambor <mjambor@suse.cz>
14397 * ipa-cp.c (create_specialized_node): Check that clones of
14398 self-recursive edges exist during IPA-CP.
14400 2018-04-19 Toon Moene <toon@moene.org>
14402 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
14405 2018-04-19 Jakub Jelinek <jakub@redhat.com>
14407 PR tree-optimization/85467
14408 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
14409 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
14410 VECTOR_CST element to type.
14412 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
14415 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
14416 * config/i386/i386.md (builtin_setjmp_setup): Removed.
14417 (builtin_longjmp): Likewise.
14418 (save_stack_nonlocal): New pattern.
14419 (restore_stack_nonlocal): Likewise.
14421 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
14424 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
14425 Replace ASM_OUTPUT_LABEL with fprintf.
14427 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
14430 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
14431 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
14432 * config/i386/i386-c.c (ix86_target_macros_internal): Also
14433 define __IBT__ and __SHSTK__ for -fcf-protection.
14434 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
14436 (ix86_trampoline_init): Likewise.
14437 (x86_output_mi_thunk): Likewise.
14438 (ix86_notrack_prefixed_insn_p): Likewise.
14439 (ix86_option_override_internal): Don't disallow -fcf-protection.
14440 * config/i386/i386.md (rdssp<mode>): Also enable for
14442 (incssp<mode>): Likewise.
14443 (nop_endbr): Likewise.
14444 * config/i386/i386.opt (mcet): Change help message to built-in
14447 (mshstk): Likewise.
14448 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
14449 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
14450 enable CET built-in functions.
14452 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
14454 * common/config/i386/i386-common.c
14455 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
14456 OPTION_MASK_ISA_MOVDIRI_UNSET,
14457 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
14458 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
14459 * config.gcc (movdirintrin.h): New header.
14460 * config/i386/cpuid.h (bit_MOVDIRI,
14461 bit_MOVDIR64B): New bits.
14462 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
14464 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
14465 (VOID, PVOID, PCVOID)): New function types.
14466 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
14467 __builtin_ia32_directstoreu_u64,
14468 __builtin_ia32_movdir64b): New builtins.
14469 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
14470 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
14472 (ix86_valid_target_attribute_inner_p): Ditto.
14473 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
14474 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
14475 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
14476 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
14477 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
14478 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
14479 (movdiri<mode>, movdir64b_<mode>): New.
14480 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
14481 * config/i386/immintrin.h: Include movdirintrin.h.
14482 * config/i386/movdirintrin.h: New file.
14483 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
14485 2018-04-19 Richard Biener <rguenther@suse.de>
14487 PR middle-end/85455
14488 * cfg.c (clear_bb_flags): When loop state says we have
14489 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
14491 2018-04-19 Richard Biener <rguenther@suse.de>
14493 PR tree-optimization/84737
14494 * tree-vect-data-refs.c (vect_copy_ref_info): New function
14495 copying restrict info.
14496 (vect_setup_realignment): Use it.
14497 * tree-vectorizer.h (vect_copy_ref_info): Declare.
14498 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
14499 the first DR to all generated stores.
14500 (vectorizable_load): Likewise for loads.
14502 2018-04-19 Jakub Jelinek <jakub@redhat.com>
14504 PR tree-optimization/85446
14505 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
14506 the integral and pointer types to have the same precision.
14508 * doc/install.texi: Document --disable-cet being the default and
14511 2018-04-18 Martin Liska <mliska@suse.cz>
14513 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
14516 2018-04-18 Martin Liska <mliska@suse.cz>
14519 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
14522 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
14523 arguments if they are comparable.
14525 2018-04-18 Martin Liska <mliska@suse.cz>
14528 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
14531 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
14534 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
14537 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
14538 ENDBR after calling __morestack.
14540 2018-04-18 David Malcolm <dmalcolm@redhat.com>
14543 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
14544 by using gcc_base_ver to generate a gcc_driver_version, and use
14545 it when generating GCC_DRIVER_NAME.
14546 * configure: Regenerate.
14548 2018-04-18 Jakub Jelinek <jakub@redhat.com>
14551 * config.gcc: Obsolete powerpc*-*-*spe*.
14553 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14556 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
14557 (stabstr_D): Change type of unum from unsigned int to
14558 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
14561 2018-04-17 Jim Wilson <jimw@sifive.com>
14564 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
14565 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
14566 Set arg_pointer_offset after using pretend_args_size.
14568 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14570 PR rtl-optimization/85431
14571 * dse.c (record_store): Ignore zero width stores.
14574 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
14575 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
14576 __builtin_stack_restore rather than after it.
14577 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
14578 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
14579 argument instead of virtual_dynamic_stack_rtx.
14581 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
14583 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
14585 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14586 Add note to error message to explain internal mapping of overloaded
14587 built-in function name to non-overloaded built-in function name.
14588 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
14591 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
14594 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
14595 where the inputs overlap with the output.
14597 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14599 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
14600 (=v, v) alternative and explicit "memory" attribute.
14601 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
14602 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
14604 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
14605 "sselog1" type instead of "sselog".
14606 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
14607 "sselog". Remove explicit "memory" attribute.
14608 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
14609 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
14611 (vec_extract_hi_v32hi): Merge all alternatives into one, use
14612 "sselog1" type instead of "sselog". Remove explicit "memory"
14614 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
14615 use "sselog1" type instead of "sselog". Remove explicit "memory"
14617 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
14618 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
14620 (vec_extract_hi_v64qi): Merge all alternatives into one, use
14621 "sselog1" type instead of "sselog". Remove explicit "memory"
14623 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
14624 use "sselog1" type instead of "sselog". Remove explicit "memory"
14628 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
14630 PR middle-end/85414
14631 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
14634 2018-04-17 Martin Jambor <mjambor@suse.cz>
14637 * ipa-cp.c (create_specialized_node): Call
14638 expand_all_artificial_thunks if necessary.
14640 2018-04-17 Martin Liska <mliska@suse.cz>
14643 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
14644 in message, remote space in between '_G' and '('.
14646 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14649 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
14650 avx512f_vmcmp<mode>3<round_saeonly_name>,
14651 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
14652 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
14653 avx512f_rndscale<mode><round_saeonly_name>,
14654 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
14655 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
14656 Use %<iptr>2 instead of %2 for -masm=intel.
14657 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
14658 avx512f_vcvttss2usi<round_saeonly_name>,
14659 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
14661 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
14662 avx512f_vcvttsd2usi<round_saeonly_name>,
14663 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
14664 Use %q1 instead of %1 for -masm=intel.
14665 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
14666 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
14667 of %3 for -masm=intel.
14668 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
14670 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
14672 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
14674 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
14675 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
14677 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
14679 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
14680 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
14681 %g1 and one with %0 and %1.
14682 (avx512er_vmrcp28<mode><round_saeonly_name>,
14683 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
14684 %1 for -masm=intel.
14685 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
14686 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
14687 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
14688 of %0 and %{%4%} for -masm=intel.
14689 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
14690 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
14691 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
14692 order of %0 and %{%5%}%{z%} for -masm=intel.
14694 2018-04-17 Jan Hubicka <jh@suse.cz>
14697 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
14699 2018-04-17 Martin Liska <mliska@suse.cz>
14702 * multiple_target.c (create_dispatcher_calls): Set apostrophes
14703 for target_clone error message. Make default implementation
14704 clone to be a local declaration.
14705 (separate_attrs): Add new argument and check for an empty
14707 (expand_target_clones): Handle it.
14708 (ipa_target_clone): Make redirection just for target_clones
14711 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
14712 Tom de Vries <tom@codesourcery.com>
14714 PR middle-end/84955
14715 * omp-expand.c (expand_oacc_for): Add dummy false branch for
14716 tiled basic blocks without omp continue statements.
14718 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
14721 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
14722 vec_extract expression as having side effects to make sure it gets
14725 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
14728 * config/i386/i386.c (get_builtin_code_for_version): Check
14731 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
14734 * config.gcc: Support "skylake".
14735 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14737 * config/i386/i386.c (m_SKYLAKE): Define.
14738 (processor_target_table): Add "skylake".
14739 (ix86_option_override_internal): Add "skylake".
14740 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
14741 PROCESSOR_CANNONLAKE.
14742 (get_builtin_code_for_version): Fix priority for
14743 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
14744 PROCESSOR_SKYLAKE-AVX512.
14745 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
14746 (processor_type): Add PROCESSOR_SKYLAKE.
14748 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
14749 Jason Merrill <jason@redhat.com>
14752 * convert.c (convert_to_integer_1): Use direct recursion for
14753 enumeral types and types with a precision less than the number
14754 of bits in their mode.
14756 2018-04-16 Julia Koval <julia.koval@intel.com>
14759 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
14760 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
14762 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
14765 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
14766 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
14767 and -mno-direct-move.
14769 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
14772 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
14773 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
14774 Ensure negative shifts result in {0}.
14776 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
14778 PR rtl-optimization/79916
14779 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
14780 regs (if any) to define how to gnerate SD moves when LRA is in
14783 2018-04-13 Jakub Jelinek <jakub@redhat.com>
14785 PR rtl-optimization/85393
14786 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
14787 * except.c (expand_dw2_landing_pad_for_region): Make static.
14788 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
14789 a label and unconditional jump to old_bb, rather than
14790 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
14793 PR rtl-optimization/85376
14794 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
14795 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
14796 instead of a specific value.
14798 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
14799 Bin Cheng <bin.cheng@arm.com>
14801 PR tree-optimization/82965
14802 PR tree-optimization/83991
14803 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
14804 by_profile_only parameter.
14805 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
14806 information if the loop was predicted to iterate too many times.
14807 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
14809 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
14812 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
14815 2018-04-13 Martin Liska <mliska@suse.cz>
14816 Jakub Jelinek <jakub@redhat.com>
14818 PR middle-end/81657
14819 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
14820 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
14821 * builtins.c (expand_builtin_memory_copy_args): Use
14822 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
14823 handle dest_addr == pc_rtx.
14825 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
14828 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
14829 asked to not generate direct moves.
14830 (fix_trunc<mode>si2_stfiwx): Similar.
14831 (fix_trunc<mode>si2_internal): Similar.
14833 2018-04-12 Jakub Jelinek <jakub@redhat.com>
14836 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
14837 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
14838 lookup if dest in some wider mode is known to be const0_rtx and
14839 if so, record permanent equivalence for it to be ZERO_EXTEND of
14840 the narrower mode destination.
14842 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
14844 * lto-streamer-out.c (output_function): Revert 259346.
14845 * omp-expand.c (expand_oacc_for): Likewise.
14847 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
14849 PR rtl-optimization/85354
14850 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
14851 * sel-sched.c (sel_global_init): ... here.
14853 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
14856 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
14857 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
14858 mode for PE-COFF targets.
14859 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
14860 (i386_pe_asm_lto_end): Likewise.
14861 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
14862 (TARGET_ASM_LTO_END): Likewise.
14863 * config/i386/winnt.c (saved_debug_info_level): New static variable.
14864 (i386_pe_asm_lto_start): New function.
14865 (i386_pe_asm_lto_end): Likewise.
14867 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
14868 Richard Biener <rguenther@suse.de>
14870 PR middle-end/84955
14871 * lto-streamer-out.c (output_function): Fix CFG loop state before
14873 * omp-expand.c (expand_oacc_for): Handle calls to internal
14874 functions like regular functions.
14876 2018-04-12 Richard Biener <rguenther@suse.de>
14879 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
14880 for the early LTO debug to properly generate references to it
14881 during DIE emission. Do not re-use that for the skeleton for
14883 (dwarf2out_early_finish): Likewise.
14885 2018-04-12 Jakub Jelinek <jakub@redhat.com>
14888 * config/i386/sse.md
14889 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
14890 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
14891 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
14892 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
14893 and output is a reg, avoid creating invalid lowpart subreg, but
14894 instead split into a 512-bit move. Don't split if not AVX512VL,
14895 input is xmm16+ reg and output is a mem.
14896 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
14897 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
14898 xmm16+ reg and output is a mem.
14900 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14902 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
14903 also for flag_dwarf2_cfi_asm.
14905 2018-04-12 Jakub Jelinek <jakub@redhat.com>
14907 PR rtl-optimization/85342
14908 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
14909 a bool scalar var inside of the loop instead. Don't try to update
14910 recog_data.operand after failed apply_change_group.
14912 2018-04-12 Tom de Vries <tom@codesourcery.com>
14915 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
14916 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
14917 array with flexible array member as array without given dimension.
14918 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
14919 argument for undefined param to true.
14921 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
14924 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
14925 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
14926 from PowerPC section.
14927 * config/rs6000/sysv4.opt (mcall-): Improve help text.
14928 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
14929 help text that is too long.
14930 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
14931 help text that is too long.
14932 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
14933 help text that is too long.
14935 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
14937 * config/alpha/alpha.md (stack_probe_internal): Rename
14938 from "probe_stack". Update all callers.
14940 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
14942 PR rtl-optimization/84566
14943 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
14944 sched_macro_fuse_insns.
14946 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
14949 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
14950 (compute_block_dependences): ... from here.
14952 2018-04-11 Jakub Jelinek <jakub@redhat.com>
14954 PR tree-optimization/85331
14955 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
14956 from int to HOST_WIDE_INT.
14958 2018-04-11 Martin Jambor <mjambor@suse.cz>
14961 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
14962 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
14963 not the same as the source val.
14964 (cgraph_edge_brings_value_p): New parameter.
14965 (gather_edges_for_value): Pass destination value to
14966 cgraph_edge_brings_value_p.
14967 (perhaps_add_new_callers): Likewise.
14968 (get_info_about_necessary_edges): Likewise and exclude values brought
14969 only by self-recursive edges.
14970 (create_specialized_node): Redirect only clones of self-calling edges.
14971 (+self_recursive_pass_through_p): New function.
14972 (find_more_scalar_values_for_callers_subset): Use it.
14973 (find_aggregate_values_for_callers_subset): Likewise.
14974 (known_aggs_to_agg_replacement_list): Removed.
14975 (decide_whether_version_node): Re-calculate known constants for all
14976 remaining context clones.
14978 2018-04-11 Richard Biener <rguenther@suse.de>
14981 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
14982 from early DWARF output.
14983 (dwarf2out_early_finish): Output line info unconditionally into
14984 early DWARF and add reference to it.
14986 2018-04-11 Jakub Jelinek <jakub@redhat.com>
14989 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
14990 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
14991 other than V2DFmode using iptr mode attribute.
14992 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
14994 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
14996 PR rtl-optimization/84659
14997 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
14999 2018-04-11 Jakub Jelinek <jakub@redhat.com>
15002 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
15004 (output_loc_list): Pass address of a dummy size variable even in the
15005 locview handling loop.
15006 (index_location_lists): Add comment on why skip_loc_list_entry can't
15009 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
15012 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
15015 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
15018 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
15019 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
15020 and -mstring-compare-inline-limit.
15022 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
15025 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
15026 for stack clash protection in a register whenever we need it to be in
15029 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
15031 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
15032 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
15034 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
15037 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
15039 (mlong-double-): Ditto.
15040 * config/rs6000/sysv4.opt (msdata=): Ditto.
15041 (mtls-size=): Ditto.
15043 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
15045 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
15046 erroneous entries for
15047 "vector int vec_ldl (int, long int *)", and
15048 "vector unsigned int vec_ldl (int, unsigned long int *)".
15049 Add comments and entries for
15050 "vector bool char vec_ldl (int, bool char *)",
15051 "vector bool short vec_ldl (int, bool short *)",
15052 "vector bool int vec_ldl (int, bool int *)",
15053 "vector bool long long vec_ldl (int, bool long long *)",
15054 "vector pixel vec_ldl (int, pixel *)",
15055 "vector long long vec_ldl (int, long long *)",
15056 "vector unsigned long long vec_ldl (int, unsigned long long *)".
15057 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
15058 type tree bool_long_long_type_node and correct definition of
15059 bool_V2DI_type_node to make reference to this new type tree.
15060 (rs6000_mangle_type): Replace erroneous reference to
15061 bool_long_type_node with bool_long_long_type_node.
15062 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
15063 comments to emphasize sign distinctions for char and int types and
15064 replace RS6000_BTI_bool_long constant with
15065 RS6000_BTI_bool_long_long constant. Also add comment to restrict
15066 use of RS6000_BTI_pixel.
15067 (bool_long_type_node): Remove this macro definition.
15068 (bool_long_long_type_node): New macro definition
15070 2018-04-10 Jakub Jelinek <jakub@redhat.com>
15072 PR rtl-optimization/85300
15073 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
15074 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
15075 simplify_unary_operation fails.
15077 2018-04-10 Martin Liska <mliska@suse.cz>
15079 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
15080 cgraph_edge and ipa_ref.
15082 2018-04-10 Jakub Jelinek <jakub@redhat.com>
15086 * config/i386/sse.md
15087 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
15088 computation of the VEC_MERGE selector from mask.
15089 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
15090 Fix decoding of the VEC_MERGE selector into mask.
15092 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
15094 PR tree-optimization/85286
15095 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
15097 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
15099 * final.c (final_1): Set insn_last_address as well as
15100 insn_current_address.
15102 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15105 * explow.c (emit_stack_probe): Call validize_mem on memory location
15106 before passing it to gen_probe_stack. Create address operand and
15107 legitimize it for the probe_stack_address case.
15109 2018-04-09 Jan Hubicka <jh@suse.cz>
15112 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
15113 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
15114 * tree.c (free_lang_data_in_type): Fix handling of binfos;
15116 (free_lang_data): Rebuild type inheritance graph.
15118 2018-04-09 Martin Sebor <msebor@redhat.com>
15120 * invoke.texi (-finline-small-functions): Mention other optimization
15122 (-findirect-inlining, -fpartial-inlining): Same.
15123 (-finline-functions-called-once): Same.
15124 (-freorder-blocks-and-partition): Same.
15126 2018-04-09 Jan Hubicka <jh@suse.cz>
15129 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
15130 jumps; choose last target that matches the criteria (i.e.
15131 no partition changes for non-crossing jumps).
15132 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
15133 support for redirecting crossing jumps to non-crossing.
15135 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
15137 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
15138 also for naked functions.
15140 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
15142 * config/arc/arc.md (add_shift): New pattern.
15143 (add_shift2): Likewise.
15144 (sub_shift): Likewise.
15145 (sub_shift_cmp0_noout): Likewise.
15146 (compare_si_ashiftsi): Likewise.
15147 (xbfu_cmp0_noout): New combine pattern.
15148 (xbfu_cmp0"): Likewise.
15149 (movsi_set_cc_insn): Place the predicable variant first.
15150 (commutative_binary_cmp0_noout): Remove clobber.
15151 (commutative_binary_cmp0): New pattern.
15152 (noncommutative_binary_cmp0): Likewise.
15153 (noncommutative_binary_cmp0_noout): Likewise.
15154 (noncommutative_binary_comparison_result_used): Removed.
15155 (rsub_cmp0): New pattern.
15156 (rsub_cmp0_noout): Likewise.
15157 (extzvsi): Changed, keep only meaningful variants.
15158 (SQH, SEZ): New iterators.
15159 (SQH_postfix): New mode attribute.
15160 (SEZ_prefix): New code attribute.
15161 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
15162 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
15163 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
15164 of numerical value.
15165 (noncommutative_operator): Check the availability of barrel
15168 2018-04-09 Richard Biener <rguenther@suse.de>
15170 PR tree-optimization/85284
15171 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
15172 Only use the niter constraining form of simple_iv when the exit
15173 is always executed.
15175 2018-04-09 Tom de Vries <tom@codesourcery.com>
15178 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
15179 (define_expand "*memory_barrier"): New define_expand.
15180 (define_insn "memory_barrier"): New insn.
15182 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15184 PR rtl-optimization/80463
15185 PR rtl-optimization/83972
15186 PR rtl-optimization/83480
15188 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
15189 correct producer for the insn.
15190 (tidy_control_flow): Fixup seqnos in case of debug insns.
15192 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15194 PR rtl-optimization/83913
15196 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
15197 different sched-times when merging exprs.
15199 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15201 PR rtl-optimization/83962
15203 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
15204 tidy_fallthru_edge and tidy_control_flow.
15206 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15208 PR rtl-optimization/83530
15210 * sel-sched.c (force_next_insn): New global variable.
15211 (remove_insn_for_debug): When force_next_insn is true, also leave only
15212 next insn in the ready list.
15213 (sel_sched_region): When the region wasn't scheduled, make another pass
15214 over it with force_next_insn set to 1.
15216 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
15218 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
15220 * config/nds32/constants.md (unspec_volatile_element): Add enum values
15221 for interrupt control.
15222 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
15223 functions for interrupt control.
15224 * config/nds32/nds32-intrinsic.md: Likewise.
15225 * config/nds32/nds32_intrinsic.h: Likewise.
15226 * config/nds32/nds32.h (nds32_builtins): Likewise.
15228 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
15230 * config/nds32/nds32.c (nds32_init_machine_status,
15231 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
15232 strict_aligned_p field.
15233 (nds32_expand_to_rtl_hook): New function.
15234 (TARGET_EXPAND_TO_RTL_HOOK): Define.
15235 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
15237 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15238 Chung-Ju Wu <jasonwucj@gmail.com>
15240 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
15241 * config/nds32/nds32-n7.md: New file.
15242 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
15243 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
15245 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
15246 * config/nds32/nds32.md (pipeline_model): Add n7.
15247 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
15248 * config/nds32/pipelines.md: Include n7 settings.
15250 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15251 Chung-Ju Wu <jasonwucj@gmail.com>
15253 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
15254 * config/nds32/nds32-e8.md: New file.
15255 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
15256 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
15258 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
15259 * config/nds32/nds32.md (pipeline_model): Add e8.
15260 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
15261 * config/nds32/pipelines.md: Include e8 settings.
15263 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15264 Chung-Ju Wu <jasonwucj@gmail.com>
15266 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
15267 * config/nds32/nds32-n8.md: New file.
15268 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
15269 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
15271 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
15272 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
15273 * config/nds32/nds32.md (pipeline_model): Add n8.
15274 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
15275 * config/nds32/pipelines.md: Include n8 settings.
15277 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15278 Chung-Ju Wu <jasonwucj@gmail.com>
15280 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
15281 * config/nds32/nds32-n9-2r1w.md: New file.
15282 * config/nds32/nds32-n9-3r2w.md: New file.
15283 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
15284 nds32_register_ports): New or modify for cpu n9.
15285 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
15287 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
15288 * config/nds32/nds32-utils.c: New file.
15289 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
15290 TARGET_MUL_SLOW): Define.
15291 * config/nds32/nds32.md (pipeline_model): New attribute.
15292 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
15293 New options that support cpu n9.
15294 * config/nds32/pipelines.md: Include n9 settings.
15295 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
15297 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
15299 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
15300 information if necessary.
15301 (output_cond_branch_compare_zero): Likewise.
15302 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
15303 (nds32_target_alignment): Refine for alignment.
15304 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
15305 (FUNCTION_BOUNDARY): Modify.
15306 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
15308 * config/nds32/nds32.opt (malways-align, malign-functions): New.
15310 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
15312 * config/nds32/constants.md (unspec_volatile_element): Add values for
15313 TLB operation and data prefetch.
15314 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
15315 functions for TLB operation and data prefetch.
15316 * config/nds32/nds32-intrinsic.md: Likewise.
15317 * config/nds32/nds32_intrinsic.h: Likewise.
15318 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
15319 (nds32_print_operand): Likewise.
15320 * config/nds32/nds32.h (nds32_builtins): Likewise.
15322 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15323 Andrew Pinski <pinsika@gcc.gnu.org>
15325 PR middle-end/82976
15326 * match.pd: Use constant_boolean_node of correct type instead of
15327 boolean_true_node or boolean_false_node for simplifying
15328 pointer comparisons to zero.
15330 2018-04-07 Jakub Jelinek <jakub@redhat.com>
15332 PR tree-optimization/80021
15333 * tree.c (verify_type_variant): Make error call in verify_variant_match
15334 translatable and remove final full stop.
15336 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15338 * config/nds32/constants.md (unspec_volatile_element): Add
15339 UNSPEC_VOLATILE_EH_RETURN.
15340 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
15341 nds32_output_stack_pop): Support dwarf exception handling process.
15342 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
15343 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
15344 exception handling process.
15345 (nds32_compute_stack_frame): Likewise.
15346 (nds32_return_addr_rtx): Likewise.
15347 (nds32_initial_elimination_offset): Likewise.
15348 (nds32_expand_prologue): Likewise.
15349 (nds32_expand_epilogue): Likewise.
15350 (nds32_dynamic_chain_address): New function.
15351 * config/nds32/nds32.h (machine_function): Add fields for dwarf
15352 exception handling.
15353 (DYNAMIC_CHAIN_ADDRESS): Define.
15354 (EH_RETURN_DATA_REGNO): Define.
15355 (EH_RETURN_STACKADJ_RTX): Define.
15356 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
15357 patterns for dwarf exception handling.
15359 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15361 * config/nds32/nds32.h: Clean up obsolete macros.
15363 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15365 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
15366 Add enum values for particular instructions.
15367 * config/nds32/nds32-intrinsic.c: Implementation of expanding
15368 particular intrinsic functions.
15369 * config/nds32/nds32-intrinsic.md: Likewise.
15370 * config/nds32/nds32_intrinsic.h: Likewise.
15371 * config/nds32/nds32.h (nds32_builtins): Likewise.
15372 * config/nds32/nds32.md (type): Add pbsad and pbsada.
15373 (btst, ave): New patterns for particular instructions.
15375 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15377 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
15378 Add enum values for atomic load/store and memory sync.
15379 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
15381 * config/nds32/nds32-intrinsic.md: Likewise.
15382 * config/nds32/nds32_intrinsic.h: Likewise.
15383 * config/nds32/nds32.h (nds32_builtins): Likewise.
15385 2018-04-07 Jakub Jelinek <jakub@redhat.com>
15387 PR tree-optimization/85257
15388 * fold-const.c (native_encode_vector): If not all elts could fit
15389 and off is -1, return 0 rather than offset.
15390 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
15391 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
15392 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
15393 adjust buffer in native_interpret_expr call.
15395 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15397 * config/nds32/constants.md (unspec_volatile_element): Add cache
15398 control enum values.
15399 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
15400 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
15401 * config/nds32/nds32.c (nds32_cctl_names): New.
15402 (nds32_print_operand): Handle cache control register names.
15403 * config/nds32/nds32.h (nds32_builtins): New enum values.
15404 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
15406 * config/nds32/nds32.md (type): Add mmu.
15407 * config/nds32/pipelines.md (simple_insn): Add mmu.
15409 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15411 * config/nds32/nds32.md (type): Remove call.
15412 * config/nds32/pipelines.md (simple_insn): Likewise.
15414 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15416 * config/nds32/constants.md (unspec_volatile_element): Add
15417 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
15418 UNSPEC_VOLATILE_FMFCFG.
15419 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
15420 description for fmfcfg and fmfcsr.
15421 (bdesc_1arg): Add fmtcsr.
15422 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
15423 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
15424 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
15425 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
15426 unspec_fmfcfg): New patterns.
15427 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
15428 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
15429 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
15430 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
15431 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
15432 __nds32__fmfcfg): Define.
15434 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15436 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
15437 intrinsic register names.
15438 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
15439 intrinsic register enum values and macros.
15441 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15443 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
15444 for load/store addressing form.
15445 (nds32_print_operand_address): Likewise.
15447 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
15450 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
15451 based on LABEL_REF. Remove useless assertion.
15452 (pic_address_needs_scratch): Fix formatting.
15453 (sparc_legitimize_pic_address): Minor tweaks.
15454 (sparc_delegitimize_address): Adjust assertion accordingly.
15455 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
15456 into symbolic_operand.
15457 (movsi_high_pic_label_ref): Likewise.
15458 (movsi_lo_sum_pic_label_ref): Likewise.
15459 (movdi_pic_label_ref): Likewise.
15460 (movdi_high_pic_label_ref): Likewise.
15461 (movdi_lo_sum_pic_label_ref): Likewise.
15463 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
15465 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
15466 custom LIB_SPEC setup.
15468 2018-04-06 Ruslan Bukin <br@bsdpad.com>
15469 Kito Cheng <kito.cheng@gmail.com>
15471 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
15472 * config/riscv/freebsd.h: New.
15474 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
15476 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
15477 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
15480 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
15481 Kito Cheng <kito.cheng@gmail.com>
15483 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
15484 nds32_output_call, nds32_symbol_binds_local_p): New functions.
15485 * config/nds32/nds32-protos.h (nds32_output_call,
15486 nds32_output_return): Declare.
15487 * config/nds32/nds32.md: Refine all the call and return patterns.
15489 2018-04-06 Jakub Jelinek <jakub@redhat.com>
15492 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
15493 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
15495 PR rtl-optimization/84872
15496 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
15497 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
15498 EDGE_CROSSING edge.
15500 2018-04-06 Tamar Christina <tamar.christina@arm.com>
15502 * expr.c (copy_blkmode_to_reg): Revert 254862.
15503 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
15505 2018-04-06 Richard Biener <rguenther@suse.de>
15507 PR middle-end/85244
15508 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
15509 after seeing a component reference with an adjacent field. Treat
15510 refs to arrays at struct end of external decls similar to
15511 refs to unconstrained commons.
15513 2018-04-06 Jakub Jelinek <jakub@redhat.com>
15516 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
15517 look through SAVE_EXPRs with non-side-effects argument. Adjust
15519 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
15522 2018-04-06 Richard Biener <rguenther@suse.de>
15524 PR middle-end/85180
15525 * alias.c (find_base_term): New wrapper around find_base_term
15526 unwinding CSELIB_VAL_PTR changes.
15527 (find_base_term): Do not restore CSELIB_VAL_PTR during the
15530 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15532 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
15534 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
15535 constant definitions.
15536 ("nop"): lr 0,0 -> nopr r0
15537 ("nop_lr0", "nop_lr1"): New insn definitions.
15539 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
15541 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
15542 NDS32_V3PUSH_AVAILABLE_P macro.
15544 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
15545 Chung-Ju Wu <jasonwucj@gmail.com>
15547 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
15548 (nds32*-*-*): Add float and fpu_config into supported_defaults.
15549 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
15550 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
15551 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
15552 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
15553 * config/nds32/constraints.md: New constraints and checking for hard
15554 float configuration.
15555 * config/nds32/iterators.md: New mode iterator and attribute for hard
15556 float configuration.
15557 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
15559 * config/nds32/nds32-fpu.md: New file.
15560 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
15561 deal with hard float code generation.
15562 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
15564 (abi_type, float_reg_number): New enum type.
15565 * config/nds32/nds32-predicates.c: New predicates for hard float.
15566 * config/nds32/nds32-protos.h: Declare functions for hard float.
15567 * config/nds32/nds32.c: Implementation for hard float configuration.
15568 * config/nds32/nds32.h: Definitions for hard float configuration.
15569 * config/nds32/nds32.md: Include hard float machine description and
15570 modify patterns for hard float configuration.
15571 * config/nds32/nds32.opt: New options for hard float configuration.
15572 * config/nds32/predicates.md: New predicates for hard float
15575 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
15577 * common/config/nds32/nds32-common.c
15578 (nds32_option_optimization_table): Enable -mreleax-hint by default.
15580 2018-04-05 Jakub Jelinek <jakub@redhat.com>
15582 PR middle-end/85195
15583 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
15584 CONSTRUCTOR_ELT (ctor, ...)->value.
15586 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
15589 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
15591 2018-04-05 Tom de Vries <tom@codesourcery.com>
15594 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
15597 2018-04-05 Shiva Chen <shiva0217@gmail.com>
15598 Kito Cheng <kito.cheng@gmail.com>
15600 * config/nds32/constraints.md (U33): Fine-tune checking condition.
15601 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
15602 * config/nds32/nds32.h (nds32_16bit_address_type): Add
15603 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
15605 2018-04-05 Shiva Chen <shiva0217@gmail.com>
15606 Kito Cheng <kito.cheng@gmail.com>
15608 * config/nds32/constraints.md (Ufe): New memory constraint.
15609 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
15610 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
15611 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
15613 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
15614 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
15616 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15618 * config/nds32/nds32.md: Use optimize_size in the condition for
15619 alu-shift instructions.
15621 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15623 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
15625 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15627 * config/nds32/nds32.md (negsi2): Refine pattern.
15629 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
15630 Chung-Ju Wu <jasonwucj@gmail.com>
15632 * config/nds32/iterators.md (shift_rotate): New code iterator.
15633 (shift): New code attribute.
15634 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
15635 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
15636 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
15637 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
15638 bit-wise operations.
15639 (andsi3, *andsi3): Ditto.
15640 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
15641 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
15642 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
15643 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
15644 nds32_ior_operand, nds32_xor_operand): New predicates.
15646 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15648 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
15649 (addsi3, subsi3): ... this.
15651 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15653 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
15655 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15657 * config/nds32/nds32.md: Adjust indention.
15659 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
15661 * config/nds32/nds32.md (feature): New attribute.
15663 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15665 * config/nds32/nds32.md (subtype): New attribute.
15667 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15670 * config/arm/arm-builtins.c (arm_expand_builtin): Change
15671 expansion to perform a bitwise AND of the argument followed by a
15672 boolean negation of the result.
15674 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
15676 PR rtl-optimization/84878
15677 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
15678 the basic block. Assert the use reference is not artificial and that
15679 it has an associated insn.
15681 2018-04-04 Michael Matz <matz@suse.de>
15683 * builtins.c (compute_objsize): Pass correct operand
15684 to array_at_struct_end_p.
15686 2018-04-04 Richard Biener <rguenther@suse.de>
15689 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
15690 from contexts for DINFO_LEVEL_TERSE and below.
15692 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
15694 * config/nds32/nds32-doubleword.md (move_<mode>): Require
15695 resiter_operand condition.
15696 * config/nds32/nds32.md (*move<mode>): Ditto.
15698 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
15699 Monk Chiang <sh.chiang04@gmail.com>
15701 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
15703 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15705 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
15707 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15708 Kito Cheng <kito.cheng@gmail.com>
15710 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
15711 nds32_cond_code_str, output_cond_branch,
15712 output_cond_branch_compare_zero, nds32_expand_cbranch,
15713 nds32_expand_cstore, nds32_expand_movcc,
15714 nds32_output_cbranchsi4_equality_zero,
15715 nds32_output_cbranchsi4_equality_reg,
15716 nds32_output_cbranchsi4_equality_reg_or_const_int,
15717 nds32_output_cbranchsi4_greater_less_zero: New functions.
15718 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
15719 nds32_expand_cstore, nds32_expand_movcc,
15720 nds32_output_cbranchsi4_equality_zero,
15721 nds32_output_cbranchsi4_equality_reg,
15722 nds32_output_cbranchsi4_equality_reg_or_const_int,
15723 nds32_output_cbranchsi4_greater_less_zero): Declare.
15724 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
15725 nds32_rimm11s_operand): New predicates.
15726 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
15727 * config/nds32/nds32.md: Rewrite all the branch and conditional move
15730 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
15732 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
15733 * config/nds32/nds32.md: Ditto.
15734 * config/nds32/pipelines.md: Ditto.
15736 2018-04-04 Richard Biener <rguenther@suse.de>
15738 PR tree-optimization/85168
15739 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
15740 propagating abnormals.
15742 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15744 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
15746 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15747 Kito Cheng <kito.cheng@gmail.com>
15749 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
15750 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
15751 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
15752 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
15753 * config/nds32/nds32.md (sibcall_internal): New.
15754 (sibcall_register): Remove.
15755 (sibcall_immediate): Remove.
15756 (sibcall_value_internal): New.
15757 (sibcall_value_register): Remove.
15758 (sibcall_value_immediate): Remove.
15759 * config/nds32/predicates.md (nds32_general_register_operand): New.
15760 (nds32_call_address_operand): New.
15762 2018-04-03 Jakub Jelinek <jakub@redhat.com>
15764 PR rtl-optimization/85167
15765 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
15766 bb_defs if *split_p, instead preinitialize it to NULL.
15768 PR tree-optimization/85156
15769 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
15770 evaluating the argument multiple times.
15772 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
15774 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
15776 (_mm_cvtpd_ps): Likewise.
15777 (_mm_cvttpd_epi32): Likewise.
15778 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
15779 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
15780 vector, pixel, and bool following altivec.h include.
15782 2018-04-03 Martin Sebor <msebor@redhat.com>
15784 * doc/extend.texi (Common Function Attributes): Clarify.
15785 (const attribute): Likewise.
15786 (pure attribute): Likewise.
15788 2018-04-03 Jakub Jelinek <jakub@redhat.com>
15791 * config/i386/i386.c (ix86_expand_vector_set): Use
15792 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
15794 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
15796 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
15797 instructions when changing rounding bits to preserve precision bits
15798 in the x87 control word.
15800 2018-04-03 Martin Liska <mliska@suse.cz>
15802 PR tree-optimization/82491
15803 * rtl.h (strip_offset_and_add): Replace += suboffset with
15804 poly_uint64 () + suboffset.
15806 2018-03-29 Martin Liska <mliska@suse.cz>
15807 Martin Jambor <mjambor@suse.cz>
15810 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
15811 param_type is not an integral or pointer type.
15813 2018-04-03 Richard Biener <rguenther@suse.de>
15815 * sese.h (recompute_all_dominators): Remove.
15817 2018-04-02 Martin Sebor <msebor@redhat.com>
15819 * doc/invoke.texi (-Wrestrict): Fix typos.
15821 2018-04-02 Jim Wilson <jimw@sifive.com>
15823 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
15824 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
15825 (<optab>di3, <optab>si3_extend): Likewise.
15826 (<optab>si3_mask, <optab>si3_mask_1): New.
15827 (<optab>di3_mask, <optab>di3_mask_1): New.
15828 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
15829 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
15830 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
15832 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
15834 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
15837 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
15839 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
15840 (nds32_canonicalize_comparison): New function.
15842 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
15843 Kito Cheng <kito.cheng@gmail.com>
15844 Kuan-Lin Chen <kuanlinchentw@gmail.com>
15846 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
15847 * config/nds32/constants.md (unspec_volatile_element): Add
15848 UNSPEC_VOLATILE_RELAX_GROUP.
15849 * config/nds32/nds32-relax-opt.c: New file.
15850 * config/nds32/nds32-predicates.c
15851 (nds32_symbol_load_store_p): New function.
15852 * config/nds32/nds32-protos.h
15853 (nds32_symbol_load_store_p): Declare function.
15854 (make_pass_nds32_relax_opt): Declare new rtl pass function.
15855 * config/nds32/nds32.c
15856 (nds32_register_pass): New function to register pass.
15857 (nds32_register_passes): New function to register passes.
15858 * config/nds32/nds32.md (relax_group): New pattern.
15859 * config/nds32/nds32.opt (mrelax-hint): New option.
15860 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
15862 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
15864 * config/nds32/t-nds32: Modify files dependency.
15866 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
15868 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
15869 (PROFILE_HOOK): Define its implementation.
15871 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
15873 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
15874 type and 32-bit size.
15876 2018-04-01 Jakub Jelinek <jakub@redhat.com>
15878 PR middle-end/85090
15879 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
15880 (V_128_256): New mode iterator.
15881 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
15882 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
15883 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
15885 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
15886 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
15888 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
15891 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
15892 NaN inputs correctly.
15894 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
15897 * config/rs6000/vsx.md (??r): New mode attribute.
15898 (*vsx_mov<mode>_64bit): Use it.
15899 (*vsx_mov<mode>_32bit): Likewise.
15901 2018-03-30 Martin Sebor <msebor@redhat.com>
15903 PR tree-optimization/84818
15904 * builtins.c (check_access): Use warning_n.
15906 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15909 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
15911 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
15914 2018-03-30 Julia Koval <julia.koval@intel.com>
15917 * x86-tune.def (movx, partial_reg_dependency): Enable for
15920 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
15922 PR inline-asm/84985
15923 * lra-constraints.c (process_alt_operands): Move setting
15924 this_alternative_matches below.
15926 2018-03-29 Martin Liska <mliska@suse.cz>
15929 * doc/invoke.texi: Document how LTO works with debug info.
15930 Describe auto-load support of binutils. Mention 'x86-64'
15931 as valid option value of -march option.
15933 2018-03-29 Jakub Jelinek <jakub@redhat.com>
15935 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
15938 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
15939 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
15940 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
15943 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
15946 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
15947 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
15948 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
15949 for RS6000_BTM_POWERPC64.
15950 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
15951 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
15952 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
15955 (DIVDEU): Likewise.
15957 2018-03-28 Carl Love <cel@us.ibm.com>
15960 2017-09-27 Carl Love <cel@us.ibm.com>
15962 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
15963 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
15964 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
15967 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15969 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
15970 instead of __vector bool.
15971 (_mm_max_pu8): Likewise.
15972 (_mm_min_pi16): Likewise.
15974 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
15977 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
15978 (DIVWEUO): Likewise.
15979 (DIVDEO): Likewise.
15980 (DIVDEUO): Likewise.
15981 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
15982 DIVWEUO and DIVDEUO.
15983 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
15984 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
15985 (div_extend): Likewise.
15986 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
15988 (__builtin_divweuo): Likewise.
15989 (__builtin_divdeo): Likewise.
15990 (__builtin_divdeuo): Likewise.
15992 2018-03-28 Jakub Jelinek <jakub@redhat.com>
15995 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
15996 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
15998 PR tree-optimization/82004
15999 * gimple-match-head.c (optimize_pow_to_exp): New function.
16000 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
16001 Don't fold to exp if optimize_pow_to_exp is false.
16003 2018-03-28 Martin Liska <mliska@suse.cz>
16006 * calls.c (initialize_argument_information): Fix trailing space.
16007 * common.opt: Fix typo and provide better explanation for
16008 -fsanitize-coverage option.
16009 * config/i386/i386.opt: Fix typo.
16011 2018-03-28 Jakub Jelinek <jakub@redhat.com>
16012 Martin Liska <mliska@suse.cz>
16015 * gimplify.c (asan_poison_variable): Don't do the check for
16016 gimplify_omp_ctxp here.
16017 (gimplify_decl_expr): Do it here.
16018 (gimplify_target_expr): Likewise.
16020 2018-03-28 Martin Liska <mliska@suse.cz>
16023 * config/i386/i386.c (ix86_function_arg_advance): Do not call
16024 chkp_type_bounds_count if MPX is not enabled.
16026 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
16028 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
16030 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
16033 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
16034 function to create the function decl for complex long double
16035 multiply and divide for -mabi=ieeelongdouble.
16036 (init_float128_ieee): Call it.
16038 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
16041 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
16042 -fcf-protection=branch -mibt.
16043 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
16045 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16048 * config/arm/arm.c (arm_valid_symbolic_address): Handle
16049 arm_word_relocations.
16051 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
16054 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
16055 extern array declarations.
16057 2018-03-27 Richard Biener <rguenther@suse.de>
16059 PR middle-end/84067
16060 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
16061 explicit single_use checks.
16063 2018-03-27 Richard Biener <rguenther@suse.de>
16065 PR tree-optimization/85082
16066 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
16069 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16071 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
16072 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
16073 Turn on fasynchronous-unwind-tables and funwind-tables.
16075 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
16078 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
16079 (*bmi_blsr_<mode>_ccz): Ditto.
16081 2018-03-26 Tom de Vries <tom@codesourcery.com>
16083 PR tree-optimization/85063
16084 * omp-general.c (offloading_function_p): New function. Factor out
16086 * omp-offload.c (pass_omp_target_link::gate): ... here.
16087 * omp-general.h (offloading_function_p): Declare.
16088 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
16089 with attribute omp declare target for offloading functions.
16091 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
16093 PR tree-optimization/84005
16094 * tree-data-ref.h (get_base_for_alignment): Declare.
16095 * tree-data-ref.c (get_base_for_alignment_1): New function.
16096 (get_base_for_alignment): Likewise.
16097 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
16098 get_base_for_alignment to find a suitable base object, instead
16099 of always using drb->base_address.
16101 2018-03-23 Jakub Jelinek <jakub@redhat.com>
16103 PR inline-asm/85022
16104 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
16105 known size by default.
16107 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
16109 PR inline-asm/85030
16110 * lra-constraints.c (process_alt_operands): Don't match BLKmode
16111 and non BLKmode operands.
16113 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16116 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
16117 Clean up attributes.
16119 2018-03-23 Richard Biener <rguenther@suse.de>
16122 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
16123 we are going to emit early debug for LTO.
16125 2018-03-23 Jakub Jelinek <jakub@redhat.com>
16127 PR inline-asm/85034
16128 * function.c (match_asm_constraints_1): Don't optimize if input
16129 doesn't satisfy general_operand predicate for output's mode.
16131 PR inline-asm/85022
16132 * alias.c (write_dependence_p): Don't require for x_canonicalized
16133 non-VOIDmode if x has VOIDmode.
16136 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
16137 just don't try to optimize it rather than assert it never happens.
16139 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
16141 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
16142 macro expansions for definition of ST_INTERNAL_<mode> and
16143 LD_INTERNAL_<mode> builtins.
16144 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
16146 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
16148 (altivec_expand_st_builtin): Likewise.
16149 (altivec_expand_builtin): Remove calls to deleted functions.
16150 (rs6000_address_for_altivec): Delete this function.
16151 * config/rs6000/vector.md: Remove expands for
16152 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
16154 2018-03-22 Sudakshina Das <sudi.das@arm.com>
16157 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
16158 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
16159 re-computing once computed.
16160 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
16161 (arm_init_machine_status): Initialize
16162 machine->static_chain_stack_bytes.
16164 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
16167 * doc/extend.texi: Add four new prototypes for vec_ld.
16168 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
16169 definitions for more logical presentation.
16170 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
16171 entries for V1TI variants of __builtin_altivec_ld builtin.
16172 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
16173 handling of V1TI variant of LVX icode pattern.
16174 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
16175 (rs6000_gimple_fold_builtin): Likewise.
16176 (altivec_init_builtins): Add code to define
16177 __builtin_altivec_lvx_v1ti function.
16179 2018-03-22 Jakub Jelinek <jakub@redhat.com>
16181 PR inline-asm/84941
16182 * function.c (match_asm_constraints_1): Don't do the optimization
16183 if input isn't a REG, SUBREG, MEM or constant.
16185 2018-03-22 Tom de Vries <tom@codesourcery.com>
16187 PR tree-optimization/84956
16188 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
16189 bb_has_abnormal_pred.
16191 2018-03-22 Jakub Jelinek <jakub@redhat.com>
16194 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
16195 DECL_INITIAL (decl) to decl at the end.
16196 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
16197 adjust the comment.
16199 2018-03-21 Joseph Myers <joseph@codesourcery.com>
16201 * doc/extend.texi (__builtin_tgmath): Document when complex
16202 integer types are treated as _Complex _Float64.
16204 2018-03-21 Tom de Vries <tom@codesourcery.com>
16206 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
16208 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16210 PR tree-optimization/84960
16211 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
16212 if it is ENTRY block, move them into single succ of ENTRY in that case.
16214 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
16216 PR tree-optimization/84811
16217 * poly-int.h (poly_span_traits): Remove the T3 parameter and
16218 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
16219 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
16220 (known_subrange_p): Update accordingly. Cast each value involved
16221 in the size comparison, rather than casting the result of the
16224 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16226 PR tree-optimization/84982
16227 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
16228 by flipping the least significant bit rather than all bits from
16229 bitpos to bitpos + bitsize - 1.
16231 2018-03-21 Nathan Sidwell <nathan@acm.org>
16233 * doc/extend.texi (Deprecated Features): Remove mention of
16234 long-deleted deprecations.
16236 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16239 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
16240 * configure: Regenerate.
16242 2018-03-21 Tom de Vries <tom@codesourcery.com>
16244 PR tree-optimization/83126
16245 * tree-parloops.c (num_phis): New function.
16246 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
16248 2018-03-21 Nathan Sidwell <nathan@acm.org>
16250 * doc/extend.texi (Deprecated Features): Update deprecated flags,
16251 mention anon-struct/union members and trailing attributes.
16253 2018-03-21 Bin Cheng <bin.cheng@arm.com>
16255 PR tree-optimization/84969
16256 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
16257 builtin memset partitions if they set different rhs values.
16259 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16261 PR rtl-optimization/84989
16262 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
16263 VEC_DUPLICATE with scalar result mode.
16265 2018-03-21 Martin Liska <mliska@suse.cz>
16268 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
16269 not intended return statement.
16271 2018-03-21 Martin Liska <mliska@suse.cz>
16274 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
16275 (chkp_find_bound_slots_1): Limit number of iterations.
16277 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
16280 * Minor grammar fixes for x86 options.
16282 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16285 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
16286 holding REG_CFA_RESTORE notes, instead turn them into a USE.
16288 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
16291 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
16292 (altivec_lvx_<mode>_1op): Likewise.
16293 (altivec_stvx_<mode>_2op): Likewise.
16294 (altivec_stvx_<mode>_1op): Likewise.
16295 (altivec_lvx_<VM2:mode>): New define_expand.
16296 (altivec_stvx_<VM2:mode>): Likewise.
16297 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
16298 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
16299 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
16300 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
16301 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
16302 (rs6000_gen_lvx): Likewise.
16303 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
16304 (altivec_expand_stv_builtin): Likewise.
16305 (altivec_expand_builtin): Likewise.
16306 * config/rs6000/vector.md: Likewise.
16308 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16311 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
16314 2018-03-20 Richard Biener <rguenther@suse.de>
16317 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
16318 sign-conversions as zero, fall back to standard scalar_stmt
16321 2018-03-20 Martin Liska <mliska@suse.cz>
16324 * predict.c (rebuild_frequencies): Handle case when we have
16325 PROFILE_ABSENT, but flag_guess_branch_prob is false.
16327 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16330 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
16331 flag_section_anchors.
16332 * varasm.c (use_blocks_for_decl_p): Remove hack for
16333 dw2_force_const_mem.
16336 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
16338 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
16339 be created, use lowpart_subreg of operands[0] rather than operands[0]
16341 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
16342 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
16343 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
16344 and n constraint instead of aarch64_shift_imm_di and Usd.
16345 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
16346 (*aarch64_<optab>_reg_minus<mode>3): ... this.
16348 2018-03-20 Sudakshina Das <sudi.das@arm.com>
16351 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
16352 to favor GPR over NEON registers.
16353 (<shift>di3_neon): Likewise.
16355 2018-03-20 Tom de Vries <tom@codesourcery.com>
16358 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
16359 (nvptx_process_pars): Emit bar.sync asap and alap.
16361 2018-03-20 Tom de Vries <tom@codesourcery.com>
16364 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
16365 seen_label if seen_label is already set.
16367 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16370 * config/i386/i386.c (fold_builtin_cpu): For features above 31
16371 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
16372 Use 1U instead of 1. Formatting fixes.
16375 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
16376 instead of TREE_TYPE (s1) for the return value.
16378 2018-03-19 Jakub Jelinek <jakub@redhat.com>
16380 PR tree-optimization/84946
16381 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
16382 bitsize + bitsize in poly_uint64 rather than poly_int64.
16385 * dwarf2asm.c: Include fold-const.c.
16386 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
16387 of decl rather than decl itself.
16389 PR rtl-optimization/84643
16390 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
16392 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
16395 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
16396 calling assemble_variable.
16398 2018-03-19 Sudakshina Das <sudi.das@arm.com>
16401 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
16402 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
16404 2018-03-19 Jim Wilson <jimw@sifive.com>
16407 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
16408 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
16409 (riscv_first_stack_step): Likewise.
16410 (riscv_option_override): Use STACK_BOUNDARY instead of
16411 MIN_STACK_BOUNDARY.
16412 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
16413 MIN_STACK_BOUNDARY.
16414 (BIGGEST_ALIGNMENT): Set to 128.
16415 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
16416 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
16419 2018-03-19 Richard Biener <rguenther@suse.de>
16421 PR tree-optimization/84933
16422 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
16423 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
16425 2018-03-19 Richard Biener <rguenther@suse.de>
16427 PR tree-optimization/84859
16428 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
16429 (cond_if_else_store_replacement): Perform sinking operation on
16430 single-store BBs regardless of MAX_STORES_TO_SINK setting.
16431 Generalize what a BB with a single eligible store is.
16433 2018-03-19 Richard Biener <rguenther@suse.de>
16435 PR tree-optimization/84929
16436 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
16437 chrec_is_positive against non-chrec arg.
16439 2018-03-19 Tamar Christina <tamar.christina@arm.com>
16442 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
16444 2018-03-18 Martin Liska <mliska@suse.cz>
16446 PR rtl-optimization/84635
16447 * regrename.c (build_def_use): Use matches_mode only when
16450 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
16452 PR tree-optimization/84913
16453 * tree-vect-loop.c (vectorizable_reduction): Don't try to
16454 vectorize chains of COND_EXPRs.
16456 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
16458 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
16460 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
16462 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
16464 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
16466 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
16468 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
16469 Kito Cheng <kito.cheng@gmail.com>
16471 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
16472 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
16473 (nds32_adjust_reg_alloc_order): New function.
16474 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
16476 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
16478 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
16479 nds32_print_operand, nds32_print_operand_address): Use
16480 HOST_WIDE_INT_PRINT_DEC instead.
16482 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
16484 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
16486 2018-03-17 Jakub Jelinek <jakub@redhat.com>
16489 * config/i386/i386.c (initial_ix86_tune_features,
16490 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
16491 unsigned long long.
16492 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
16493 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
16494 rather than 1u << ix86_tune. Formatting fix.
16495 (ix86_option_override_internal): Change ix86_arch_mask from
16496 unsigned int to unsigned HOST_WIDE_INT, initialize to
16497 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
16498 (ix86_function_specific_restore): Likewise.
16500 2018-03-16 Jakub Jelinek <jakub@redhat.com>
16503 * postreload.c (reload_combine_recognize_pattern): Perform
16504 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
16505 truncate_int_for_mode the result for the destination's mode.
16508 * hsa-gen.c (mem_type_for_type): Fix comment typo.
16509 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
16511 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
16514 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
16517 * lra-assigns.c (lra_split_hard_reg_for): Don't use
16518 regno_allocno_class_array and sorted_pseudos.
16519 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
16520 insns where regno is used.
16522 2018-03-16 Martin Liska <mliska@suse.cz>
16525 * multiple_target.c (create_dispatcher_calls): Redirect
16526 reference in the symbol table.
16528 2018-03-16 Martin Liska <mliska@suse.cz>
16531 * multiple_target.c (create_dispatcher_calls): Redirect also
16534 2018-03-16 Jakub Jelinek <jakub@redhat.com>
16538 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
16539 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
16540 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
16542 2018-03-16 Julia Koval <julia.koval@intel.com>
16544 * doc/invoke.texi (Skylake Server): Add CLWB.
16545 Cannonlake): Remove CLWB.
16547 2018-03-16 Jakub Jelinek <jakub@redhat.com>
16549 PR tree-optimization/84841
16550 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
16552 (FLOAT_ONE_CONST_TYPE): Define.
16553 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
16554 (sort_by_operand_rank): Put entries with higher constant_type last
16555 rather than first to match comments.
16557 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
16559 * config/nios2/nios2.md (movsi_internal): Fix thinko in
16562 2018-03-15 Jakub Jelinek <jakub@redhat.com>
16565 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
16566 check and use address of target always.
16568 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
16571 * config/i386/i386.c (indirect_thunk_needed): Update comments.
16572 (indirect_thunk_bnd_needed): Likewise.
16573 (indirect_thunks_used): Likewise.
16574 (indirect_thunks_bnd_used): Likewise.
16575 (indirect_return_needed): New.
16576 (indirect_return_bnd_needed): Likewise.
16577 (output_indirect_thunk_function): Add a bool argument for
16579 (output_indirect_thunk_function): Don't generate alias for
16580 function return thunk.
16581 (ix86_code_end): Call output_indirect_thunk_function to generate
16582 function return thunks.
16583 (ix86_output_function_return): Set indirect_return_bnd_needed
16584 and indirect_return_needed instead of indirect_thunk_bnd_needed
16585 and indirect_thunk_needed.
16587 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
16589 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
16590 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
16591 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
16593 2018-03-15 David Malcolm <dmalcolm@redhat.com>
16594 Paul Hua <paul.hua.gm@gmail.com>
16597 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
16599 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
16601 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
16602 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
16603 resp. SFmode cases.
16605 2018-03-15 Tamar Christina <tamar.christina@arm.com>
16608 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
16609 instead of GET_MODE_SIZE when comparing Units.
16611 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
16614 * varasm.c (hash_section): Return an unchangeble hash value
16615 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
16616 Return !aarch64_can_use_per_function_literal_pools_p ().
16618 2018-03-15 Jakub Jelinek <jakub@redhat.com>
16621 * optabs.c (emit_conditional_move): Pass address of cmode's copy
16622 rather than address of cmode as last argument to prepare_cmp_insn.
16624 2018-03-15 Julia Koval <julia.koval@intel.com>
16626 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
16627 F_AVX512VNNI, F_AVX512BITALG): New.
16629 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
16632 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
16633 insn for floating-point loads and stores.
16635 2018-03-14 Carl Love <cel@us.ibm.com>
16637 * config/rs6000/rs6000-c.c: Add macro definitions for
16638 ALTIVEC_BUILTIN_VEC_PERMXOR.
16639 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
16640 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
16641 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
16642 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
16644 * config/doc/extend.texi: Add prototypes for vec_permxor.
16646 2018-03-14 David Malcolm <dmalcolm@redhat.com>
16649 * diagnostic-show-locus.c (class layout_point): Convert m_line
16650 from int to linenum_type.
16651 (line_span::comparator): Use linenum "compare" function when
16652 comparing line numbers.
16653 (test_line_span): New function.
16654 (layout_range::contains_point): Convert param "row" from int to
16656 (layout_range::intersects_line_p): Likewise.
16657 (layout::will_show_line_p): Likewise.
16658 (layout::print_source_line): Likewise.
16659 (layout::should_print_annotation_line_p): Likewise.
16660 (layout::print_annotation_line): Likewise.
16661 (layout::print_leading_fixits): Likewise.
16662 (layout::annotation_line_showed_range_p): Likewise.
16663 (struct line_corrections): Likewise for field m_row.
16664 (line_corrections::line_corrections): Likewise for param "row".
16665 (layout::print_trailing_fixits): Likewise.
16666 (layout::get_state_at_point): Likewise.
16667 (layout::get_x_bound_for_row): Likewise.
16668 (layout::print_line): Likewise.
16669 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
16670 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
16671 * input.c (selftest::test_linenum_comparisons): New function.
16672 (selftest::input_c_tests): Call it.
16673 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
16674 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
16675 * selftest.h (ASSERT_GT): New macro.
16676 (ASSERT_GT_AT): New macro.
16677 (ASSERT_LT): New macro.
16678 (ASSERT_LT_AT): New macro.
16680 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
16682 PR rtl-optimization/84780
16683 * combine.c (distribute_links): Don't make a link based on pc_rtx.
16685 2018-03-14 Martin Liska <mliska@suse.cz>
16687 * tree.c (record_node_allocation_statistics): Use
16688 get_stats_node_kind.
16689 (get_stats_node_kind): New function extracted from
16690 record_node_allocation_statistics.
16691 (free_node): Use get_stats_node_kind.
16693 2018-03-14 Richard Biener <rguenther@suse.de>
16695 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
16696 that the value-set of ANTIC_IN doesn't grow.
16699 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
16701 (BB_VISITED_WITH_VISITED_SUCCS): New define.
16702 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
16704 2018-03-14 Julia Koval <julia.koval@intel.com>
16706 * config.gcc (icelake-client, icelake-server): New.
16708 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
16709 (initial_ix86_arch_features): Ditto.
16710 (PTA_SKYLAKE): Add SGX.
16711 (PTA_ICELAKE): Remove.
16712 (PTA_ICELAKE_CLIENT): New.
16713 (PTA_ICELAKE_SERVER): New.
16714 (ix86_option_override_internal): Split up icelake on icelake client and
16716 (get_builtin_code_for_version): Ditto.
16717 (fold_builtin_cpu): Ditto.
16718 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
16719 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
16720 * config/i386/i386.h (processor_type): Ditto.
16721 * doc/invoke.texi: Ditto.
16723 2018-03-14 Jakub Jelinek <jakub@redhat.com>
16726 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
16727 INTEGER_CST offset, add it together with bitpos / 8 and
16728 sign extend based on POINTER_SIZE.
16732 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
16735 * config/i386/constraints.md (Yc): New register constraint.
16736 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
16737 Use Yc constraint for alternative 2 of operand 0. Remove
16738 preferred_for_speed attribute.
16740 2018-03-14 Richard Biener <rguenther@suse.de>
16742 PR tree-optimization/84830
16743 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
16744 with the old one to avoid oscillations.
16746 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
16749 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
16751 (assign_by_spills): Return a flag of reload assignment failure.
16752 Do not process the reload assignment failures. Do not spill other
16753 reload pseudos if they has the same reg class. Update n if
16755 (lra_assign): Add a return arg. Set up from the result of
16756 assign_by_spills call.
16757 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
16758 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
16759 usage_insns if it is not NULL.
16760 (spill_hard_reg_in_range): New function.
16761 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
16762 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
16763 function prototypes.
16764 (lra_assign): Change prototype.
16765 * lra.c (lra): Add code to deal with fails by splitting hard reg
16768 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
16770 * config/riscv/riscv.opt (mrelax): New option.
16771 * config/riscv/riscv.c (riscv_file_start): Emit ".option
16772 "norelax" when riscv_mrelax is disabled.
16773 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
16775 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16778 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
16779 reassociation for int modes.
16781 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
16783 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
16784 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
16786 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
16787 * config/aarch64/aarch64-sve.md
16788 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
16789 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
16790 (*extend<mode><Vwide>2): Rename to...
16791 (aarch64_sve_extend<mode><Vwide>2): ...this.
16792 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
16793 renaming the old pattern to...
16794 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
16796 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
16797 define_expand, renaming the old pattern to...
16798 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
16799 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
16800 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
16801 account when deciding which SVE instruction the optab should use.
16802 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
16804 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
16806 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
16807 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
16808 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
16809 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
16810 (tlsdesc_small_<mode>): Turn a define_expand and use
16811 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
16812 (tlsdesc_small_advsimd_<mode>): ...this.
16813 (tlsdesc_small_sve_<mode>): New pattern.
16815 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
16817 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
16818 (UNSPEC_UMUL_HIGHPART): New constants.
16819 (MUL_HIGHPART): New int iteraor.
16820 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
16821 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
16823 (*<su>mul<mode>3_highpart): New define_insn.
16825 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
16828 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
16831 2018-03-13 Martin Liska <mliska@suse.cz>
16834 * (sem_item_optimizer::sem_item_optimizer): Initialize new
16836 (sem_item_optimizer::~sem_item_optimizer): Release it.
16837 (sem_item_optimizer::merge_classes): Register variable aliases.
16838 (sem_item_optimizer::fixup_pt_set): New function.
16839 (sem_item_optimizer::fixup_points_to_sets): Likewise.
16840 * ipa-icf.h: Declare new variables and functions.
16842 2018-03-13 Jakub Jelinek <jakub@redhat.com>
16844 PR middle-end/84834
16845 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
16846 integer_pow2p@2 and test integer_pow2p in condition.
16847 (A < 0 ? C : 0): Similarly for @1.
16849 PR middle-end/84831
16850 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
16851 characters starting at p contain '\0' character, don't look beyond
16855 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
16856 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
16859 * reg-stack.c (change_stack): Change update_end var from int to
16860 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
16861 also call set_block_for_insn on the newly added insns and rescan.
16864 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
16865 on the last operand.
16868 * tree.c (stabilize_reference_1): Return save_expr (e) for
16869 STATEMENT_LIST even if it doesn't have side-effects.
16871 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
16873 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
16875 2018-03-12 Renlin Li <renlin.li@arm.com>
16877 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
16878 aarch64_output_scalar_simd_mov_immediate.
16880 2018-03-12 Martin Sebor <msebor@redhat.com>
16882 PR tree-optimization/83456
16883 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
16884 for perfectly overlapping calls to memcpy.
16885 (gimple_fold_builtin_memory_chk): Same.
16886 (gimple_fold_builtin_strcpy): Handle no-warning.
16887 (gimple_fold_builtin_stxcpy_chk): Same.
16888 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
16890 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
16892 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
16893 parameter. Use it for SFmode.
16894 (rs6000_function_arg_advance_1): Adjust.
16895 (rs6000_function_arg): Adjust.
16896 (rs6000_gimplify_va_arg): Pass false for that new parameter.
16898 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
16900 PR rtl-optimization/84169
16901 PR rtl-optimization/84780
16902 * combine.c (can_combine_p): Check for a 2-insn combination whether
16903 the destination register is used between the two insns, too.
16905 2018-03-12 Richard Biener <rguenther@suse.de>
16907 PR tree-optimization/84803
16908 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
16909 for refs DR analysis didn't process.
16911 2018-03-12 Richard Biener <rguenther@suse.de>
16913 PR tree-optimization/84777
16914 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
16915 force-vectorize loops ignore whether we are optimizing for size.
16917 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
16919 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
16920 (TARGET_MD_ASM_ADJUST): Define.
16922 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
16923 Kito Cheng <kito.cheng@gmail.com>
16924 Chung-Ju Wu <jasonwucj@gmail.com>
16926 * config/nds32/nds32.c (nds32_compute_stack_frame,
16927 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
16928 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
16929 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
16930 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
16931 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
16932 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
16933 * config/nds32/nds32.md (prologue, epilogue): Use macro
16934 NDS32_V3PUSH_AVAILABLE_P to do checking.
16936 2018-03-11 Jakub Jelinek <jakub@redhat.com>
16939 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
16940 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
16941 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
16942 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
16943 addition of most attributes on !orig_type_die or the attribute not
16944 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
16946 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
16947 Chung-Ju Wu <jasonwucj@gmail.com>
16949 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
16950 __NDS32_VH__ macro.
16951 * config/nds32/nds32.opt (mvh): New option.
16953 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
16954 Chung-Ju Wu <jasonwucj@gmail.com>
16956 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
16958 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
16959 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
16962 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
16963 Chung-Ju Wu <jasonwucj@gmail.com>
16965 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
16967 * config/nds32/nds32-multiple.md (strlensi): New pattern.
16968 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
16970 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
16971 Kito Cheng <kito.cheng@gmail.com>
16972 Chung-Ju Wu <jasonwucj@gmail.com>
16974 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
16975 UNSPEC_FFMISM and UNSPEC_FLMISM.
16976 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
16977 for ffb, ffmism and flmism.
16978 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
16979 (unspec_ffmism): Ditto.
16980 (unspec_flmism): Ditto.
16981 (nds32_expand_builtin_impl): Check if string extension is available.
16982 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
16983 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
16985 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
16988 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
16991 * lra-assigns.c (assign_by_spills): Return a flag of reload
16992 assignment failure. Do not process the reload assignment
16993 failures. Do not spill other reload pseudos if they has the same
16995 (lra_assign): Add a return arg. Set up from the result of
16996 assign_by_spills call.
16997 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
16998 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
16999 usage_insns if it is not NULL.
17000 (spill_hard_reg_in_range): New function.
17001 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
17002 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
17003 function prototypes.
17004 (lra_assign): Change prototype.
17005 * lra.c (lra): Add code to deal with fails by splitting hard reg
17008 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
17011 * config/i386/i386.opt: Replace Enforcment with Enforcement.
17013 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
17016 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
17017 (dw_val_node): Add val_symbolic_view.
17018 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
17019 (symview_upper_bound): New.
17020 (new_line_info_table): Initialize symviews_since_reset.
17021 (dwarf2out_source_line): Count symviews_since_reset and set
17022 symview_upper_bound.
17023 (dw_val_equal_p): Handle symview.
17024 (add_AT_symview): New.
17025 (print_dw_val): Handle symview.
17026 (attr_checksum, attr_checksum_ordered): Likewise.
17027 (same_dw_val_p, size_of_die): Likewise.
17028 (value_format, output_die): Likewise.
17029 (add_high_low_attributes): Use add_AT_symview for entry_view.
17030 (dwarf2out_finish): Reset symview_upper_bound, clear
17033 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
17036 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
17037 Add strict argument and use it.
17038 (rs6000_split_multireg_move): Update for new strict argument.
17039 (mem_operand_gpr): Disallow all non-offsettable addresses.
17040 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
17042 2018-03-09 Jakub Jelinek <jakub@redhat.com>
17045 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
17046 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
17047 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
17050 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
17051 decl, use remap_type if we want to use the type.
17053 2018-03-09 Martin Sebor <msebor@redhat.com>
17055 PR tree-optimization/84526
17056 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
17058 (builtin_access::generic_overlap): Be prepared to handle non-array
17061 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
17063 PR rtl-optimization/84682
17064 * lra-constraints.c (process_address_1): Check is_address flag
17065 for address constraints.
17066 (process_alt_operands): Likewise.
17067 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
17068 preprocess_constraints.
17069 * recog.h (preprocess_constraints): Add oploc parameter.
17071 * recog.c (preprocess_constraints): Test address_operand for
17072 CT_ADDRESS constraints.
17074 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
17077 * lra-assigns.c (assign_by_spills): Return a flag of reload
17078 assignment failure. Do not process the reload assignment
17079 failures. Do not spill other reload pseudos if they has the same
17081 (lra_assign): Add a return arg. Set up from the result of
17082 assign_by_spills call.
17083 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
17084 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
17085 usage_insns if it is not NULL.
17086 (spill_hard_reg_in_range): New function.
17087 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
17088 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
17089 function prototypes.
17090 (lra_assign): Change prototype.
17091 * lra.c (lra): Add code to deal with fails by splitting hard reg
17094 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17097 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
17098 Accept complain bool parameter. Only emit errors if it is true.
17099 (arm_parse_cpu_option_name): Likewise.
17100 (arm_target_thumb_only): Adjust callers of the above.
17101 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
17102 prototype to take a default true bool parameter.
17103 (arm_parse_arch_option_name): Likewise.
17105 2018-03-09 David Malcolm <dmalcolm@redhat.com>
17106 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
17110 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
17111 * configure: Regenerate.
17112 * configure.ac ("linker --version-script option"): New.
17113 ("linker soname option"): New.
17115 2018-03-09 Richard Biener <rguenther@suse.de>
17117 PR tree-optimization/84775
17118 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
17119 immediate uses of predicate stmts and mark them modified.
17122 PR tree-optimization/84178
17123 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
17125 (version_loop_for_if_conversion): Delay update_ssa call.
17126 (tree_if_conversion): Delay update_ssa until after predicate
17129 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
17132 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
17133 when the function accesses prior frames.
17135 2018-03-08 Jakub Jelinek <jakub@redhat.com>
17138 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
17139 gen_llsym, otherwise call maybe_gen_llsym.
17141 PR inline-asm/84742
17142 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
17143 has ',' character inside of it.
17145 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17148 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
17149 as clobbering CC_REGNUM.
17151 2018-03-08 Richard Biener <rguenther@suse.de>
17153 PR middle-end/84552
17154 * tree-scalar-evolution.c: Include tree-into-ssa.h.
17155 (follow_copies_to_constant): Do not follow SSA names registered
17158 2018-03-08 Richard Biener <rguenther@suse.de>
17160 PR tree-optimization/84178
17161 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
17163 (version_loop_for_if_conversion): Delay update_ssa call.
17164 (tree_if_conversion): Delay update_ssa until after predicate
17167 2018-03-08 David Malcolm <dmalcolm@redhat.com>
17169 PR tree-optimization/84178
17170 * tree-if-conv.c (release_bb_predicate): Remove the
17171 the assertion that the stmts have NULL use_ops.
17172 Discard the statements, asserting that they haven't
17173 yet been added to a BB.
17175 2018-03-08 Richard Biener <rguenther@suse.de>
17177 PR tree-optimization/84746
17178 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
17179 (phi_translate): Pass in destination ANTIC_OUT set.
17180 (phi_translate_1): Likewise. For a simplified result lookup
17181 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
17182 (phi_translate_set): Adjust.
17183 (do_pre_regular_insertion): Likewise.
17184 (do_pre_partial_partial_insertion): Likewise.
17186 2018-03-08 Martin Liska <mliska@suse.cz>
17188 PR gcov-profile/84735
17189 * doc/gcov.texi: Document usage of profile files.
17190 * gcov-io.h: Document changes in the format.
17192 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
17196 * dwarf2out.c (struct dw_line_info_table): Update comments for
17198 (FORCE_RESET_NEXT_VIEW): New.
17199 (FORCE_RESETTING_VIEW_P): New.
17200 (RESETTING_VIEW_P): Check for -1 too.
17201 (ZERO_VIEW_P): Likewise.
17202 (new_line_info_table): Force-reset next view.
17203 (dwarf2out_begin_function): Likewise.
17204 (dwarf2out_source_line): Simplify zero_view_p initialization.
17205 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
17206 view directly. Omit view when omitting .loc at line 0.
17208 2018-03-08 Jakub Jelinek <jakub@redhat.com>
17210 PR tree-optimization/84740
17211 * tree-switch-conversion.c (process_switch): Call build_constructors
17212 only if info.phi_count is non-zero.
17214 PR tree-optimization/84739
17215 * tree-tailcall.c (find_tail_calls): Check call arguments against
17216 DECL_ARGUMENTS (current_function_decl) rather than
17217 DECL_ARGUMENTS (func) when checking for tail recursion.
17219 2018-03-07 Jakub Jelinek <jakub@redhat.com>
17221 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
17222 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
17223 Volker Reichelt's entry and add entries for people that perform
17224 GCC fuzzy testing and report numerous bugs.
17226 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
17229 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
17230 readonly data in sdata, if that is disabled.
17231 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
17232 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
17233 -mreadonly-in-sdata option.
17235 2018-03-07 Martin Sebor <msebor@redhat.com>
17237 PR tree-optimization/84468
17238 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
17239 basic block when looking for nul assignment.
17241 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
17244 * except.h (output_function_exception_table): Adjust prototype.
17245 * except.c (output_function_exception_table): Remove FNNAME parameter
17246 and add SECTION parameter. Ouput one part of the table at a time.
17247 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
17248 the first part of the exception table and emit unwind directives.
17249 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
17250 (i386_pe_seh_cold_init): Likewise.
17251 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
17252 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
17253 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
17254 (ix86_output_call_insn): Emit a nop in one more case for SEH.
17255 * config/i386/winnt.c: Include except.h.
17256 (struct seh_frame_state): Add reg_offset, after_prologue and
17257 in_cold_section fields.
17258 (i386_pe_seh_end_prologue): Set seh->after_prologue.
17259 (i386_pe_seh_cold_init): New function.
17260 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
17261 to seh->in_cold_section.
17262 (seh_emit_push): Record the offset of the push.
17263 (seh_emit_save): Record the offet of the save.
17264 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
17265 Test seh->after_prologue to disregard the epilogue.
17266 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
17267 (i386_pe_end_cold_function): New function.
17269 2018-03-07 Jakub Jelinek <jakub@redhat.com>
17272 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
17273 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
17276 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
17278 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
17279 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
17281 PR middle-end/84723
17282 * multiple_target.c: Include tree-inline.h and intl.h.
17283 (expand_target_clones): Diagnose and fail if node->definition and
17284 !tree_versionable_function_p (node->decl).
17286 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
17288 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
17290 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
17291 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17292 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
17294 2018-03-06 Jakub Jelinek <jakub@redhat.com>
17297 * combine.c (try_combine): Use reg_or_subregno instead of handling
17298 just paradoxical SUBREGs and REGs.
17300 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
17302 * config/arc/arc.c (arc_finalize_pic): Remove function.
17303 (arc_must_save_register): We use single base PIC register, remove
17304 checks to save/restore the PIC register.
17305 (arc_expand_prologue): Likewise.
17306 * config/arc/arc-protos.h (arc_set_default_type_attributes):
17308 (arc_verify_short): Likewise.
17309 (arc_attr_type): Likewise.
17310 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
17311 (walk_stores): Likewise.
17312 (arc_address_cost): Make it static.
17313 (arc_verify_short): Likewise.
17314 (branch_dest): Likewise.
17315 (arc_attr_type): Likewise.
17316 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
17317 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
17318 (arc_final_prescan_insn): Remove inserting the nops due to
17319 hardware hazards. It is done in reorg step.
17320 (insn_length_variant_t): Remove.
17321 (insn_length_parameters_t): Likewise.
17322 (arc_insn_length_parameters): Likewise.
17323 (arc_get_insn_variants): Likewise.
17324 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
17326 2018-03-06 Jakub Jelinek <jakub@redhat.com>
17328 PR inline-asm/84683
17329 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
17332 PR tree-optimization/84687
17333 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
17335 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
17337 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17339 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
17340 Rename to ppc_speculation_barrier.
17341 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
17342 __builtin_ppc_speculation_barrier.
17344 2018-03-05 Jakub Jelinek <jakub@redhat.com>
17347 * combine.c (combine_simplify_rtx): Don't try to simplify if
17348 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
17351 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
17353 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
17354 to 32 bytes when compiling for POWER9.
17356 2018-03-05 Jakub Jelinek <jakub@redhat.com>
17359 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
17360 regparm >= 3 with no arg reg available also for calls with
17361 flag_force_indirect_call. Pass decl to ix86_function_regparm.
17364 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
17366 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
17368 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
17371 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
17373 2018-03-05 Richard Biener <rguenther@suse.de>
17375 PR tree-optimization/84486
17376 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
17377 When inserting a __builtin_assume_aligned call set the LHS
17378 SSA name alignment info accordingly.
17380 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
17382 PR tree-optimization/84114
17383 * config/aarch64/aarch64.c (aarch64_reassociation_width)
17384 Avoid reassociation of FLOAT_MODE addition.
17386 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
17388 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
17389 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
17390 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
17391 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
17392 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
17393 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
17394 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
17396 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
17397 __builtin_ia32_wbinvd): New builtins.
17398 (SPECIAL_ARGS2): New.
17399 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
17400 (SPECIAL_ARGS2): New.
17401 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
17402 (ix86_valid_target_attribute_inner_p): Ditto.
17403 (ix86_init_mmx_sse_builtins): Add special_args2.
17404 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
17405 TARGET_WBNOINVD_P): New.
17406 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
17407 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
17408 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
17409 * config/i386/immintrin.h (_wbinvd): New intrinsic.
17410 * config/i386/pconfigintrin.h: New file.
17411 * config/i386/wbnoinvdintrin.h: Ditto.
17412 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
17414 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
17416 2018-03-05 Richard Biener <rguenther@suse.de>
17418 PR tree-optimization/84670
17419 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
17421 (BB_VISITED_WITH_VISITED_SUCCS): New define.
17422 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
17423 (compute_antic_aux): Only assert the number of values in ANTIC_IN
17424 doesn't grow if all successors (recursively) were visited at least
17427 2018-03-05 Richard Biener <rguenther@suse.de>
17429 PR tree-optimization/84650
17430 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
17431 if executed in the loop pipeline.
17433 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
17435 * doc/configfiles.texi (Configuration Files): Move info about
17436 conditionalizing $target-protos.h to...
17437 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
17438 differs from $target-protos.h.
17440 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
17441 Chung-Ju Wu <jasonwucj@gmail.com>
17443 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
17444 * config/nds32/nds32-multiple.md (setmemsi): Define.
17445 * config/nds32/nds32-memory-manipulation.c
17446 (nds32_gen_dup_4_byte_to_word_value): New.
17447 (emit_setmem_word_loop): New.
17448 (emit_setmem_byte_loop): New.
17449 (nds32_expand_setmem_loop): New.
17450 (nds32_expand_setmem_loop_v3m): New.
17451 (nds32_expand_setmem_unroll): New.
17452 (nds32_expand_setmem): New.
17454 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
17455 Chung-Ju Wu <jasonwucj@gmail.com>
17457 * config/nds32/nds32-memory-manipulation.c
17458 (nds32_emit_load_store): New.
17459 (nds32_emit_post_inc_load_store): New.
17460 (nds32_emit_mem_move): New.
17461 (nds32_emit_mem_move_block): New.
17462 (nds32_expand_movmemsi_loop_unknown_size): New.
17463 (nds32_expand_movmemsi_loop_known_size): New.
17464 (nds32_expand_movmemsi_loop): New.
17465 (nds32_expand_movmemsi_unroll): New.
17466 (nds32_expand_movmemqi): Rename ...
17467 (nds32_expand_movmemsi): ... to this.
17468 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
17469 (movmemsi): ... to this.
17470 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
17471 (nds32_expand_movmemsi): ... to this.
17473 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
17474 Monk Chiang <sh.chiang04@gmail.com>
17475 Chung-Ju Wu <jasonwucj@gmail.com>
17477 * config/nds32/nds32-protos.h
17478 (nds32_expand_load_multiple): New arguments.
17479 (nds32_expand_store_multiple): Ditto.
17480 (nds32_valid_multiple_load_store): Rename ...
17481 (nds32_valid_multiple_load_store_p): ... to this.
17482 * config/nds32/nds32-memory-manipulation.c
17483 (nds32_expand_load_multiple): Refine implementation.
17484 (nds32_expand_store_multiple): Ditto.
17485 * config/nds32/nds32-multiple.md
17486 (load_multiple): Update nds32_expand_load_multiple interface.
17487 (store_multiple): Update nds32_expand_store_multiple interface.
17488 * config/nds32/nds32-predicates.c
17489 (nds32_valid_multiple_load_store): Rename ...
17490 (nds32_valid_multiple_load_store_p): ... to this and refine
17492 * config/nds32/predicates.md
17493 (nds32_load_multiple_and_update_address_operation): New predicate.
17494 (nds32_store_multiple_and_update_address_operation): New predicate.
17496 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
17497 Chung-Ju Wu <jasonwucj@gmail.com>
17499 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
17500 (combo): New attribute.
17501 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
17503 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
17505 * config/nds32/nds32.opt: Change -mcmodel= default value.
17507 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
17508 Monk Chiang <sh.chiang04@gmail.com>
17509 Chung-Ju Wu <jasonwucj@gmail.com>
17511 * config/nds32/constants.md (unspec_element): New enum.
17512 * config/nds32/constraints.md (Umw): New constraint.
17513 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
17514 * config/nds32/nds32-intrinsic.md: Likewise.
17515 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
17516 (nds32_valid_smw_lwm_base_p): New.
17517 (nds32_output_smw_single_word): New.
17518 (nds32_output_lmw_single_word): New.
17519 (nds32_expand_unaligned_load): New.
17520 (nds32_expand_unaligned_store): New.
17521 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
17522 (nds32_output_smw_single_word): Declare.
17523 (nds32_output_lmw_single_word): Declare.
17524 (nds32_expand_unaligned_load): Declare.
17525 (nds32_expand_unaligned_store): Declare.
17526 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
17527 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
17528 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
17529 NDS32_BUILTIN_UASTORE_DW.
17530 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
17533 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
17534 Kito Cheng <kito.cheng@gmail.com>
17535 Chung-Ju Wu <jasonwucj@gmail.com>
17537 * config/nds32/nds32-intrinsic.c
17538 (nds32_expand_builtin_null_ftype_reg): Delete.
17539 (nds32_expand_builtin_reg_ftype_imm): Ditto.
17540 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
17541 (nds32_read_argument): New.
17542 (nds32_legitimize_target): Ditto.
17543 (nds32_legitimize_argument): Ditto.
17544 (nds32_check_constant_argument): Ditto.
17545 (nds32_expand_unop_builtin): Ditto.
17546 (nds32_expand_unopimm_builtin): Ditto.
17547 (nds32_expand_binop_builtin): Ditto.
17548 (nds32_builtin_decl_impl): Ditto.
17549 (builtin_description): Ditto.
17550 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
17551 (nds32_init_builtins_impl): Ditto.
17552 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
17553 (nds32_builtin_decl): New.
17554 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
17555 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
17557 2018-03-02 Jeff Law <law@redhat.com>
17559 * reorg.c (stop_search_p): Handle DEBUG_INSN.
17560 (redundant_insn, fill_simple_delay_slots): Likewise.
17561 (fill_slots_from_thread): Likewise.
17562 * resource.c (mark_referenced_resources): Likewise.
17563 (mark_set_resources, find_dead_or_set_registers): Likewise.
17565 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17567 * substring-locations.h (format_warning_va): Formatting fix for
17568 ATTRIBUTE_GCC_DIAG.
17569 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
17571 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
17572 * substring-locations.c: Include intl.h.
17573 (format_warning_va): Turned into small wrapper around
17574 format_warning_n_va, renamed to ...
17575 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
17576 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
17578 (format_warning_at_substring_n): New function.
17579 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
17580 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
17581 format_warning_at_substring with just a shorter name instead of
17582 const function pointer.
17583 (fmtwarn_n): New function.
17584 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
17585 appropriate, get rid of all the fmtstr temporaries, move conditionals
17586 with G_() wrapped string literals directly into fmtwarn arguments,
17587 cast dir.len to (int), formatting fixes.
17589 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
17591 * doc/invoke.texi: Remove "Cilk Plus" references.
17593 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17594 Richard Biener <rguenther@suse.de>
17597 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
17598 for error or warning attributes if CALL_FROM_THUNK_P is set.
17601 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17604 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
17605 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
17608 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
17609 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
17611 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
17612 instead of -1U in last predictors element's probability member.
17614 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
17617 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
17618 arguments if they are comparable.
17620 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
17622 PR tree-optimization/84634
17623 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
17624 masks and masked_loop_p with a single loop_masks, making sure it's
17625 null for bb vectorization.
17627 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
17629 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
17630 (vect_analyze_data_ref_access): Use loop->safe_len rather than
17631 loop->force_vectorize to check whether there is no alias.
17633 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17636 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
17638 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
17640 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
17641 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
17642 instead of a loop around prev_real_insn.
17643 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
17646 PR inline-asm/84625
17647 * config/i386/i386.c (ix86_print_operand): Use conditional
17648 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
17651 2018-03-02 Richard Biener <rguenther@suse.de>
17653 PR tree-optimization/84427
17654 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
17655 (bitmap_set_subtract_values): Rewrite to handle multiple
17658 (prune_clobbered_mems): Likewise.
17659 (phi_translate): Take edge instead of pred/phiblock.
17660 (phi_translate_1): Likewise.
17661 (phi_translate_set): Likewise. Insert all translated
17662 exprs for a value into the set, keeping possibly multiple
17663 expressions per value.
17664 (compute_antic_aux): Adjust for phi_translate changes.
17665 When intersecting union the expressions and prune those
17666 not in the final value set, keeping possibly multiple
17667 expressions per value. Do not use value-insertion
17668 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
17669 all expressions. Add verification that the value-sets
17670 only shrink during iteration.
17671 (compute_partial_antic_aux): Adjust for the phi_translate changes.
17672 (do_pre_regular_insertion): Likewise.
17673 (do_pre_partial_partial_insertion): Likewise.
17675 2018-03-02 Richard Biener <rguenther@suse.de>
17678 * config/darwin.c (saved_debug_info_level): New static global.
17679 (darwin_asm_lto_start): Disable debug info generation for LTO out.
17680 (darwin_asm_lto_end): Restore debug info generation settings.
17682 2018-03-01 Martin Liska <mliska@suse.cz>
17685 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
17686 volatile arguments.
17688 2018-03-01 Richard Biener <rguenther@suse.de>
17691 * dwarf2out.c (gen_variable_die): Properly handle late VLA
17692 type annotation with LTO when debug was disabled at compile-time.
17694 2018-03-01 Matthew Fortune <mfortune@gmail.com>
17696 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
17698 (mips_final_postscan_insn): Likewise.
17700 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
17702 PR rtl-optimization/84528
17703 * alias.c (init_alias_target): Add commentary.
17704 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
17705 a unique base value if the frame pointer is not eliminated
17706 to the stack pointer.
17708 2018-03-01 Tom de Vries <tom@codesourcery.com>
17710 PR rtl-optimization/83327
17711 * lra-int.h (hard_regs_spilled_into): Declare.
17712 * lra.c (hard_regs_spilled_into): Define.
17713 (init_reg_info): Init hard_regs_spilled_into.
17714 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
17715 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
17716 (process_bb_lives): Handle hard_regs_spilled_into.
17717 (lra_create_live_ranges_1): Before doing liveness propagation, clear
17718 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
17720 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
17722 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
17723 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
17724 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
17725 * config/rs6000/aix72.h: New file.
17727 2018-02-28 Jakub Jelinek <jakub@redhat.com>
17729 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
17730 instead of warning_at with conditional singular and plural messages
17734 * stor-layout.c (update_alignment_for_field): For
17735 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
17736 && !DECL_PACKED (field), do the alignment update, just use
17737 only desired_align instead of MAX (type_align, desired_align)
17739 (place_field): Don't do known_align < desired_align handling
17740 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
17741 is non-NULL, instead do it after rli->prev_field handling and
17742 only if not within a bitfield word. For DECL_PACKED (field)
17743 use type_align of BITS_PER_UNIT.
17745 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
17747 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
17748 superfluous parentheses and trailing spaces.
17750 2018-02-28 Richard Biener <rguenther@suse.de>
17752 PR tree-optimization/84584
17753 * graphite-scop-detection.c (scop_detection::add_scop): Discard
17754 SCoPs with fake exit edge.
17756 2018-02-28 Martin Liska <mliska@suse.cz>
17759 * timevar.c (timer::print): Fix format to properly print 100%
17762 2018-02-28 Richard Biener <rguenther@suse.de>
17764 PR middle-end/84607
17765 * genmatch.c (capture_info::walk_match): Do not mark
17766 captured expressions without operands as expr_p given
17767 they act more like predicates and should be subject to
17768 "lost tail" side-effect preserving.
17770 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
17772 PR rtl-optimization/81611
17773 * auto-inc-dec.c (attempt_change): Move dead note from
17774 mem_insn if it's the next use of regno
17775 (find_address): Take address use of reg holding
17776 non-incremented value. Add parm to limit search to the named
17778 (merge_in_block): Attempt to use a mem insn that is the next
17779 use of the original regno.
17781 2018-02-27 Martin Sebor <msebor@redhat.com>
17784 * doc/invoke.texi (-Wmissing-attributes): New option.
17785 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
17787 2018-02-27 Martin Sebor <msebor@redhat.com>
17789 PR translation/84207
17790 * diagnostic-core.h (warning_n, error_n, inform_n): Change
17791 n argument to unsigned HOST_WIDE_INT.
17792 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
17793 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
17794 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
17795 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
17797 2018-02-27 Richard Biener <rguenther@suse.de>
17799 PR tree-optimization/84512
17800 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
17801 Do not use the estimate returned from record_stmt_cost for
17802 the scalar iteration cost but sum properly using add_stmt_cost.
17804 2018-02-27 Richard Biener <rguenther@suse.de>
17806 PR tree-optimization/84466
17807 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
17808 Adjust last change to less strictly validate use operands.
17810 2018-02-27 Martin Liska <mliska@suse.cz>
17812 PR gcov-profile/84548
17813 * gcov.c (process_file): Allow partial overlap and consider it
17814 also as group functions.
17815 (output_lines): Properly calculate range of lines for a group.
17817 2018-02-27 Martin Liska <mliska@suse.cz>
17819 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
17820 'ggc' suffixes. Change first column width.
17821 (timer::print): Fix formatting of the column.
17823 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
17825 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
17826 preserve inline entry blocks for the sake of debug inline
17827 entry point markers alone.
17828 (remove_unused_locals): Suggest in comments a better place to
17829 force the preservation of inline entry blocks that are
17830 otherwise unused, but do not preserve them.
17832 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
17834 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
17836 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
17839 * config/i386/constraints.md (Bs): Replace
17840 ix86_indirect_branch_register with
17841 TARGET_INDIRECT_BRANCH_REGISTER.
17843 * config/i386/i386.md (indirect_jump): Likewise.
17844 (tablejump): Likewise.
17845 (*sibcall_memory): Likewise.
17846 (*sibcall_value_memory): Likewise.
17847 Peepholes of indirect call and jump via memory: Likewise.
17848 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
17849 (*sibcall_value_GOT_32): Likewise.
17850 * config/i386/predicates.md (indirect_branch_operand): Likewise.
17851 (GOT_memory_operand): Likewise.
17852 (call_insn_operand): Likewise.
17853 (sibcall_insn_operand): Likewise.
17854 (GOT32_symbol_operand): Likewise.
17855 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
17857 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
17859 PR rtl-optimization/83496
17860 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
17861 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
17862 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
17863 redundant insn, if any.
17864 (relax_delay_slots): Likewise.
17865 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
17867 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
17869 PR tree-optimization/83965
17870 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
17871 that grouped statements are part of a reduction chain. Return
17872 true if the statement is not marked as a reduction itself but
17873 is part of a group.
17874 (vect_recog_dot_prod_pattern): Don't check whether the statement
17875 is part of a group here.
17876 (vect_recog_sad_pattern): Likewise.
17877 (vect_recog_widen_sum_pattern): Likewise.
17879 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
17882 * final.c (rest_of_clean_state): Also look for calls inside sequences.
17884 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
17887 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
17889 (ix86_output_indirect_function_return): New prototype.
17890 (ix86_split_simple_return_pop_internal): Likewise.
17891 * config/i386/i386.c (indirect_return_via_cx): New.
17892 (indirect_return_via_cx_bnd): Likewise.
17893 (indirect_thunk_name): Handle return va CX_REG.
17894 (output_indirect_thunk_function): Create alias for
17895 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
17896 (ix86_output_indirect_jmp): Remove the bool argument.
17897 (ix86_output_indirect_function_return): New function.
17898 (ix86_split_simple_return_pop_internal): Likewise.
17899 * config/i386/i386.md (*indirect_jump): Don't pass false
17900 to ix86_output_indirect_jmp.
17901 (*tablejump_1): Likewise.
17902 (simple_return_pop_internal): Change it to define_insn_and_split.
17903 Call ix86_split_simple_return_pop_internal to split it for
17904 -mfunction-return=.
17905 (simple_return_indirect_internal): Call
17906 ix86_output_indirect_function_return instead of
17907 ix86_output_indirect_jmp.
17909 2018-02-26 Jakub Jelinek <jakub@redhat.com>
17912 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
17913 memset and value initialization afterwards.
17915 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
17917 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
17919 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17922 * common/config/aarch64/aarch64-common.c
17923 (aarch_option_optimization_table[]): Switch
17924 off fomit-frame-pointer
17926 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
17927 Chung-Ju Wu <jasonwucj@gmail.com>
17929 * config/nds32/nds32-multiple.md (load_multiple): Disallow
17931 (store_multiple): Ditto.
17933 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
17935 * config.gcc: Add --with-cpu support for nds32 target.
17936 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
17937 * config/nds32/nds32.opt: Add -mcpu= option.
17939 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
17941 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
17942 isel=yes): Warn for these deprecated options.
17944 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
17946 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
17947 ISA_2_5_MASKS_EMBEDDED.
17949 2018-02-23 Jakub Jelinek <jakub@redhat.com>
17951 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
17952 p->max as pointers rather than using iterative_hash_expr.
17954 2018-02-23 Carl Love <cel@us.ibm.com>
17956 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
17957 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
17959 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
17960 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
17961 P8V_BUILTIN_VEC_VUNSIGNED2.
17963 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
17966 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
17967 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
17969 * lra-constraints.c (curr_insn_transform): Set up
17970 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
17972 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
17973 * lra-eliminations.c (spill_pseudos): Ditto.
17974 (process_insn_for_elimination): Ditto.
17975 * lra-lives.c (reg_early_clobber_p): Use the new macros.
17976 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
17977 LRA_NON_CLOBBERED_ALT.
17979 2018-02-22 Martin Sebor <msebor@redhat.com>
17981 PR tree-optimization/84480
17982 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
17983 to maybe_diag_stxncpy_trunc. Call it.
17984 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
17985 from gimple_fold_builtin_strcpy. Print inlining stack.
17986 (handle_builtin_stxncpy): Print inlining stack.
17987 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
17989 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
17992 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
17993 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
17994 and -fcheck-pointer-bounds are used together.
17995 (indirect_thunk_prefix): New enum.
17996 (indirect_thunk_need_prefix): New function.
17997 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
17998 "_nt" instead of "_bnd" for NOTRACK prefix.
17999 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
18000 (output_indirect_thunk_function): Likewise.
18002 (ix86_code_end): Update output_indirect_thunk_function calls.
18003 (ix86_output_indirect_branch_via_reg): Replace
18004 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
18005 (ix86_output_indirect_branch_via_push): Likewise.
18006 (ix86_output_function_return): Likewise.
18007 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
18008 incompatible with -fcf-protection=branch and
18009 -fcheck-pointer-bounds.
18011 2018-02-22 Steve Ellcey <sellcey@cavium.com>
18014 * config/aarch64/aarch64.c (aarch64_print_address_internal):
18015 Change gcc_assert call to output_operand_lossage.
18017 2018-02-22 Steve Ellcey <sellcey@cavium.com>
18019 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
18021 2018-02-22 DJ Delorie <dj@redhat.com>
18022 Sebastian Perta <sebastian.perta@renesas.com>
18023 Oleg Endo <olegendo@gcc.gnu.org>
18025 * config/rx/rx.c (rx_rtx_costs): New function.
18026 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
18028 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
18030 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
18032 2018-02-22 Martin Liska <mliska@suse.cz>
18035 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
18036 Add "native" as a possible value.
18038 2018-02-22 Martin Liska <mliska@suse.cz>
18041 * config/i386/i386.c (ix86_option_override_internal):
18042 Add "native" as a possible value for -march and -mtune.
18044 2018-02-22 Jakub Jelinek <jakub@redhat.com>
18047 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
18048 to all type variants.
18050 PR tree-optimization/84503
18051 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
18052 width as info->bitpos + info->bitsize - start.
18053 (merged_store_group::merge_overlapping): Simplify width computation.
18054 (check_no_overlap): New function.
18055 (imm_store_chain_info::try_coalesce_bswap): Compute expected
18056 start + width and last_order of the group, fail if check_no_overlap
18058 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
18059 to group if check_no_overlap fails.
18061 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
18063 * config/rs6000/altivec.md: Delete contraint arguments to
18064 define_expand, define_split, and define_peephole2, and in
18065 define_insn_and_split if always unused.
18066 * config/rs6000/darwin.md: Ditto.
18067 * config/rs6000/dfp.md: Ditto.
18068 * config/rs6000/rs6000.md: Ditto.
18069 * config/rs6000/sync.md: Ditto.
18070 * config/rs6000/vector.md: Ditto.
18071 * config/rs6000/vsx.md: Ditto.
18073 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
18075 * config/rs6000/altivec.md: Write output control strings as braced
18076 blocks instead of double-quoted strings.
18077 * config/rs6000/darwin.md: Ditto.
18078 * config/rs6000/rs6000.md: Ditto.
18079 * config/rs6000/vector.md: Ditto.
18080 * config/rs6000/vsx.md: Ditto.
18082 2018-02-21 Jason Merrill <jason@redhat.com>
18084 PR c++/84314 - ICE with templates and fastcall attribute.
18085 * attribs.c (build_type_attribute_qual_variant): Remove assert.
18087 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
18089 * ipa-cp.c (determine_versionability): Fix comment typos.
18091 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
18094 * ipa-cp.c (determine_versionability): Do not version functions caling
18097 2018-02-21 Martin Liska <mliska@suse.cz>
18100 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
18101 Add "native" as a possible value.
18102 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
18103 the macro when native cpu detection is available.
18105 2018-02-21 Martin Liska <mliska@suse.cz>
18108 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
18109 Add "native" as a possible value.
18110 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
18111 when native cpu detection is available.
18113 2018-02-21 Jakub Jelinek <jakub@redhat.com>
18114 Martin Sebor <msebor@redhat.com>
18116 PR tree-optimization/84478
18117 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
18119 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
18120 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
18121 support which is conservatively correct, for 2 only stay conservative
18122 for maxlen. Formatting and comment capitalization fixes. Add STRICT
18123 argument to the 2 argument get_range_strlen, adjust 6 arg
18124 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
18126 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
18127 (gimple_fold_builtin_strlen): Pass true as last argument to
18130 2018-02-20 Martin Sebor <msebor@redhat.com>
18132 PR middle-end/84095
18133 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
18134 (builtin_memref::set_base_and_offset): Same. Handle inner references.
18135 (builtin_memref::builtin_memref): Factor out parts into
18136 set_base_and_offset and call it.
18138 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
18140 PR middle-end/84406
18141 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
18142 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
18143 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
18144 search at the associated MODE_INT.
18146 2018-02-20 Jeff Law <law@redhat.com>
18148 PR middle-end/82123
18149 PR tree-optimization/81592
18150 PR middle-end/79257
18151 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
18152 for range data rather than using global data.
18153 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
18154 range data rather than using global data.
18155 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
18156 pass it to children as needed.
18157 (struct directive::fmtresult): Similarly.
18158 (struct directive::set_width): Similarly.
18159 (struct directive::set_precision): Similarly.
18160 (format_integer, format_directive, parse_directive): Similarly.
18161 (format_none): Accept unnamed vr_values parameter.
18162 (format_percent, format_floating, format_character): Similarly.
18163 (format_string, format_plain): Similarly.
18164 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
18165 the EVRP range analyzer for range data rather than using global data.
18166 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
18167 gimple-ssa-evrp-analyze.h
18168 (class sprintf_dom_walker): Add after_dom_children member function.
18169 Add evrp_range_analyzer member.
18170 (sprintf_dom_walker::before_dom_children): Call into the EVRP
18171 range analyzer as needed.
18172 (sprintf_dom_walker::after_dom_children): New member function.
18173 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
18175 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
18176 (evrp_range_analyzer::pop_to_marker): Likewise.
18178 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
18180 PR tree-optimization/84419
18181 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
18182 with the required type if its current type is compatible but
18185 2018-02-20 Jakub Jelinek <jakub@redhat.com>
18187 PR middle-end/82004
18188 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
18189 after vectorization.
18191 2018-02-20 Martin Liska <mliska@suse.cz>
18194 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
18195 possible values if we don't have a hint.
18197 2018-02-20 Martin Liska <mliska@suse.cz>
18201 * final.c (shorten_branches): Build align_tab array with one
18203 * opts.c (finish_options): Add alignment option limit check.
18204 (MAX_CODE_ALIGN): Likewise.
18205 (MAX_CODE_ALIGN_VALUE): Likewise.
18206 * doc/invoke.texi: Document maximum allowed option value for
18207 all -falign-* options.
18209 2018-02-19 Jakub Jelinek <jakub@redhat.com>
18212 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
18213 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
18214 * var-tracking.c (emit_note_insn_var_location): Remove all references
18215 to NOTE_INSN_CALL_ARG_LOCATION.
18216 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
18217 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
18218 Use copy_rtx_if_shared.
18219 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
18220 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
18221 (dwarf2out_var_location): Remove handling of
18222 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
18224 * final.c (final_scan_insn): Remove all references to
18225 NOTE_INSN_CALL_ARG_LOCATION.
18226 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
18227 before dumping final insns.
18228 * except.c (emit_note_eh_region_end): Remove all references to
18229 NOTE_INSN_CALL_ARG_LOCATION.
18230 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
18231 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
18232 * config/arc/arc.c (hwloop_optimize): Likewise.
18233 * config/arm/arm.c (create_fix_barrier): Likewise.
18234 * config/s390/s390.c (s390_chunkify_start): Likewise.
18235 * config/sh/sh.c (find_barrier): Likewise.
18236 * config/i386/i386.c (rest_of_insert_endbranch,
18237 ix86_seh_fixup_eh_fallthru): Likewise.
18238 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
18239 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
18240 * config/frv/frv.c (frv_function_prologue): Likewise.
18241 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
18243 (note_outside_basic_block_p): Remove all references to
18244 NOTE_INSN_CALL_ARG_LOCATION.
18245 * gengtype.c (adjust_field_rtx_def): Likewise.
18246 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
18248 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
18249 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
18252 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
18255 PR tree-optimization/84452
18256 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
18257 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
18260 2018-02-19 Martin Liska <mliska@suse.cz>
18263 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
18265 2018-02-19 Martin Liska <mliska@suse.cz>
18266 Richard Sandiford <richard.sandiford@linaro.org>
18268 PR tree-optimization/82491
18269 * gimple-fold.c (get_base_constructor): Make earlier bail out
18272 2018-02-19 Carl Love <cel@us.ibm.com>
18274 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
18275 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
18277 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
18278 P8V_BUILTIN_VEC_NEG.
18280 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
18282 * config/rl78/rl78.md (movdf): New define expand.
18284 2018-02-19 Martin Liska <mliska@suse.cz>
18287 * doc/invoke.texi: Fix typo.
18288 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
18290 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
18292 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
18293 handle rs6000_single_float and rs6000_double_float specially for
18296 2018-02-16 Jeff Law <law@redhat.com>
18298 * config/rx/rx.c (add_pop_cfi_notes): New function.;
18299 (pop_regs): Use it.
18301 2018-02-16 Jakub Jelinek <jakub@redhat.com>
18304 * ipa-inline.c (inline_small_functions): Fix a typo.
18306 2018-02-16 Nathan Sidwell <nathan@acm.org>
18308 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
18310 2018-02-16 Carl Love <cel@us.ibm.com>
18312 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
18313 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
18314 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
18315 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
18316 expansion to P8V_BUILTIN_VEC_FLOAT2.
18318 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
18320 PR rtl-optimization/70023
18321 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
18322 src_regno into account.
18324 2018-02-16 Carl Love <cel@us.ibm.com>
18326 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
18327 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
18328 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
18329 * config/rs6000/rs6000.c: Remove case statements for
18330 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
18331 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
18332 and P9V_BUILTIN_VEC_VINSERT4B.
18333 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
18334 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
18335 * config/rs6000/vsx.md:
18336 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
18339 2018-02-16 Carl Love <cel@us.ibm.com>
18341 * config/rs6000/altivec.h: Add builtin names vec_extract4b
18343 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
18345 * config/rs6000/rs6000-c.c: Add the definitions for
18346 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
18347 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
18348 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
18349 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
18350 definition for insert4b and define insn *insert3b_internal.
18351 * doc/extend.texi: Add documentation for vec_extract4b.
18353 2018-02-16 Nathan Sidwell <nathan@acm.org>
18355 * doc/extend.texi (Backwards Compatibility): Mention friend
18356 injection. Note for-scope is deprecated.
18357 * doc/invoke.texi (-ffriend-injection): Deprecate.
18359 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
18361 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
18362 that moved to I2, also allow destinations that are a paradoxical
18363 subreg (instead of a normal reg).
18365 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
18368 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
18371 2018-02-16 Richard Biener <rguenther@suse.de>
18373 PR tree-optimization/84037
18374 PR tree-optimization/84016
18376 * config/i386/i386.c (ix86_builtin_vectorization_cost):
18377 Adjust vec_construct for the fact we need additional higher latency
18378 128bit inserts for AVX256 and AVX512 vector builds.
18379 (ix86_add_stmt_cost): Scale vector construction cost for
18382 2018-02-16 Richard Biener <rguenther@suse.de>
18384 PR tree-optimization/84417
18385 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
18386 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
18387 (non_rewritable_lvalue_p): Likewise, use poly-ints.
18389 2018-02-16 Martin Liska <mliska@suse.cz>
18392 * internal-fn.def (ASAN_CHECK): Set proper flags.
18393 (ASAN_MARK): Likewise.
18395 2018-02-16 Julia Koval <julia.koval@intel.com>
18397 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
18398 from PTA_CANNONLAKE.
18400 2018-02-16 Jakub Jelinek <jakub@redhat.com>
18403 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
18404 Use ++iter rather than iter++ for std::list iterators.
18405 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
18406 defer deleting them until all nodes in the forest are processed. Do
18407 free even leaf nodes. Change to_process into auto_vec.
18410 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
18411 * vec.h (vec_default_construct): Use memset instead of placement new
18412 if BROKEN_VALUE_INITIALIZATION is defined.
18413 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
18414 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
18417 PR rtl-optimization/83723
18418 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
18419 * lra.c (lra_substitute_pseudo): Likewise. If true, use
18420 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
18422 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
18424 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
18426 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
18428 PR rtl-optimization/81443
18429 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
18430 from inner REGs to paradoxical SUBREGs.
18432 2018-02-16 Richard Biener <rguenther@suse.de>
18434 PR tree-optimization/84399
18435 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
18436 For operands we can analyze at their definition make sure we can
18437 analyze them at each use as well.
18439 2018-02-16 Richard Biener <rguenther@suse.de>
18441 PR tree-optimization/84190
18442 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
18443 volatile accesses if the decl isn't volatile.
18445 2018-02-15 Jason Merrill <jason@redhat.com>
18447 PR c++/84314 - ICE with templates and fastcall attribute.
18448 * attribs.c (build_type_attribute_qual_variant): Don't clobber
18449 TYPE_CANONICAL on an existing type.
18451 2018-02-15 Jakub Jelinek <jakub@redhat.com>
18453 PR tree-optimization/84383
18454 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
18455 dstoff nor call operand_equal_p if dstbase is NULL.
18457 PR tree-optimization/84334
18458 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
18459 also a CONSTANT_CLASS_P, punt.
18461 2018-02-14 Jim Wilson <jimw@sifive.com>
18463 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
18464 first SMALL_OPERAND check. New local min_second_step. Move assert
18465 to where locals are set. Add TARGET_RVC support.
18466 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
18468 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
18470 * doc/invoke.texi: Correct -Wformat-overflow code sample.
18472 2018-02-14 Martin Sebor <msebor@redhat.com>
18474 PR tree-optimization/83698
18475 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
18476 arrays constrain the offset range to their bounds.
18477 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
18478 (builtin_access::overlap): Avoid setting the size of overlap if it's
18480 (maybe_diag_overlap): Also consider arrays when deciding what values
18481 of offsets to include in diagnostics.
18483 2018-02-14 Martin Sebor <msebor@redhat.com>
18486 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
18487 that correspond to the kind of a declaration.
18489 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
18492 * config/pa/pa.md: Load address of PIC label using the linkage table
18493 if the label is nonlocal.
18495 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
18497 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
18498 warning message if user requests -maltivec=be.
18499 * doc/invoke.texi: Document deprecation of -maltivec=be.
18501 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
18504 * config/rs6000/rs6000-c.c: Update definitions for
18505 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
18506 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
18508 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
18511 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
18512 add _get_ssp intrinsics. Remove argument from
18513 __builtin_ia32_rdssp[d|q].
18514 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
18515 * config/i386/i386-builtin.def: Remove argument from
18516 __builtin_ia32_rdssp[d|q].
18517 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
18518 ix86_expand_special_args_builtin for _rdssp[d|q].
18519 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
18520 Clear register before usage.
18521 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
18522 Add documentation for new _get_ssp and _inc_ssp intrinsics.
18524 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
18526 PR tree-optimization/84357
18527 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
18528 operand 1 of an ARRAY_REF too.
18530 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
18533 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
18534 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
18536 (set_of_reg): New struct.
18537 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
18538 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
18539 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
18541 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
18542 Split into bitclr, bitset, bitinvert patterns if appropriate.
18543 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
18544 use rx_fuse_in_memory_bitop.
18545 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
18546 to named insn, correct maximum insn length.
18548 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
18551 * machmode.def: Define a complex mode for PARTIAL_INT.
18552 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
18554 * doc/rtl.texi: Document CSPImode.
18555 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
18557 (msp430_hard_regno_nregs_with_padding): Likewise.
18559 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
18562 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
18564 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
18566 PR rtl-optimization/84169
18567 * combine.c (try_combine): New variable split_i2i3. Set it to true if
18568 we generated a parallel as new i3 and we split that to new i2 and i3
18569 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
18570 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
18571 those to i2, not i1. Partially rewrite this scan code.
18573 2018-02-13 Jakub Jelinek <jakub@redhat.com>
18576 * stor-layout.c (place_field): For variable length fields, adjust
18577 offset_align afterwards not just based on the field's alignment,
18578 but also on the size.
18580 PR middle-end/84309
18581 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
18582 of exps and logs in the use_exp2 case.
18584 2018-02-13 Jeff Law <law@redhat.com>
18586 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
18587 entry for "vector".
18589 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
18592 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
18596 * common.opt (gas-loc-support, gas-locview-support): New.
18597 (ginline-points, ginternal-reset-location-views): New.
18598 * doc/invoke.texi: Document them. Use @itemx where intended.
18599 (gvariable-location-views): Adjust.
18600 * target.def (reset_location_view): New.
18601 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
18602 (TARGET_RESET_LOCATION_VIEW): New.
18603 * doc/tm.texi: Rebuilt.
18604 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
18605 (dwarf2out_default_as_locview_support): New.
18606 (output_asm_line_debug_info): Use option variables.
18607 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
18608 (output_loc_list): Likewise.
18609 (add_high_low_attributes): Check option variables.
18610 Don't output entry view attribute in strict mode.
18611 (gen_inlined_subroutine_die): Check option variables.
18612 (dwarf2out_inline_entry): Likewise.
18613 (init_sections_and_labels): Likewise.
18614 (dwarf2out_early_finish): Likewise.
18615 (maybe_reset_location_view): New, from...
18616 (dwarf2out_var_location): ... here. Call it.
18617 * debug.h (dwarf2out_default_as_loc_support): Declare.
18618 (dwarf2out_default_as_locview_support): Declare.
18619 * hooks.c (hook_int_rtx_insn_0): New.
18620 * hooks.h (hook_int_rtx_insn_0): Declare.
18621 * toplev.c (process_options): Take -gas-loc-support and
18622 -gas-locview-support from dwarf2out. Enable
18623 -gvariable-location-views by default only with locview
18624 assembler support. Enable -ginternal-reset-location-views by
18625 default only if the target defines the corresponding hook.
18626 Enable -ginline-points by default if location views are
18627 enabled; force it disabled if statement frontiers are
18629 * tree-inline.c (expand_call_inline): Check option variables.
18630 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
18632 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
18634 PR tree-optimization/84321
18635 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
18636 handling. Also check whether the anti-range contains any values
18637 that satisfy the mask; switch to a VR_RANGE if not.
18639 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
18642 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
18644 2018-02-13 Martin Jambor <mjambor@suse.cz>
18647 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
18648 of call statements, also set location of a load to a temporary.
18650 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
18652 * config/rl78/rl78.c (add_vector_labels): New function.
18653 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
18654 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
18655 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
18656 which checks that no arguments are passed.
18657 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
18658 * doc/extend.texi: Documentation for the new attribute.
18660 2018-02-13 Andreas Schwab <schwab@suse.de>
18662 * config/riscv/linux.h (CPP_SPEC): Define.
18664 2018-02-13 Jakub Jelinek <jakub@redhat.com>
18667 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
18668 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
18669 OPTION_MASK_ISA_AES as first argument to def_builtin_const
18670 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
18671 instead of OPTION_MASK_ISA_PCLMUL as first argument to
18672 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
18673 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
18674 temporarily for AES and PCLMUL builtins.
18676 PR tree-optimization/84339
18677 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
18678 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
18681 PR middle-end/84309
18682 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
18683 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
18684 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
18686 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
18688 * omp-simd-clone.h: New file.
18689 * omp-simd-clone.c: Include omp-simd-clone.h.
18690 (expand_simd_clones): No longer static.
18691 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
18692 cgraph.h and omp-simd-clone.h.
18693 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
18694 (vect_recog_widen_shift_pattern): Formatting fix.
18695 (vect_pattern_recog_1): Don't check optab for calls.
18698 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
18699 operands[2] into a REG before using gen_lowpart on it.
18701 2018-02-12 Jeff Law <law@redhat.com>
18704 * config/sh/sh.c (find_barrier): Consider a sibling call
18707 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
18708 successfully back substituting a reg.
18710 2018-02-12 Richard Biener <rguenther@suse.de>
18712 PR tree-optimization/84037
18713 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
18714 parameter, move visited init to caller.
18715 (vect_slp_analyze_operations): Separate cost from validity
18716 check, initialize visited once for all instances.
18717 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
18719 * tree-vect-stmts.c (vect_model_simple_cost): Make early
18721 (vect_model_promotion_demotion_cost): Likewise.
18722 (vectorizable_bswap): Guard cost modeling with !slp_node
18723 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
18725 (vectorizable_call): Likewise.
18726 (vectorizable_conversion): Likewise.
18727 (vectorizable_assignment): Likewise.
18728 (vectorizable_shift): Likewise.
18729 (vectorizable_operation): Likewise.
18730 (vectorizable_store): Likewise.
18731 (vectorizable_load): Likewise.
18732 (vectorizable_condition): Likewise.
18733 (vectorizable_comparison): Likewise.
18735 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
18738 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
18739 (ASAN_MARK): Fix fnspec to account for return value, change pointer
18740 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
18742 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
18744 PR middle-end/83665
18745 * params.def (inline-min-speedup): Increase from 8 to 15.
18746 (max-inline-insns-auto): Decrease from 40 to 30.
18747 * ipa-split.c (consider_split): Add some buffer for function to
18748 be considered inlining candidate.
18749 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
18752 2018-02-12 Richard Biener <rguenther@suse.de>
18754 PR tree-optimization/84037
18755 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
18756 matched stmts if we cannot swap the non-matched ones.
18758 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
18760 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
18761 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
18762 _mm_maskz_scalef_round_ss): New intrinsics.
18763 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
18764 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
18765 __builtin_ia32_scalefss_round): Remove.
18766 (__builtin_ia32_scalefsd_mask_round,
18767 __builtin_ia32_scalefss_mask_round): New intrinsics.
18768 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
18769 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
18770 ((match_operand:VF_128 2 "<round_nimm_predicate>"
18771 "<round_constraint>")): Changed to ...
18772 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
18773 "<round_scalar_constraint>")): ... this.
18774 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
18775 %0, %1, %2<round_op3>}"): Changed to ...
18776 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
18777 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
18778 %2<round_scalar_mask_op3>}"): ... this.
18779 * config/i386/subst.md (round_scalar_nimm_predicate): New.
18781 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
18783 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
18784 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
18785 (_mm_maskz_sqrt_round_ss): New intrinsics.
18786 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
18787 (__builtin_ia32_sqrtsd_mask_round)
18788 (__builtin_ia32_sqrtss_mask_round): New builtins.
18789 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
18790 (__builtin_ia32_sqrtss_round): Remove.
18791 (__builtin_ia32_sqrtsd_mask_round)
18792 (__builtin_ia32_sqrtss_mask_round): New builtins.
18793 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
18794 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
18795 ((match_operand:VF_128 1 "vector_operand"
18796 "xBm,<round_constraint>")): Changed to ...
18797 ((match_operand:VF_128 1 "vector_operand"
18798 "xBm,<round_scalar_constraint>")): ... this.
18799 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
18800 %0, %2, %<iptr>1<round_op3>}): Changed to ...
18801 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
18802 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
18803 %<iptr>1<round_scalar_mask_op3>}): ... this.
18804 ((set_attr "prefix" "<round_prefix>")): Changed to ...
18805 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
18807 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
18810 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
18811 Cast vec_cmpeq result to correct type.
18812 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
18813 Cast vec_cmpgt result to correct type.
18815 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
18817 * final.c (final_scan_insn_1): Renamed from...
18818 (final_scan_insn): ... this. New wrapper, to recover
18819 seen from the outermost call in recursive ones.
18820 * config/sparc/sparc.c (output_return): Drop seen from call.
18821 (output_sibcall): Likewise.
18822 * config/visium/visium.c (output_branch): Likewise.
18824 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
18826 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
18829 2018-02-10 Alan Modra <amodra@gmail.com>
18832 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
18833 Specify LR as an input.
18835 2018-02-10 Jakub Jelinek <jakub@redhat.com>
18838 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
18839 remove_member_access_dummy_vars): New functions.
18840 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
18841 lower_omp_1, execute_lower_omp): Use them.
18843 PR rtl-optimization/84308
18844 * shrink-wrap.c (spread_components): Release todo vector.
18846 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
18848 PR rtl-optimization/57193
18849 * ira-color.c (struct allocno_color_data): Add member
18850 conflict_allocno_hard_prefs.
18851 (update_conflict_allocno_hard_prefs): New.
18852 (bucket_allocno_compare_func): Add a preference based on
18853 conflict_allocno_hard_prefs.
18854 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
18855 (color_allocnos): Remove a dead code. Initiate
18856 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
18858 2018-02-09 Jakub Jelinek <jakub@redhat.com>
18861 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
18862 constraint from =wa to wa. Avoid a subreg on the output operand,
18863 instead use a pseudo and subreg it in a move.
18864 (p9_xxbrd_<mode>): Changed to ...
18865 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
18866 (p9_xxbrd_v2df): New expander.
18867 (p9_xxbrw_<mode>): Changed to ...
18868 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
18869 (p9_xxbrw_v4sf): New expander.
18871 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
18873 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
18875 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
18878 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
18879 multiply in 32-bit mode.
18880 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
18881 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
18884 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
18886 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
18887 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
18888 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
18889 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
18891 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
18894 * dwarf2out.c (is_trivial_indirect_ref): New function.
18895 (dwarf2out_late_global_decl): Do not generate a location
18896 attribute for variables that have a non-trivial DECL_VALUE_EXPR
18897 and that are not defined in the current unit.
18899 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
18901 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
18902 instead of a libcall for UNORDERED.
18904 2018-02-09 Tamar Christina <tamar.christina@arm.com>
18907 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
18908 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
18910 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18913 * config/s390/s390.c (s390_set_current_function): Invoke
18914 s390_indirect_branch_settings also if fndecl didn't change.
18916 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
18918 * config/rs6000/rs6000.md (blockage): Set length to zero.
18920 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
18922 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
18924 2018-02-09 Jakub Jelinek <jakub@redhat.com>
18927 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
18928 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
18932 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
18933 PARALLEL incoming that failed vt_get_decl_and_offset check.
18935 PR middle-end/84237
18936 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
18937 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
18939 (get_variable_section): For decls in named .bss* sections pass true as
18940 second argument to bss_initializer_p.
18942 2018-02-09 Marek Polacek <polacek@redhat.com>
18943 Jakub Jelinek <jakub@redhat.com>
18946 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
18947 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
18948 Sync some changes from cxx_fold_indirect_ref.
18950 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
18952 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
18954 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
18955 (BLOCK_INLINE_ENTRY_LABEL): New.
18956 (dwarf2out_var_location): Disregard inline entry markers.
18957 (inline_entry_data): New struct.
18958 (inline_entry_data_hasher): New hashtable type.
18959 (inline_entry_data_hasher::hash): New.
18960 (inline_entry_data_hasher::equal): New.
18961 (inline_entry_data_table): New variable.
18962 (add_high_low_attributes): Add DW_AT_entry_pc and
18963 DW_AT_GNU_entry_view attributes if a pending entry is found
18964 in inline_entry_data_table. Add old entry_pc attribute only
18965 if debug nonbinding markers are disabled.
18966 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
18967 markers are enabled.
18968 (block_within_block_p, dwarf2out_inline_entry): New.
18969 (dwarf2out_finish): Check that no entries remained in
18970 inline_entry_data_table.
18971 * final.c (reemit_insn_block_notes): Handle inline entry notes.
18972 (final_scan_insn, notice_source_line): Likewise.
18973 (rest_of_clean_state): Skip inline entry markers.
18974 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
18976 * gimple.c (gimple_build_debug_inline_entry): New.
18977 * gimple.h (enum gimple_debug_subcode): Add
18978 GIMPLE_DEBUG_INLINE_ENTRY.
18979 (gimple_build_debug_inline_entry): Declare.
18980 (gimple_debug_inline_entry_p): New.
18981 (gimple_debug_nonbind_marker_p): Adjust.
18982 * insn-notes.def (INLINE_ENTRY): New.
18983 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
18984 inline entry marker notes.
18985 (print_insn): Likewise.
18986 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
18987 (INSN_DEBUG_MARKER_KIND): Likewise.
18988 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
18989 * tree-inline.c (expand_call_inline): Build and insert
18990 debug_inline_entry stmt.
18991 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
18992 inline entry blocks early, if nonbind markers are enabled.
18993 (dump_scope_block): Dump fragment info.
18994 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
18995 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
18996 (gimple_build_debug_inline_entry): New.
18997 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
18998 Enable/disable inline entry points too.
18999 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
19000 (DEBUG_INSN): Describe inline entry markers.
19002 * common.opt (gvariable-location-views): New.
19003 (gvariable-location-views=incompat5): New.
19004 * config.in: Rebuilt.
19005 * configure: Rebuilt.
19006 * configure.ac: Test assembler for view support.
19007 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
19008 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
19009 * dwarf2out.c (var_loc_view): New typedef.
19010 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
19011 (dwarf2out_locviews_in_attribute): New.
19012 (dwarf2out_locviews_in_loclist): New.
19013 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
19014 (enum dw_line_info_opcode): Add LI_adv_address.
19015 (struct dw_line_info_table): Add view.
19016 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
19017 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
19018 (zero_view_p): New variable.
19019 (ZERO_VIEW_P): New macro.
19020 (output_asm_line_debug_info): New.
19021 (struct var_loc_node): Add view.
19022 (add_AT_view_list, AT_loc_list): New.
19023 (add_var_loc_to_decl): Add view param. Test it against last.
19024 (new_loc_list): Add view params. Record them.
19025 (AT_loc_list_ptr): Handle loc and view lists.
19026 (view_list_to_loc_list_val_node): New.
19027 (print_dw_val): Handle dw_val_class_view_list.
19028 (size_of_die): Likewise.
19029 (value_format): Likewise.
19030 (loc_list_has_views): New.
19031 (gen_llsym): Set vl_symbol too.
19032 (maybe_gen_llsym, skip_loc_list_entry): New.
19033 (dwarf2out_maybe_output_loclist_view_pair): New.
19034 (output_loc_list): Output view list or entries too.
19035 (output_view_list_offset): New.
19036 (output_die): Handle dw_val_class_view_list.
19037 (output_dwarf_version): New.
19038 (output_compilation_unit_header): Use it.
19039 (output_skeleton_debug_sections): Likewise.
19040 (output_rnglists, output_line_info): Likewise.
19041 (output_pubnames, output_aranges): Update version comments.
19042 (output_one_line_info_table): Output view numbers in asm comments.
19043 (dw_loc_list): Determine current endview, pass it to new_loc_list.
19044 Call maybe_gen_llsym.
19045 (loc_list_from_tree_1): Adjust.
19046 (add_AT_location_description): Create view list attribute if
19047 needed, check it's absent otherwise.
19048 (convert_cfa_to_fb_loc_list): Adjust.
19049 (maybe_emit_file): Call output_asm_line_debug_info for test.
19050 (dwarf2out_var_location): Reset views as needed. Precompute
19051 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
19052 attribute. Set view.
19053 (new_line_info_table): Reset next view.
19054 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
19055 (dwarf2out_source_line): Likewise. Output view resets and labels to
19056 the assembler, or select appropriate line info opcodes.
19057 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
19058 (optimize_string_length): Catch it. Adjust.
19059 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
19060 dw_val_class_view_list, and remove it if no longer needed.
19061 (hash_loc_list): Hash view numbers.
19062 (loc_list_hasher::equal): Compare them.
19063 (optimize_location_lists): Check whether a view list symbol is
19064 needed, and whether the locview attribute is present, and
19065 whether they match. Remove the locview attribute if no longer
19067 (index_location_lists): Call skip_loc_list_entry for test.
19068 (dwarf2out_finish): Call output_asm_line_debug_info for test.
19069 Use output_dwarf_version.
19070 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
19071 (struct dw_val_node): Add val_view_list.
19072 * final.c (SEEN_NEXT_VIEW): New.
19073 (set_next_view_needed): New.
19074 (clear_next_view_needed): New.
19075 (maybe_output_next_view): New.
19076 (final_start_function): Rename to...
19077 (final_start_function_1): ... this. Take pointer to FIRST,
19078 add SEEN parameter. Emit param bindings in the initial view.
19079 (final_start_function): Reintroduce SEEN-less interface.
19080 (final): Rename to...
19081 (final_1): ... this. Take SEEN parameter. Output final pending
19082 next view at the end.
19083 (final): Reintroduce seen-less interface.
19084 (final_scan_insn): Output pending next view before switching
19085 sections or ending a block. Mark the next view as needed when
19086 outputting variable locations. Notify debug backend of section
19087 changes, and of location view changes.
19088 (rest_of_handle_final): Adjust.
19089 * toplev.c (process_options): Autodetect value for debug variable
19090 location views option. Warn on incompat5 without -gdwarf-5.
19091 * doc/invoke.texi (gvariable-location-views): New.
19092 (gvariable-location-views=incompat5): New.
19093 (gno-variable-location-views): New.
19095 2018-02-08 David Malcolm <dmalcolm@redhat.com>
19097 PR tree-optimization/84136
19098 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
19099 that the result of find_edge is non-NULL.
19101 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
19104 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
19105 storing integer register in SImode. Fix cost of 256 and 512
19106 byte aligned SSE register store.
19108 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
19110 * config/i386/i386.c (ix86_multiplication_cost): Fix
19111 multiplication cost for TARGET_AVX512DQ.
19113 2018-02-08 Marek Polacek <polacek@redhat.com>
19115 PR tree-optimization/84238
19116 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
19119 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19121 PR tree-optimization/84265
19122 * tree-vect-stmts.c (vectorizable_store): Don't treat
19123 VMAT_CONTIGUOUS accesses as grouped.
19124 (vectorizable_load): Likewise.
19126 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19128 PR tree-optimization/81635
19129 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
19130 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
19131 (test_round_for_mask): New functions.
19132 (wide_int_cc_tests): Call test_round_for_mask.
19133 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
19134 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
19135 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
19136 range returned by get_range_info.
19138 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
19141 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
19142 * symtab.c: Include builtins.h
19143 (symtab_node::output_to_lto_symbol_table_p): Move here
19144 from lto-streamer-out.c:output_symbol_p.
19145 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
19146 (output_symbol_p): Move all logic to symtab.c
19147 (produce_symtab): Update.
19149 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19151 * config/s390/s390-opts.h (enum indirect_branch): Define.
19152 * config/s390/s390-protos.h (s390_return_addr_from_memory)
19153 (s390_indirect_branch_via_thunk)
19154 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
19155 (enum s390_indirect_branch_type): Define.
19156 * config/s390/s390.c (struct s390_frame_layout, struct
19157 machine_function): Remove.
19158 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
19159 (indirect_branch_table_label_no, indirect_branch_table_name):
19161 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
19162 (enum s390_indirect_branch_option): Define.
19163 (s390_return_addr_from_memory): New function.
19164 (s390_handle_string_attribute): New function.
19165 (s390_attribute_table): Add new attribute handler.
19166 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
19167 (s390_indirect_branch_via_thunk): New function.
19168 (s390_indirect_branch_via_inline_thunk): New function.
19169 (s390_function_ok_for_sibcall): When jumping via thunk disallow
19170 sibling call optimization for non z10 compiles.
19171 (s390_emit_call): Force indirect branch target to be a single
19172 register. Add r1 clobber for non-z10 compiles.
19173 (s390_emit_epilogue): Emit return jump via return_use expander.
19174 (s390_reorg): Handle JUMP_INSNs as execute targets.
19175 (s390_option_override_internal): Perform validity checks for the
19176 new command line options.
19177 (s390_indirect_branch_attrvalue): New function.
19178 (s390_indirect_branch_settings): New function.
19179 (s390_set_current_function): Invoke s390_indirect_branch_settings.
19180 (s390_output_indirect_thunk_function): New function.
19181 (s390_code_end): Implement target hook.
19182 (s390_case_values_threshold): Implement target hook.
19183 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
19185 * config/s390/s390.h (struct s390_frame_layout)
19186 (struct machine_function): Move here from s390.c.
19187 (TARGET_INDIRECT_BRANCH_NOBP_RET)
19188 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
19189 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
19190 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
19191 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
19192 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
19193 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
19194 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
19195 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
19196 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
19197 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
19198 (mnemonic attribute): Add values which aren't recognized
19200 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
19201 pattern for branch conversion. Fix mnemonic attribute.
19202 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
19203 indirect branch via thunk if requested.
19204 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
19205 ("*indirect_jump"): Disable for branch conversion using out of
19207 ("indirect_jump_via_thunk<mode>_z10")
19208 ("indirect_jump_via_thunk<mode>")
19209 ("indirect_jump_via_inlinethunk<mode>_z10")
19210 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
19211 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
19212 ("casesi_jump_via_inlinethunk<mode>_z10")
19213 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
19214 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
19215 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
19216 ("*indirect2_jump"): Disable for branch conversion.
19217 ("casesi_jump"): Turn into expander and expand patterns for branch
19219 ("return_use"): New expander.
19220 ("*return"): Emit return via thunk and rename it to ...
19221 ("*return<mode>"): ... this one.
19222 * config/s390/s390.opt: Add new options and and enum for the
19225 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19227 * lra-constraints.c (match_reload): Unconditionally use
19228 gen_lowpart_SUBREG, rather than selecting between that
19229 and equivalent gen_rtx_SUBREG code.
19231 2018-02-08 Richard Biener <rguenther@suse.de>
19233 PR tree-optimization/84233
19234 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
19235 changed flag instead of boguously re-using phi_inserted.
19237 2018-02-08 Martin Jambor <mjambor@suse.cz>
19239 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
19240 static local variables.
19242 2018-02-08 Richard Biener <rguenther@suse.de>
19244 PR tree-optimization/84278
19245 * tree-vect-stmts.c (vectorizable_store): When looking for
19246 smaller vector types to perform grouped strided loads/stores
19247 make sure the mode is supported by the target.
19248 (vectorizable_load): Likewise.
19250 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
19252 * config/aarch64/aarch64.c (aarch64_components_for_bb):
19253 Increase LDP/STP opportunities by adding adjacent callee-saves.
19255 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
19257 PR rtl-optimization/84068
19258 PR rtl-optimization/83459
19259 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
19261 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
19263 PR tree-optimization/84224
19264 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
19265 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
19266 non-zero arguments.
19268 2018-02-07 Iain Sandoe <iain@codesourcery.com>
19271 * config/rs6000/altivec.md (*restore_world): Remove LR use.
19272 * config/rs6000/predicates.md (restore_world_operation): Adjust op
19273 count, remove one USE.
19275 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19277 * doc/install.texi (Configuration): Document the
19278 --with-long-double-format={ibm,ieee} PowerPC configuration
19282 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
19283 Convert from define_expand to be define_insn_and_split. Rework
19284 float/double/_Float128 conversions to QI/HI/SImode to work with
19285 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
19286 conversions to QI/HImode types did a store and then a load to
19287 truncate the value. For conversions to VSX registers, don't split
19288 the insn, instead emit the code directly. Use the code iterator
19289 any_fix to combine signed and unsigned conversions.
19290 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
19291 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
19292 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
19293 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
19294 (fix_<mode>di2_hw): Likewise.
19295 (fixuns_<mode>di2_hw): Likewise.
19296 (fix_<mode>si2_hw): Likewise.
19297 (fixuns_<mode>si2_hw): Likewise.
19298 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
19299 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
19300 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
19301 fix<uns>_trunc<SFDF:mode>si2_p8.
19302 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
19303 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19304 (fix<uns>_<mode>_mem): Likewise.
19305 (fctiw<u>z_<mode>_mem): Likewise.
19306 (fix<uns>_<mode>_mem): Likewise.
19307 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
19308 the register allocator from doing a direct move to the GPRs to do
19309 a store, and instead use the ISA 3.0 store byte/half-word from
19310 vector register instruction. For IEEE 128-bit floating point,
19311 also optimize stores of 32-bit ints.
19312 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
19314 2018-02-07 Alan Hayward <alan.hayward@arm.com>
19316 * genextract.c (push_pathstr_operand): New function to support
19318 (walk_rtx): Call push_pathstr_operand.
19319 (print_path): Support [a-zA-Z].
19321 2018-02-07 Richard Biener <rguenther@suse.de>
19323 PR tree-optimization/84037
19324 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
19325 (cse_and_gimplify_to_preheader): Declare.
19326 (vect_get_place_in_interleaving_chain): Likewise.
19327 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
19329 (_loop_vec_info::~_loop_vec_info): Delete it.
19330 (cse_and_gimplify_to_preheader): New function.
19331 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
19332 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
19333 (vectorizable_load): Likewise. For grouped stores always base
19334 the IV on the first element.
19335 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
19336 condition before gimplifying.
19338 2018-02-07 Jakub Jelinek <jakub@redhat.com>
19340 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
19341 *DIV_EXPR and *MOD_EXPR.
19343 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
19346 * config/i386/i386.c (ix86_option_override_internal): Mask out
19347 the CF_SET bit when checking -fcf-protection.
19349 2018-02-07 Tom de Vries <tom@codesourcery.com>
19352 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
19355 2018-02-07 Richard Biener <rguenther@suse.de>
19357 PR tree-optimization/84204
19358 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
19361 PR tree-optimization/84205
19362 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
19363 special-case isl_ast_op_zdiv_r.
19365 PR tree-optimization/84223
19366 * graphite-scop-detection.c (gather_bbs::before_dom_children):
19367 Only add conditions from within the region.
19368 (gather_bbs::after_dom_children): Adjust.
19370 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
19373 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
19374 * config/avr/avr.md: Only post-reload split REG-REG moves if
19375 either register is GENERAL_REG_P.
19377 2018-02-07 Jakub Jelinek <jakub@redhat.com>
19379 PR tree-optimization/84235
19380 * tree-ssa-scopedtables.c
19381 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
19382 if the subtraction is performed in floating point type where NaNs are
19383 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
19384 build 1. Formatting fix.
19386 2018-02-06 Jakub Jelinek <jakub@redhat.com>
19389 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
19390 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
19391 and skip it regardless of bb boundaries. Use CALL_P macro,
19392 don't test INSN_P (insn) together with CALL_P or JUMP_P check
19393 unnecessarily, formatting fix.
19395 2018-02-06 Michael Collison <michael.collison@arm.com>
19397 * config/arm/thumb2.md:
19398 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
19399 (*thumb_mov_notscc): Ditto.
19401 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
19404 * config/rs6000/rs6000.md (su code attribute): Use "u" for
19405 unsigned_fix, not "s".
19407 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19409 * configure.ac (gcc_fn_eh_frame_ro): New function.
19410 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
19411 correct .eh_frame permissions.
19412 * configure: Regenerate.
19414 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
19416 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
19417 irrelevant options.
19419 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19421 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19422 Display warning message for -mno-speculate-indirect-jumps.
19424 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
19426 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
19428 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
19430 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
19432 PR tree-optimization/84225
19433 * tree-eh.c (find_trapping_overflow): Only call
19434 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
19436 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
19439 * config/i386/i386.c: Reimplement the check of possible options
19440 -mibt/-mshstk conbination. Change error messages.
19441 * doc/invoke.texi: Fix a typo: remove extra '='.
19443 2018-02-06 Marek Polacek <polacek@redhat.com>
19445 PR tree-optimization/84228
19446 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
19448 2018-02-06 Tamar Christina <tamar.christina@arm.com>
19451 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
19452 emitted arch directives.
19453 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
19454 __ARM_FEATURE_COPROC before changing architectures.
19456 2018-02-06 Richard Biener <rguenther@suse.de>
19458 * config/i386/i386.c (print_reg): Fix typo.
19459 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
19461 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
19463 * configure: Regenerate.
19465 2018-02-05 Martin Sebor <msebor@redhat.com>
19467 PR tree-optimization/83369
19468 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
19471 2018-02-05 Martin Liska <mliska@suse.cz>
19473 * doc/invoke.texi: Cherry-pick upstream r323995.
19475 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
19477 * ira.c (ira_init_register_move_cost): Adjust comment.
19479 2018-02-05 Martin Liska <mliska@suse.cz>
19481 PR gcov-profile/84137
19482 * doc/gcov.texi: Fix typo in documentation.
19484 2018-02-05 Martin Liska <mliska@suse.cz>
19486 PR gcov-profile/83879
19487 * doc/gcov.texi: Document necessity of --dynamic-list-data when
19488 using dlopen functionality.
19490 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
19492 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
19493 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
19494 _mm_maskz_range_ss, _mm_mask_range_round_ss,
19495 _mm_maskz_range_round_ss): New intrinsics.
19496 (__builtin_ia32_rangesd128_round)
19497 (__builtin_ia32_rangess128_round): Remove.
19498 (__builtin_ia32_rangesd128_mask_round,
19499 __builtin_ia32_rangess128_mask_round): New builtins.
19500 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
19501 __builtin_ia32_rangess128_round): Remove.
19502 (__builtin_ia32_rangesd128_mask_round,
19503 __builtin_ia32_rangess128_mask_round): New builtins.
19504 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
19505 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
19506 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
19507 "<round_saeonly_constraint>")): Changed to ...
19508 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
19509 "<round_saeonly_scalar_constraint>")): ... this.
19510 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
19511 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
19512 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
19513 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
19514 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
19516 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
19518 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
19520 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
19521 Remove all values except native, 8540 and 8548.
19523 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
19525 * config/i386/i386.c (ix86_output_function_return): Pass
19526 INVALID_REGNUM, instead of -1, as invalid register number to
19527 indirect_thunk_name and output_indirect_thunk.
19529 2018-02-02 Julia Koval <julia.koval@intel.com>
19531 * config.gcc: Add -march=icelake.
19532 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
19533 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
19534 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
19535 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
19536 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
19537 (processor_target_table): Add icelake.
19538 (ix86_option_override_internal): Handle new PTAs.
19539 (get_builtin_code_for_version): Handle icelake.
19540 (M_INTEL_COREI7_ICELAKE): New.
19541 (fold_builtin_cpu): Handle icelake.
19542 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
19543 * doc/invoke.texi: Add -march=icelake.
19545 2018-02-02 Julia Koval <julia.koval@intel.com>
19547 * config/i386/i386.c (ix86_option_override_internal): Change flags type
19548 to wide_int_bitmask.
19549 * wide-int-bitmask.h: New.
19551 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
19554 * config/i386/i386.md: Replace Pmode with word_mode in
19555 builtin_setjmp_setup and builtin_longjmp to support x32.
19557 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
19561 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
19563 (rs6000_supported_cpu_names): New static variable.
19564 (linux_cpu_translation_table): Likewise.
19565 (elf_platform) <cpu>: Define new static variable and use it.
19566 Translate kernel AT_PLATFORM name to canonical name if needed.
19567 Error if platform name is unknown.
19569 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
19572 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
19574 2018-02-01 Jeff Law <law@redhat.com>
19577 * config/i386/i386.c (release_scratch_register_on_entry): Add new
19578 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
19579 the scratch if RELEASE_VIA_POP is false.
19580 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
19581 If we have to save a temporary register, decrement SIZE appropriately.
19582 Pass new arguments to release_scratch_register_on_entry.
19583 (ix86_adjust_stack_and_probe): Likewise.
19584 (ix86_emit_probe_stack_range): Pass new arguments to
19585 release_scratch_register_on_entry.
19587 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
19589 PR rtl-optimization/84157
19590 * combine.c (change_zero_ext): Use REG_P predicate in
19591 front of HARD_REGISTER_P predicate.
19593 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
19595 * config/avr/avr.c (avr_option_override): Move disabling of
19596 -fdelete-null-pointer-checks to...
19597 * common/config/avr/avr-common.c (avr_option_optimization_table):
19600 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19602 PR tree-optimization/81635
19603 * tree-data-ref.c (split_constant_offset_1): For types that
19604 wrap on overflow, try to use range info to prove that wrapping
19607 2018-02-01 Renlin Li <renlin.li@arm.com>
19610 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
19611 TAILCALL_ADDR_REGS.
19612 (aarch64_register_move_cost): Likewise.
19613 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
19614 TAILCALL_ADDR_REGS.
19615 (REG_CLASS_NAMES): Likewise.
19616 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
19617 TAILCALL_ADDR_REGS. Remove IP registers.
19618 * config/aarch64/aarch64.md (Ucs): Update register constraint.
19620 2018-02-01 Richard Biener <rguenther@suse.de>
19622 * domwalk.h (dom_walker::dom_walker): Add additional constructor
19623 for specifying RPO order and allow NULL for that.
19624 * domwalk.c (dom_walker::dom_walker): Likewise.
19625 (dom_walker::walk): Handle NULL RPO order.
19626 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
19628 (rewrite_update_dom_walker): Likewise.
19629 (mark_def_dom_walker): Likewise.
19631 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19633 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
19634 (aarch64_maybe_expand_sve_subreg_move): Declare.
19635 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
19636 * config/aarch64/predicates.md (aarch64_any_register_operand): New
19638 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
19639 that are semantically a reverse operation.
19640 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
19641 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
19642 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
19644 (aarch64_can_change_mode_class): For big-endian, forbid changes
19645 between two SVE modes if they have different element sizes.
19647 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19649 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
19650 the TImode handling for big-endian targets.
19652 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19654 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
19655 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
19657 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
19658 Remove BSWAP handing for big-endian targets and use the form of
19659 LD1RQ appropariate for the mode.
19661 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19663 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
19664 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
19665 duplicated element.
19667 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19670 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
19671 check for operands that need to go through aarch64_sve_reload_be.
19673 2018-02-01 Jakub Jelinek <jakub@redhat.com>
19675 PR tree-optimization/81661
19676 PR tree-optimization/84117
19677 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
19678 * tree-eh.c: Include gimplify.h.
19679 (find_trapping_overflow, replace_trapping_overflow,
19680 rewrite_to_non_trapping_overflow): New functions.
19681 * tree-vect-loop.c: Include tree-eh.h.
19682 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
19683 * tree-data-ref.c: Include tree-eh.h.
19684 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
19686 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
19688 PR rtl-optimization/84123
19689 * combine.c (change_zero_ext): Check if hard register satisfies
19690 can_change_dest_mode before calling gen_lowpart_SUBREG.
19692 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
19695 * ira.c (ira_init_register_move_cost): Remove assert.
19697 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
19699 PR rtl-optimization/84071
19700 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
19701 * doc/tm.texi: Regenerate.
19703 2018-01-31 Richard Biener <rguenther@suse.de>
19705 PR tree-optimization/84132
19706 * tree-data-ref.c (analyze_miv_subscript): Properly
19707 check whether evolution_function_is_affine_multivariate_p
19708 before calling gcd_of_steps_may_divide_p.
19710 2018-01-31 Julia Koval <julia.koval@intel.com>
19713 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
19714 * config/i386/i386.md (rdpid_rex64) New.
19715 (rdpid): Make 32bit only.
19717 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
19720 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
19721 an IDENTIFIER_NODE for FUNCTION_TYPE's.
19723 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
19726 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
19728 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
19730 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
19732 PR rtl-optimization/84071
19733 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
19734 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
19736 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
19738 * config/arc/arc.c (arc_handle_aux_attribute): New function.
19739 (arc_attribute_table): Add 'aux' attribute.
19740 (arc_in_small_data_p): Consider aux like variables.
19741 (arc_is_aux_reg_p): New function.
19742 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
19743 (arc_get_aux_arg): New function.
19744 (prepare_move_operands): Handle aux-register access.
19745 (arc_handle_aux_attribute): New function.
19746 * doc/extend.texi (ARC Variable attributes): Add subsection.
19748 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
19750 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
19751 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
19752 (arc_attribute_table): Add 'uncached' attribute.
19753 (arc_print_operand): Print '.di' flag for uncached memory
19755 (arc_in_small_data_p): Do not consider for small data the uncached
19757 (arc_is_uncached_mem_p): New function.
19758 * config/arc/predicates.md (compact_store_memory_operand): Check
19759 for uncached memory accesses.
19760 (nonvol_nonimm_operand): Likewise.
19761 * doc/extend.texi (ARC Type Attribute): New subsection.
19763 2018-01-31 Jakub Jelinek <jakub@redhat.com>
19766 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
19767 falign-loops=): Add Optimization flag.
19769 2018-01-30 Jeff Law <law@redhat.com>
19772 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
19773 INT_REGISTERS_SAVED. Check it prior to calling
19774 get_scratch_register_on_entry.
19775 (ix86_adjust_stack_and_probe): Similarly.
19776 (ix86_emit_probe_stack_range): Similarly.
19777 (ix86_expand_prologue): Corresponding changes.
19779 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19782 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
19783 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
19785 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
19788 * lra-constraints.c (curr_insn_transform): Process AND in the
19791 2018-01-30 Jakub Jelinek <jakub@redhat.com>
19793 PR rtl-optimization/83986
19794 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
19795 dependence against last_pending_memory_flush in addition to
19796 pending_jump_insns.
19798 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
19800 PR tree-optimization/81611
19801 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
19804 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19807 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
19810 2018-01-30 Richard Biener <rguenther@suse.de>
19811 Jakub Jelinek <jakub@redhat.com>
19813 PR tree-optimization/84111
19814 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
19815 inner loops added during recursion, as they don't have up-to-date
19818 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
19821 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
19822 (can_inline_edge_by_limits_p): ... here.
19823 (can_early_inline_edge_p, check_callers,
19824 update_caller_keys, update_callee_keys, recursive_inlining,
19825 add_new_edges_to_heap, speculation_useful_p,
19826 inline_small_functions,
19827 inline_small_functions, flatten_function,
19828 inline_to_all_callers_1): Update.
19830 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
19832 * profile-count.c (profile_count::combine_with_ipa_count): Handle
19835 2018-01-30 Richard Biener <rguenther@suse.de>
19837 PR tree-optimization/83008
19838 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
19839 invariant and constant vector uses in stmts when they need
19840 more than one stmt.
19842 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19845 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
19846 * configure: Regenerate.
19848 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
19850 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
19852 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
19853 Use gen_rtx_REG rather than gen_lowpart.
19855 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
19857 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
19858 rather than 0 when creating partial subregs.
19860 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
19862 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
19865 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19868 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
19869 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
19870 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
19871 flags. This restores the settings used before the 2017-07-24.
19872 Turning off pre increment/decrement/modify allows IVOPTS to
19873 optimize DF/SF loops where the index is an int.
19875 2018-01-29 Richard Biener <rguenther@suse.de>
19876 Kelvin Nilsen <kelvin@gcc.gnu.org>
19879 * tree-vect-stmts.c (vectorizable_call): Don't call
19880 targetm.vectorize_builtin_md_vectorized_function if callee is
19883 2018-01-22 Carl Love <cel@us.ibm.com>
19885 * doc/extend.tex: Fix typo in second arg in
19886 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
19888 2018-01-29 Richard Biener <rguenther@suse.de>
19890 PR tree-optimization/84086
19891 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
19892 (flush_ssaname_freelist): When SSA names were released reset
19893 the SCEV hash table.
19895 2018-01-29 Richard Biener <rguenther@suse.de>
19897 PR tree-optimization/84057
19898 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
19899 removed paths when removing edges.
19901 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
19903 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
19904 -mfunction-return=@var{choice}.
19906 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
19908 PR diagnostic/84034
19909 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
19910 Handle CR like TAB.
19911 (layout::print_source_line): Likewise.
19912 (test_get_line_width_without_trailing_whitespace): Add test cases.
19914 2018-01-27 Jakub Jelinek <jakub@redhat.com>
19916 PR middle-end/84040
19917 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
19920 2018-01-26 Jim Wilson <jimw@sifive.com>
19922 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
19924 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
19927 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19929 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
19930 and CMP + SUB-immediate -> SUBS.
19932 2018-01-26 Martin Sebor <msebor@redhat.com>
19934 PR tree-optimization/83896
19935 * tree-ssa-strlen.c (get_string_len): Rename...
19936 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
19937 Avoid assuming length is constant.
19938 (handle_char_store): Use HOST_WIDE_INT for string length.
19940 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
19943 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
19944 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
19946 2018-01-26 Richard Biener <rguenther@suse.de>
19948 PR rtl-optimization/84003
19949 * dse.c (record_store): Only record redundant stores when
19950 the earlier store aliases at least all accesses the later one does.
19952 2018-01-26 Jakub Jelinek <jakub@redhat.com>
19954 PR rtl-optimization/83985
19955 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
19956 REG_CFA_RESTORE insns.
19957 (delete_unmarked_insns): Don't ignore separate shrink wrapping
19958 REG_CFA_RESTORE insns here.
19961 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
19962 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
19964 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
19966 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
19967 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
19968 (arc_init): Likewise.
19969 (arc_override_options): Likewise.
19970 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
19972 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
19974 * config/arc/arc.h (TARGET_DBNZ): Define.
19975 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
19976 properly set the tune attribute.
19977 (dbnz): Use TARGET_DBNZ guard.
19978 * config/arc/arc.opt (mtune): Add core3 option.
19980 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
19982 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
19983 recognize new pic like addresses.
19984 (arc_delegitimize_address): Clean up.
19986 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
19988 * config/arc/arc-arches.def: Option mrf16 valid for all
19990 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
19991 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
19992 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
19993 * config/arc/arc-tables.opt: Regenerate.
19994 * config/arc/arc.c (arc_conditional_register_usage): Handle
19995 reduced register file case.
19996 (arc_file_start): Set must have build attributes.
19997 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
19998 mrf16 option value.
19999 * config/arc/arc.opt (mrf16): Add new option.
20000 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
20001 * config/arc/genmultilib.awk: Handle new mrf16 option.
20002 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
20003 * config/arc/t-multilib: Regenerate.
20004 * doc/invoke.texi (ARC Options): Document mrf16 option.
20006 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20008 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
20009 * config/arc/arc.c (arc_handle_secure_attribute): New function.
20010 (arc_attribute_table): Add 'secure_call' attribute.
20011 (arc_print_operand): Print secure call operand.
20012 (arc_function_ok_for_sibcall): Don't optimize tail calls when
20014 (arc_is_secure_call_p): New function. * config/arc/arc.md
20015 (call_i): Add support for sjli instruction.
20016 (call_value_i): Likewise.
20017 * config/arc/constraints.md (Csc): New constraint.
20019 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20020 John Eric Martin <John.Martin@emmicro-us.com>
20022 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
20023 * config/arc/arc.c (_arc_jli_section): New struct.
20024 (arc_jli_section): New type.
20025 (rc_jli_sections): New static variable.
20026 (arc_handle_jli_attribute): New function.
20027 (arc_attribute_table): Add jli_always and jli_fixed attribute.
20028 (arc_file_end): New function.
20029 (TARGET_ASM_FILE_END): Define.
20030 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
20031 (arc_add_jli_section): New function.
20032 (jli_call_scan): Likewise.
20033 (arc_reorg): Call jli_call_scan.
20034 (arc_output_addsi): Remove 'S' from printing asm operand.
20035 (arc_is_jli_call_p): New function.
20036 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
20038 (movhi_insn): Likewise.
20039 (movsi_insn): Likewise.
20040 (movsi_set_cc_insn): Likewise.
20041 (loadqi_update): Likewise.
20042 (load_zeroextendqisi_update): Likewise.
20043 (load_signextendqisi_update): Likewise.
20044 (loadhi_update): Likewise.
20045 (load_zeroextendhisi_update): Likewise.
20046 (load_signextendhisi_update): Likewise.
20047 (loadsi_update): Likewise.
20048 (loadsf_update): Likewise.
20049 (movsicc_insn): Likewise.
20050 (bset_insn): Likewise.
20051 (bxor_insn): Likewise.
20052 (bclr_insn): Likewise.
20053 (bmsk_insn): Likewise.
20054 (bicsi3_insn): Likewise.
20055 (cmpsi_cc_c_insn): Likewise.
20056 (movsi_ne): Likewise.
20057 (movsi_cond_exec): Likewise.
20058 (clrsbsi2): Likewise.
20059 (norm_f): Likewise.
20069 (call_i): Remove 'S' asm letter, add jli instruction.
20070 (call_value_i): Likewise.
20071 * config/arc/arc.op (mjli-always): New option.
20072 * config/arc/constraints.md (Cji): New constraint.
20073 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
20075 (subsf3_fpx): Likewise.
20076 (mulsf3_fpx): Likewise.
20077 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
20079 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
20080 function attrbutes.
20081 * doc/invoke.texi (ARC): Document mjli-always option.
20083 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
20085 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
20086 avoid addition with 0 and use incw and decw where possible.
20088 2018-01-26 Richard Biener <rguenther@suse.de>
20090 PR tree-optimization/81082
20091 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
20092 association if it requires casting to unsigned.
20093 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
20094 from fold_plusminus_mult_expr to catch important cases late when
20095 range info is available.
20097 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20099 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
20100 * configure.ac (hidden_linkonce): New test.
20101 * configure: Regenerate.
20102 * config.in: Regenerate.
20104 2018-01-26 Julia Koval <julia.koval@intel.com>
20106 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
20107 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
20108 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
20109 _mm_mask_bitshuffle_epi64_mask): Fix type.
20110 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
20111 USI_FTYPE_V4DI_V4DI_USI): Remove.
20112 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
20113 __builtin_ia32_vpshufbitqmb256_mask,
20114 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
20115 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
20116 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
20118 2018-01-26 Alan Modra <amodra@gmail.com>
20121 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
20122 UNSPEC_VBPERMQ. Sort other unspecs.
20124 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
20126 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
20128 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
20130 PR middle-end/83055
20131 * predict.c (drop_profile): Do not push/pop cfun; update also
20133 (handle_missing_profiles): Fix logic looking for zero profiles.
20135 2018-01-25 Jakub Jelinek <jakub@redhat.com>
20137 PR middle-end/83977
20138 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
20139 on functions with #pragma omp declare simd or functions with simd
20141 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
20142 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
20143 Remove trailing \n from warning_at calls.
20145 2018-01-25 Tom de Vries <tom@codesourcery.com>
20148 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
20149 for neutered workers.
20151 2018-01-24 Joseph Myers <joseph@codesourcery.com>
20154 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
20155 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
20157 2018-01-24 Jeff Law <law@redhat.com>
20160 * i386.c (get_probe_interval): Move to earlier point.
20161 (ix86_compute_frame_layout): If -fstack-clash-protection and
20162 the frame is larger than the probe interval, then use pushes
20163 to save registers rather than reg->mem moves.
20164 (ix86_expand_prologue): Remove conditional for int_registers_saved
20167 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
20170 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
20171 min/max for never referenced object.
20173 2018-01-24 Jakub Jelinek <jakub@redhat.com>
20175 PR middle-end/83977
20176 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
20178 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
20179 attributes from DECL_ATTRIBUTES (decl) without affecting
20180 DECL_ATTRIBUTES (current_function_decl).
20181 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
20182 functions with non-NULL DECL_ABSTRACT_ORIGIN.
20184 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
20186 PR tree-optimization/83979
20187 * fold-const.c (fold_comparison): Use constant_boolean_node
20188 instead of boolean_{true,false}_node.
20190 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
20192 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
20195 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20197 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
20198 Simplify the clause that sets the length attribute.
20199 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
20200 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
20201 clause that sets the length attribute.
20202 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
20204 2018-01-24 Tom de Vries <tom@codesourcery.com>
20207 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
20208 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
20209 Add strict parameter.
20210 (prevent_branch_around_nothing): Insert dummy insn between branch to
20211 label and label with no ptx insn inbetween.
20212 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
20214 2018-01-24 Tom de Vries <tom@codesourcery.com>
20217 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
20218 for neutered threads in warp.
20219 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
20221 2018-01-24 Richard Biener <rguenther@suse.de>
20223 PR tree-optimization/83176
20224 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
20227 2018-01-24 Richard Biener <rguenther@suse.de>
20229 PR tree-optimization/82819
20230 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
20231 code generating pluses that are no-ops in the target precision.
20233 2018-01-24 Richard Biener <rguenther@suse.de>
20235 PR middle-end/84000
20236 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
20238 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
20240 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
20241 to merge probabilities.
20242 * predict.c (probably_never_executed): Also mark as cold functions
20243 with global 0 profile and guessed local profile.
20244 * profile-count.c (profile_probability::combine_with_count): New
20246 * profile-count.h (profile_probability::operator*,
20247 profile_probability::operator*=, profile_probability::operator/,
20248 profile_probability::operator/=): Reduce precision to adjusted
20249 and set value to guessed on contradictory divisions.
20250 (profile_probability::combine_with_freq): Remove.
20251 (profile_probability::combine_wiht_count): Declare.
20252 (profile_count::force_nonzero):: Set to adjusted.
20253 (profile_count::probability_in):: Set quality to adjusted.
20254 * tree-ssa-tail-merge.c (replace_block_by): Use
20255 combine_with_count.
20257 2018-01-23 Andrew Waterman <andrew@sifive.com>
20258 Jim Wilson <jimw@sifive.com>
20260 * config/riscv/riscv.c (riscv_stack_boundary): New.
20261 (riscv_option_override): Set riscv_stack_boundary. Handle
20262 riscv_preferred_stack_boundary_arg.
20263 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
20264 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
20265 (STACK_BOUNDARY): Set to riscv_stack_boundary.
20266 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
20267 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
20268 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
20270 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
20273 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
20274 of struct ix86_frame.
20275 (ix86_expand_epilogue): Likewise. Add a local variable for
20276 the reg_save_offset field in struct ix86_frame.
20278 2018-01-23 Bin Cheng <bin.cheng@arm.com>
20280 PR tree-optimization/82604
20281 * tree-loop-distribution.c (enum partition_kind): New enum item
20282 PKIND_PARTIAL_MEMSET.
20283 (partition_builtin_p): Support above new enum item.
20284 (generate_code_for_partition): Ditto.
20285 (compute_access_range): Differentiate cases that equality can be
20286 proven at all loops, the innermost loops or no loops.
20287 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
20288 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
20289 (finalize_partitions, distribute_loop): Don't fuse partition of
20290 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
20291 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
20292 parloop is enabled.
20294 2018-01-23 Martin Liska <mliska@suse.cz>
20296 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
20297 order to ignore the predictor.
20298 (PRED_POLYMORPHIC_CALL): Likewise.
20299 (PRED_RECURSIVE_CALL): Likewise.
20301 2018-01-23 Martin Liska <mliska@suse.cz>
20303 * tree-profile.c (tree_profiling): Print function header to
20304 aware reader which function we are working on.
20305 * value-prof.c (gimple_find_values_to_profile): Do not print
20306 not interesting value histograms.
20308 2018-01-23 Martin Liska <mliska@suse.cz>
20310 * profile-count.h (enum profile_quality): Add
20311 profile_uninitialized as the first value. Do not number values
20312 as they are zero based.
20313 (profile_count::verify): Update sanity check.
20314 (profile_probability::verify): Likewise.
20316 2018-01-23 Nathan Sidwell <nathan@acm.org>
20318 * doc/invoke.texi (ffor-scope): Deprecate.
20320 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20322 PR tree-optimization/83510
20323 * domwalk.c (set_all_edges_as_executable): New function.
20324 (dom_walker::dom_walker): Convert bool param
20325 "skip_unreachable_blocks" to enum reachability. Move setup of
20326 edge flags to set_all_edges_as_executable and only do it when
20327 reachability is REACHABLE_BLOCKS.
20328 * domwalk.h (enum dom_walker::reachability): New enum.
20329 (dom_walker::dom_walker): Convert bool param
20330 "skip_unreachable_blocks" to enum reachability.
20331 (set_all_edges_as_executable): New decl.
20332 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
20333 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
20335 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
20336 but converting true to REACHABLE_BLOCKS.
20337 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
20339 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
20340 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
20341 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
20343 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
20344 if check_all_array_refs will be called.
20346 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20348 * tree.c (selftest::test_location_wrappers): Add more test
20351 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20353 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
20354 (selftest::test_bit_in_range): Likewise.
20356 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
20359 * doc/sourcebuild.texi (vect_float): Say that the selector
20360 only describes the situation when -funsafe-math-optimizations is on.
20361 (vect_float_strict): Document.
20363 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
20365 PR tree-optimization/83965
20366 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
20367 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
20368 instead of checking only for a reduction.
20369 (vect_recog_widen_sum_pattern): Likewise.
20371 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
20373 * predict.c (probably_never_executed): Only use precise profile info.
20374 (compute_function_frequency): Skip after inlining hack since we now
20375 have quality checking.
20377 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
20379 * profile-count.h (profile_probability::very_unlikely,
20380 profile_probability::unlikely, profile_probability::even): Set
20381 precision to guessed.
20383 2018-01-23 Richard Biener <rguenther@suse.de>
20385 PR tree-optimization/83963
20386 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
20387 Properly terminate dominator walk when crossing the exit edge not
20388 when visiting its source block.
20390 2018-01-23 Jakub Jelinek <jakub@redhat.com>
20393 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
20394 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
20396 2018-01-22 Jakub Jelinek <jakub@redhat.com>
20398 PR tree-optimization/83957
20399 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
20400 semicolon after for body surrounded by braces.
20402 PR tree-optimization/83081
20403 * profile-count.h (profile_probability::split): New method.
20404 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
20405 Use profile_probability::split.
20406 (do_compare_rtx_and_jump): Fix adjustment of probabilities
20407 when splitting a single conditional jump into 2.
20409 2018-01-22 David Malcolm <dmalcolm@redhat.com>
20411 PR tree-optimization/69452
20412 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
20415 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20417 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
20418 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
20419 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
20421 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20423 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
20425 * config/rl78/rl78.md (movdi): New define_expand.
20426 * config/rl78/rl78.c (rl78_split_movdi): New function.
20428 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
20431 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
20433 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
20434 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
20435 128-bit to produce an UNSPEC move to get the double word with the
20436 signbit and then a shift directly to do signbit.
20437 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
20438 implementation with a new version that just does either a direct
20439 move or a regular move. Move memory interface to separate insns.
20440 Move insns so they are next to the expander.
20441 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
20442 with signbit move. Split big and little endian case.
20443 (signbit<mode>2_dm_mem_le): Likewise.
20444 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
20445 (signbit<mode>2_dm2): Likewise.
20447 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20449 * config/rl78/rl78.md (anddi3): New define_expand.
20451 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20453 * config/rl78/rl78.md (umindi3): New define_expand.
20455 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20457 * config/rl78/rl78.md (smindi3): New define_expand.
20459 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20461 * config/rl78/rl78.md (smaxdi3): New define_expand.
20463 2018-01-22 Carl Love <cel@us.ibm.com>
20465 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
20466 LVX_V1TI): Add macro expansion.
20467 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
20468 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
20469 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
20470 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
20471 Change check to determine if the instruction is a byte reversing
20472 entry. Fix typo in comment.
20473 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
20474 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
20475 Add def_builtin calls for new builtins.
20476 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
20477 Add define_insn expansion.
20479 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20481 * config/rl78/rl78.md (umaxdi3): New define_expand.
20483 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20485 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
20486 for non-QImode registers.
20488 2018-01-22 Richard Biener <rguenther@suse.de>
20490 PR tree-optimization/83963
20491 * graphite-scop-detection.c (scop_detection::get_sese): Delay
20492 including the loop exit block.
20493 (scop_detection::merge_sese): Likewise.
20494 (scop_detection::add_scop): Do it here instead.
20496 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20498 * doc/sourcebuild.texi (arm_softfloat): Document.
20500 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
20503 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
20504 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
20505 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
20507 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20508 David Edelsohn <dje.gcc@gmail.com>
20511 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
20512 Change "crset eq" to "crset 2".
20513 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
20514 (*call_indirect_aix<mode>_nospec): Likewise.
20515 (*call_value_indirect_aix<mode>_nospec): Likewise.
20516 (*call_indirect_elfv2<mode>_nospec): Likewise.
20517 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
20518 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
20519 change assembly output from . to $.
20520 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
20521 (indirect_jump<mode>_nospec): Change assembly output from . to $.
20522 (*tablejump<mode>_internal1_nospec): Likewise.
20524 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
20527 * config/sh/sh_optimize_sett_clrt.cc:
20528 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
20530 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
20532 PR tree-optimization/83940
20533 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
20534 offset_dt to vect_constant_def rather than vect_unknown_def_type.
20535 (vect_check_load_store_mask): Add a mask_dt_out parameter and
20536 use it to pass back the definition type.
20537 (vect_check_store_rhs): Likewise rhs_dt_out.
20538 (vect_build_gather_load_calls): Add a mask_dt argument and use
20539 it instead of a call to vect_is_simple_use.
20540 (vectorizable_store): Update calls to vect_check_load_store_mask
20541 and vect_check_store_rhs. Use the dt returned by the latter instead
20542 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
20543 instead of calls to vect_is_simple_use. Pass the scalar rather
20544 than the vector operand to vect_is_simple_use when handling
20545 second and subsequent copies of an rhs value.
20546 (vectorizable_load): Update calls to vect_check_load_store_mask
20547 and vect_build_gather_load_calls. Use the cached mask_dt and
20548 gs_info.offset_dt instead of calls to vect_is_simple_use.
20550 2018-01-20 Jakub Jelinek <jakub@redhat.com>
20552 PR middle-end/83945
20553 * tree-emutls.c: Include gimplify.h.
20554 (lower_emutls_2): New function.
20555 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
20556 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
20557 it before further processing.
20560 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
20561 UINTVAL (trueop1) instead of INTVAL (op1).
20563 2018-01-19 Jakub Jelinek <jakub@redhat.com>
20567 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
20568 INCOMING_FRAME_SP_OFFSET if not defined.
20569 (scan_trace): Add ENTRY argument. If true and
20570 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
20571 emit a note to adjust the CFA offset.
20572 (create_cfi_notes): Adjust scan_trace callers.
20573 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
20574 INCOMING_FRAME_SP_OFFSET in the CIE.
20575 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
20576 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
20578 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
20579 * doc/tm.texi: Regenerated.
20581 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20583 PR rtl-optimization/83147
20584 * lra-constraints.c (remove_inheritance_pseudos): Use
20585 lra_substitute_pseudo_within_insn.
20587 2018-01-19 Tom de Vries <tom@codesourcery.com>
20588 Cesar Philippidis <cesar@codesourcery.com>
20591 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
20593 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
20596 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
20597 spaces for function labels.
20599 2018-01-19 Martin Liska <mliska@suse.cz>
20601 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
20602 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
20603 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
20604 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
20605 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
20606 (PRED_CONST_RETURN): Change from 69 to 65.
20607 (PRED_NULL_RETURN): Change from 91 to 71.
20608 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
20609 (PRED_LOOP_GUARD): Change from 66 to 73.
20611 2018-01-19 Martin Liska <mliska@suse.cz>
20613 * predict.c (predict_insn_def): Add new assert.
20614 (struct branch_predictor): Change type to signed integer.
20615 (test_prediction_value_range): Amend test to cover
20616 PROB_UNINITIALIZED.
20617 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
20618 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
20619 (PRED_LOOP_ITERATIONS_MAX): Likewise.
20620 (PRED_LOOP_IV_COMPARE): Likewise.
20621 * predict.h (PROB_UNINITIALIZED): Define new constant.
20623 2018-01-19 Martin Liska <mliska@suse.cz>
20625 * predict.c (dump_prediction): Add new format for
20626 analyze_brprob.py script which is enabled with -details
20628 * profile-count.h (precise_p): New function.
20630 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
20632 PR tree-optimization/83922
20633 * tree-vect-loop.c (vect_verify_full_masking): Return false if
20634 there are no statements that need masking.
20635 (vect_active_double_reduction_p): New function.
20636 (vect_analyze_loop_operations): Use it when handling phis that
20637 are not in the loop header.
20639 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
20641 PR tree-optimization/83914
20642 * tree-vect-loop.c (vectorizable_induction): Don't convert
20643 init_expr or apply the peeling adjustment for inductions
20644 that are nested within the vectorized loop.
20646 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20648 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
20651 2018-01-18 Jakub Jelinek <jakub@redhat.com>
20655 * function.h (gimplify_parameters): Add gimple_seq * argument.
20656 * function.c: Include gimple.h and options.h.
20657 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
20658 for the added local temporaries if needed.
20659 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
20660 if there are any parameter cleanups, wrap whole body into a
20661 try/finally with the cleanups.
20663 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
20666 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20667 Use GET_MODE_CLASS for scalar floating point.
20669 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
20673 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
20674 Fix call of call_cgraph_insertion_hooks.
20676 2018-01-18 Martin Sebor <msebor@redhat.com>
20678 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
20680 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
20683 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
20686 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
20689 * common.opt: (-ffile-prefix-map): New option.
20690 * opts.c (common_handle_option): Defer it.
20691 * opts-global.c (handle_common_deferred_options): Handle it.
20692 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
20693 * file-prefix-map.h: New file.
20694 (remap_debug_filename, add_debug_prefix_map): ...here.
20695 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
20696 * final.c (debug_prefix_map, add_debug_prefix_map
20697 remap_debug_filename): Move to...
20698 * file-prefix-map.c: New file.
20699 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
20700 generalize, get rid of alloca(), use strrchr() instead of strchr().
20701 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
20702 Implement in terms of add_prefix_map().
20703 (remap_macro_filename, remap_debug_filename): Implement in term of
20705 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
20706 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
20707 * dbxout.c: Include file-prefix-map.h.
20708 * varasm.c: Likewise.
20709 * vmsdbgout.c: Likewise.
20710 * xcoffout.c: Likewise.
20711 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
20712 * doc/cppopts.texi (-fmacro-prefix-map): Document.
20713 * doc/invoke.texi (-ffile-prefix-map): Document.
20714 (-fdebug-prefix-map): Update description.
20716 2018-01-18 Martin Liska <mliska@suse.cz>
20718 * config/i386/i386.c (indirect_thunk_name): Document that also
20720 (output_indirect_thunk): Document why both instructions
20721 (pause and lfence) are generated.
20723 2018-01-18 Richard Biener <rguenther@suse.de>
20725 PR tree-optimization/83887
20726 * graphite-scop-detection.c
20727 (scop_detection::get_nearest_dom_with_single_entry): Remove.
20728 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
20729 (scop_detection::merge_sese): Re-implement with a flood-fill
20730 algorithm that properly finds a SESE region if it exists.
20732 2018-01-18 Jakub Jelinek <jakub@redhat.com>
20735 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
20736 pointer_diff optimizations use view_convert instead of convert.
20738 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20740 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
20741 Generate different code for -mno-speculate-indirect-jumps.
20742 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
20743 (*call_indirect_aix<mode>): Disable for
20744 -mno-speculate-indirect-jumps.
20745 (*call_indirect_aix<mode>_nospec): New define_insn.
20746 (*call_value_indirect_aix<mode>): Disable for
20747 -mno-speculate-indirect-jumps.
20748 (*call_value_indirect_aix<mode>_nospec): New define_insn.
20749 (*sibcall_nonlocal_sysv<mode>): Generate different code for
20750 -mno-speculate-indirect-jumps.
20751 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
20753 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
20755 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
20756 long double type, set the flags for noting the default long double
20757 type, even if we don't pass or return a long double type.
20759 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
20762 * ipa-inline.c (flatten_function): Do not overwrite final inlining
20765 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
20767 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
20768 support for merge[hl].
20769 (fold_mergehl_helper): New helper function.
20770 (tree-vector-builder.h): New #include for tree_vector_builder usage.
20771 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
20772 (altivec_vmrglw_direct): Add xxmrglw insn.
20774 2018-01-17 Andrew Waterman <andrew@sifive.com>
20776 * config/riscv/riscv.c (riscv_conditional_register_usage): If
20777 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
20779 2018-01-17 David Malcolm <dmalcolm@redhat.com>
20782 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
20783 call the lto_location_cache before reading the
20784 DECL_SOURCE_LOCATION of the types.
20786 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
20787 Richard Sandiford <richard.sandiford@linaro.org>
20789 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
20790 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
20791 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
20792 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
20793 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
20795 * config/aarch64/constraints.md (aarch64_movti_operand):
20797 * config/aarch64/predicates.md (Uti): Add new constraint.
20799 2018-01-17 Carl Love <cel@us.ibm.com>
20801 * config/rs6000/vsx.md (define_expand xl_len_r,
20802 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
20803 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
20805 (define_expand, define_insn): Move the shift left from the
20806 define_insn to the define_expand for lxvl and stxvl instructions.
20807 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
20808 and XL_LEN_R definitions to PURE.
20810 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
20812 * config/i386/i386.c (indirect_thunk_name): Declare regno
20813 as unsigned int. Compare regno with INVALID_REGNUM.
20814 (output_indirect_thunk): Ditto.
20815 (output_indirect_thunk_function): Ditto.
20816 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
20817 in the call to output_indirect_thunk_function.
20819 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
20821 PR middle-end/83884
20822 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
20823 rather than the size of inner_type to determine the stack slot size
20824 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
20826 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
20829 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
20832 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
20834 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
20835 endian Linux systems to optionally enable multilibs for selecting
20836 the long double type if the user configured an explicit type.
20837 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
20838 have no long double multilibs if not defined.
20839 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
20840 warn if the user used -mabi={ieee,ibm}longdouble and we built
20841 multilibs for long double.
20842 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
20843 appropriate multilib option.
20844 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
20846 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
20847 for building long double multilibs.
20848 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
20850 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
20852 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
20855 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
20857 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
20860 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
20863 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
20866 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
20868 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
20869 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
20871 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
20873 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
20874 different rtl trees depending on TARGET_64BIT.
20875 (rs6000_gen_lvx): Likewise.
20877 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
20879 * config/visium/visium.md (nop): Tweak comment.
20880 (hazard_nop): Likewise.
20882 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20884 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
20885 -mspeculate-indirect-jumps.
20886 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
20887 for -mno-speculate-indirect-jumps.
20888 (*call_indirect_elfv2<mode>_nospec): New define_insn.
20889 (*call_value_indirect_elfv2<mode>): Disable for
20890 -mno-speculate-indirect-jumps.
20891 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
20892 (indirect_jump): Emit different RTL for
20893 -mno-speculate-indirect-jumps.
20894 (*indirect_jump<mode>): Disable for
20895 -mno-speculate-indirect-jumps.
20896 (*indirect_jump<mode>_nospec): New define_insn.
20897 (tablejump): Emit different RTL for
20898 -mno-speculate-indirect-jumps.
20899 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
20900 (tablejumpsi_nospec): New define_expand.
20901 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
20902 (tablejumpdi_nospec): New define_expand.
20903 (*tablejump<mode>_internal1): Disable for
20904 -mno-speculate-indirect-jumps.
20905 (*tablejump<mode>_internal1_nospec): New define_insn.
20906 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
20909 2018-01-16 Artyom Skrobov tyomitch@gmail.com
20911 * caller-save.c (insert_save): Drop unnecessary parameter. All
20914 2018-01-16 Jakub Jelinek <jakub@redhat.com>
20915 Richard Biener <rguenth@suse.de>
20918 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
20919 return early, inline manually is_gimple_sizepos. Make sure if we
20920 call gimplify_expr we don't end up with a gimple constant.
20921 * tree.c (variably_modified_type_p): Don't return true for
20922 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
20923 * gimplify.h (is_gimple_sizepos): Remove.
20925 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
20927 PR tree-optimization/83857
20928 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
20929 vectorizable_live_operation for pure SLP statements.
20930 (vectorizable_live_operation): Handle PHIs.
20932 2018-01-16 Richard Biener <rguenther@suse.de>
20934 PR tree-optimization/83867
20935 * tree-vect-stmts.c (vect_transform_stmt): Precompute
20936 nested_in_vect_loop_p since the scalar stmt may get invalidated.
20938 2018-01-16 Jakub Jelinek <jakub@redhat.com>
20941 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
20942 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
20943 If off is not INTEGER_CST, issue a may not be aligned warning
20944 rather than isn't aligned. Use isn%'t rather than isn't.
20945 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
20947 <case MULT_EXPR>: Improve the case when bottom and one of the
20948 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
20949 operand, in that case check if the other operand is multiple of
20950 bottom divided by the INTEGER_CST operand.
20952 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
20955 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
20956 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
20957 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
20958 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
20959 * config/pa/pa.c (pa_function_arg_advance): Likewise.
20960 (pa_function_arg, pa_arg_partial_bytes): Likewise.
20961 (pa_function_arg_size): New function.
20963 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
20965 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
20966 in a separate statement.
20968 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
20970 PR tree-optimization/83847
20971 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
20972 group gathers and scatters.
20974 2018-01-16 Jakub Jelinek <jakub@redhat.com>
20976 PR rtl-optimization/86620
20977 * params.def (max-sched-ready-insns): Bump minimum value to 1.
20979 PR rtl-optimization/83213
20980 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
20981 to last if both are JUMP_INSNs.
20983 PR tree-optimization/83843
20984 * gimple-ssa-store-merging.c
20985 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
20986 store_immediate_info for bswap/nop orig_stores.
20988 2018-01-15 Andrew Waterman <andrew@sifive.com>
20990 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
20992 <UDIV>: Increase cost if !TARGET_DIV.
20994 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
20996 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
20997 (define_attr "cr_logical_3op"): New.
20998 (cceq_ior_compare): Adjust.
20999 (cceq_ior_compare_complement): Adjust.
21000 (*cceq_rev_compare): Adjust.
21001 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
21002 (is_cracked_insn): Adjust.
21003 (insn_must_be_first_in_group): Adjust.
21004 * config/rs6000/40x.md: Adjust.
21005 * config/rs6000/440.md: Adjust.
21006 * config/rs6000/476.md: Adjust.
21007 * config/rs6000/601.md: Adjust.
21008 * config/rs6000/603.md: Adjust.
21009 * config/rs6000/6xx.md: Adjust.
21010 * config/rs6000/7450.md: Adjust.
21011 * config/rs6000/7xx.md: Adjust.
21012 * config/rs6000/8540.md: Adjust.
21013 * config/rs6000/cell.md: Adjust.
21014 * config/rs6000/e300c2c3.md: Adjust.
21015 * config/rs6000/e500mc.md: Adjust.
21016 * config/rs6000/e500mc64.md: Adjust.
21017 * config/rs6000/e5500.md: Adjust.
21018 * config/rs6000/e6500.md: Adjust.
21019 * config/rs6000/mpc.md: Adjust.
21020 * config/rs6000/power4.md: Adjust.
21021 * config/rs6000/power5.md: Adjust.
21022 * config/rs6000/power6.md: Adjust.
21023 * config/rs6000/power7.md: Adjust.
21024 * config/rs6000/power8.md: Adjust.
21025 * config/rs6000/power9.md: Adjust.
21026 * config/rs6000/rs64.md: Adjust.
21027 * config/rs6000/titan.md: Adjust.
21029 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21031 * config/i386/predicates.md (indirect_branch_operand): Rewrite
21032 ix86_indirect_branch_register logic.
21034 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21036 * config/i386/constraints.md (Bs): Update
21037 ix86_indirect_branch_register check. Don't check
21038 ix86_indirect_branch_register with GOT_memory_operand.
21040 * config/i386/predicates.md (GOT_memory_operand): Don't check
21041 ix86_indirect_branch_register here.
21042 (GOT32_symbol_operand): Likewise.
21044 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21046 * config/i386/predicates.md (constant_call_address_operand):
21047 Rewrite ix86_indirect_branch_register logic.
21048 (sibcall_insn_operand): Likewise.
21050 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21052 * config/i386/constraints.md (Bs): Replace
21053 ix86_indirect_branch_thunk_register with
21054 ix86_indirect_branch_register.
21056 * config/i386/i386.md (indirect_jump): Likewise.
21057 (tablejump): Likewise.
21058 (*sibcall_memory): Likewise.
21059 (*sibcall_value_memory): Likewise.
21060 Peepholes of indirect call and jump via memory: Likewise.
21061 * config/i386/i386.opt: Likewise.
21062 * config/i386/predicates.md (indirect_branch_operand): Likewise.
21063 (GOT_memory_operand): Likewise.
21064 (call_insn_operand): Likewise.
21065 (sibcall_insn_operand): Likewise.
21066 (GOT32_symbol_operand): Likewise.
21068 2018-01-15 Jakub Jelinek <jakub@redhat.com>
21070 PR middle-end/83837
21071 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
21072 type rather than type addr's type points to.
21073 (expand_omp_atomic_mutex): Likewise.
21074 (expand_omp_atomic): Likewise.
21076 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21079 * config/i386/i386.c (output_indirect_thunk_function): Use
21080 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
21081 for __x86_return_thunk.
21083 2018-01-15 Richard Biener <rguenther@suse.de>
21085 PR middle-end/83850
21086 * expmed.c (extract_bit_field_1): Fix typo.
21088 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21091 * config/arm/iterators.md (VF): New mode iterator.
21092 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
21093 Remove integer-related logic from pattern.
21094 (neon_vabd<mode>_3): Likewise.
21096 2018-01-15 Jakub Jelinek <jakub@redhat.com>
21098 PR middle-end/82694
21099 * common.opt (fstrict-overflow): No longer an alias.
21100 (fwrapv-pointer): New option.
21101 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
21102 also for pointer types based on flag_wrapv_pointer.
21103 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
21104 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
21105 opts->x_flag_wrapv got set.
21106 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
21107 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
21108 POINTER_TYPE_OVERFLOW_UNDEFINED.
21109 * match.pd: Likewise in address comparison pattern.
21110 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
21112 2018-01-15 Richard Biener <rguenther@suse.de>
21115 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
21116 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
21117 Reset type names to their identifier if their TYPE_DECL doesn't
21118 have linkage (and thus is used for ODR and devirt).
21119 (save_debug_info_for_decl): Remove.
21120 (save_debug_info_for_type): Likewise.
21121 (add_tree_to_fld_list): Adjust.
21122 * tree-pretty-print.c (dump_generic_node): Make dumping of
21123 type names more robust.
21125 2018-01-15 Richard Biener <rguenther@suse.de>
21127 * BASE-VER: Bump to 8.0.1.
21129 2018-01-14 Martin Sebor <msebor@redhat.com>
21132 * builtins.c (check_access): Avoid warning when the no-warning bit
21135 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
21137 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
21138 * ira-color (allocno_hard_regs_compare): Likewise.
21140 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
21143 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
21144 Use .pushsection/.popsection.
21146 2018-01-14 Martin Sebor <msebor@redhat.com>
21149 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
21151 2018-01-14 Jakub Jelinek <jakub@redhat.com>
21153 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
21154 entry from extra_headers.
21155 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
21156 extra_headers, make the list bitwise identical to the i?86-*-* one.
21158 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21160 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
21161 -mcmodel=large with -mindirect-branch=thunk,
21162 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
21163 -mfunction-return=thunk-extern.
21164 * doc/invoke.texi: Document -mcmodel=large is incompatible with
21165 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
21166 -mfunction-return=thunk and -mfunction-return=thunk-extern.
21168 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21170 * config/i386/i386.c (print_reg): Print the name of the full
21171 integer register without '%'.
21172 (ix86_print_operand): Handle 'V'.
21173 * doc/extend.texi: Document 'V' modifier.
21175 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21177 * config/i386/constraints.md (Bs): Disallow memory operand for
21178 -mindirect-branch-register.
21180 * config/i386/predicates.md (indirect_branch_operand): Likewise.
21181 (GOT_memory_operand): Likewise.
21182 (call_insn_operand): Likewise.
21183 (sibcall_insn_operand): Likewise.
21184 (GOT32_symbol_operand): Likewise.
21185 * config/i386/i386.md (indirect_jump): Call convert_memory_address
21186 for -mindirect-branch-register.
21187 (tablejump): Likewise.
21188 (*sibcall_memory): Likewise.
21189 (*sibcall_value_memory): Likewise.
21190 Disallow peepholes of indirect call and jump via memory for
21191 -mindirect-branch-register.
21192 (*call_pop): Replace m with Bw.
21193 (*call_value_pop): Likewise.
21194 (*sibcall_pop_memory): Replace m with Bs.
21195 * config/i386/i386.opt (mindirect-branch-register): New option.
21196 * doc/invoke.texi: Document -mindirect-branch-register option.
21198 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21200 * config/i386/i386-protos.h (ix86_output_function_return): New.
21201 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
21202 set function_return_type.
21203 (indirect_thunk_name): Add ret_p to indicate thunk for function
21205 (output_indirect_thunk_function): Pass false to
21206 indirect_thunk_name.
21207 (ix86_output_indirect_branch_via_reg): Likewise.
21208 (ix86_output_indirect_branch_via_push): Likewise.
21209 (output_indirect_thunk_function): Create alias for function
21210 return thunk if regno < 0.
21211 (ix86_output_function_return): New function.
21212 (ix86_handle_fndecl_attribute): Handle function_return.
21213 (ix86_attribute_table): Add function_return.
21214 * config/i386/i386.h (machine_function): Add
21215 function_return_type.
21216 * config/i386/i386.md (simple_return_internal): Use
21217 ix86_output_function_return.
21218 (simple_return_internal_long): Likewise.
21219 * config/i386/i386.opt (mfunction-return=): New option.
21220 (indirect_branch): Mention -mfunction-return=.
21221 * doc/extend.texi: Document function_return function attribute.
21222 * doc/invoke.texi: Document -mfunction-return= option.
21224 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21226 * config/i386/i386-opts.h (indirect_branch): New.
21227 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
21228 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
21229 with local indirect jump when converting indirect call and jump.
21230 (ix86_set_indirect_branch_type): New.
21231 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
21232 (indirectlabelno): New.
21233 (indirect_thunk_needed): Likewise.
21234 (indirect_thunk_bnd_needed): Likewise.
21235 (indirect_thunks_used): Likewise.
21236 (indirect_thunks_bnd_used): Likewise.
21237 (INDIRECT_LABEL): Likewise.
21238 (indirect_thunk_name): Likewise.
21239 (output_indirect_thunk): Likewise.
21240 (output_indirect_thunk_function): Likewise.
21241 (ix86_output_indirect_branch_via_reg): Likewise.
21242 (ix86_output_indirect_branch_via_push): Likewise.
21243 (ix86_output_indirect_branch): Likewise.
21244 (ix86_output_indirect_jmp): Likewise.
21245 (ix86_code_end): Call output_indirect_thunk_function if needed.
21246 (ix86_output_call_insn): Call ix86_output_indirect_branch if
21248 (ix86_handle_fndecl_attribute): Handle indirect_branch.
21249 (ix86_attribute_table): Add indirect_branch.
21250 * config/i386/i386.h (machine_function): Add indirect_branch_type
21251 and has_local_indirect_jump.
21252 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
21254 (tablejump): Likewise.
21255 (*indirect_jump): Use ix86_output_indirect_jmp.
21256 (*tablejump_1): Likewise.
21257 (simple_return_indirect_internal): Likewise.
21258 * config/i386/i386.opt (mindirect-branch=): New option.
21259 (indirect_branch): New.
21262 (thunk-inline): Likewise.
21263 (thunk-extern): Likewise.
21264 * doc/extend.texi: Document indirect_branch function attribute.
21265 * doc/invoke.texi: Document -mindirect-branch= option.
21267 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
21270 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
21272 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
21274 * ipa-inline.c (want_inline_small_function_p): Return false if
21275 inlining has already failed with CIF_FINAL_ERROR.
21276 (update_caller_keys): Call want_inline_small_function_p before
21278 (update_callee_keys): Likewise.
21280 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21282 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
21284 (rs6000_quadword_masked_address_p): Likewise.
21285 (quad_aligned_load_p): Likewise.
21286 (quad_aligned_store_p): Likewise.
21287 (const_load_sequence_p): Add comment to describe the outer-most loop.
21288 (mimic_memory_attributes_and_flags): New function.
21289 (rs6000_gen_stvx): Likewise.
21290 (replace_swapped_aligned_store): Likewise.
21291 (rs6000_gen_lvx): Likewise.
21292 (replace_swapped_aligned_load): Likewise.
21293 (replace_swapped_load_constant): Capitalize argument name in
21294 comment describing this function.
21295 (rs6000_analyze_swaps): Add a third pass to search for vector loads
21296 and stores that access quad-word aligned addresses and replace
21297 with stvx or lvx instructions when appropriate.
21298 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
21299 New function prototype.
21300 (rs6000_quadword_masked_address_p): Likewise.
21301 (rs6000_gen_lvx): Likewise.
21302 (rs6000_gen_stvx): Likewise.
21303 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
21304 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
21305 when memory address is aligned.
21306 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
21307 this split to select lvx instruction when memory address is aligned.
21308 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
21309 instruction when memory address is aligned.
21310 (*vsx_le_perm_load_v16qi): Likewise.
21311 (four unnamed splitters): Modify to select the stvx instruction
21312 when memory is aligned.
21314 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
21316 * predict.c (determine_unlikely_bbs): Handle correctly BBs
21317 which appears in the queue multiple times.
21319 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21320 Alan Hayward <alan.hayward@arm.com>
21321 David Sherwood <david.sherwood@arm.com>
21323 * tree-vectorizer.h (vec_lower_bound): New structure.
21324 (_loop_vec_info): Add check_nonzero and lower_bounds.
21325 (LOOP_VINFO_CHECK_NONZERO): New macro.
21326 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
21327 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
21328 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
21329 fields. Make seg_len the distance travelled, not including the
21331 (dr_direction_indicator): Declare.
21332 (dr_zero_step_indicator): Likewise.
21333 (dr_known_forward_stride_p): Likewise.
21334 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
21336 (runtime_alias_check_p): Allow runtime alias checks with
21338 (operator ==): Compare access_size and align.
21339 (prune_runtime_alias_test_list): Rework for new distinction between
21340 the access_size and seg_len.
21341 (create_intersect_range_checks_index): Likewise. Cope with polynomial
21343 (get_segment_min_max): New function.
21344 (create_intersect_range_checks): Use it.
21345 (dr_step_indicator): New function.
21346 (dr_direction_indicator): Likewise.
21347 (dr_zero_step_indicator): Likewise.
21348 (dr_known_forward_stride_p): Likewise.
21349 * tree-loop-distribution.c (data_ref_segment_size): Return
21350 DR_STEP * (niters - 1).
21351 (compute_alias_check_pairs): Update call to the dr_with_seg_len
21353 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
21354 (vect_preserves_scalar_order_p): New function, split out from...
21355 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
21356 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
21357 (vect_vfa_access_size): New function.
21358 (vect_vfa_align): Likewise.
21359 (vect_compile_time_alias): Take access_size_a and access_b arguments.
21360 (dump_lower_bound): New function.
21361 (vect_check_lower_bound): Likewise.
21362 (vect_small_gap_p): Likewise.
21363 (vectorizable_with_step_bound_p): Likewise.
21364 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
21365 depencies if the vectorization factor is 1. Convert the checks
21366 for nonzero steps into checks on the bounds of DR_STEP. Try using
21367 a bunds check for variable steps if the minimum required step is
21368 relatively small. Update calls to the dr_with_seg_len
21369 constructor and to vect_compile_time_alias.
21370 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
21372 (vect_loop_versioning): Call it.
21373 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
21375 (vect_estimate_min_profitable_iters): Account for any bounds checks.
21377 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21378 Alan Hayward <alan.hayward@arm.com>
21379 David Sherwood <david.sherwood@arm.com>
21381 * doc/sourcebuild.texi (vect_scatter_store): Document.
21382 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
21384 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
21386 * genopinit.c (main): Add supports_vec_scatter_store and
21387 supports_vec_scatter_store_cached to target_optabs.
21388 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
21389 IFN_MASK_SCATTER_STORE.
21390 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
21392 * internal-fn.h (internal_store_fn_p): Declare.
21393 (internal_fn_stored_value_index): Likewise.
21394 * internal-fn.c (scatter_store_direct): New macro.
21395 (expand_scatter_store_optab_fn): New function.
21396 (direct_scatter_store_optab_supported_p): New macro.
21397 (internal_store_fn_p): New function.
21398 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
21399 IFN_MASK_SCATTER_STORE.
21400 (internal_fn_mask_index): Likewise.
21401 (internal_fn_stored_value_index): New function.
21402 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
21403 for scatter stores.
21404 * optabs-query.h (supports_vec_scatter_store_p): Declare.
21405 * optabs-query.c (supports_vec_scatter_store_p): New function.
21406 * tree-vectorizer.h (vect_get_store_rhs): Declare.
21407 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
21408 true for scatter stores.
21409 (vect_gather_scatter_fn_p): Handle scatter stores too.
21410 (vect_check_gather_scatter): Consider using scatter stores if
21411 supports_vec_scatter_store_p.
21412 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
21413 scatter stores too.
21414 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
21415 internal_fn_stored_value_index.
21416 (check_load_store_masking): Handle scatter stores too.
21417 (vect_get_store_rhs): Make public.
21418 (vectorizable_call): Use internal_store_fn_p.
21419 (vectorizable_store): Handle scatter store internal functions.
21420 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
21421 when deciding whether the end of the group has been reached.
21422 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
21423 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
21424 (mask_scatter_store<mode>): New insns.
21426 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21427 Alan Hayward <alan.hayward@arm.com>
21428 David Sherwood <david.sherwood@arm.com>
21430 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
21431 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
21432 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
21434 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
21435 Use vect_truncate_gather_scatter_offset if we can't treat the
21436 operation as a normal gather load or scatter store.
21437 (get_group_load_store_type): Take the gather_scatter_info
21438 as argument. Try using a gather load or scatter store for
21439 single-element groups.
21440 (get_load_store_type): Update calls to get_group_load_store_type
21441 and vect_use_strided_gather_scatters_p.
21443 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21444 Alan Hayward <alan.hayward@arm.com>
21445 David Sherwood <david.sherwood@arm.com>
21447 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
21448 optional tree argument.
21449 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
21451 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
21452 but continue to use the current value as a fallback.
21453 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
21454 to compare the updates.
21455 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
21456 (get_load_store_type): Use it when handling a strided access.
21457 (vect_get_strided_load_store_ops): New function.
21458 (vect_get_data_ptr_increment): Likewise.
21459 (vectorizable_load): Handle strided gather loads. Always pass
21460 a step to vect_create_data_ref_ptr and bump_vector_ptr.
21462 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21463 Alan Hayward <alan.hayward@arm.com>
21464 David Sherwood <david.sherwood@arm.com>
21466 * doc/md.texi (gather_load@var{m}): Document.
21467 (mask_gather_load@var{m}): Likewise.
21468 * genopinit.c (main): Add supports_vec_gather_load and
21469 supports_vec_gather_load_cached to target_optabs.
21470 * optabs-tree.c (init_tree_optimization_optabs): Use
21471 ggc_cleared_alloc to allocate target_optabs.
21472 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
21473 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
21475 * internal-fn.h (internal_load_fn_p): Declare.
21476 (internal_gather_scatter_fn_p): Likewise.
21477 (internal_fn_mask_index): Likewise.
21478 (internal_gather_scatter_fn_supported_p): Likewise.
21479 * internal-fn.c (gather_load_direct): New macro.
21480 (expand_gather_load_optab_fn): New function.
21481 (direct_gather_load_optab_supported_p): New macro.
21482 (direct_internal_fn_optab): New function.
21483 (internal_load_fn_p): Likewise.
21484 (internal_gather_scatter_fn_p): Likewise.
21485 (internal_fn_mask_index): Likewise.
21486 (internal_gather_scatter_fn_supported_p): Likewise.
21487 * optabs-query.c (supports_at_least_one_mode_p): New function.
21488 (supports_vec_gather_load_p): Likewise.
21489 * optabs-query.h (supports_vec_gather_load_p): Declare.
21490 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
21491 and memory_type field.
21492 (NUM_PATTERNS): Bump to 15.
21493 * tree-vect-data-refs.c: Include internal-fn.h.
21494 (vect_gather_scatter_fn_p): New function.
21495 (vect_describe_gather_scatter_call): Likewise.
21496 (vect_check_gather_scatter): Try using internal functions for
21497 gather loads. Recognize existing calls to a gather load function.
21498 (vect_analyze_data_refs): Consider using gather loads if
21499 supports_vec_gather_load_p.
21500 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
21501 (vect_get_gather_scatter_offset_type): Likewise.
21502 (vect_convert_mask_for_vectype): Likewise.
21503 (vect_add_conversion_to_patterm): Likewise.
21504 (vect_try_gather_scatter_pattern): Likewise.
21505 (vect_recog_gather_scatter_pattern): New pattern recognizer.
21506 (vect_vect_recog_func_ptrs): Add it.
21507 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
21508 internal_fn_mask_index and internal_gather_scatter_fn_p.
21509 (check_load_store_masking): Take the gather_scatter_info as an
21510 argument and handle gather loads.
21511 (vect_get_gather_scatter_ops): New function.
21512 (vectorizable_call): Check internal_load_fn_p.
21513 (vectorizable_load): Likewise. Handle gather load internal
21515 (vectorizable_store): Update call to check_load_store_masking.
21516 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
21517 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
21518 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
21519 (aarch64_gather_scale_operand_d): New predicates.
21520 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
21521 (mask_gather_load<mode>): New insns.
21523 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21524 Alan Hayward <alan.hayward@arm.com>
21525 David Sherwood <david.sherwood@arm.com>
21527 * optabs.def (fold_left_plus_optab): New optab.
21528 * doc/md.texi (fold_left_plus_@var{m}): Document.
21529 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
21530 * internal-fn.c (fold_left_direct): Define.
21531 (expand_fold_left_optab_fn): Likewise.
21532 (direct_fold_left_optab_supported_p): Likewise.
21533 * fold-const-call.c (fold_const_fold_left): New function.
21534 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
21535 * tree-parloops.c (valid_reduction_p): New function.
21536 (gather_scalar_reductions): Use it.
21537 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
21538 (vect_finish_replace_stmt): Declare.
21539 * tree-vect-loop.c (fold_left_reduction_fn): New function.
21540 (needs_fold_left_reduction_p): New function, split out from...
21541 (vect_is_simple_reduction): ...here. Accept reductions that
21542 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
21543 (vect_force_simple_reduction): Also store the reduction type in
21544 the assignment's STMT_VINFO_REDUC_TYPE.
21545 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
21546 (merge_with_identity): New function.
21547 (vect_expand_fold_left): Likewise.
21548 (vectorize_fold_left_reduction): Likewise.
21549 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
21550 scalar phi in place for it. Check for target support and reject
21551 cases that would reassociate the operation. Defer the transform
21552 phase to vectorize_fold_left_reduction.
21553 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
21554 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
21555 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
21557 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21559 * tree-if-conv.c (predicate_mem_writes): Remove redundant
21560 call to ifc_temp_var.
21562 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21563 Alan Hayward <alan.hayward@arm.com>
21564 David Sherwood <david.sherwood@arm.com>
21566 * target.def (legitimize_address_displacement): Take the original
21567 offset as a poly_int.
21568 * targhooks.h (default_legitimize_address_displacement): Update
21570 * targhooks.c (default_legitimize_address_displacement): Likewise.
21571 * doc/tm.texi: Regenerate.
21572 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
21573 as an argument, moving assert of ad->disp == ad->disp_term to...
21574 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
21575 Try calling targetm.legitimize_address_displacement before expanding
21576 the address rather than afterwards, and adjust for the new interface.
21577 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
21578 Match the new hook interface. Handle SVE addresses.
21579 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
21580 new hook interface.
21582 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21584 * Makefile.in (OBJS): Add early-remat.o.
21585 * target.def (select_early_remat_modes): New hook.
21586 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
21587 * doc/tm.texi: Regenerate.
21588 * targhooks.h (default_select_early_remat_modes): Declare.
21589 * targhooks.c (default_select_early_remat_modes): New function.
21590 * timevar.def (TV_EARLY_REMAT): New timevar.
21591 * passes.def (pass_early_remat): New pass.
21592 * tree-pass.h (make_pass_early_remat): Declare.
21593 * early-remat.c: New file.
21594 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
21596 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
21598 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21599 Alan Hayward <alan.hayward@arm.com>
21600 David Sherwood <david.sherwood@arm.com>
21602 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
21603 vfm1 with a bound_epilog parameter.
21604 (vect_do_peeling): Update calls accordingly, and move the prologue
21605 call earlier in the function. Treat the base bound_epilog as 0 for
21606 fully-masked loops and retain vf - 1 for other loops. Add 1 to
21607 this base when peeling for gaps.
21608 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
21609 with fully-masked loops.
21610 (vect_estimate_min_profitable_iters): Handle the single peeled
21611 iteration in that case.
21613 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21614 Alan Hayward <alan.hayward@arm.com>
21615 David Sherwood <david.sherwood@arm.com>
21617 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
21618 single-element interleaving even if the size is not a power of 2.
21619 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
21620 accesses for single-element interleaving if the group size is
21623 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21624 Alan Hayward <alan.hayward@arm.com>
21625 David Sherwood <david.sherwood@arm.com>
21627 * doc/md.texi (fold_extract_last_@var{m}): Document.
21628 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
21629 * optabs.def (fold_extract_last_optab): New optab.
21630 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
21631 * internal-fn.c (fold_extract_direct): New macro.
21632 (expand_fold_extract_optab_fn): Likewise.
21633 (direct_fold_extract_optab_supported_p): Likewise.
21634 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
21635 * tree-vect-loop.c (vect_model_reduction_cost): Handle
21636 EXTRACT_LAST_REDUCTION.
21637 (get_initial_def_for_reduction): Do not create an initial vector
21638 for EXTRACT_LAST_REDUCTION reductions.
21639 (vectorizable_reduction): Leave the scalar phi in place for
21640 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
21641 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
21642 epilogue code for EXTRACT_LAST_REDUCTION and defer the
21643 transform phase to vectorizable_condition.
21644 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
21646 (vect_finish_stmt_generation): ...here.
21647 (vect_finish_replace_stmt): New function.
21648 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
21649 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
21651 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
21653 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21654 Alan Hayward <alan.hayward@arm.com>
21655 David Sherwood <david.sherwood@arm.com>
21657 * doc/md.texi (extract_last_@var{m}): Document.
21658 * optabs.def (extract_last_optab): New optab.
21659 * internal-fn.def (EXTRACT_LAST): New internal function.
21660 * internal-fn.c (cond_unary_direct): New macro.
21661 (expand_cond_unary_optab_fn): Likewise.
21662 (direct_cond_unary_optab_supported_p): Likewise.
21663 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
21664 loops using EXTRACT_LAST.
21665 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
21666 (extract_last_<mode>): ...this optab.
21667 (vec_extract<mode><Vel>): Update accordingly.
21669 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21670 Alan Hayward <alan.hayward@arm.com>
21671 David Sherwood <david.sherwood@arm.com>
21673 * target.def (empty_mask_is_expensive): New hook.
21674 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
21675 * doc/tm.texi: Regenerate.
21676 * targhooks.h (default_empty_mask_is_expensive): Declare.
21677 * targhooks.c (default_empty_mask_is_expensive): New function.
21678 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
21679 if the target says that empty masks are expensive.
21680 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
21682 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
21684 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21685 Alan Hayward <alan.hayward@arm.com>
21686 David Sherwood <david.sherwood@arm.com>
21688 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
21689 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
21690 (vect_use_loop_mask_for_alignment_p): New function.
21691 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
21692 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
21693 niters_skip argument. Make sure that the first niters_skip elements
21694 of the first iteration are inactive.
21695 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
21696 Update call to vect_set_loop_masks_directly.
21697 (get_misalign_in_elems): New function, split out from...
21698 (vect_gen_prolog_loop_niters): ...here.
21699 (vect_update_init_of_dr): Take a code argument that specifies whether
21700 the adjustment should be added or subtracted.
21701 (vect_update_init_of_drs): Likewise.
21702 (vect_prepare_for_masked_peels): New function.
21703 (vect_do_peeling): Skip prologue peeling if we're using a mask
21704 instead. Update call to vect_update_inits_of_drs.
21705 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
21707 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
21708 alignment. Do not include the number of peeled iterations in
21709 the minimum threshold in that case.
21710 (vectorizable_induction): Adjust the start value down by
21711 LOOP_VINFO_MASK_SKIP_NITERS iterations.
21712 (vect_transform_loop): Call vect_prepare_for_masked_peels.
21713 Take the number of skipped iterations into account when calculating
21715 * tree-vect-stmts.c (vect_gen_while_not): New function.
21717 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21718 Alan Hayward <alan.hayward@arm.com>
21719 David Sherwood <david.sherwood@arm.com>
21721 * doc/sourcebuild.texi (vect_fully_masked): Document.
21722 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
21723 default value to 0.
21724 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
21726 (vect_analyze_loop_2): ...here. Don't check the vectorization
21727 factor against the number of loop iterations if the loop is
21730 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21731 Alan Hayward <alan.hayward@arm.com>
21732 David Sherwood <david.sherwood@arm.com>
21734 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
21735 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
21736 (dump_groups): Update accordingly.
21737 (iv_use::mem_type): New member variable.
21738 (address_p): New function.
21739 (record_use): Add a mem_type argument and initialize the new
21741 (record_group_use): Add a mem_type argument. Use address_p.
21742 Remove obsolete null checks of base_object. Update call to record_use.
21743 (find_interesting_uses_op): Update call to record_group_use.
21744 (find_interesting_uses_cond): Likewise.
21745 (find_interesting_uses_address): Likewise.
21746 (get_mem_type_for_internal_fn): New function.
21747 (find_address_like_use): Likewise.
21748 (find_interesting_uses_stmt): Try find_address_like_use before
21749 calling find_interesting_uses_op.
21750 (addr_offset_valid_p): Use the iv mem_type field as the type
21751 of the addressed memory.
21752 (add_autoinc_candidates): Likewise.
21753 (get_address_cost): Likewise.
21754 (split_small_address_groups_p): Use address_p.
21755 (split_address_groups): Likewise.
21756 (add_iv_candidate_for_use): Likewise.
21757 (autoinc_possible_for_pair): Likewise.
21758 (rewrite_groups): Likewise.
21759 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
21760 (determine_group_iv_cost): Update after split of USE_ADDRESS.
21761 (get_alias_ptr_type_for_ptr_address): New function.
21762 (rewrite_use_address): Rewrite address uses in calls that were
21763 identified by find_address_like_use.
21765 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21766 Alan Hayward <alan.hayward@arm.com>
21767 David Sherwood <david.sherwood@arm.com>
21769 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
21771 * gimple-expr.h (is_gimple_addressable: Likewise.
21772 * gimple-expr.c (is_gimple_address): Likewise.
21773 * internal-fn.c (expand_call_mem_ref): New function.
21774 (expand_mask_load_optab_fn): Use it.
21775 (expand_mask_store_optab_fn): Likewise.
21777 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21778 Alan Hayward <alan.hayward@arm.com>
21779 David Sherwood <david.sherwood@arm.com>
21781 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
21782 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
21783 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
21784 (cond_umax@var{mode}): Document.
21785 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
21786 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
21787 (cond_umin_optab, cond_umax_optab): New optabs.
21788 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
21789 (COND_IOR, COND_XOR): New internal functions.
21790 * internal-fn.h (get_conditional_internal_fn): Declare.
21791 * internal-fn.c (cond_binary_direct): New macro.
21792 (expand_cond_binary_optab_fn): Likewise.
21793 (direct_cond_binary_optab_supported_p): Likewise.
21794 (get_conditional_internal_fn): New function.
21795 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
21796 Cope with reduction statements that are vectorized as calls rather
21798 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
21799 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
21800 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
21801 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
21802 (UNSPEC_COND_EOR): New unspecs.
21803 (optab): Add mappings for them.
21804 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
21805 (sve_int_op, sve_fp_op): New int attributes.
21807 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21808 Alan Hayward <alan.hayward@arm.com>
21809 David Sherwood <david.sherwood@arm.com>
21811 * optabs.def (while_ult_optab): New optab.
21812 * doc/md.texi (while_ult@var{m}@var{n}): Document.
21813 * internal-fn.def (WHILE_ULT): New internal function.
21814 * internal-fn.h (direct_internal_fn_supported_p): New override
21815 that takes two types as argument.
21816 * internal-fn.c (while_direct): New macro.
21817 (expand_while_optab_fn): New function.
21818 (convert_optab_supported_p): Likewise.
21819 (direct_while_optab_supported_p): New macro.
21820 * wide-int.h (wi::udiv_ceil): New function.
21821 * tree-vectorizer.h (rgroup_masks): New structure.
21822 (vec_loop_masks): New typedef.
21823 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
21824 and fully_masked_p.
21825 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
21826 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
21827 (vect_max_vf): New function.
21828 (slpeel_make_loop_iterate_ntimes): Delete.
21829 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
21830 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
21831 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
21832 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
21833 internal-fn.h, stor-layout.h and optabs-query.h.
21834 (vect_set_loop_mask): New function.
21835 (add_preheader_seq): Likewise.
21836 (add_header_seq): Likewise.
21837 (interleave_supported_p): Likewise.
21838 (vect_maybe_permute_loop_masks): Likewise.
21839 (vect_set_loop_masks_directly): Likewise.
21840 (vect_set_loop_condition_masked): Likewise.
21841 (vect_set_loop_condition_unmasked): New function, split out from
21842 slpeel_make_loop_iterate_ntimes.
21843 (slpeel_make_loop_iterate_ntimes): Rename to..
21844 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
21845 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
21846 (vect_do_peeling): Update call accordingly.
21847 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
21849 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
21850 mask_compare_type, can_fully_mask_p and fully_masked_p.
21851 (release_vec_loop_masks): New function.
21852 (_loop_vec_info): Use it to free the loop masks.
21853 (can_produce_all_loop_masks_p): New function.
21854 (vect_get_max_nscalars_per_iter): Likewise.
21855 (vect_verify_full_masking): Likewise.
21856 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
21857 retries, and free the mask rgroups before retrying. Check loop-wide
21858 reasons for disallowing fully-masked loops. Make the final decision
21859 about whether use a fully-masked loop or not.
21860 (vect_estimate_min_profitable_iters): Do not assume that peeling
21861 for the number of iterations will be needed for fully-masked loops.
21862 (vectorizable_reduction): Disable fully-masked loops.
21863 (vectorizable_live_operation): Likewise.
21864 (vect_halve_mask_nunits): New function.
21865 (vect_double_mask_nunits): Likewise.
21866 (vect_record_loop_mask): Likewise.
21867 (vect_get_loop_mask): Likewise.
21868 (vect_transform_loop): Handle the case in which the final loop
21869 iteration might handle a partial vector. Call vect_set_loop_condition
21870 instead of slpeel_make_loop_iterate_ntimes.
21871 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
21872 (check_load_store_masking): New function.
21873 (prepare_load_store_mask): Likewise.
21874 (vectorizable_store): Handle fully-masked loops.
21875 (vectorizable_load): Likewise.
21876 (supportable_widening_operation): Use vect_halve_mask_nunits for
21878 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
21879 (vect_gen_while): New function.
21880 * config/aarch64/aarch64.md (umax<mode>3): New expander.
21881 (aarch64_uqdec<mode>): New insn.
21883 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21884 Alan Hayward <alan.hayward@arm.com>
21885 David Sherwood <david.sherwood@arm.com>
21887 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
21888 (reduc_xor_scal_optab): New optabs.
21889 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
21890 (reduc_xor_scal_@var{m}): Document.
21891 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
21892 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
21893 internal functions.
21894 * fold-const-call.c (fold_const_call): Handle them.
21895 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
21896 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
21897 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
21898 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
21899 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
21900 (UNSPEC_XORV): New unspecs.
21901 (optab): Add entries for them.
21902 (BITWISEV): New int iterator.
21903 (bit_reduc_op): New int attributes.
21905 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21906 Alan Hayward <alan.hayward@arm.com>
21907 David Sherwood <david.sherwood@arm.com>
21909 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
21910 * internal-fn.def (VEC_SHL_INSERT): New internal function.
21911 * optabs.def (vec_shl_insert_optab): New optab.
21912 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
21913 (duplicate_and_interleave): Likewise.
21914 * tree-vect-loop.c: Include internal-fn.h.
21915 (neutral_op_for_slp_reduction): New function, split out from
21916 get_initial_defs_for_reduction.
21917 (get_initial_def_for_reduction): Handle option 2 for variable-length
21918 vectors by loading the neutral value into a vector and then shifting
21919 the initial value into element 0.
21920 (get_initial_defs_for_reduction): Replace the code argument with
21921 the neutral value calculated by neutral_op_for_slp_reduction.
21922 Use gimple_build_vector for constant-length vectors.
21923 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
21924 but the first group_size elements have a neutral value.
21925 Use duplicate_and_interleave otherwise.
21926 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
21927 Update call to get_initial_defs_for_reduction. Handle SLP
21928 reductions for variable-length vectors by creating one vector
21929 result for each scalar result, with the elements associated
21930 with other scalar results stubbed out with the neutral value.
21931 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
21932 Require IFN_VEC_SHL_INSERT for double reductions on
21933 variable-length vectors, or SLP reductions that have
21934 a neutral value. Require can_duplicate_and_interleave_p
21935 support for variable-length unchained SLP reductions if there
21936 is no neutral value, such as for MIN/MAX reductions. Also require
21937 the number of vector elements to be a multiple of the number of
21938 SLP statements when doing variable-length unchained SLP reductions.
21939 Update call to vect_create_epilog_for_reduction.
21940 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
21941 and remove initial values.
21942 (duplicate_and_interleave): Make public.
21943 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
21944 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
21946 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21947 Alan Hayward <alan.hayward@arm.com>
21948 David Sherwood <david.sherwood@arm.com>
21950 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
21951 (can_duplicate_and_interleave_p): New function.
21952 (vect_get_and_check_slp_defs): Take the vector of statements
21953 rather than just the current one. Remove excess parentheses.
21954 Restriction rejectinon of vect_constant_def and vect_external_def
21955 for variable-length vectors to boolean types, or types for which
21956 can_duplicate_and_interleave_p is false.
21957 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
21958 (duplicate_and_interleave): New function.
21959 (vect_get_constant_vectors): Use gimple_build_vector for
21960 constant-length vectors and suitable variable-length constant
21961 vectors. Use duplicate_and_interleave for other variable-length
21962 vectors. Don't defer the update when inserting new statements.
21964 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21965 Alan Hayward <alan.hayward@arm.com>
21966 David Sherwood <david.sherwood@arm.com>
21968 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
21969 min_profitable_iters doesn't go negative.
21971 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21972 Alan Hayward <alan.hayward@arm.com>
21973 David Sherwood <david.sherwood@arm.com>
21975 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
21976 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
21977 * optabs.def (vec_mask_load_lanes_optab): New optab.
21978 (vec_mask_store_lanes_optab): Likewise.
21979 * internal-fn.def (MASK_LOAD_LANES): New internal function.
21980 (MASK_STORE_LANES): Likewise.
21981 * internal-fn.c (mask_load_lanes_direct): New macro.
21982 (mask_store_lanes_direct): Likewise.
21983 (expand_mask_load_optab_fn): Handle masked operations.
21984 (expand_mask_load_lanes_optab_fn): New macro.
21985 (expand_mask_store_optab_fn): Handle masked operations.
21986 (expand_mask_store_lanes_optab_fn): New macro.
21987 (direct_mask_load_lanes_optab_supported_p): Likewise.
21988 (direct_mask_store_lanes_optab_supported_p): Likewise.
21989 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
21991 (vect_load_lanes_supported): Likewise.
21992 * tree-vect-data-refs.c (strip_conversion): New function.
21993 (can_group_stmts_p): Likewise.
21994 (vect_analyze_data_ref_accesses): Use it instead of checking
21995 for a pair of assignments.
21996 (vect_store_lanes_supported): Take a masked_p parameter.
21997 (vect_load_lanes_supported): Likewise.
21998 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
21999 vect_store_lanes_supported and vect_load_lanes_supported.
22000 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
22001 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
22002 parameter. Don't allow gaps for masked accesses.
22003 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
22004 and vect_load_lanes_supported.
22005 (get_load_store_type): Take a masked_p parameter and update
22006 call to get_group_load_store_type.
22007 (vectorizable_store): Update call to get_load_store_type.
22008 Handle IFN_MASK_STORE_LANES.
22009 (vectorizable_load): Update call to get_load_store_type.
22010 Handle IFN_MASK_LOAD_LANES.
22012 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22013 Alan Hayward <alan.hayward@arm.com>
22014 David Sherwood <david.sherwood@arm.com>
22016 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
22018 * config/aarch64/aarch64-protos.h
22019 (aarch64_sve_struct_memory_operand_p): Declare.
22020 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
22021 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
22022 (VPRED, vpred): Handle SVE structure modes.
22023 * config/aarch64/constraints.md (Utx): New constraint.
22024 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
22025 (aarch64_sve_struct_nonimmediate_operand): New predicates.
22026 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
22027 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
22028 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
22029 structure modes. Split into pieces after RA.
22030 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
22031 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
22033 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
22034 SVE structure modes.
22035 (aarch64_classify_address): Likewise.
22036 (sizetochar): Move earlier in file.
22037 (aarch64_print_operand): Handle SVE register lists.
22038 (aarch64_array_mode): New function.
22039 (aarch64_sve_struct_memory_operand_p): Likewise.
22040 (TARGET_ARRAY_MODE): Redefine.
22042 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22043 Alan Hayward <alan.hayward@arm.com>
22044 David Sherwood <david.sherwood@arm.com>
22046 * target.def (array_mode): New target hook.
22047 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
22048 * doc/tm.texi: Regenerate.
22049 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
22050 * hooks.c (hook_optmode_mode_uhwi_none): New function.
22051 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
22052 targetm.array_mode.
22053 * stor-layout.c (mode_for_array): Likewise. Support polynomial
22056 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22057 Alan Hayward <alan.hayward@arm.com>
22058 David Sherwood <david.sherwood@arm.com>
22060 * fold-const.c (fold_binary_loc): Check the argument types
22061 rather than the result type when testing for a vector operation.
22063 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22065 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
22066 * doc/tm.texi: Regenerate.
22068 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22069 Alan Hayward <alan.hayward@arm.com>
22070 David Sherwood <david.sherwood@arm.com>
22072 * doc/invoke.texi (-msve-vector-bits=): Document new option.
22073 (sve): Document new AArch64 extension.
22074 * doc/md.texi (w): Extend the description of the AArch64
22075 constraint to include SVE vectors.
22076 (Upl, Upa): Document new AArch64 predicate constraints.
22077 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
22079 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
22080 (msve-vector-bits=): New option.
22081 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
22082 SVE when these are disabled.
22083 (sve): New extension.
22084 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
22085 modes. Adjust their number of units based on aarch64_sve_vg.
22086 (MAX_BITSIZE_MODE_ANY_MODE): Define.
22087 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
22088 aarch64_addr_query_type.
22089 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
22090 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
22091 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
22092 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
22093 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
22094 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
22095 (aarch64_simd_imm_zero_p): Delete.
22096 (aarch64_check_zero_based_sve_index_immediate): Declare.
22097 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
22098 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
22099 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
22100 (aarch64_sve_float_mul_immediate_p): Likewise.
22101 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
22102 rather than an rtx.
22103 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
22104 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
22105 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
22106 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
22107 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
22108 (aarch64_regmode_natural_size): Likewise.
22109 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
22110 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
22112 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
22113 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
22114 for VG and the SVE predicate registers.
22115 (V_ALIASES): Add a "z"-prefixed alias.
22116 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
22117 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
22118 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
22119 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
22120 (REG_CLASS_NAMES): Add entries for them.
22121 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
22122 and the predicate registers.
22123 (aarch64_sve_vg): Declare.
22124 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
22125 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
22126 (REGMODE_NATURAL_SIZE): Define.
22127 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
22129 * config/aarch64/aarch64.c: Include cfgrtl.h.
22130 (simd_immediate_info): Add a constructor for series vectors,
22131 and an associated step field.
22132 (aarch64_sve_vg): New variable.
22133 (aarch64_dbx_register_number): Handle VG and the predicate registers.
22134 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
22135 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
22136 (VEC_ANY_DATA, VEC_STRUCT): New constants.
22137 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
22138 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
22139 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
22140 (aarch64_get_mask_mode): New functions.
22141 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
22142 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
22143 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
22144 predicate modes and predicate registers. Explicitly restrict
22145 GPRs to modes of 16 bytes or smaller. Only allow FP registers
22146 to store a vector mode if it is recognized by
22147 aarch64_classify_vector_mode.
22148 (aarch64_regmode_natural_size): New function.
22149 (aarch64_hard_regno_caller_save_mode): Return the original mode
22151 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
22152 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
22153 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
22154 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
22156 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
22157 does not overlap dest if the function is frame-related. Handle
22159 (aarch64_split_add_offset): New function.
22160 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
22161 them aarch64_add_offset.
22162 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
22163 and update call to aarch64_sub_sp.
22164 (aarch64_add_cfa_expression): New function.
22165 (aarch64_expand_prologue): Pass extra temporary registers to the
22166 functions above. Handle the case in which we need to emit new
22167 DW_CFA_expressions for registers that were originally saved
22168 relative to the stack pointer, but now have to be expressed
22169 relative to the frame pointer.
22170 (aarch64_output_mi_thunk): Pass extra temporary registers to the
22172 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
22173 IP0 and IP1 values for SVE frames.
22174 (aarch64_expand_vec_series): New function.
22175 (aarch64_expand_sve_widened_duplicate): Likewise.
22176 (aarch64_expand_sve_const_vector): Likewise.
22177 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
22178 Handle SVE constants. Use emit_move_insn to move a force_const_mem
22179 into the register, rather than emitting a SET directly.
22180 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
22181 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
22182 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
22183 (offset_9bit_signed_scaled_p): New functions.
22184 (aarch64_replicate_bitmask_imm): New function.
22185 (aarch64_bitmask_imm): Use it.
22186 (aarch64_cannot_force_const_mem): Reject expressions involving
22187 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
22188 (aarch64_classify_index): Handle SVE indices, by requiring
22189 a plain register index with a scale that matches the element size.
22190 (aarch64_classify_address): Handle SVE addresses. Assert that
22191 the mode of the address is VOIDmode or an integer mode.
22192 Update call to aarch64_classify_symbol.
22193 (aarch64_classify_symbolic_expression): Update call to
22194 aarch64_classify_symbol.
22195 (aarch64_const_vec_all_in_range_p): New function.
22196 (aarch64_print_vector_float_operand): Likewise.
22197 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
22198 "vN" for FP registers with SVE modes. Handle (const ...) vectors
22199 and the FP immediates 1.0 and 0.5.
22200 (aarch64_print_address_internal): Handle SVE addresses.
22201 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
22202 (aarch64_regno_regclass): Handle predicate registers.
22203 (aarch64_secondary_reload): Handle big-endian reloads of SVE
22205 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
22206 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
22207 (aarch64_convert_sve_vector_bits): New function.
22208 (aarch64_override_options): Use it to handle -msve-vector-bits=.
22209 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
22210 rather than an rtx.
22211 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
22212 Handle SVE vector and predicate modes. Accept VL-based constants
22213 that need only one temporary register, and VL offsets that require
22214 no temporary registers.
22215 (aarch64_conditional_register_usage): Mark the predicate registers
22216 as fixed if SVE isn't available.
22217 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
22218 Return true for SVE vector and predicate modes.
22219 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
22220 rather than an unsigned int. Handle SVE modes.
22221 (aarch64_preferred_simd_mode): Update call accordingly. Handle
22223 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
22225 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
22226 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
22227 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
22228 (aarch64_sve_float_mul_immediate_p): New functions.
22229 (aarch64_sve_valid_immediate): New function.
22230 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
22231 Explicitly reject structure modes. Check for INDEX constants.
22232 Handle PTRUE and PFALSE constants.
22233 (aarch64_check_zero_based_sve_index_immediate): New function.
22234 (aarch64_simd_imm_zero_p): Delete.
22235 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
22236 vector modes. Accept constants in the range of CNT[BHWD].
22237 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
22238 ask for an Advanced SIMD mode.
22239 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
22240 (aarch64_simd_vector_alignment): Handle SVE predicates.
22241 (aarch64_vectorize_preferred_vector_alignment): New function.
22242 (aarch64_simd_vector_alignment_reachable): Use it instead of
22244 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
22245 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
22247 (MAX_VECT_LEN): Delete.
22248 (expand_vec_perm_d): Add a vec_flags field.
22249 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
22250 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
22251 (aarch64_evpc_ext): Don't apply a big-endian lane correction
22253 (aarch64_evpc_rev): Rename to...
22254 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
22255 (aarch64_evpc_rev_global): New function.
22256 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
22257 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
22259 (aarch64_evpc_sve_tbl): New function.
22260 (aarch64_expand_vec_perm_const_1): Update after rename of
22261 aarch64_evpc_rev. Handle SVE permutes too, trying
22262 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
22263 than aarch64_evpc_tbl.
22264 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
22265 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
22266 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
22267 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
22268 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
22269 (aarch64_expand_sve_vcond): New functions.
22270 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
22271 of aarch64_vector_mode_p.
22272 (aarch64_dwarf_poly_indeterminate_value): New function.
22273 (aarch64_compute_pressure_classes): Likewise.
22274 (aarch64_can_change_mode_class): Likewise.
22275 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
22276 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
22277 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
22278 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
22279 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
22280 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
22281 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
22282 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
22284 (Dn, Dl, Dr): Accept const as well as const_vector.
22285 (Dz): Likewise. Compare against CONST0_RTX.
22286 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
22287 of "vector" where appropriate.
22288 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
22289 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
22290 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
22291 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
22292 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
22293 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
22294 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
22295 (v_int_equiv): Extend to SVE modes.
22296 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
22298 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
22299 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
22300 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
22301 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
22302 (SVE_COND_FP_CMP): New int iterators.
22303 (perm_hilo): Handle the new unpack unspecs.
22304 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
22306 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
22307 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
22308 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
22309 (aarch64_equality_operator, aarch64_constant_vector_operand)
22310 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
22311 (aarch64_sve_nonimmediate_operand): Likewise.
22312 (aarch64_sve_general_operand): Likewise.
22313 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
22314 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
22315 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
22316 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
22317 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
22318 (aarch64_sve_float_arith_immediate): Likewise.
22319 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
22320 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
22321 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
22322 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
22323 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
22324 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
22325 (aarch64_sve_float_arith_operand): Likewise.
22326 (aarch64_sve_float_arith_with_sub_operand): Likewise.
22327 (aarch64_sve_float_mul_operand): Likewise.
22328 (aarch64_sve_vec_perm_operand): Likewise.
22329 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
22330 (aarch64_mov_operand): Accept const_poly_int and const_vector.
22331 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
22332 as well as const_vector.
22333 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
22334 in file. Use CONST0_RTX and CONSTM1_RTX.
22335 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
22336 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
22337 Use aarch64_simd_imm_zero.
22338 * config/aarch64/aarch64-sve.md: New file.
22339 * config/aarch64/aarch64.md: Include it.
22340 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
22341 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
22342 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
22343 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
22344 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
22345 (sve): New attribute.
22346 (enabled): Disable instructions with the sve attribute unless
22348 (movqi, movhi): Pass CONST_POLY_INT operaneds through
22349 aarch64_expand_mov_immediate.
22350 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
22351 CNT[BHSD] immediates.
22352 (movti): Split CONST_POLY_INT moves into two halves.
22353 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
22354 Split additions that need a temporary here if the destination
22355 is the stack pointer.
22356 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
22357 (*add<mode>3_poly_1): New instruction.
22358 (set_clobber_cc): New expander.
22360 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22362 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
22363 parameter and use it instead of GET_MODE_SIZE (innermode). Use
22364 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
22365 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
22366 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
22367 Change innermode from fixed_mode_size to machine_mode.
22368 (simplify_subreg): Update call accordingly. Handle a constant-sized
22369 subreg of a variable-length CONST_VECTOR.
22371 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22372 Alan Hayward <alan.hayward@arm.com>
22373 David Sherwood <david.sherwood@arm.com>
22375 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
22376 (add_offset_to_base): New function, split out from...
22377 (create_mem_ref): ...here. When handling a scale other than 1,
22378 check first whether the address is valid without the offset.
22379 Add it into the base if so, leaving the index and scale as-is.
22381 2018-01-12 Jakub Jelinek <jakub@redhat.com>
22384 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
22385 fold_for_warn before checking if arg2 is INTEGER_CST.
22387 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
22389 * config/rs6000/predicates.md (load_multiple_operation): Delete.
22390 (store_multiple_operation): Delete.
22391 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
22392 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
22393 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
22394 guarded by TARGET_STRING.
22395 (rs6000_output_load_multiple): Delete.
22396 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22397 OPTION_MASK_STRING / TARGET_STRING handling.
22398 (print_operand) <'N', 'O'>: Add comment that these are unused now.
22399 (const rs6000_opt_masks) <"string">: Change mask to 0.
22400 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
22401 (MASK_STRING): Delete.
22402 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
22404 (load_multiple): Delete.
22411 (store_multiple): Delete.
22418 (movmemsi_8reg): Delete.
22419 (corresponding unnamed define_insn): Delete.
22420 (movmemsi_6reg): Delete.
22421 (corresponding unnamed define_insn): Delete.
22422 (movmemsi_4reg): Delete.
22423 (corresponding unnamed define_insn): Delete.
22424 (movmemsi_2reg): Delete.
22425 (corresponding unnamed define_insn): Delete.
22426 (movmemsi_1reg): Delete.
22427 (corresponding unnamed define_insn): Delete.
22428 * config/rs6000/rs6000.opt (mno-string): New.
22429 (mstring): Replace by deprecation warning stub.
22430 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
22432 2018-01-12 Jakub Jelinek <jakub@redhat.com>
22434 * regrename.c (regrename_do_replace): If replacing the same
22435 reg multiple times, try to reuse last created gen_raw_REG.
22438 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
22439 main to workaround a bug in GDB.
22441 2018-01-12 Tom de Vries <tom@codesourcery.com>
22444 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
22446 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
22448 PR rtl-optimization/80481
22449 * ira-color.c (get_cap_member): New function.
22450 (allocnos_conflict_by_live_ranges_p): Use it.
22451 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
22452 (setup_slot_coalesced_allocno_live_ranges): Ditto.
22454 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
22457 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
22458 (*saddl_se_1): Ditto.
22459 (*ssubsi_1): Ditto.
22460 (*ssubl_se_1): Ditto.
22462 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
22464 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
22465 rather than wi::to_widest for DR_INITs.
22466 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
22467 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
22468 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
22470 (vect_analyze_group_access_1): Note that here.
22472 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
22474 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
22475 polynomial type sizes.
22477 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
22479 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
22480 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
22481 (gimple_add_tmp_var): Likewise.
22483 2018-01-12 Martin Liska <mliska@suse.cz>
22485 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
22486 (gimple_alloc_sizes): Likewise.
22487 (dump_gimple_statistics): Use PRIu64 in printf format.
22488 * gimple.h: Change uint64_t to int.
22490 2018-01-12 Martin Liska <mliska@suse.cz>
22492 * tree-core.h: Use uint64_t instead of int.
22493 * tree.c (tree_node_counts): Likewise.
22494 (tree_node_sizes): Likewise.
22495 (dump_tree_statistics): Use PRIu64 in printf format.
22497 2018-01-12 Martin Liska <mliska@suse.cz>
22499 * Makefile.in: As qsort_chk is implemented in vec.c, add
22500 vec.o to linkage of gencfn-macros.
22501 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
22502 passing the info to record_node_allocation_statistics.
22503 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
22505 * ggc-common.c (struct ggc_usage): Add operator== and use
22506 it in operator< and compare function.
22507 * mem-stats.h (struct mem_usage): Likewise.
22508 * vec.c (struct vec_usage): Remove operator< and compare
22509 function. Can be simply inherited.
22511 2018-01-12 Martin Jambor <mjambor@suse.cz>
22514 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
22515 * tree-ssa-math-opts.c: Include domwalk.h.
22516 (convert_mult_to_fma_1): New function.
22517 (fma_transformation_info): New type.
22518 (fma_deferring_state): Likewise.
22519 (cancel_fma_deferring): New function.
22520 (result_of_phi): Likewise.
22521 (last_fma_candidate_feeds_initial_phi): Likewise.
22522 (convert_mult_to_fma): Added deferring logic, split actual
22523 transformation to convert_mult_to_fma_1.
22524 (math_opts_dom_walker): New type.
22525 (math_opts_dom_walker::after_dom_children): New method, body moved
22526 here from pass_optimize_widening_mul::execute, added deferring logic
22528 (pass_optimize_widening_mul::execute): Moved most of code to
22529 math_opts_dom_walker::after_dom_children.
22530 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
22531 * config/i386/i386.c (ix86_option_override_internal): Added
22532 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
22534 2018-01-12 Richard Biener <rguenther@suse.de>
22537 * dwarf2out.c (gen_variable_die): Do not reset old_die for
22538 inline instance vars.
22540 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
22543 * config/rx/rx.c (rx_is_restricted_memory_address):
22544 Handle SUBREG case.
22546 2018-01-12 Richard Biener <rguenther@suse.de>
22548 PR tree-optimization/80846
22549 * target.def (split_reduction): New target hook.
22550 * targhooks.c (default_split_reduction): New function.
22551 * targhooks.h (default_split_reduction): Declare.
22552 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
22553 target requests first reduce vectors by combining low and high
22555 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
22556 (get_vectype_for_scalar_type_and_size): Export.
22557 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
22558 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
22559 * doc/tm.texi: Regenerate.
22560 * config/i386/i386.c (ix86_split_reduction): Implement
22561 TARGET_VECTORIZE_SPLIT_REDUCTION.
22563 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
22566 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
22567 in PIC mode except for TARGET_VXWORKS_RTP.
22568 * config/sparc/sparc.c: Include cfgrtl.h.
22569 (TARGET_INIT_PIC_REG): Define.
22570 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
22571 (sparc_pic_register_p): New predicate.
22572 (sparc_legitimate_address_p): Use it.
22573 (sparc_legitimize_pic_address): Likewise.
22574 (sparc_delegitimize_address): Likewise.
22575 (sparc_mode_dependent_address_p): Likewise.
22576 (gen_load_pcrel_sym): Remove 4th parameter.
22577 (load_got_register): Adjust call to above. Remove obsolete stuff.
22578 (sparc_expand_prologue): Do not call load_got_register here.
22579 (sparc_flat_expand_prologue): Likewise.
22580 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
22581 (sparc_use_pseudo_pic_reg): New function.
22582 (sparc_init_pic_reg): Likewise.
22583 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
22584 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
22586 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
22588 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22589 Add item for branch_cost.
22591 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
22593 PR rtl-optimization/83565
22594 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
22595 not extend the result to a larger mode for rotate operations.
22596 (num_sign_bit_copies1): Likewise.
22598 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22601 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
22603 Use values-Xc.o for -pedantic.
22604 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
22606 2018-01-12 Martin Liska <mliska@suse.cz>
22609 * ipa-devirt.c (final_warning_record::grow_type_warnings):
22611 (possible_polymorphic_call_targets): Use it.
22612 (ipa_devirt): Likewise.
22614 2018-01-12 Martin Liska <mliska@suse.cz>
22616 * profile-count.h (enum profile_quality): Use 0 as invalid
22617 enum value of profile_quality.
22619 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
22621 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
22622 -mext-string options.
22624 2018-01-12 Richard Biener <rguenther@suse.de>
22626 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
22627 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
22628 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
22630 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
22632 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
22634 * configure.ac (--with-long-double-format): Add support for the
22635 configuration option to change the default long double format on
22637 * config.gcc (powerpc*-linux*-*): Likewise.
22638 * configure: Regenerate.
22639 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
22640 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
22641 used without modification.
22643 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22645 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
22646 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
22647 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
22648 MISC_BUILTIN_SPEC_BARRIER.
22649 (rs6000_init_builtins): Likewise.
22650 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
22652 (speculation_barrier): New define_insn.
22653 * doc/extend.texi: Document __builtin_speculation_barrier.
22655 2018-01-11 Jakub Jelinek <jakub@redhat.com>
22658 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
22659 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
22660 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
22662 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
22663 integral modes instead of "ss" and "sd".
22664 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
22665 vectors with 32-bit and 64-bit elements.
22666 (vecdupssescalarmodesuffix): New mode attribute.
22667 (vec_dup<mode>): Use it.
22669 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
22672 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
22673 frame if argument is passed on stack.
22675 2018-01-11 Jakub Jelinek <jakub@redhat.com>
22678 * ree.c (combine_reaching_defs): Optimize also
22679 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
22680 reg2=any_extend(exp); reg1=reg2;, formatting fix.
22682 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
22684 PR middle-end/83189
22685 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
22687 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
22689 PR middle-end/83718
22690 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
22691 after they are computed.
22693 2018-01-11 Bin Cheng <bin.cheng@arm.com>
22695 PR tree-optimization/83695
22696 * gimple-loop-linterchange.cc
22697 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
22698 reset cached scev information after interchange.
22699 (pass_linterchange::execute): Remove call to scev_reset_htab.
22701 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22703 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
22704 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
22705 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
22706 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
22707 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
22708 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
22709 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
22710 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
22711 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
22712 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
22713 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
22714 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
22715 (V_lane_reg): Likewise.
22716 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
22718 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
22719 (vfmal_lane_low<mode>_intrinsic,
22720 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
22721 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
22722 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
22723 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
22724 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
22725 vfmsl_lane_high<mode>_intrinsic): New define_insns.
22727 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22729 * config/arm/arm-cpus.in (fp16fml): New feature.
22730 (ALL_SIMD): Add fp16fml.
22731 (armv8.2-a): Add fp16fml as an option.
22732 (armv8.3-a): Likewise.
22733 (armv8.4-a): Add fp16fml as part of fp16.
22734 * config/arm/arm.h (TARGET_FP16FML): Define.
22735 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
22737 * config/arm/arm-modes.def (V2HF): Define.
22738 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
22739 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
22740 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
22741 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
22742 vfmsl_low, vfmsl_high): New set of builtins.
22743 * config/arm/iterators.md (PLUSMINUS): New code iterator.
22744 (vfml_op): New code attribute.
22745 (VFMLHALVES): New int iterator.
22746 (VFML, VFMLSEL): New mode attributes.
22747 (V_reg): Define mapping for V2HF.
22748 (V_hi, V_lo): New mode attributes.
22749 (VF_constraint): Likewise.
22750 (vfml_half, vfml_half_selector): New int attributes.
22751 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
22753 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
22754 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
22756 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
22757 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
22758 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
22759 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
22761 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
22762 Document new effective target and option set.
22764 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22766 * config/arm/arm-cpus.in (armv8_4): New feature.
22767 (ARMv8_4a): New fgroup.
22768 (armv8.4-a): New arch.
22769 * config/arm/arm-tables.opt: Regenerate.
22770 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
22771 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
22772 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
22773 Add matching rules for -march=armv8.4-a and extensions.
22774 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
22776 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
22779 * config/rx/rx.md (BW): New mode attribute.
22780 (sync_lock_test_and_setsi): Add mode suffix to insn output.
22782 2018-01-11 Richard Biener <rguenther@suse.de>
22784 PR tree-optimization/83435
22785 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
22786 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
22787 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
22789 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22790 Alan Hayward <alan.hayward@arm.com>
22791 David Sherwood <david.sherwood@arm.com>
22793 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
22795 (aarch64_classify_address): Initialize it. Track polynomial offsets.
22796 (aarch64_print_address_internal): Use it to check for a zero offset.
22798 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22799 Alan Hayward <alan.hayward@arm.com>
22800 David Sherwood <david.sherwood@arm.com>
22802 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
22803 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
22804 Return a poly_int64 rather than a HOST_WIDE_INT.
22805 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
22806 rather than a HOST_WIDE_INT.
22807 * config/aarch64/aarch64.h (aarch64_frame): Protect with
22808 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
22809 hard_fp_offset, frame_size, initial_adjust, callee_offset and
22810 final_offset from HOST_WIDE_INT to poly_int64.
22811 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
22812 to_constant when getting the number of units in an Advanced SIMD
22814 (aarch64_builtin_vectorized_function): Check for a constant number
22816 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
22818 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
22819 attribute instead of GET_MODE_NUNITS.
22820 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
22821 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
22822 GET_MODE_SIZE for fixed-size registers.
22823 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
22824 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
22825 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
22826 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
22827 (aarch64_print_operand, aarch64_print_address_internal)
22828 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
22829 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
22830 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
22831 Handle polynomial GET_MODE_SIZE.
22832 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
22833 wider than SImode without modification.
22834 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
22835 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
22836 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
22837 passing and returning SVE modes.
22838 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
22839 rather than GEN_INT.
22840 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
22841 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
22842 (aarch64_allocate_and_probe_stack_space): Likewise.
22843 (aarch64_layout_frame): Cope with polynomial offsets.
22844 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
22845 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
22846 polynomial offsets.
22847 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
22848 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
22849 poly_int64 rather than a HOST_WIDE_INT.
22850 (aarch64_get_separate_components, aarch64_process_components)
22851 (aarch64_expand_prologue, aarch64_expand_epilogue)
22852 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
22853 (aarch64_anchor_offset): New function, split out from...
22854 (aarch64_legitimize_address): ...here.
22855 (aarch64_builtin_vectorization_cost): Handle polynomial
22856 TYPE_VECTOR_SUBPARTS.
22857 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
22859 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
22860 number of elements from the PARALLEL rather than the mode.
22861 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
22862 rather than GET_MODE_BITSIZE.
22863 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
22864 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
22865 (aarch64_expand_vec_perm_const_1): Handle polynomial
22866 d->perm.length () and d->perm elements.
22867 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
22868 Apply to_constant to d->perm elements.
22869 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
22870 polynomial CONST_VECTOR_NUNITS.
22871 (aarch64_move_pointer): Take amount as a poly_int64 rather
22873 (aarch64_progress_pointer): Avoid temporary variable.
22874 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
22875 the mode attribute instead of GET_MODE.
22877 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22878 Alan Hayward <alan.hayward@arm.com>
22879 David Sherwood <david.sherwood@arm.com>
22881 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
22882 x exists before using it.
22883 (aarch64_add_constant_internal): Rename to...
22884 (aarch64_add_offset_1): ...this. Replace regnum with separate
22885 src and dest rtxes. Handle the case in which they're different,
22886 including when the offset is zero. Replace scratchreg with an rtx.
22887 Use 2 additions if there is no spare register into which we can
22888 move a 16-bit constant.
22889 (aarch64_add_constant): Delete.
22890 (aarch64_add_offset): Replace reg with separate src and dest
22891 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
22892 Use aarch64_add_offset_1.
22893 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
22894 an rtx rather than an int. Take the delta as a poly_int64
22895 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
22896 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
22897 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
22898 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
22899 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
22900 and aarch64_add_sp.
22901 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
22902 aarch64_add_constant.
22904 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22906 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22907 Use scalar_float_mode.
22909 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22911 * config/aarch64/aarch64-simd.md
22912 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
22913 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
22914 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
22915 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
22916 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
22917 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
22918 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
22919 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
22920 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
22921 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
22923 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22926 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
22927 targ_options->x_arm_arch_string is non NULL.
22929 2018-01-11 Tamar Christina <tamar.christina@arm.com>
22931 * config/aarch64/aarch64.h
22932 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
22934 2018-01-11 Sudakshina Das <sudi.das@arm.com>
22937 * expmed.c (emit_store_flag_force): Swap if const op0
22938 and change VOIDmode to mode of op0.
22940 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22942 PR rtl-optimization/83761
22943 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
22944 than bytes to mode_for_size.
22946 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
22948 PR middle-end/83189
22949 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
22950 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
22953 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
22955 PR middle-end/83575
22956 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
22957 when in layout mode.
22958 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
22959 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
22962 2018-01-10 Michael Collison <michael.collison@arm.com>
22964 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
22965 * config/aarch64/aarch64-option-extension.def: Add
22966 AARCH64_OPT_EXTENSION of 'fp16fml'.
22967 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
22968 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
22969 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
22970 * config/aarch64/constraints.md (Ui7): New constraint.
22971 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
22972 (VFMLA_SEL_W): Ditto.
22975 (VFMLA16_LOW): New int iterator.
22976 (VFMLA16_HIGH): Ditto.
22977 (UNSPEC_FMLAL): New unspec.
22978 (UNSPEC_FMLSL): Ditto.
22979 (UNSPEC_FMLAL2): Ditto.
22980 (UNSPEC_FMLSL2): Ditto.
22981 (f16mac): New code attribute.
22982 * config/aarch64/aarch64-simd-builtins.def
22983 (aarch64_fmlal_lowv2sf): Ditto.
22984 (aarch64_fmlsl_lowv2sf): Ditto.
22985 (aarch64_fmlalq_lowv4sf): Ditto.
22986 (aarch64_fmlslq_lowv4sf): Ditto.
22987 (aarch64_fmlal_highv2sf): Ditto.
22988 (aarch64_fmlsl_highv2sf): Ditto.
22989 (aarch64_fmlalq_highv4sf): Ditto.
22990 (aarch64_fmlslq_highv4sf): Ditto.
22991 (aarch64_fmlal_lane_lowv2sf): Ditto.
22992 (aarch64_fmlsl_lane_lowv2sf): Ditto.
22993 (aarch64_fmlal_laneq_lowv2sf): Ditto.
22994 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
22995 (aarch64_fmlalq_lane_lowv4sf): Ditto.
22996 (aarch64_fmlsl_lane_lowv4sf): Ditto.
22997 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
22998 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
22999 (aarch64_fmlal_lane_highv2sf): Ditto.
23000 (aarch64_fmlsl_lane_highv2sf): Ditto.
23001 (aarch64_fmlal_laneq_highv2sf): Ditto.
23002 (aarch64_fmlsl_laneq_highv2sf): Ditto.
23003 (aarch64_fmlalq_lane_highv4sf): Ditto.
23004 (aarch64_fmlsl_lane_highv4sf): Ditto.
23005 (aarch64_fmlalq_laneq_highv4sf): Ditto.
23006 (aarch64_fmlsl_laneq_highv4sf): Ditto.
23007 * config/aarch64/aarch64-simd.md:
23008 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
23009 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
23010 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
23011 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
23012 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
23013 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
23014 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
23015 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
23016 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
23017 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
23018 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
23019 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
23020 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
23021 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
23022 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
23023 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
23024 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
23025 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
23026 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
23027 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
23028 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
23029 (vfmlsl_low_u32): Ditto.
23030 (vfmlalq_low_u32): Ditto.
23031 (vfmlslq_low_u32): Ditto.
23032 (vfmlal_high_u32): Ditto.
23033 (vfmlsl_high_u32): Ditto.
23034 (vfmlalq_high_u32): Ditto.
23035 (vfmlslq_high_u32): Ditto.
23036 (vfmlal_lane_low_u32): Ditto.
23037 (vfmlsl_lane_low_u32): Ditto.
23038 (vfmlal_laneq_low_u32): Ditto.
23039 (vfmlsl_laneq_low_u32): Ditto.
23040 (vfmlalq_lane_low_u32): Ditto.
23041 (vfmlslq_lane_low_u32): Ditto.
23042 (vfmlalq_laneq_low_u32): Ditto.
23043 (vfmlslq_laneq_low_u32): Ditto.
23044 (vfmlal_lane_high_u32): Ditto.
23045 (vfmlsl_lane_high_u32): Ditto.
23046 (vfmlal_laneq_high_u32): Ditto.
23047 (vfmlsl_laneq_high_u32): Ditto.
23048 (vfmlalq_lane_high_u32): Ditto.
23049 (vfmlslq_lane_high_u32): Ditto.
23050 (vfmlalq_laneq_high_u32): Ditto.
23051 (vfmlslq_laneq_high_u32): Ditto.
23052 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
23053 (AARCH64_FL_FOR_ARCH8_4): New.
23054 (AARCH64_ISA_F16FML): New ISA flag.
23055 (TARGET_F16FML): New feature flag for fp16fml.
23056 (doc/invoke.texi): Document new fp16fml option.
23058 2018-01-10 Michael Collison <michael.collison@arm.com>
23060 * config/aarch64/aarch64-builtins.c:
23061 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
23062 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23063 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
23064 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
23065 (AARCH64_ISA_SHA3): New ISA flag.
23066 (TARGET_SHA3): New feature flag for sha3.
23067 * config/aarch64/iterators.md (sha512_op): New int attribute.
23068 (CRYPTO_SHA512): New int iterator.
23069 (UNSPEC_SHA512H): New unspec.
23070 (UNSPEC_SHA512H2): Ditto.
23071 (UNSPEC_SHA512SU0): Ditto.
23072 (UNSPEC_SHA512SU1): Ditto.
23073 * config/aarch64/aarch64-simd-builtins.def
23074 (aarch64_crypto_sha512hqv2di): New builtin.
23075 (aarch64_crypto_sha512h2qv2di): Ditto.
23076 (aarch64_crypto_sha512su0qv2di): Ditto.
23077 (aarch64_crypto_sha512su1qv2di): Ditto.
23078 (aarch64_eor3qv8hi): Ditto.
23079 (aarch64_rax1qv2di): Ditto.
23080 (aarch64_xarqv2di): Ditto.
23081 (aarch64_bcaxqv8hi): Ditto.
23082 * config/aarch64/aarch64-simd.md:
23083 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
23084 (aarch64_crypto_sha512su0qv2di): Ditto.
23085 (aarch64_crypto_sha512su1qv2di): Ditto.
23086 (aarch64_eor3qv8hi): Ditto.
23087 (aarch64_rax1qv2di): Ditto.
23088 (aarch64_xarqv2di): Ditto.
23089 (aarch64_bcaxqv8hi): Ditto.
23090 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
23091 (vsha512h2q_u64): Ditto.
23092 (vsha512su0q_u64): Ditto.
23093 (vsha512su1q_u64): Ditto.
23094 (veor3q_u16): Ditto.
23095 (vrax1q_u64): Ditto.
23096 (vxarq_u64): Ditto.
23097 (vbcaxq_u16): Ditto.
23098 * config/arm/types.md (crypto_sha512): New type attribute.
23099 (crypto_sha3): Ditto.
23100 (doc/invoke.texi): Document new sha3 option.
23102 2018-01-10 Michael Collison <michael.collison@arm.com>
23104 * config/aarch64/aarch64-builtins.c:
23105 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
23106 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23107 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
23108 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
23109 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
23110 (AARCH64_ISA_SM4): New ISA flag.
23111 (TARGET_SM4): New feature flag for sm4.
23112 * config/aarch64/aarch64-simd-builtins.def
23113 (aarch64_sm3ss1qv4si): Ditto.
23114 (aarch64_sm3tt1aq4si): Ditto.
23115 (aarch64_sm3tt1bq4si): Ditto.
23116 (aarch64_sm3tt2aq4si): Ditto.
23117 (aarch64_sm3tt2bq4si): Ditto.
23118 (aarch64_sm3partw1qv4si): Ditto.
23119 (aarch64_sm3partw2qv4si): Ditto.
23120 (aarch64_sm4eqv4si): Ditto.
23121 (aarch64_sm4ekeyqv4si): Ditto.
23122 * config/aarch64/aarch64-simd.md:
23123 (aarch64_sm3ss1qv4si): Ditto.
23124 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
23125 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
23126 (aarch64_sm4eqv4si): Ditto.
23127 (aarch64_sm4ekeyqv4si): Ditto.
23128 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
23129 (sm3part_op): Ditto.
23130 (CRYPTO_SM3TT): Ditto.
23131 (CRYPTO_SM3PART): Ditto.
23132 (UNSPEC_SM3SS1): New unspec.
23133 (UNSPEC_SM3TT1A): Ditto.
23134 (UNSPEC_SM3TT1B): Ditto.
23135 (UNSPEC_SM3TT2A): Ditto.
23136 (UNSPEC_SM3TT2B): Ditto.
23137 (UNSPEC_SM3PARTW1): Ditto.
23138 (UNSPEC_SM3PARTW2): Ditto.
23139 (UNSPEC_SM4E): Ditto.
23140 (UNSPEC_SM4EKEY): Ditto.
23141 * config/aarch64/constraints.md (Ui2): New constraint.
23142 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
23143 * config/arm/types.md (crypto_sm3): New type attribute.
23144 (crypto_sm4): Ditto.
23145 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
23146 (vsm3tt1aq_u32): Ditto.
23147 (vsm3tt1bq_u32): Ditto.
23148 (vsm3tt2aq_u32): Ditto.
23149 (vsm3tt2bq_u32): Ditto.
23150 (vsm3partw1q_u32): Ditto.
23151 (vsm3partw2q_u32): Ditto.
23152 (vsm4eq_u32): Ditto.
23153 (vsm4ekeyq_u32): Ditto.
23154 (doc/invoke.texi): Document new sm4 option.
23156 2018-01-10 Michael Collison <michael.collison@arm.com>
23158 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
23159 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
23160 (AARCH64_FL_FOR_ARCH8_4): New.
23161 (AARCH64_FL_V8_4): New flag.
23162 (doc/invoke.texi): Document new armv8.4-a option.
23164 2018-01-10 Michael Collison <michael.collison@arm.com>
23166 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23167 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
23168 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
23169 * config/aarch64/aarch64-option-extension.def: Add
23170 AARCH64_OPT_EXTENSION of 'sha2'.
23171 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
23172 (crypto): Disable sha2 and aes if crypto disabled.
23173 (crypto): Enable aes and sha2 if enabled.
23174 (simd): Disable sha2 and aes if simd disabled.
23175 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
23177 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
23178 (TARGET_SHA2): New feature flag for sha2.
23179 (TARGET_AES): New feature flag for aes.
23180 * config/aarch64/aarch64-simd.md:
23181 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
23182 conditional on TARGET_AES.
23183 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
23184 (aarch64_crypto_sha1hsi): Make pattern conditional
23186 (aarch64_crypto_sha1hv4si): Ditto.
23187 (aarch64_be_crypto_sha1hv4si): Ditto.
23188 (aarch64_crypto_sha1su1v4si): Ditto.
23189 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
23190 (aarch64_crypto_sha1su0v4si): Ditto.
23191 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
23192 (aarch64_crypto_sha256su0v4si): Ditto.
23193 (aarch64_crypto_sha256su1v4si): Ditto.
23194 (doc/invoke.texi): Document new aes and sha2 options.
23196 2018-01-10 Martin Sebor <msebor@redhat.com>
23198 PR tree-optimization/83781
23199 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
23202 2018-01-11 Martin Sebor <msebor@gmail.com>
23203 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23205 PR tree-optimization/83501
23206 PR tree-optimization/81703
23208 * tree-ssa-strlen.c (get_string_cst): Rename...
23209 (get_string_len): ...to this. Handle global constants.
23210 (handle_char_store): Adjust.
23212 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
23213 Jim Wilson <jimw@sifive.com>
23215 * config/riscv/riscv-protos.h (riscv_output_return): New.
23216 * config/riscv/riscv.c (struct machine_function): New naked_p field.
23217 (riscv_attribute_table, riscv_output_return),
23218 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
23219 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
23220 (riscv_compute_frame_info): Only compute frame->mask if not a naked
23222 (riscv_expand_prologue): Add early return for naked function.
23223 (riscv_expand_epilogue): Likewise.
23224 (riscv_function_ok_for_sibcall): Return false for naked function.
23225 (riscv_set_current_function): New.
23226 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
23227 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
23228 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
23229 * doc/extend.texi (RISC-V Function Attributes): New.
23231 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
23233 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
23234 check for 128-bit long double before checking TCmode.
23235 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
23236 128-bit long doubles before checking TFmode or TCmode.
23237 (FLOAT128_IBM_P): Likewise.
23239 2018-01-10 Martin Sebor <msebor@redhat.com>
23241 PR tree-optimization/83671
23242 * builtins.c (c_strlen): Unconditionally return zero for the empty
23244 Use -Warray-bounds for warnings.
23245 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
23246 for non-constant array indices with COMPONENT_REF, arrays of
23247 arrays, and pointers to arrays.
23248 (gimple_fold_builtin_strlen): Determine and set length range for
23249 non-constant character arrays.
23251 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
23253 PR middle-end/81897
23254 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
23257 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
23259 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
23261 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
23264 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
23265 VECTOR_MEM_ALTIVEC_OR_VSX_P.
23266 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
23267 indexed_or_indirect_operand predicate.
23268 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
23269 (*vsx_le_perm_load_v8hi): Likewise.
23270 (*vsx_le_perm_load_v16qi): Likewise.
23271 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
23272 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
23273 (*vsx_le_perm_store_v8hi): Likewise.
23274 (*vsx_le_perm_store_v16qi): Likewise.
23275 (eight unnamed splitters): Likewise.
23277 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
23279 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
23280 * config/rs6000/emmintrin.h: Likewise.
23281 * config/rs6000/mmintrin.h: Likewise.
23282 * config/rs6000/xmmintrin.h: Likewise.
23284 2018-01-10 David Malcolm <dmalcolm@redhat.com>
23287 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
23289 * tree.c (tree_nop_conversion): Return true for location wrapper
23291 (maybe_wrap_with_location): New function.
23292 (selftest::check_strip_nops): New function.
23293 (selftest::test_location_wrappers): New function.
23294 (selftest::tree_c_tests): Call it.
23295 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
23296 (maybe_wrap_with_location): New decl.
23297 (EXPR_LOCATION_WRAPPER_P): New macro.
23298 (location_wrapper_p): New inline function.
23299 (tree_strip_any_location_wrapper): New inline function.
23301 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
23304 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
23305 stack_realign_offset for the largest alignment of stack slot
23307 (ix86_find_max_used_stack_alignment): New function.
23308 (ix86_finalize_stack_frame_flags): Use it. Set
23309 max_used_stack_alignment if we don't realign stack.
23310 * config/i386/i386.h (machine_function): Add
23311 max_used_stack_alignment.
23313 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
23315 * config/arm/arm.opt (-mbranch-cost): New option.
23316 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
23319 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
23322 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
23323 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
23325 2018-01-10 Richard Biener <rguenther@suse.de>
23328 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
23329 early out so it also covers the case where we have a non-NULL
23332 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
23334 PR tree-optimization/83753
23335 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
23336 for non-strided grouped accesses if the number of elements is 1.
23338 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
23341 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
23342 * i386.h (TARGET_USE_GATHER): Define.
23343 * x86-tune.def (X86_TUNE_USE_GATHER): New.
23345 2018-01-10 Martin Liska <mliska@suse.cz>
23348 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
23349 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
23351 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
23352 CLEANUP_NO_PARTITIONING is not set.
23354 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
23356 * doc/rtl.texi: Remove documentation of (const ...) wrappers
23357 for vectors, as a partial revert of r254296.
23358 * rtl.h (const_vec_p): Delete.
23359 (const_vec_duplicate_p): Don't test for vector CONSTs.
23360 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
23361 * expmed.c (make_tree): Likewise.
23364 * common.md (E, F): Use CONSTANT_P instead of checking for
23366 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
23367 checking for CONST_VECTOR.
23369 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
23371 PR middle-end/83575
23372 * predict.c (force_edge_cold): Handle in more sane way edges
23373 with no prediction.
23375 2018-01-09 Carl Love <cel@us.ibm.com>
23377 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
23379 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
23380 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
23381 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
23382 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
23383 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
23384 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
23385 * config/rs6000/rs6000-protos.h: Add extern defition for
23386 rs6000_generate_float2_double_code.
23387 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
23389 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
23390 (float2_v2df): Add define_expand.
23392 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
23395 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
23396 op_mode in the force_to_mode call.
23398 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
23400 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
23401 instead of checking each element individually.
23402 (aarch64_evpc_uzp): Likewise.
23403 (aarch64_evpc_zip): Likewise.
23404 (aarch64_evpc_ext): Likewise.
23405 (aarch64_evpc_rev): Likewise.
23406 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
23407 instead of checking each element individually. Return true without
23409 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
23410 whether all selected elements come from the same input, instead of
23411 checking each element individually. Remove calls to gen_rtx_REG,
23412 start_sequence and end_sequence and instead assert that no rtl is
23415 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
23417 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
23418 order of HIGH and CONST checks.
23420 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
23422 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
23423 if the destination isn't an SSA_NAME.
23425 2018-01-09 Richard Biener <rguenther@suse.de>
23427 PR tree-optimization/83668
23428 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
23430 (canonicalize_loop_form): ... here, renamed from ...
23431 (canonicalize_loop_closed_ssa_form): ... this and amended to
23432 swap successor edges for loop exit blocks to make us use
23433 the RPO order we need for initial schedule generation.
23435 2018-01-09 Joseph Myers <joseph@codesourcery.com>
23437 PR tree-optimization/64811
23438 * match.pd: When optimizing comparisons with Inf, avoid
23439 introducing or losing exceptions from comparisons with NaN.
23441 2018-01-09 Martin Liska <mliska@suse.cz>
23444 * asan.c (shadow_mem_size): Add gcc_assert.
23446 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
23448 Don't save registers in main().
23451 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
23452 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
23453 * config/avr/avr.c (avr_set_current_function): Don't error if
23454 naked, OS_task or OS_main are specified at the same time.
23455 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
23457 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
23459 * common/config/avr/avr-common.c (avr_option_optimization_table):
23460 Switch on -mmain-is-OS_task for optimizing compilations.
23462 2018-01-09 Richard Biener <rguenther@suse.de>
23464 PR tree-optimization/83572
23465 * graphite.c: Include cfganal.h.
23466 (graphite_transform_loops): Connect infinite loops to exit
23467 and remove fake edges at the end.
23469 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
23471 * ipa-inline.c (edge_badness): Revert accidental checkin.
23473 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
23476 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
23477 symbols; not inline clones.
23479 2018-01-09 Jakub Jelinek <jakub@redhat.com>
23482 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
23483 hard registers. Formatting fixes.
23485 PR preprocessor/83722
23486 * gcc.c (try_generate_repro): Pass
23487 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
23488 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
23491 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
23492 Kito Cheng <kito.cheng@gmail.com>
23494 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
23495 (riscv_leaf_function_p): Delete.
23496 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
23498 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23500 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
23502 (do_ifelse): New function.
23503 (do_isel): New function.
23504 (do_sub3): New function.
23505 (do_add3): New function.
23506 (do_load_mask_compare): New function.
23507 (do_overlap_load_compare): New function.
23508 (expand_compare_loop): New function.
23509 (expand_block_compare): Call expand_compare_loop() when appropriate.
23510 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
23511 option description.
23512 (-mblock-compare-inline-loop-limit): New option.
23514 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23517 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
23518 Reverse order of second and third operands in first alternative.
23519 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
23520 of first and second elements in UNSPEC_VPERMR vector.
23521 (altivec_expand_vec_perm_le): Likewise.
23523 2018-01-08 Jeff Law <law@redhat.com>
23525 PR rtl-optimizatin/81308
23526 * tree-switch-conversion.c (cfg_altered): New file scoped static.
23527 (process_switch): If group_case_labels makes a change, then set
23529 (pass_convert_switch::execute): If a switch is converted, then
23530 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
23532 PR rtl-optimization/81308
23533 * recog.c (split_all_insns): Conditionally cleanup the CFG after
23536 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
23538 PR target/83663 - Revert r255946
23539 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
23540 generation for cases where splatting a value is not useful.
23541 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
23542 across a vec_duplicate and a paradoxical subreg forming a vector
23543 mode to a vec_concat.
23545 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23547 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
23548 -march=armv8.3-a variants.
23549 * config/arm/t-multilib: Likewise.
23550 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
23552 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23554 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
23556 (cceq_ior_compare_complement): Give it a name so I can use it, and
23557 change boolean_or_operator predicate to boolean_operator so it can
23558 be used to generate a crand.
23559 (eqne): New code iterator.
23560 (bd/bd_neg): New code_attrs.
23561 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
23562 a single define_insn.
23563 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
23564 decrement (bdnzt/bdnzf/bdzt/bdzf).
23565 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
23566 with the new names of the branch decrement patterns, and added the
23567 names of the branch decrement conditional patterns.
23569 2018-01-08 Richard Biener <rguenther@suse.de>
23571 PR tree-optimization/83563
23572 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
23575 2018-01-08 Richard Biener <rguenther@suse.de>
23577 PR middle-end/83713
23578 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
23580 2018-01-08 Richard Biener <rguenther@suse.de>
23582 PR tree-optimization/83685
23583 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
23584 references to abnormals.
23586 2018-01-08 Richard Biener <rguenther@suse.de>
23589 * dwarf2out.c (output_indirect_strings): Handle empty
23590 skeleton_debug_str_hash.
23591 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
23593 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
23595 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
23596 (emit_store_direct): Likewise.
23597 (arc_trampoline_adjust_address): Likewise.
23598 (arc_asm_trampoline_template): New function.
23599 (arc_initialize_trampoline): Use asm_trampoline_template.
23600 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
23601 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
23602 * config/arc/arc.md (flush_icache): Delete pattern.
23604 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
23606 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
23607 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
23610 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
23613 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
23614 by not USED_FOR_TARGET.
23615 (make_pass_resolve_sw_modes): Likewise.
23617 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
23619 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
23622 2018-01-08 Richard Biener <rguenther@suse.de>
23624 PR middle-end/83580
23625 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
23627 2018-01-08 Richard Biener <rguenther@suse.de>
23629 PR middle-end/83517
23630 * match.pd ((t * 2) / 2) -> t): Add missing :c.
23632 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
23634 PR middle-end/81897
23635 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
23636 basic blocks with a small number of successors.
23637 (convert_control_dep_chain_into_preds): Improve handling of
23639 (dump_predicates): Split apart into...
23640 (dump_pred_chain): ...here...
23641 (dump_pred_info): ...and here.
23642 (can_one_predicate_be_invalidated_p): Add debugging printfs.
23643 (can_chain_union_be_invalidated_p): Improve check for invalidation
23645 (uninit_uses_cannot_happen): Avoid unnecessary if
23646 convert_control_dep_chain_into_preds yielded nothing.
23648 2018-01-06 Martin Sebor <msebor@redhat.com>
23650 PR tree-optimization/83640
23651 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
23652 subtracting negative offset from size.
23653 (builtin_access::overlap): Adjust offset bounds of the access to fall
23654 within the size of the object if possible.
23656 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
23658 PR rtl-optimization/83699
23659 * expmed.c (extract_bit_field_1): Restrict the vector usage of
23660 extract_bit_field_as_subreg to cases in which the extracted
23661 value is also a vector.
23663 * lra-constraints.c (process_alt_operands): Test for the equivalence
23664 substitutions when detecting a possible reload cycle.
23666 2018-01-06 Jakub Jelinek <jakub@redhat.com>
23669 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
23670 by default if flag_selective_schedling{,2}. Formatting fixes.
23672 PR rtl-optimization/83682
23673 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
23674 if it has non-VECTOR_MODE element mode.
23675 (vec_duplicate_p): Likewise.
23677 PR middle-end/83694
23678 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
23679 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
23681 2018-01-05 Jakub Jelinek <jakub@redhat.com>
23684 * config/i386/i386-builtin.def
23685 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
23686 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
23687 Require also OPTION_MASK_ISA_AVX512F in addition to
23688 OPTION_MASK_ISA_GFNI.
23689 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
23690 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
23691 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
23692 to OPTION_MASK_ISA_GFNI.
23693 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
23694 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
23695 OPTION_MASK_ISA_AVX512BW.
23696 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
23697 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
23698 addition to OPTION_MASK_ISA_GFNI.
23699 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
23700 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
23701 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
23702 to OPTION_MASK_ISA_GFNI.
23703 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
23704 a requirement for all ISAs rather than any of them with a few
23706 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
23708 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
23709 bitmasks to be enabled with 3 exceptions, instead of requiring any
23710 enabled ISA with lots of exceptions.
23711 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
23712 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
23713 Change avx512bw in isa attribute to avx512f.
23714 * config/i386/sgxintrin.h: Add license boilerplate.
23715 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
23716 to __AVX512F__ and __AVX512VL to __AVX512VL__.
23717 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
23718 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
23720 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
23721 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
23722 temporarily sse2 rather than sse if not enabled already.
23725 * config/i386/sse.md (VI248_VLBW): Rename to ...
23726 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
23727 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
23728 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
23729 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
23730 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
23731 mode iterator instead of VI248_VLBW.
23733 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
23735 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
23736 (record_modified): Skip clobbers; add debug output.
23737 (param_change_prob): Use sreal frequencies.
23739 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
23741 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
23742 punt for user-aligned variables.
23744 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
23746 * tree-chrec.c (chrec_contains_symbols): Return true for
23749 2018-01-05 Sudakshina Das <sudi.das@arm.com>
23752 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
23753 of (x|y) == x for BICS pattern.
23755 2018-01-05 Jakub Jelinek <jakub@redhat.com>
23757 PR tree-optimization/83605
23758 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
23759 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
23762 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
23764 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
23765 * config/epiphany/rtems.h: New file.
23767 2018-01-04 Jakub Jelinek <jakub@redhat.com>
23768 Uros Bizjak <ubizjak@gmail.com>
23771 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
23772 QIreg_operand instead of register_operand predicate.
23773 * config/i386/i386.c (ix86_rop_should_change_byte_p,
23774 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
23775 comments instead of -fmitigate[-_]rop.
23777 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23780 * cgraphunit.c (symbol_table::compile): Switch to text_section
23781 before calling assembly_start debug hook.
23782 * run-rtl-passes.c (run_rtl_passes): Likewise.
23785 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23787 * tree-vrp.c (extract_range_from_binary_expr_1): Check
23788 range_int_cst_p rather than !symbolic_range_p before calling
23789 extract_range_from_multiplicative_op_1.
23791 2018-01-04 Jeff Law <law@redhat.com>
23793 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
23794 redundant test in assertion.
23796 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23798 * doc/rtl.texi: Document machine_mode wrapper classes.
23800 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23802 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
23803 using tree_to_uhwi.
23805 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23807 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
23808 the VEC_PERM_EXPR fold to fail.
23810 2018-01-04 Jakub Jelinek <jakub@redhat.com>
23813 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
23814 to switched_sections.
23816 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23819 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
23820 test for d.testing.
23822 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
23825 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
23826 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
23828 2018-01-04 Jakub Jelinek <jakub@redhat.com>
23831 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
23832 is BLKmode and bitpos not zero or mode change is needed.
23834 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23837 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
23840 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
23843 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
23844 instead of MULT rtx. Update all corresponding splitters.
23845 (*saddl_se): Ditto.
23846 (*ssub<modesuffix>): Ditto.
23847 (*ssubl_se): Ditto.
23848 (*cmp_sadd_di): Update split patterns.
23849 (*cmp_sadd_si): Ditto.
23850 (*cmp_sadd_sidi): Ditto.
23851 (*cmp_ssub_di): Ditto.
23852 (*cmp_ssub_si): Ditto.
23853 (*cmp_ssub_sidi): Ditto.
23854 * config/alpha/predicates.md (const23_operand): New predicate.
23855 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
23856 Look for ASHIFT, not MULT inner operand.
23857 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
23859 2018-01-04 Martin Liska <mliska@suse.cz>
23861 PR gcov-profile/83669
23862 * gcov.c (output_intermediate_file): Add version to intermediate
23864 * doc/gcov.texi: Document new field 'version' in intermediate
23865 file format. Fix location of '-k' option of gcov command.
23867 2018-01-04 Martin Liska <mliska@suse.cz>
23870 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
23872 2018-01-04 Jakub Jelinek <jakub@redhat.com>
23874 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
23876 2018-01-03 Martin Sebor <msebor@redhat.com>
23878 PR tree-optimization/83655
23879 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
23880 checking calls with invalid arguments.
23882 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23884 * tree-vect-stmts.c (vect_get_store_rhs): New function.
23885 (vectorizable_mask_load_store): Delete.
23886 (vectorizable_call): Return false for masked loads and stores.
23887 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
23888 instead of gimple_assign_rhs1.
23889 (vectorizable_load): Handle IFN_MASK_LOAD.
23890 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
23892 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23894 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
23896 (vectorizable_mask_load_store): ...here.
23897 (vectorizable_load): ...and here.
23899 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23901 * tree-vect-stmts.c (vect_build_all_ones_mask)
23902 (vect_build_zero_merge_argument): New functions, split out from...
23903 (vectorizable_load): ...here.
23905 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23907 * tree-vect-stmts.c (vect_check_store_rhs): New function,
23909 (vectorizable_mask_load_store): ...here.
23910 (vectorizable_store): ...and here.
23912 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23914 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
23916 (vectorizable_mask_load_store): ...here.
23918 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23920 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
23921 (vect_model_store_cost): Take a vec_load_store_type instead of a
23923 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
23924 (vect_model_store_cost): Take a vec_load_store_type instead of a
23926 (vectorizable_mask_load_store): Update accordingly.
23927 (vectorizable_store): Likewise.
23928 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
23930 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23932 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
23933 IFN_MASK_LOAD calls here rather than...
23934 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
23936 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23937 Alan Hayward <alan.hayward@arm.com>
23938 David Sherwood <david.sherwood@arm.com>
23940 * expmed.c (extract_bit_field_1): For vector extracts,
23941 fall back to extract_bit_field_as_subreg if vec_extract
23944 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23945 Alan Hayward <alan.hayward@arm.com>
23946 David Sherwood <david.sherwood@arm.com>
23948 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
23949 they are variable or constant sized.
23950 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
23951 slots for constant-sized data.
23953 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23954 Alan Hayward <alan.hayward@arm.com>
23955 David Sherwood <david.sherwood@arm.com>
23957 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
23958 handling COND_EXPRs with boolean comparisons, try to find a better
23959 basis for the mask type than the boolean itself.
23961 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23963 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
23964 is calculated and how it can be overridden.
23965 * genmodes.c (max_bitsize_mode_any_mode): New variable.
23966 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
23968 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
23971 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23972 Alan Hayward <alan.hayward@arm.com>
23973 David Sherwood <david.sherwood@arm.com>
23975 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
23976 Remove the mode argument.
23977 (aarch64_simd_valid_immediate): Remove the mode and inverse
23979 * config/aarch64/iterators.md (bitsize): New iterator.
23980 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
23981 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
23982 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
23983 aarch64_simd_valid_immediate.
23984 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
23985 (aarch64_reg_or_bic_imm): Likewise.
23986 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
23987 with an insn_type enum and msl with a modifier_type enum.
23988 Replace element_width with a scalar_mode. Change the shift
23989 to unsigned int. Add constructors for scalar_float_mode and
23990 scalar_int_mode elements.
23991 (aarch64_vect_float_const_representable_p): Delete.
23992 (aarch64_can_const_movi_rtx_p)
23993 (aarch64_simd_scalar_immediate_valid_for_move)
23994 (aarch64_simd_make_constant): Update call to
23995 aarch64_simd_valid_immediate.
23996 (aarch64_advsimd_valid_immediate_hs): New function.
23997 (aarch64_advsimd_valid_immediate): Likewise.
23998 (aarch64_simd_valid_immediate): Remove mode and inverse
23999 arguments. Rewrite to use the above. Use const_vec_duplicate_p
24000 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
24001 and aarch64_float_const_representable_p on the result.
24002 (aarch64_output_simd_mov_immediate): Remove mode argument.
24003 Update call to aarch64_simd_valid_immediate and use of
24004 simd_immediate_info.
24005 (aarch64_output_scalar_simd_mov_immediate): Update call
24008 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24009 Alan Hayward <alan.hayward@arm.com>
24010 David Sherwood <david.sherwood@arm.com>
24012 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
24013 (mode_nunits): Likewise CONST_MODE_NUNITS.
24014 * machmode.def (ADJUST_NUNITS): Document.
24015 * genmodes.c (mode_data::need_nunits_adj): New field.
24016 (blank_mode): Update accordingly.
24017 (adj_nunits): New variable.
24018 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
24020 (emit_mode_size_inline): Set need_bytesize_adj for all modes
24021 listed in adj_nunits.
24022 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
24023 listed in adj_nunits. Don't emit case statements for such modes.
24024 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
24025 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
24026 nothing if adj_nunits is nonnull.
24027 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
24028 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
24029 (emit_mode_fbit): Update use of print_maybe_const_decl.
24030 (emit_move_size): Likewise. Treat the array as non-const
24032 (emit_mode_adjustments): Handle adj_nunits.
24034 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24036 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
24037 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
24038 (VECTOR_MODES): Use it.
24039 (make_vector_modes): Take the prefix as an argument.
24041 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24042 Alan Hayward <alan.hayward@arm.com>
24043 David Sherwood <david.sherwood@arm.com>
24045 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
24046 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
24047 for MODE_VECTOR_BOOL.
24048 * machmode.def (VECTOR_BOOL_MODE): Document.
24049 * genmodes.c (VECTOR_BOOL_MODE): New macro.
24050 (make_vector_bool_mode): New function.
24051 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
24053 * lto-streamer-in.c (lto_input_mode_table): Likewise.
24054 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
24056 * stor-layout.c (int_mode_for_mode): Likewise.
24057 * tree.c (build_vector_type_for_mode): Likewise.
24058 * varasm.c (output_constant_pool_2): Likewise.
24059 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
24060 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
24061 for MODE_VECTOR_BOOL.
24062 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
24063 of mode class checks.
24064 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
24065 instead of a list of mode class checks.
24066 (expand_vector_scalar_condition): Likewise.
24067 (type_for_widest_vector_mode): Handle BImode as an inner mode.
24069 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24070 Alan Hayward <alan.hayward@arm.com>
24071 David Sherwood <david.sherwood@arm.com>
24073 * machmode.h (mode_size): Change from unsigned short to
24075 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
24076 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
24077 or if measurement_type is not polynomial.
24078 (fixed_size_mode::includes_p): Check for constant-sized modes.
24079 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
24080 return a poly_uint16 rather than an unsigned short.
24081 (emit_mode_size): Change the type of mode_size from unsigned short
24082 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
24083 (emit_mode_adjustments): Cope with polynomial vector sizes.
24084 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24086 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24088 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
24089 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
24090 * caller-save.c (setup_save_areas): Likewise.
24091 (replace_reg_with_saved_mem): Likewise.
24092 * calls.c (emit_library_call_value_1): Likewise.
24093 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
24094 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
24095 (gen_lowpart_for_combine): Likewise.
24096 * convert.c (convert_to_integer_1): Likewise.
24097 * cse.c (equiv_constant, cse_insn): Likewise.
24098 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
24099 (cselib_subst_to_values): Likewise.
24100 * dce.c (word_dce_process_block): Likewise.
24101 * df-problems.c (df_word_lr_mark_ref): Likewise.
24102 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
24103 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
24104 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
24105 (rtl_for_decl_location): Likewise.
24106 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
24107 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
24108 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
24109 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
24110 (expand_expr_real_1): Likewise.
24111 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
24112 (pad_below): Likewise.
24113 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
24114 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
24115 * ira.c (get_subreg_tracking_sizes): Likewise.
24116 * ira-build.c (ira_create_allocno_objects): Likewise.
24117 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
24118 (ira_sort_regnos_for_alter_reg): Likewise.
24119 * ira-costs.c (record_operand_costs): Likewise.
24120 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
24121 (resolve_simple_move): Likewise.
24122 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
24123 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
24124 (lra_constraints): Likewise.
24125 (CONST_POOL_OK_P): Reject variable-sized modes.
24126 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
24127 (add_pseudo_to_slot, lra_spill): Likewise.
24128 * omp-low.c (omp_clause_aligned_alignment): Likewise.
24129 * optabs-query.c (get_best_extraction_insn): Likewise.
24130 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
24131 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
24132 (expand_mult_highpart, valid_multiword_target_p): Likewise.
24133 * recog.c (offsettable_address_addr_space_p): Likewise.
24134 * regcprop.c (maybe_mode_change): Likewise.
24135 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
24136 * regrename.c (build_def_use): Likewise.
24137 * regstat.c (dump_reg_info): Likewise.
24138 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
24139 (find_reloads, find_reloads_subreg_address): Likewise.
24140 * reload1.c (eliminate_regs_1): Likewise.
24141 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
24142 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
24143 (simplify_binary_operation_1, simplify_subreg): Likewise.
24144 * targhooks.c (default_function_arg_padding): Likewise.
24145 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
24146 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
24147 (verify_gimple_assign_ternary): Likewise.
24148 * tree-inline.c (estimate_move_cost): Likewise.
24149 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
24150 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
24151 (get_address_cost_ainc): Likewise.
24152 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
24153 (vect_supportable_dr_alignment): Likewise.
24154 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24155 (vectorizable_reduction): Likewise.
24156 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
24157 (vectorizable_operation, vectorizable_load): Likewise.
24158 * tree.c (build_same_sized_truth_vector_type): Likewise.
24159 * valtrack.c (cleanup_auto_inc_dec): Likewise.
24160 * var-tracking.c (emit_note_insn_var_location): Likewise.
24161 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
24162 (ADDR_VEC_ALIGN): Likewise.
24164 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24165 Alan Hayward <alan.hayward@arm.com>
24166 David Sherwood <david.sherwood@arm.com>
24168 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
24170 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
24171 or if measurement_type is polynomial.
24172 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
24173 * combine.c (make_extraction): Likewise.
24174 * dse.c (find_shift_sequence): Likewise.
24175 * dwarf2out.c (mem_loc_descriptor): Likewise.
24176 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
24177 (extract_bit_field, extract_low_bits): Likewise.
24178 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
24179 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
24180 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
24181 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
24182 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
24183 * reload.c (find_reloads): Likewise.
24184 * reload1.c (alter_reg): Likewise.
24185 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
24186 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
24187 * tree-if-conv.c (predicate_mem_writes): Likewise.
24188 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
24189 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
24190 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
24191 * valtrack.c (dead_debug_insert_temp): Likewise.
24192 * varasm.c (mergeable_constant_section): Likewise.
24193 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
24195 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24196 Alan Hayward <alan.hayward@arm.com>
24197 David Sherwood <david.sherwood@arm.com>
24199 * expr.c (expand_assignment): Cope with polynomial mode sizes
24200 when assigning to a CONCAT.
24202 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24203 Alan Hayward <alan.hayward@arm.com>
24204 David Sherwood <david.sherwood@arm.com>
24206 * machmode.h (mode_precision): Change from unsigned short to
24208 (mode_to_precision): Return a poly_uint16 rather than an unsigned
24210 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
24211 or if measurement_type is not polynomial.
24212 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
24213 in which the mode is already known to be a scalar_int_mode.
24214 * genmodes.c (emit_mode_precision): Change the type of mode_precision
24215 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
24217 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24218 for GET_MODE_PRECISION.
24219 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24220 for GET_MODE_PRECISION.
24221 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
24223 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
24224 (expand_field_assignment, make_extraction): Likewise.
24225 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
24226 (get_last_value): Likewise.
24227 * convert.c (convert_to_integer_1): Likewise.
24228 * cse.c (cse_insn): Likewise.
24229 * expr.c (expand_expr_real_1): Likewise.
24230 * lra-constraints.c (simplify_operand_subreg): Likewise.
24231 * optabs-query.c (can_atomic_load_p): Likewise.
24232 * optabs.c (expand_atomic_load): Likewise.
24233 (expand_atomic_store): Likewise.
24234 * ree.c (combine_reaching_defs): Likewise.
24235 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
24236 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
24237 * tree.h (type_has_mode_precision_p): Likewise.
24238 * ubsan.c (instrument_si_overflow): Likewise.
24240 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24241 Alan Hayward <alan.hayward@arm.com>
24242 David Sherwood <david.sherwood@arm.com>
24244 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
24245 polynomial numbers of units.
24246 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
24247 (valid_vector_subparts_p): New function.
24248 (build_vector_type): Remove temporary shim and take the number
24249 of units as a poly_uint64 rather than an int.
24250 (build_opaque_vector_type): Take the number of units as a
24251 poly_uint64 rather than an int.
24252 * tree.c (build_vector_from_ctor): Handle polynomial
24253 TYPE_VECTOR_SUBPARTS.
24254 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
24255 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
24256 (build_vector_from_val): If the number of units is variable,
24257 use build_vec_duplicate_cst for constant operands and
24258 VEC_DUPLICATE_EXPR otherwise.
24259 (make_vector_type): Remove temporary is_constant ().
24260 (build_vector_type, build_opaque_vector_type): Take the number of
24261 units as a poly_uint64 rather than an int.
24262 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
24264 * cfgexpand.c (expand_debug_expr): Likewise.
24265 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
24266 (store_constructor, expand_expr_real_1): Likewise.
24267 (const_scalar_mask_from_tree): Likewise.
24268 * fold-const-call.c (fold_const_reduction): Likewise.
24269 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
24270 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
24271 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
24272 (fold_relational_const): Likewise.
24273 (native_interpret_vector): Likewise. Change the size from an
24274 int to an unsigned int.
24275 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
24276 TYPE_VECTOR_SUBPARTS.
24277 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
24278 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
24279 duplicating a non-constant operand into a variable-length vector.
24280 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
24281 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
24282 * ipa-icf.c (sem_variable::equals): Likewise.
24283 * match.pd: Likewise.
24284 * omp-simd-clone.c (simd_clone_subparts): Likewise.
24285 * print-tree.c (print_node): Likewise.
24286 * stor-layout.c (layout_type): Likewise.
24287 * targhooks.c (default_builtin_vectorization_cost): Likewise.
24288 * tree-cfg.c (verify_gimple_comparison): Likewise.
24289 (verify_gimple_assign_binary): Likewise.
24290 (verify_gimple_assign_ternary): Likewise.
24291 (verify_gimple_assign_single): Likewise.
24292 * tree-pretty-print.c (dump_generic_node): Likewise.
24293 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
24294 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
24295 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
24296 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
24297 (vect_shift_permute_load_chain): Likewise.
24298 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
24299 (expand_vector_condition, optimize_vector_constructor): Likewise.
24300 (lower_vec_perm, get_compute_type): Likewise.
24301 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24302 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
24303 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
24304 (vect_recog_mask_conversion_pattern): Likewise.
24305 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
24306 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
24307 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
24308 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
24309 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
24310 (vectorizable_shift, vectorizable_operation, vectorizable_store)
24311 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
24312 (supportable_widening_operation): Likewise.
24313 (supportable_narrowing_operation): Likewise.
24314 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
24316 * varasm.c (output_constant): Likewise.
24318 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24319 Alan Hayward <alan.hayward@arm.com>
24320 David Sherwood <david.sherwood@arm.com>
24322 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
24323 so that both the length == 3 and length != 3 cases set up their
24324 own permute vectors. Add comments explaining why we know the
24325 number of elements is constant.
24326 (vect_permute_load_chain): Likewise.
24328 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24329 Alan Hayward <alan.hayward@arm.com>
24330 David Sherwood <david.sherwood@arm.com>
24332 * machmode.h (mode_nunits): Change from unsigned char to
24334 (ONLY_FIXED_SIZE_MODES): New macro.
24335 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
24336 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
24337 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
24339 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
24340 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
24341 or if measurement_type is not polynomial.
24342 * genmodes.c (ZERO_COEFFS): New macro.
24343 (emit_mode_nunits_inline): Make mode_nunits_inline return a
24345 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
24346 Use ZERO_COEFFS when emitting initializers.
24347 * data-streamer.h (bp_pack_poly_value): New function.
24348 (bp_unpack_poly_value): Likewise.
24349 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24350 for GET_MODE_NUNITS.
24351 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24352 for GET_MODE_NUNITS.
24353 * tree.c (make_vector_type): Remove temporary shim and make
24354 the real function take the number of units as a poly_uint64
24355 rather than an int.
24356 (build_vector_type_for_mode): Handle polynomial nunits.
24357 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
24358 * emit-rtl.c (const_vec_series_p_1): Likewise.
24359 (gen_rtx_CONST_VECTOR): Likewise.
24360 * fold-const.c (test_vec_duplicate_folding): Likewise.
24361 * genrecog.c (validate_pattern): Likewise.
24362 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
24363 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
24364 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
24365 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
24366 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
24367 * rtlanal.c (subreg_get_info): Likewise.
24368 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
24369 (vect_grouped_load_supported): Likewise.
24370 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
24371 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
24372 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
24373 (simplify_const_unary_operation, simplify_binary_operation_1)
24374 (simplify_const_binary_operation, simplify_ternary_operation)
24375 (test_vector_ops_duplicate, test_vector_ops): Likewise.
24376 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
24377 instead of CONST_VECTOR_NUNITS.
24378 * varasm.c (output_constant_pool_2): Likewise.
24379 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
24380 explicit-encoded elements in the XVEC for variable-length vectors.
24382 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24384 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
24386 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24387 Alan Hayward <alan.hayward@arm.com>
24388 David Sherwood <david.sherwood@arm.com>
24390 * coretypes.h (fixed_size_mode): Declare.
24391 (fixed_size_mode_pod): New typedef.
24392 * builtins.h (target_builtins::x_apply_args_mode)
24393 (target_builtins::x_apply_result_mode): Change type to
24394 fixed_size_mode_pod.
24395 * builtins.c (apply_args_size, apply_result_size, result_vector)
24396 (expand_builtin_apply_args_1, expand_builtin_apply)
24397 (expand_builtin_return): Update accordingly.
24399 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24401 * cse.c (hash_rtx_cb): Hash only the encoded elements.
24402 * cselib.c (cselib_hash_rtx): Likewise.
24403 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
24404 CONST_VECTOR encoding.
24406 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24407 Jeff Law <law@redhat.com>
24410 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
24411 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
24412 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
24413 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
24416 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
24417 explicitly probe *sp in a noreturn function if there were any callee
24418 register saves or frame pointer is needed.
24420 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24423 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
24424 BLKmode for ternary, binary or unary expressions.
24427 * var-tracking.c (delete_vta_debug_insn): New inline function.
24428 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
24429 insns from get_insns () to NULL instead of each bb separately.
24430 Use delete_vta_debug_insn. No longer static.
24431 (vt_debug_insns_local, variable_tracking_main_1): Adjust
24432 delete_vta_debug_insns callers.
24433 * rtl.h (delete_vta_debug_insns): Declare.
24434 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
24435 instead of variable_tracking_main.
24437 2018-01-03 Martin Sebor <msebor@redhat.com>
24439 PR tree-optimization/83603
24440 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
24441 arguments past the endof the argument list in functions declared
24442 without a prototype.
24443 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
24444 Avoid checking when arguments are null.
24446 2018-01-03 Martin Sebor <msebor@redhat.com>
24449 * doc/extend.texi (attribute const): Fix a typo.
24450 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
24451 issuing -Wsuggest-attribute for void functions.
24453 2018-01-03 Martin Sebor <msebor@redhat.com>
24455 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
24456 offset_int::from instead of wide_int::to_shwi.
24457 (maybe_diag_overlap): Remove assertion.
24458 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
24459 * gimple-ssa-sprintf.c (format_directive): Same.
24460 (parse_directive): Same.
24461 (sprintf_dom_walker::compute_format_length): Same.
24462 (try_substitute_return_value): Same.
24464 2018-01-03 Jeff Law <law@redhat.com>
24466 PR middle-end/83654
24467 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
24468 non-constant residual for zero at runtime and avoid probing in
24469 that case. Reorganize code for trailing problem to mirror handling
24472 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24474 PR tree-optimization/83501
24475 * tree-ssa-strlen.c (get_string_cst): New.
24476 (handle_char_store): Call get_string_cst.
24478 2018-01-03 Martin Liska <mliska@suse.cz>
24480 PR tree-optimization/83593
24481 * tree-ssa-strlen.c: Include tree-cfg.h.
24482 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
24483 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
24484 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
24486 (strlen_dom_walker::before_dom_children): Call
24487 gimple_purge_dead_eh_edges. Dump tranformation with details
24489 (strlen_dom_walker::before_dom_children): Update call by adding
24490 new argument cleanup_eh.
24491 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
24493 2018-01-03 Martin Liska <mliska@suse.cz>
24496 * cif-code.def (VARIADIC_THUNK): New enum value.
24497 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
24500 2018-01-03 Jan Beulich <jbeulich@suse.com>
24502 * sse.md (mov<mode>_internal): Tighten condition for when to use
24503 vmovdqu<ssescalarsize> for TI and OI modes.
24505 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24507 Update copyright years.
24509 2018-01-03 Martin Liska <mliska@suse.cz>
24512 * ipa-visibility.c (function_and_variable_visibility): Skip
24513 functions with noipa attribure.
24515 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24517 * gcc.c (process_command): Update copyright notice dates.
24518 * gcov-dump.c (print_version): Ditto.
24519 * gcov.c (print_version): Ditto.
24520 * gcov-tool.c (print_version): Ditto.
24521 * gengtype.c (create_file): Ditto.
24522 * doc/cpp.texi: Bump @copying's copyright year.
24523 * doc/cppinternals.texi: Ditto.
24524 * doc/gcc.texi: Ditto.
24525 * doc/gccint.texi: Ditto.
24526 * doc/gcov.texi: Ditto.
24527 * doc/install.texi: Ditto.
24528 * doc/invoke.texi: Ditto.
24530 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24532 * vector-builder.h (vector_builder::m_full_nelts): Change from
24533 unsigned int to poly_uint64.
24534 (vector_builder::full_nelts): Update prototype accordingly.
24535 (vector_builder::new_vector): Likewise.
24536 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
24537 (vector_builder::operator ==): Likewise.
24538 (vector_builder::finalize): Likewise.
24539 * int-vector-builder.h (int_vector_builder::int_vector_builder):
24540 Take the number of elements as a poly_uint64 rather than an
24542 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
24543 from unsigned int to poly_uint64.
24544 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
24545 (vec_perm_indices::new_vector): Likewise.
24546 (vec_perm_indices::length): Likewise.
24547 (vec_perm_indices::nelts_per_input): Likewise.
24548 (vec_perm_indices::input_nelts): Likewise.
24549 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
24550 number of elements per input as a poly_uint64 rather than an
24551 unsigned int. Use the original encoding for variable-length
24552 vectors, rather than clamping each individual element.
24553 For the second and subsequent elements in each pattern,
24554 clamp the step and base before clamping their sum.
24555 (vec_perm_indices::series_p): Handle polynomial element counts.
24556 (vec_perm_indices::all_in_range_p): Likewise.
24557 (vec_perm_indices_to_tree): Likewise.
24558 (vec_perm_indices_to_rtx): Likewise.
24559 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
24560 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
24561 (tree_vector_builder::new_binary_operation): Handle polynomial
24562 element counts. Return false if we need to know the number
24563 of elements at compile time.
24564 * fold-const.c (fold_vec_perm): Punt if the number of elements
24565 isn't known at compile time.
24567 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24569 * vec-perm-indices.h (vec_perm_builder): Change element type
24570 from HOST_WIDE_INT to poly_int64.
24571 (vec_perm_indices::element_type): Update accordingly.
24572 (vec_perm_indices::clamp): Handle polynomial element_types.
24573 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
24574 (vec_perm_indices::all_in_range_p): Likewise.
24575 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
24577 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
24578 polynomial vec_perm_indices element types.
24579 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
24580 * fold-const.c (fold_vec_perm): Likewise.
24581 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
24582 * tree-vect-generic.c (lower_vec_perm): Likewise.
24583 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
24584 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
24585 element type to HOST_WIDE_INT.
24587 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24588 Alan Hayward <alan.hayward@arm.com>
24589 David Sherwood <david.sherwood@arm.com>
24591 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
24592 rather than an int. Use plus_constant.
24593 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
24594 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
24596 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24597 Alan Hayward <alan.hayward@arm.com>
24598 David Sherwood <david.sherwood@arm.com>
24600 * calls.c (emit_call_1, expand_call): Change struct_value_size from
24601 a HOST_WIDE_INT to a poly_int64.
24603 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24604 Alan Hayward <alan.hayward@arm.com>
24605 David Sherwood <david.sherwood@arm.com>
24607 * calls.c (load_register_parameters): Cope with polynomial
24608 mode sizes. Require a constant size for BLKmode parameters
24609 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
24610 forces a parameter to be padded at the lsb end in order to
24611 fill a complete number of words, require the parameter size
24612 to be ordered wrt UNITS_PER_WORD.
24614 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24615 Alan Hayward <alan.hayward@arm.com>
24616 David Sherwood <david.sherwood@arm.com>
24618 * reload1.c (spill_stack_slot_width): Change element type
24619 from unsigned int to poly_uint64_pod.
24620 (alter_reg): Treat mode sizes as polynomial.
24622 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24623 Alan Hayward <alan.hayward@arm.com>
24624 David Sherwood <david.sherwood@arm.com>
24626 * reload.c (complex_word_subreg_p): New function.
24627 (reload_inner_reg_of_subreg, push_reload): Use it.
24629 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24630 Alan Hayward <alan.hayward@arm.com>
24631 David Sherwood <david.sherwood@arm.com>
24633 * lra-constraints.c (process_alt_operands): Reject matched
24634 operands whose sizes aren't ordered.
24635 (match_reload): Refer to this check here.
24637 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24638 Alan Hayward <alan.hayward@arm.com>
24639 David Sherwood <david.sherwood@arm.com>
24641 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
24642 that the mode size is in the set {1, 2, 4, 8, 16}.
24644 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24645 Alan Hayward <alan.hayward@arm.com>
24646 David Sherwood <david.sherwood@arm.com>
24648 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
24649 Use plus_constant instead of gen_rtx_PLUS.
24651 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24652 Alan Hayward <alan.hayward@arm.com>
24653 David Sherwood <david.sherwood@arm.com>
24655 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
24656 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
24657 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
24658 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
24659 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
24660 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
24661 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
24662 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
24663 * config/i386/i386.c (ix86_push_rounding): ...this new function.
24664 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
24666 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
24667 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
24668 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
24669 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
24670 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
24671 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
24672 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
24673 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
24674 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
24675 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
24677 * expr.c (emit_move_resolve_push): Treat the input and result
24678 of PUSH_ROUNDING as a poly_int64.
24679 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
24680 (emit_push_insn): Likewise.
24681 * lra-eliminations.c (mark_not_eliminable): Likewise.
24682 * recog.c (push_operand): Likewise.
24683 * reload1.c (elimination_effects): Likewise.
24684 * rtlanal.c (nonzero_bits1): Likewise.
24685 * calls.c (store_one_arg): Likewise. Require the padding to be
24686 known at compile time.
24688 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24689 Alan Hayward <alan.hayward@arm.com>
24690 David Sherwood <david.sherwood@arm.com>
24692 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
24693 Use plus_constant instead of gen_rtx_PLUS.
24695 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24696 Alan Hayward <alan.hayward@arm.com>
24697 David Sherwood <david.sherwood@arm.com>
24699 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
24700 rather than an int.
24702 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24703 Alan Hayward <alan.hayward@arm.com>
24704 David Sherwood <david.sherwood@arm.com>
24706 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
24707 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
24708 via stack temporaries. Treat the mode size as polynomial too.
24710 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24711 Alan Hayward <alan.hayward@arm.com>
24712 David Sherwood <david.sherwood@arm.com>
24714 * expr.c (expand_expr_real_2): When handling conversions involving
24715 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
24716 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
24717 as a poly_uint64 too.
24719 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24720 Alan Hayward <alan.hayward@arm.com>
24721 David Sherwood <david.sherwood@arm.com>
24723 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
24725 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24726 Alan Hayward <alan.hayward@arm.com>
24727 David Sherwood <david.sherwood@arm.com>
24729 * combine.c (can_change_dest_mode): Handle polynomial
24730 REGMODE_NATURAL_SIZE.
24731 * expmed.c (store_bit_field_1): Likewise.
24732 * expr.c (store_constructor): Likewise.
24733 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
24734 and polynomial REGMODE_NATURAL_SIZE.
24735 (gen_lowpart_common): Likewise.
24736 * reginfo.c (record_subregs_of_mode): Likewise.
24737 * rtlanal.c (read_modify_subreg_p): Likewise.
24739 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24740 Alan Hayward <alan.hayward@arm.com>
24741 David Sherwood <david.sherwood@arm.com>
24743 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
24744 numbers of elements.
24746 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24747 Alan Hayward <alan.hayward@arm.com>
24748 David Sherwood <david.sherwood@arm.com>
24750 * match.pd: Cope with polynomial numbers of vector elements.
24752 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24753 Alan Hayward <alan.hayward@arm.com>
24754 David Sherwood <david.sherwood@arm.com>
24756 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
24757 in a POINTER_PLUS_EXPR.
24759 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24760 Alan Hayward <alan.hayward@arm.com>
24761 David Sherwood <david.sherwood@arm.com>
24763 * omp-simd-clone.c (simd_clone_subparts): New function.
24764 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
24765 (ipa_simd_modify_function_body): Likewise.
24767 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24768 Alan Hayward <alan.hayward@arm.com>
24769 David Sherwood <david.sherwood@arm.com>
24771 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
24772 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
24773 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
24774 (expand_vector_condition, vector_element): Likewise.
24775 (subparts_gt): New function.
24776 (get_compute_type): Use subparts_gt.
24777 (count_type_subparts): Delete.
24778 (expand_vector_operations_1): Use subparts_gt instead of
24779 count_type_subparts.
24781 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24782 Alan Hayward <alan.hayward@arm.com>
24783 David Sherwood <david.sherwood@arm.com>
24785 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
24786 (vect_compile_time_alias): ...this new function. Do the calculation
24787 on poly_ints rather than trees.
24788 (vect_prune_runtime_alias_test_list): Update call accordingly.
24790 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24791 Alan Hayward <alan.hayward@arm.com>
24792 David Sherwood <david.sherwood@arm.com>
24794 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
24796 (vect_schedule_slp_instance): Likewise.
24798 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24799 Alan Hayward <alan.hayward@arm.com>
24800 David Sherwood <david.sherwood@arm.com>
24802 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
24803 constant and extern definitions for variable-length vectors.
24804 (vect_get_constant_vectors): Note that the number of units
24805 is known to be constant.
24807 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24808 Alan Hayward <alan.hayward@arm.com>
24809 David Sherwood <david.sherwood@arm.com>
24811 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
24812 of units as polynomial. Choose between WIDE and NARROW based
24815 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24816 Alan Hayward <alan.hayward@arm.com>
24817 David Sherwood <david.sherwood@arm.com>
24819 * tree-vect-stmts.c (simd_clone_subparts): New function.
24820 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
24822 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24823 Alan Hayward <alan.hayward@arm.com>
24824 David Sherwood <david.sherwood@arm.com>
24826 * tree-vect-stmts.c (vectorizable_call): Treat the number of
24827 vectors as polynomial. Use build_index_vector for
24828 IFN_GOMP_SIMD_LANE.
24830 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24831 Alan Hayward <alan.hayward@arm.com>
24832 David Sherwood <david.sherwood@arm.com>
24834 * tree-vect-stmts.c (get_load_store_type): Treat the number of
24835 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
24836 for variable-length vectors.
24837 (vectorizable_mask_load_store): Treat the number of units as
24838 polynomial, asserting that it is constant if the condition has
24839 already been enforced.
24840 (vectorizable_store, vectorizable_load): Likewise.
24842 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24843 Alan Hayward <alan.hayward@arm.com>
24844 David Sherwood <david.sherwood@arm.com>
24846 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
24847 of units as polynomial. Punt if we can't tell at compile time
24848 which vector contains the final result.
24850 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24851 Alan Hayward <alan.hayward@arm.com>
24852 David Sherwood <david.sherwood@arm.com>
24854 * tree-vect-loop.c (vectorizable_induction): Treat the number
24855 of units as polynomial. Punt on SLP inductions. Use an integer
24856 VEC_SERIES_EXPR for variable-length integer reductions. Use a
24857 cast of such a series for variable-length floating-point
24860 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24861 Alan Hayward <alan.hayward@arm.com>
24862 David Sherwood <david.sherwood@arm.com>
24864 * tree.h (build_index_vector): Declare.
24865 * tree.c (build_index_vector): New function.
24866 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
24867 of units as polynomial, forcibly converting it to a constant if
24868 vectorizable_reduction has already enforced the condition.
24869 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
24870 to create a {1,2,3,...} vector.
24871 (vectorizable_reduction): Treat the number of units as polynomial.
24872 Choose vectype_in based on the largest scalar element size rather
24873 than the smallest number of units. Enforce the restrictions
24876 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24877 Alan Hayward <alan.hayward@arm.com>
24878 David Sherwood <david.sherwood@arm.com>
24880 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
24881 number of units as polynomial.
24883 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24884 Alan Hayward <alan.hayward@arm.com>
24885 David Sherwood <david.sherwood@arm.com>
24887 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
24888 * target.def (autovectorize_vector_sizes): Return the vector sizes
24889 by pointer, using vector_sizes rather than a bitmask.
24890 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
24891 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
24892 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
24894 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
24895 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
24896 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
24897 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
24898 * omp-general.c (omp_max_vf): Likewise.
24899 * omp-low.c (omp_clause_aligned_alignment): Likewise.
24900 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
24901 * tree-vect-loop.c (vect_analyze_loop): Likewise.
24902 * tree-vect-slp.c (vect_slp_bb): Likewise.
24903 * doc/tm.texi: Regenerate.
24904 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
24906 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
24907 the vector size as a poly_uint64 rather than an unsigned int.
24908 (current_vector_size): Change from an unsigned int to a poly_uint64.
24909 (get_vectype_for_scalar_type): Update accordingly.
24910 * tree.h (build_truth_vector_type): Take the size and number of
24911 units as a poly_uint64 rather than an unsigned int.
24912 (build_vector_type): Add a temporary overload that takes
24913 the number of units as a poly_uint64 rather than an unsigned int.
24914 * tree.c (make_vector_type): Likewise.
24915 (build_truth_vector_type): Take the number of units as a poly_uint64
24916 rather than an unsigned int.
24918 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24919 Alan Hayward <alan.hayward@arm.com>
24920 David Sherwood <david.sherwood@arm.com>
24922 * target.def (get_mask_mode): Take the number of units and length
24923 as poly_uint64s rather than unsigned ints.
24924 * targhooks.h (default_get_mask_mode): Update accordingly.
24925 * targhooks.c (default_get_mask_mode): Likewise.
24926 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
24927 * doc/tm.texi: Regenerate.
24929 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24930 Alan Hayward <alan.hayward@arm.com>
24931 David Sherwood <david.sherwood@arm.com>
24933 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
24934 * omp-general.c (omp_max_vf): Likewise.
24935 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
24936 (expand_omp_simd): Handle polynomial safelen.
24937 * omp-low.c (omplow_simd_context): Add a default constructor.
24938 (omplow_simd_context::max_vf): Change from int to poly_uint64.
24939 (lower_rec_simd_input_clauses): Update accordingly.
24940 (lower_rec_input_clauses): Likewise.
24942 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24943 Alan Hayward <alan.hayward@arm.com>
24944 David Sherwood <david.sherwood@arm.com>
24946 * tree-vectorizer.h (vect_nunits_for_cost): New function.
24947 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
24948 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
24949 (vect_analyze_slp_cost): Likewise.
24950 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
24951 (vect_model_load_cost): Likewise.
24953 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24954 Alan Hayward <alan.hayward@arm.com>
24955 David Sherwood <david.sherwood@arm.com>
24957 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
24958 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
24959 from an unsigned int * to a poly_uint64_pod *.
24960 (calculate_unrolling_factor): New function.
24961 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
24963 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24964 Alan Hayward <alan.hayward@arm.com>
24965 David Sherwood <david.sherwood@arm.com>
24967 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
24968 from an unsigned int to a poly_uint64.
24969 (_loop_vec_info::slp_unrolling_factor): Likewise.
24970 (_loop_vec_info::vectorization_factor): Change from an int
24972 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
24973 (vect_get_num_vectors): New function.
24974 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
24975 (vect_get_num_copies): Use vect_get_num_vectors.
24976 (vect_analyze_data_ref_dependences): Change max_vf from an int *
24977 to an unsigned int *.
24978 (vect_analyze_data_refs): Change min_vf from an int * to a
24980 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
24981 than an unsigned HOST_WIDE_INT.
24982 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
24983 (vect_analyze_data_ref_dependence): Change max_vf from an int *
24984 to an unsigned int *.
24985 (vect_analyze_data_ref_dependences): Likewise.
24986 (vect_compute_data_ref_alignment): Handle polynomial vf.
24987 (vect_enhance_data_refs_alignment): Likewise.
24988 (vect_prune_runtime_alias_test_list): Likewise.
24989 (vect_shift_permute_load_chain): Likewise.
24990 (vect_supportable_dr_alignment): Likewise.
24991 (dependence_distance_ge_vf): Take the vectorization factor as a
24992 poly_uint64 rather than an unsigned HOST_WIDE_INT.
24993 (vect_analyze_data_refs): Change min_vf from an int * to a
24995 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
24996 vfm1 as a poly_uint64 rather than an int. Make the same change
24997 for the returned bound_scalar.
24998 (vect_gen_vector_loop_niters): Handle polynomial vf.
24999 (vect_do_peeling): Likewise. Update call to
25000 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
25001 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
25003 * tree-vect-loop.c (vect_determine_vectorization_factor)
25004 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
25005 (vect_get_known_peeling_cost): Likewise.
25006 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
25007 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
25008 (vect_transform_loop): Likewise. Use the lowest possible VF when
25009 updating the upper bounds of the loop.
25010 (vect_min_worthwhile_factor): Make static. Return an unsigned int
25011 rather than an int.
25012 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
25013 polynomial unroll factors.
25014 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
25015 (vect_make_slp_decision): Likewise.
25016 (vect_supported_load_permutation_p): Likewise, and polynomial
25018 (vect_analyze_slp_cost): Handle polynomial vf.
25019 (vect_slp_analyze_node_operations): Likewise.
25020 (vect_slp_analyze_bb_1): Likewise.
25021 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
25022 than an unsigned HOST_WIDE_INT.
25023 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
25024 (vectorizable_load): Handle polynomial vf.
25025 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
25027 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
25029 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25030 Alan Hayward <alan.hayward@arm.com>
25031 David Sherwood <david.sherwood@arm.com>
25033 * match.pd: Handle bit operations involving three constants
25034 and try to fold one pair.
25036 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25038 * tree-vect-loop-manip.c: Include gimple-fold.h.
25039 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
25040 niters_maybe_zero parameters. Handle other cases besides a step of 1.
25041 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
25042 Add a path that uses a step of VF instead of 1, but disable it
25044 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
25045 and niters_no_overflow parameters. Update calls to
25046 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
25047 Create a new SSA name if the latter choses to use a ste other
25048 than zero, and return it via niters_vector_mult_vf_var.
25049 * tree-vect-loop.c (vect_transform_loop): Update calls to
25050 vect_do_peeling, vect_gen_vector_loop_niters and
25051 slpeel_make_loop_iterate_ntimes.
25052 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
25053 (vect_gen_vector_loop_niters): Update declarations after above changes.
25055 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
25057 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
25058 128-bit round to integer instructions.
25059 (ceil<mode>2): Likewise.
25060 (btrunc<mode>2): Likewise.
25061 (round<mode>2): Likewise.
25063 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25065 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
25066 unaligned VSX load/store on P8/P9.
25067 (expand_block_clear): Allow the use of unaligned VSX
25068 load/store on P8/P9.
25070 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25072 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
25074 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
25075 swap associated with both a load and a store.
25077 2018-01-02 Andrew Waterman <andrew@sifive.com>
25079 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
25080 * config/riscv/riscv.md (clear_cache): Use it.
25082 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
25084 * web.c: Remove out-of-date comment.
25086 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25088 * expr.c (fixup_args_size_notes): Check that any existing
25089 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
25090 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
25091 (emit_single_push_insn): ...here.
25093 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25095 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
25096 (const_vector_encoded_nelts): New function.
25097 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
25098 (const_vector_int_elt, const_vector_elt): Declare.
25099 * emit-rtl.c (const_vector_int_elt_1): New function.
25100 (const_vector_elt): Likewise.
25101 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
25102 of CONST_VECTOR_ELT.
25104 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25106 * expr.c: Include rtx-vector-builder.h.
25107 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
25108 directly on the tree encoding.
25109 (const_vector_from_tree): Likewise.
25110 * optabs.c: Include rtx-vector-builder.h.
25111 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
25112 sequence of "u" values.
25113 * vec-perm-indices.c: Include rtx-vector-builder.h.
25114 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
25115 directly on the vec_perm_indices encoding.
25117 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25119 * doc/rtl.texi (const_vector): Describe new encoding scheme.
25120 * Makefile.in (OBJS): Add rtx-vector-builder.o.
25121 * rtx-vector-builder.h: New file.
25122 * rtx-vector-builder.c: Likewise.
25123 * rtl.h (rtx_def::u2): Add a const_vector field.
25124 (CONST_VECTOR_NPATTERNS): New macro.
25125 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
25126 (CONST_VECTOR_DUPLICATE_P): Likewise.
25127 (CONST_VECTOR_STEPPED_P): Likewise.
25128 (CONST_VECTOR_ENCODED_ELT): Likewise.
25129 (const_vec_duplicate_p): Check for a duplicated vector encoding.
25130 (unwrap_const_vec_duplicate): Likewise.
25131 (const_vec_series_p): Check for a non-duplicated vector encoding.
25132 Say that the function only returns true for integer vectors.
25133 * emit-rtl.c: Include rtx-vector-builder.h.
25134 (gen_const_vec_duplicate_1): Delete.
25135 (gen_const_vector): Call gen_const_vec_duplicate instead of
25136 gen_const_vec_duplicate_1.
25137 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
25138 (gen_const_vec_duplicate): Use rtx_vector_builder.
25139 (gen_const_vec_series): Likewise.
25140 (gen_rtx_CONST_VECTOR): Likewise.
25141 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
25142 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
25143 Build a new vector rather than modifying a CONST_VECTOR in-place.
25144 (handle_special_swappables): Update call accordingly.
25145 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
25146 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
25147 Build a new vector rather than modifying a CONST_VECTOR in-place.
25148 (handle_special_swappables): Update call accordingly.
25150 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25152 * simplify-rtx.c (simplify_const_binary_operation): Use
25153 CONST_VECTOR_ELT instead of XVECEXP.
25155 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25157 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
25158 the selector elements to be different from the data elements
25159 if the selector is a VECTOR_CST.
25160 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
25161 ssizetype for the selector.
25163 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25165 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
25166 before testing each element individually.
25167 * tree-vect-generic.c (lower_vec_perm): Likewise.
25169 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25171 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
25172 * selftest-run-tests.c (selftest::run_tests): Call it.
25173 * vector-builder.h (vector_builder::operator ==): New function.
25174 (vector_builder::operator !=): Likewise.
25175 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
25176 (vec_perm_indices::all_from_input_p): New function.
25177 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
25178 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
25179 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
25180 instead of reading the VECTOR_CST directly. Detect whether both
25181 vector inputs are the same before constructing the vec_perm_indices,
25182 and update the number of inputs argument accordingly. Use the
25183 utility functions added above. Only construct sel2 if we need to.
25185 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25187 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
25188 the broadcast of the low byte.
25189 (expand_mult_highpart): Use an explicit encoding for the permutes.
25190 * optabs-query.c (can_mult_highpart_p): Likewise.
25191 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25192 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25193 (vectorizable_bswap): Likewise.
25194 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
25195 explicit encoding for the power-of-2 permutes.
25196 (vect_permute_store_chain): Likewise.
25197 (vect_grouped_load_supported): Likewise.
25198 (vect_permute_load_chain): Likewise.
25200 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25202 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
25203 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
25204 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
25205 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
25206 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
25207 (vect_gen_perm_mask_any): Likewise.
25209 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25211 * int-vector-builder.h: New file.
25212 * vec-perm-indices.h: Include int-vector-builder.h.
25213 (vec_perm_indices): Redefine as an int_vector_builder.
25214 (auto_vec_perm_indices): Delete.
25215 (vec_perm_builder): Redefine as a stand-alone class.
25216 (vec_perm_indices::vec_perm_indices): New function.
25217 (vec_perm_indices::clamp): Likewise.
25218 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
25219 (vec_perm_indices::new_vector): New function.
25220 (vec_perm_indices::new_expanded_vector): Update for new
25221 vec_perm_indices class.
25222 (vec_perm_indices::rotate_inputs): New function.
25223 (vec_perm_indices::all_in_range_p): Operate directly on the
25224 encoded form, without computing elided elements.
25225 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
25226 encoding. Update for new vec_perm_indices class.
25227 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
25228 the given vec_perm_builder.
25229 (expand_vec_perm_var): Update vec_perm_builder constructor.
25230 (expand_mult_highpart): Use vec_perm_builder instead of
25231 auto_vec_perm_indices.
25232 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
25233 vec_perm_indices instead of auto_vec_perm_indices. Use a single
25234 or double series encoding as appropriate.
25235 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
25236 vec_perm_indices instead of auto_vec_perm_indices.
25237 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
25238 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
25239 (vect_permute_store_chain): Likewise.
25240 (vect_grouped_load_supported): Likewise.
25241 (vect_permute_load_chain): Likewise.
25242 (vect_shift_permute_load_chain): Likewise.
25243 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
25244 (vect_transform_slp_perm_load): Likewise.
25245 (vect_schedule_slp_instance): Likewise.
25246 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25247 (vectorizable_mask_load_store): Likewise.
25248 (vectorizable_bswap): Likewise.
25249 (vectorizable_store): Likewise.
25250 (vectorizable_load): Likewise.
25251 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
25252 vec_perm_indices instead of auto_vec_perm_indices. Use
25253 tree_to_vec_perm_builder to read the vector from a tree.
25254 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
25255 vec_perm_builder instead of a vec_perm_indices.
25256 (have_whole_vector_shift): Use vec_perm_builder and
25257 vec_perm_indices instead of auto_vec_perm_indices. Leave the
25258 truncation to calc_vec_perm_mask_for_shift.
25259 (vect_create_epilog_for_reduction): Likewise.
25260 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
25261 from auto_vec_perm_indices to vec_perm_indices.
25262 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
25263 instead of changing individual elements.
25264 (aarch64_vectorize_vec_perm_const): Use new_vector to install
25265 the vector in d.perm.
25266 * config/arm/arm.c (expand_vec_perm_d::perm): Change
25267 from auto_vec_perm_indices to vec_perm_indices.
25268 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
25269 instead of changing individual elements.
25270 (arm_vectorize_vec_perm_const): Use new_vector to install
25271 the vector in d.perm.
25272 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
25273 Update vec_perm_builder constructor.
25274 (rs6000_expand_interleave): Likewise.
25275 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
25276 (rs6000_expand_interleave): Likewise.
25278 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25280 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
25281 to qimode could truncate the indices.
25282 * optabs.c (expand_vec_perm_var): Likewise.
25284 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25286 * Makefile.in (OBJS): Add vec-perm-indices.o.
25287 * vec-perm-indices.h: New file.
25288 * vec-perm-indices.c: Likewise.
25289 * target.h (vec_perm_indices): Replace with a forward class
25291 (auto_vec_perm_indices): Move to vec-perm-indices.h.
25292 * optabs.h: Include vec-perm-indices.h.
25293 (expand_vec_perm): Delete.
25294 (selector_fits_mode_p, expand_vec_perm_var): Declare.
25295 (expand_vec_perm_const): Declare.
25296 * target.def (vec_perm_const_ok): Replace with...
25297 (vec_perm_const): ...this new hook.
25298 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
25299 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
25300 * doc/tm.texi: Regenerate.
25301 * optabs.def (vec_perm_const): Delete.
25302 * doc/md.texi (vec_perm_const): Likewise.
25303 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
25304 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
25305 expand_vec_perm for constant permutation vectors. Assert that
25306 the mode of variable permutation vectors is the integer equivalent
25307 of the mode that is being permuted.
25308 * optabs-query.h (selector_fits_mode_p): Declare.
25309 * optabs-query.c: Include vec-perm-indices.h.
25310 (selector_fits_mode_p): New function.
25311 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
25312 is defined, instead of checking whether the vec_perm_const_optab
25313 exists. Use targetm.vectorize.vec_perm_const instead of
25314 targetm.vectorize.vec_perm_const_ok. Check whether the indices
25315 fit in the vector mode before using a variable permute.
25316 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
25317 vec_perm_indices instead of an rtx.
25318 (expand_vec_perm): Replace with...
25319 (expand_vec_perm_const): ...this new function. Take the selector
25320 as a vec_perm_indices rather than an rtx. Also take the mode of
25321 the selector. Update call to shift_amt_for_vec_perm_mask.
25322 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
25323 Use vec_perm_indices::new_expanded_vector to expand the original
25324 selector into bytes. Check whether the indices fit in the vector
25325 mode before using a variable permute.
25326 (expand_vec_perm_var): Make global.
25327 (expand_mult_highpart): Use expand_vec_perm_const.
25328 * fold-const.c: Includes vec-perm-indices.h.
25329 * tree-ssa-forwprop.c: Likewise.
25330 * tree-vect-data-refs.c: Likewise.
25331 * tree-vect-generic.c: Likewise.
25332 * tree-vect-loop.c: Likewise.
25333 * tree-vect-slp.c: Likewise.
25334 * tree-vect-stmts.c: Likewise.
25335 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
25337 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
25338 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
25339 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
25340 (aarch64_vectorize_vec_perm_const): ...this new function.
25341 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25342 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25343 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
25344 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
25345 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25346 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25347 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
25349 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
25350 check for NEON modes.
25351 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
25352 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
25353 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
25354 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
25356 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
25357 the old VEC_PERM_CONST conditions.
25358 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
25359 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
25360 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
25361 (ia64_vectorize_vec_perm_const_ok): Merge into...
25362 (ia64_vectorize_vec_perm_const): ...this new function.
25363 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
25364 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
25365 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
25366 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
25367 * config/mips/mips.c (mips_expand_vec_perm_const)
25368 (mips_vectorize_vec_perm_const_ok): Merge into...
25369 (mips_vectorize_vec_perm_const): ...this new function.
25370 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
25371 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
25372 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
25373 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
25374 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
25375 (rs6000_expand_vec_perm_const): Delete.
25376 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
25378 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25379 (altivec_expand_vec_perm_const_le): Take each operand individually.
25380 Operate on constant selectors rather than rtxes.
25381 (altivec_expand_vec_perm_const): Likewise. Update call to
25382 altivec_expand_vec_perm_const_le.
25383 (rs6000_expand_vec_perm_const): Delete.
25384 (rs6000_vectorize_vec_perm_const_ok): Delete.
25385 (rs6000_vectorize_vec_perm_const): New function.
25386 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
25387 an element count and rtx array.
25388 (rs6000_expand_extract_even): Update call accordingly.
25389 (rs6000_expand_interleave): Likewise.
25390 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
25391 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
25392 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
25393 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
25394 (rs6000_expand_vec_perm_const): Delete.
25395 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25396 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25397 (altivec_expand_vec_perm_const_le): Take each operand individually.
25398 Operate on constant selectors rather than rtxes.
25399 (altivec_expand_vec_perm_const): Likewise. Update call to
25400 altivec_expand_vec_perm_const_le.
25401 (rs6000_expand_vec_perm_const): Delete.
25402 (rs6000_vectorize_vec_perm_const_ok): Delete.
25403 (rs6000_vectorize_vec_perm_const): New function. Remove stray
25404 reference to the SPE evmerge intructions.
25405 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
25406 an element count and rtx array.
25407 (rs6000_expand_extract_even): Update call accordingly.
25408 (rs6000_expand_interleave): Likewise.
25409 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
25410 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
25412 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25414 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25416 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
25417 vector mode and that that mode matches the mode of the data
25419 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
25420 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
25421 directly using expand_vec_perm_1 when forcing selectors into
25423 (expand_vec_perm_var): New function, split out from expand_vec_perm.
25425 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25427 * optabs-query.h (can_vec_perm_p): Delete.
25428 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
25429 * optabs-query.c (can_vec_perm_p): Split into...
25430 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
25431 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
25432 particular selector is valid.
25433 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
25434 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
25435 (vect_grouped_load_supported): Likewise.
25436 (vect_shift_permute_load_chain): Likewise.
25437 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
25438 (vect_transform_slp_perm_load): Likewise.
25439 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25440 (vectorizable_bswap): Likewise.
25441 (vect_gen_perm_mask_checked): Likewise.
25442 * fold-const.c (fold_ternary_loc): Likewise. Don't take
25443 implementations of variable permutation vectors into account
25444 when deciding which selector to use.
25445 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
25446 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
25447 with a false third argument.
25448 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
25449 to test whether the constant selector is valid and can_vec_perm_var_p
25450 to test whether a variable selector is valid.
25452 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25454 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
25455 * optabs-query.c (can_vec_perm_p): Likewise.
25456 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
25457 instead of vec_perm_indices.
25458 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
25459 (vect_gen_perm_mask_checked): Likewise,
25460 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
25461 (vect_gen_perm_mask_checked): Likewise,
25463 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25465 * optabs-query.h (qimode_for_vec_perm): Declare.
25466 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
25467 (qimode_for_vec_perm): ...this new function.
25468 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
25470 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25472 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
25473 does not have a conditional at the top.
25475 2018-01-02 Richard Biener <rguenther@suse.de>
25477 * ipa-inline.c (big_speedup_p): Fix expression.
25479 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
25482 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
25485 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
25489 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
25490 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
25491 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
25492 cond_taken_branch_cost 3->4.
25494 2018-01-01 Jakub Jelinek <jakub@redhat.com>
25496 PR tree-optimization/83581
25497 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
25498 TODO_cleanup_cfg if any changes have been made.
25500 PR middle-end/83608
25501 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
25502 convert_modes if target mode has the right side, but different mode
25505 PR middle-end/83609
25506 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
25507 last argument when extracting from CONCAT. If either from_real or
25508 from_imag is NULL, use expansion through memory. If result is not
25509 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
25510 the parts directly to inner mode, if even that fails, use expansion
25513 PR middle-end/83623
25514 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
25515 check for bswap in mode rather than HImode and use that in expand_unop
25518 Copyright (C) 2018 Free Software Foundation, Inc.
25520 Copying and distribution of this file, with or without modification,
25521 are permitted in any medium without royalty provided the copyright
25522 notice and this notice are preserved.