1 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3 PR tree-optimization/87259
5 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
6 execute_cse_reciprocals_1 has tried transforming.
8 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
10 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
11 VR_VARYING for PLUS/MINUS_EXPR.
13 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
15 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
18 2018-09-14 Richard Biener <rguenther@suse.de>
21 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
22 bits for the merged partition.
24 2018-09-13 Martin Sebor <msebor@redhat.com>
25 Bernd Edlinger <bernd.edlinger@hotmail.de>
27 * builtins.h (c_srlen): Add argument.
28 * builtins.c (warn_string_no_nul): New function.
29 (c_strlen): Add argument and use it. Update recursive calls.
30 Pass DECL argument to string_constant to get info on non
31 terminated strings. Update *NONSTR as needed.
32 (fold_builtin_strlen): Add argument to calls to c_strlen.
33 Warn for unterminated arrays.
34 (warn_string_no_null): Add prototype.
35 * expr.c (string_constant): Update arguments. Update recursive
36 calls appropriately. Detect missing NUL terminator and outermost
37 declaration its missing in.
38 Improve checks for arrays with nonzero lower bound or elements
39 that are not a single byte. Simplify offset computation.
40 Simplify checks for non-NUL terminated strings.
41 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
42 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
44 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
46 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
48 * fold-const.c (c_getstr): Fix function comment. Remove unused third
49 argument. Fix range checks.
50 * fold-const.h (c_getstr): Adjust protoype.
51 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
52 string is constant but contains no NUL byte.
54 * expr.c (string_constant): Adjust function comment.
55 Remove bogus check for zero termination.
57 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
59 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
60 (get_constant_size): Don't make STRING_CSTs larger than they are.
61 (check_string_literal): New check function for STRING_CSTs.
62 (output_constant): Use it.
64 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
67 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
69 2018-09-13 Richard Biener <rguenther@suse.de>
71 PR tree-optimization/87263
72 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
73 (struct unwind_state): Add max_rpo field.
74 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
75 Compute max_rpo, the max RPO number a block can be backwards reached
76 from. Re-write non-iterating mode to a RPO ordered worklist approach,
77 separating it from the iterating mode.
79 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
81 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
82 (rfs_decision): New scheduling decision.
84 2018-09-13 Richard Biener <rguenther@suse.de>
87 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
88 (vn_nary_op_insert_pieces_predicated): Do not write useless
89 valid_dominated_by_p entry outside of the allocated storage.
91 2018-09-13 Omar Sandoval <osandov@osandov.com>
92 Tom de Vries <tdevries@suse.de>
95 * dwarf2out.c (is_c): New function.
96 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
98 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
101 * config/aarch64/aarch64.md (*aarch64_bfxil):
103 * config/aarch64/constraints.md (Ulc): Define.
104 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
106 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
109 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
111 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
112 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
113 aarch64_layout_frame call.
114 (aarch64_expand_epilogue): Likewise.
115 (aarch64_initial_elimination_offset): Likewise.
116 (aarch64_get_separate_components): Likewise.
117 (aarch64_use_return_insn_p): Likewise.
118 (aarch64_layout_frame): Remove unneeded check.
120 2018-09-13 Jakub Jelinek <jakub@redhat.com>
122 * configure.ac: Only append
123 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
124 gcc_config_arguments if it was never reconfigured or last reconfigure
125 was with different arguments.
126 * configure: Regenerated.
128 2018-09-13 Jakub Jelinek <jakub@redhat.com>
129 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
132 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
133 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
135 2018-09-13 Jakub Jelinek <jakub@redhat.com>
137 PR tree-optimization/87287
138 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
139 X % C == 0 to X % (unsigned) C == 0 optimization to ...
140 * match.pd (X % C == 0): ... here. New optimization.
142 2018-09-12 Jakub Jelinek <jakub@redhat.com>
145 * expr.h (maybe_optimize_mod_cmp): Declare.
146 * expr.c (mod_inv): New function.
147 (maybe_optimize_mod_cmp): New function.
148 (do_store_flag): Use it.
149 * cfgexpand.c (expand_gimple_cond): Likewise.
151 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
152 Julian Brown <julian@codesourcery.com>
155 * gimplify.c (gimplify_scan_omp_clauses): Set
156 target_firstprivatize_array_bases in OpenACC parallel and kernels
157 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
158 OpenACC data regions.
160 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
162 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
163 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
165 2018-09-12 Richard Biener <rguenther@suse.de>
167 PR tree-optimization/87280
168 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
169 edge but unreachable target.
170 (do_rpo_vn): For conservatively handling a PHI only mark
171 the backedge executable but not the block reachable.
173 2018-09-12 Richard Biener <rguenther@suse.de>
175 PR tree-optimization/87266
176 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
179 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
181 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
183 ("trunc<BFP:mode><DFP_ALL:mode>2")
184 ("trunc<DFP_ALL:mode><BFP:mode>2")
185 ("extend<BFP:mode><DFP_ALL:mode>2")
186 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
187 according to the target operand type.
189 2018-09-12 Jakub Jelinek <jakub@redhat.com>
190 Andreas Krebbel <krebbel@linux.ibm.com>
192 PR tree-optimization/86844
193 * gimple-ssa-store-merging.c
194 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
195 there are any overlapping stores in between them, make sure they are
196 also coalesced or we give up completely.
198 2018-09-12 Jakub Jelinek <jakub@redhat.com>
201 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
202 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
204 2018-09-12 Tom de Vries <tdevries@suse.de>
206 * common.opt (gdescribe-dies): Add option.
207 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
208 attribute for artifical and nameless decls.
209 (dwarf2out_register_external_die): Add description attribute to
210 external reference die.
211 (add_desc_attribute): New functions.
212 (gen_subprogram_die): Add description attribute to
213 DW_TAG_call_site_parameter.
214 * tree-pretty-print.c (print_generic_expr_to_str): New function.
215 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
216 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
218 (@item -gdescribe-dies): Add.
220 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
222 * tree-vrp.c (vrp_shift_undefined_p): Remove.
223 (extract_range_from_binary_expr_1: Call
224 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
225 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
228 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
230 * gimple-ssa-warn-alloca.c
231 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
232 field for ALLOCA_BOUND_*_LARGE.
234 2018-09-11 Nathan Sidwell <nathan@acm.org>
236 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
238 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
240 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
241 for clobbers. Remove obsolete comment.
243 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
245 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
246 mpxchk, mpxld and mpxst types.
247 (define_attr length_immediate): Remove all processing of mpx types.
248 (define_attr prefix_0f): Ditto.
249 (define_attr memory): Ditto.
251 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
253 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
254 (log<mode>2): Change operand 1 predicate to general_operand.
255 Extend operand 1 to XFmode and generate logxf3 insn.
256 (log10<mode>2): Change operand 1 predicate to general_operand.
257 Extend operand 1 to XFmode and generate log10xf3 insn.
258 (log2<mode>2): Change operand 1 predicate to general_operand.
259 Extend operand 1 to XFmode and generate log2xf3 insn.
260 (fyl2xp1_extend<mode>xf3_i387): Remove.
261 (log1p<mode>2): Change operand 1 predicate to general_operand.
262 Extend operand 1 to XFmode and generate log1pxf3 insn.
263 (fxtract_extend<mode>xf3_i387): Remove.
264 (logb<mode>2): Change operand 1 predicate to general_operand.
265 Extend operand 1 to XFmode and generate logbxf3 insn.
266 (ilogb<mode>2): Change operand 1 predicate to general_operand.
267 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
268 (significand<mode>2): Change operand 1 predicate to general_operand.
269 Extend operand 1 to XFmode and generate significandxf3 insn.
271 2018-09-11 Nathan Sidwell <nathan@acm.org>
273 * gcc.c (perror_with_name, pfatal_with_name): Delete.
274 (load_specs): Use fatal_error.
275 (DELETE_IF_ORDINARY, process_command): Use error.
276 (execute, run_attempt): Use fatal_error.
278 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
280 * diagnostic-core.h (sorry_at): New prototype.
281 * diagnostic.c (sorry_at): New function.
283 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
285 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
286 by zero as VR_UNDEFINED.
288 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
290 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
291 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
292 (<sincos>mode2): New expander.
293 (sincos_extend<mode>xf3_i387): Remove insn pattern.
294 (sincos -> sin, cos splitters): Remove splitter patterns.
295 (sincos<mode>3): Change operand 2 predicate to general_operand.
296 Extend operand 2 to XFmode and generate sincosxf3 insn.
297 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
298 Change operand 3 predicate to const1_operand.
299 (fptan_extend<mode>xf4_i387): Remove insn pattern.
300 (tanxf2): Update operands in the call to fptanxf4_i387.
301 (tan<mode>2): Change operand 1 predicate to general_operand.
302 Extend operand 1 to XFmode and generate tanxf3 insn.
303 (atan2xf3): Rename from *fpatanxf3_i387.
304 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
305 (atan2xf3): Remove expander.
306 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
307 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
308 (atan<mode>2): Change operand 1 predicate to general_operand.
309 Extend operand 1 to XFmode and generate atanxf3 insn.
311 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
313 * config/i386/i386.md (x87/SSE constant load splitter): Use
314 memory_operand instead of nonimmediate_operand for input operand
317 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
319 * config/i386/i386.md (float partial SSE register stall splitter): Move
320 splitter near its instruction pattern.
321 (float_extend partial SSE register stall splitter): Ditto.
322 (float_truncate partial SSE register stall splitter): Ditto.
324 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
327 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
328 to speculation_safe_value_not_needed.
331 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
332 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
333 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
334 leaf_function_p, instead use has_hard_reg_initial_val.
336 2018-09-09 Nathan Sidwell <nathan@acm.org>
338 * gcc.h (pfatal_with_name): Don't declare here.
339 * gcc.c (pfatal_with_name): Make static.
341 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
343 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
346 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
348 PR rtl-optimization/85458
349 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
350 priority hook to reduce the priority of EXPR.
352 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
354 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
355 DImode for x87 on 32bit targets. Conditionally disable x87 modes
356 with X87_ENABLE_FLOAT. Remove preparation code.
357 (*float<SWI48:mode><MODEF:mode>2): Rename from
358 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
359 math using "enabled" attribute.
360 (*floatdi<MODEF:mode>2_i387): Rename from
361 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
362 enable for 32bit targets only.
363 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
365 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
366 as operand 1 predicate. Rewrite as define_insn_and_split.
367 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
369 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
371 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
372 to fallthru to FLOAT case.
374 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
377 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
378 around folding of vec_sl to handle out of range shift values.
380 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
382 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
383 Update callers to gen_fix_trunc<mode>_i387_fisttp
384 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
385 nonimmediate_operand.
386 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
387 and corresponding splitters.
388 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
389 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
390 (fix_truncdi_i387_with_temp): Remove insn pattern
391 and corresponding splitters.
392 (fix_trunc<mode>_i387): Change operand 0 predicate to
393 nonimmediate_operand.
394 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
395 and corresponding splitters.
396 (*fistdi2_1): Remove.
398 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
399 (lrintxfdi2): Remove expander. Reimplement as define_insn.
400 (*fist<mode>2_1): Remove.
401 (fist<mode>2): Ditto.
402 (fist<mode>2_with_temp): Remove insn pattern and corresponding
404 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
405 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
406 (fistdi2_<rounding>): Change operand 0 predicate to
407 nonimmediate_operand.
408 (fistdi2_<rounding>_with_temp): Remove insn pattern
409 and corresponding splitters.
410 (fist<mode>2_<rounding>): Change operand 0 predicate to
411 nonimmediate_operand.
412 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
413 and corresponding splitters.
415 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
417 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
419 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
422 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
424 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
425 early gimple folding of vec_splat().
426 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
427 * gimple-fold.h: Add an extern define for tree_vec_extract().
429 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
431 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
432 wrappers around TREE_TYPE comparisons.
434 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
437 * config/s390/predicates.md: Add nonsym_memory_operand.
438 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
439 contains a SYMBOL_REF, load it into an intermediate pseudo.
440 (s390_emit_compare_and_swap): Legitimize operand.
441 * config/s390/s390.md: Use the new nonsym_memory_operand
442 with UNSPECV_CAS patterns.
444 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
447 * config/s390/s390-passes.def: New file.
448 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
450 (make_pass_s390_early_mach): Add declaration.
451 * config/s390/s390.c (make_pass_s390_early_mach):
452 (s390_option_override): Remove dynamic registration.
453 * config/s390/t-s390: Add s390-passes.def.
455 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
457 * config/s390/s390.c (s390_decompose_constant_pool_ref):
458 Remove UNSPEC_LTREL_BASE check.
459 (annotate_constant_pool_refs): Likewise.
460 (find_constant_pool_ref): Likewise.
461 (find_ltrel_base): Removed.
462 (replace_ltrel_base): Removed.
463 (s390_mainpool_finish): Remove replace_ltrel_base call.
464 (s390_chunkify_start): Remove pending LTREL_BASE logic.
465 (s390_chunkify_finish): Remove replace_ltrel_base call.
466 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
468 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
471 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
472 to speculation_safe_value_not_needed.
474 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
475 Bernd Schmidt <bernds_cb1@t-online.de>
477 * config/nvptx/nvptx-opts.h: New file.
478 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
479 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
481 (TARGET_SM35): New macro.
482 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
484 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
487 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
489 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
491 * config/i386/i386.md (truncdfsf2): Remove expander.
492 (truncdfsf2_with_temp): Ditto.
493 (truncxf<mode>2): Ditto.
494 (*truncdfsf_fast_mixed): Remove insn pattern.
495 (*truncdfsf_fast_i387): Ditto.
496 (*truncdfsf_mixed): Ditto.
497 (*truncdfsf_i387): Ditto.
498 (*truncdfsf2_i387_1): Ditto.
499 (*truncxfsf2_mixed): Ditto.
500 (*truncxfdf2_mixed): Ditto.
501 (*truncxf<mode>2_i387_noop): Ditto. Update callers
502 to call gen_truncxf<mode>2 instead.
503 (*truncxf<mode>2_i387): Remove.
504 (reg->reg splitters): Remove splitter pattern.
505 (reg->mem splitters): Ditto.
507 (truncdfsf2): New insn pattern.
508 (truncxf<mode>2): Ditto.
510 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
512 * tree-ssa-math-opts.c (is_mult_by): New function.
513 (is_square_of): Use the above.
514 (optimize_recip_sqrt): New function.
515 (pass_cse_reciprocals::execute): Use the above.
517 2018-09-05 Richard Biener <rguenther@suse.de>
520 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
521 to zero-init the emplaced vec.
523 2018-09-05 Martin Liska <mliska@suse.cz>
525 PR tree-optimization/87205
526 * tree-switch-conversion.c (pass_lower_switch::execute):
527 Group cases for switch statements.
529 2018-09-05 Richard Biener <rguenther@suse.de>
531 PR tree-optimization/87217
532 * tree-ssa-sccvn.c (vuse_valueize): New.
533 (vn_reference_lookup_pieces): Use it.
534 (vn_reference_lookup): Likewise.
536 2018-09-05 Nathan Sidwell <nathan@acm.org>
539 * stor-layout.c (place_field): Scan forwards to check last
540 bitfield when ms_bitfield_placement is in effect.
542 2018-09-05 Richard Biener <rguenther@suse.de>
545 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
548 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
549 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
551 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
552 * config.gcc (extra_objs): Build it.
553 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
555 * config/aarch64/aarch64-passes.def
556 (pass_tag_collision_avoidance): New pass.
557 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
558 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
559 (aarch64_classify_address): Remove static qualifier.
560 (aarch64_address_info, aarch64_address_type): Move to...
561 * config/aarch64/aarch64-protos.h: ... here.
562 (make_pass_tag_collision_avoidance): New function.
563 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
566 2018-09-05 Martin Liska <mliska@suse.cz>
568 * doc/gcov.texi: Update documentation of humar
570 * gcov.c (format_count): Print one decimal place, it provides
571 more fine number of situations like '1G' vs. '1.4G'.
573 2018-09-05 Martin Liska <mliska@suse.cz>
576 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
577 * optc-gen.awk: Allow 'Var' for Deprecated options in order
578 to generate a MASK value.
580 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
583 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
584 if frame pointer isn't used.
585 (compute_frame_pointer_to_fb_displacement): Likewise.
587 2018-09-04 Jakub Jelinek <jakub@redhat.com>
590 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
591 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
592 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
593 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
594 and OPTION_MASK_ISA_XSAVEC_UNSET.
596 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
598 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
599 XOR operations in NAND case.
601 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
603 * wide-int-range.cc (wide_int_range_convert): New.
604 * wide-int-range.h (wide_int_range_convert): New.
605 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
606 code into wide_int_range_convert.
607 (extract_range_into_wide_ints): Do not munge anti range constants
608 into the entire domain. Just return the range back.
610 2018-09-04 Martin Liska <mliska@suse.cz>
612 * genmatch.c (output_line_directive): Add new argument
614 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
616 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
618 * doc/invoke.texi (Option Summary): Add whitespace.
620 * doc/invoke.texi (Option Summary): Add -Waligned-new.
622 2018-09-04 Richard Biener <rguenther@suse.de>
624 PR tree-optimization/87211
625 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
626 backedge value we're supposed to treat as VARYING also number
627 the PHI to VARYING in case it got a different value-number already.
629 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
631 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
632 (extract_range_from_binary_expr_1): Do not call
633 vrp_can_optimize_bit_op.
634 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
636 (wide_int_range_get_mask_and_bounds): New.
637 (wide_int_range_optimize_bit_op): New.
638 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
639 (wide_int_range_bit_and): Same.
640 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
641 (wide_int_range_optimize_bit_op): New.
642 (wide_int_range_get_mask_and_bounds): New.
644 2018-09-04 Richard Biener <rguenther@suse.de>
646 PR tree-optimization/87176
647 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
648 variable. When value-numbering a virtual PHI node make sure
649 to not value-number to the backedge value.
651 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
653 * doc/extend.texi (Long Long, Hex Floats): Document support for
654 long long and hex floats in more recent versions of ISO C++.
656 2018-09-03 Richard Biener <rguenther@suse.de>
658 PR tree-optimization/87177
659 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
662 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
664 * bb-reorder.c (edge_order): Convert to C-qsort-style
665 tri-state comparator.
666 (reorder_basic_blocks_simple): Change std::stable_sort to gcc_stablesort.
668 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
670 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
671 tri-state comparator.
672 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
674 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
676 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
677 (mergesort): ... here as maximum count for using netsort.
678 (gcc_qsort): Set nlim to 3 if stable sort is requested.
679 (gcc_stablesort): New.
680 * system.h (gcc_stablesort): Declare.
682 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
684 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
685 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
686 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
688 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
690 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
691 lxsdx and stxsdx alternatives.
692 (*mov<mode>_hardfloat64): Ditto.
693 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
695 2018-09-03 Richard Biener <rguenther@suse.de>
697 PR tree-optimization/87200
698 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
701 2018-09-03 Martin Liska <mliska@suse.cz>
703 PR tree-optimization/87201
704 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
705 Fix parenthesis in an expression.
707 2018-09-03 Richard Biener <rguenther@suse.de>
709 PR tree-optimization/87197
710 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
711 visited. CSE the VN_INFO hashtable lookup.
713 PR tree-optimization/87169
714 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
715 iterating make sure there's no extra backedges from irreducible
716 regions feeding the header. Mark the destination block
719 2018-09-03 Martin Liska <mliska@suse.cz>
722 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
723 * common/common-targhooks.c (default_get_valid_option_values):
725 * common/common-targhooks.h (default_get_valid_option_values):
727 * common/config/i386/i386-common.c: Move processor_target_table
729 (ix86_get_valid_option_values): New function.
730 (TARGET_GET_VALID_OPTION_VALUES): New macro.
731 * config/i386/i386.c (struct ptt): Move to i386-common.c.
732 (PTA_*): Move all defined masks into i386-common.c.
733 (ix86_function_specific_restore): Use new processor_cost_table.
734 * config/i386/i386.h (struct ptt): Moved from i386.c.
735 (struct pta): Likewise.
736 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
737 * doc/tm.texi.in: Likewise.
738 * opt-suggestions.c (option_proposer::suggest_option):
739 Pass prefix to build_option_suggestions.
740 (option_proposer::get_completions): Likewise.
741 (option_proposer::build_option_suggestions): Use the new target
743 * opts.c (struct option_help_tuple): New struct.
744 (print_filtered_help): Use the new target hook.
746 2018-09-03 Martin Liska <mliska@suse.cz>
749 * predict.c (set_even_probabilities): Add likely_edges
750 argument and handle cases where we have precisely one
752 (combine_predictions_for_bb): Catch also likely_edges.
753 (tree_predict_by_opcode): Handle gswitch statements.
754 * tree-cfg.h (find_case_label_for_value): New declaration.
755 (find_taken_edge_switch_expr): Likewise.
756 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
757 Find pivot in decision tree based on probabily, not by number of
760 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
762 * doc/standards.texi (Standards): Update Objective-C reference.
764 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
766 * doc/install.texi (Prerequisites): Update link for MPC.
768 2018-09-01 Michael Matz <matz@suse.de>
770 PR tree-optimization/87074
771 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
772 PHIs for outer-loop uses.
774 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
776 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
777 * doc/invoke.texi (C Dialect Options): Ditto.
779 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
781 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
783 2018-08-31 Richard Biener <rguenther@suse.de>
785 PR tree-optimization/87168
786 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
787 (rpo_elim::eliminate_avail): When OP was not visited it must
790 2018-08-31 David Malcolm <dmalcolm@redhat.com>
792 * tree-vrp.c (copy_value_range): Convert param "from" from
793 "value_range *" to "const value_range *".
794 (range_is_null): Likewise for param "vr".
795 (range_int_cst_p): Likewise.
796 (range_int_cst_singleton_p): Likewise.
797 (symbolic_range_p): Likewise.
798 (value_ranges_intersect_p): Likewise for both params.
799 (value_range_nonnegative_p): Likewise for param "vr".
800 (value_range_constant_singleton): Likewise.
801 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
802 (extract_range_into_wide_ints): Likewise for param "vr".
803 (extract_range_from_multiplicative_op): Likewise for params "vr0"
805 (vrp_can_optimize_bit_op): Likewise.
806 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
808 (extract_range_from_unary_expr): Likewise.
809 (debug_value_range): Likewise for param "vr".
810 (value_range::dump): Add "const" qualifier.
811 (vrp_prop::check_array_ref): Convert local "vr" from
812 "value_range *" to "const value_range *".
813 (vrp_prop::check_mem_ref): Likewise.
814 (vrp_prop::visit_stmt): Likewise for local "old_vr".
815 (vrp_intersect_ranges_1): Likewise for param "vr_1".
816 (vrp_intersect_ranges): Likewise.
817 (simplify_stmt_for_jump_threading): Likewise for local "vr".
818 (vrp_prop::vrp_finalize): Likewise.
819 * tree-vrp.h (value_range::dump): Add "const" qualifier.
820 (vrp_intersect_ranges): Add "const" qualifier to params as above.
821 (extract_range_from_unary_expr): Likewise.
822 (value_range_constant_singleton): Likewise.
823 (symbolic_range_p): Likewise.
824 (copy_value_range): Likewise.
825 (extract_range_from_binary_expr_1): Likewise.
826 (range_int_cst_p): Likewise.
827 (vrp_set_zero_nonzero_bits): Likewise.
828 (range_int_cst_singleton_p): Likewise.
830 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
832 * config/aarch64/arm_neon.h (vabsd_s64): New.
833 (vnegd_s64): Likewise.
835 2018-08-31 Martin Jambor <mjambor@suse.cz>
837 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
839 2018-08-31 Martin Liska <mliska@suse.cz>
841 * ipa-icf.c (sem_item::add_type): Use
842 sem_item::m_type_hash_cache.
843 * ipa-icf.h: Move the cache from sem_item_optimizer
846 2018-08-31 Nathan Sidwell <nathan@acm.org>
848 * doc/extend.texi (Backwards Compatibility): Remove implicit
849 extern C leeway of () being (...).
851 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
853 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
855 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
859 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
861 2018-08-31 Jakub Jelinek <jakub@redhat.com>
864 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
865 gen_int_mode. Formatting fixes.
867 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
869 * target.def (custom_function_descriptors): Improve documentation.
870 * doc/tm.texi.in (Trampolines): Expand discussion of function
871 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
872 beginning of the section.
873 * doc/tm.texi: Regenerated.
875 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
877 * cfg.h (class auto_edge_flag): Spell out the template-id of the
878 base class in the initializer list. This is a workaround for
879 building with older GCC.
880 (class auto_bb_flag): Likewise.
882 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
884 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
885 (altivec_vcmpequ<VI_char>_p): Remove star.
886 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
888 (expand_strncmp_vec_sequence): New function.
889 (emit_final_str_compare_vec): New function.
890 (expand_strn_compare): Add support for vector strncmp.
891 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
892 length specification to bytes.
893 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
894 (vcmpnezb_p): New pattern.
895 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
896 for option -mstring-compare-inline-limit.
898 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
900 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
901 (PTA_SKYLAKE): Add PTA_AES.
902 (PTA_GOLDMONT): Likewise.
904 2018-08-29 Jan Hubicka <jh@suse.cz>
907 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
908 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
910 2018-08-29 Jan Hubicka <jh@suse.cz>
912 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
914 (hash_tree): Do not visit TYPE_STUB_DECL.
915 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
916 stream TYPE_STUB_DECL.
917 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
918 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
919 after free_lang_data.
920 (type_in_anonymous_namespace_p): Likewise.
922 2018-08-29 Jan Hubicka <jh@suse.cz>
924 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
925 comment that it has to be even number.
926 (class sreal): Change m_sig type to int32_t.
927 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
928 int64_t for temporary calculations.
929 (sreal_verify_basics): Drop one bit from minimum and maximum.
931 2018-08-30 Richard Biener <rguenther@suse.de>
933 PR tree-optimization/87147
934 * tree-ssa-sccvn.c (SSA_VISITED): New function.
935 (visit_phi): When the degenerate result is from the backedge and
936 we didn't visit its definition yet drop to VARYING.
937 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
939 2018-08-29 Jan Hubicka <jh@suse.cz>
941 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
943 (hash_tree): Likewise.
945 2018-08-29 Jan Hubicka <jh@suse.cz>
947 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
948 and TYPE_NEXT_REF_TO.
950 2018-08-29 Jan Hubicka <jh@suse.cz>
952 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
953 comment that it has to be even number.
954 (class sreal): Change m_sig type to int32_t.
955 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
956 int64_t for temporary calculations.
957 (sreal_verify_basics): Drop one bit from minimum and maximum.
959 2018-08-30 Tamar Christina <tamar.christina@arm.com>
961 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
963 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
966 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
967 if to_add is negative.
969 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
972 * builtins.c (c_strlen): Improve range checks.
974 2018-08-29 Martin Sebor <msebor@redhat.com>
975 Jeff Law <law@redhat.com>
977 PR tree-optimization/86714
978 PR tree-optimization/86711
979 * builtins.c (c_strlen): Add arguments to call to string_constant.
980 * expr.c (string_constant): Add argument. Detect missing nul
981 terminator and outermost declaration it's missing in.
982 * expr.h (string_constant): Add argument.
983 * fold-const.c (read_from_constant_string): Add arguments to call to
985 (c_getstr): Likewise.
986 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
988 * tree-ssa-strlen.c (get_stridx): Likewise.
990 2018-08-29 Jan Hubicka <jh@suse.cz>
992 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
993 Do not stream DECL_VINDEX.
994 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
995 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
996 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
998 2018-08-29 Richard Biener <rguenther@suse.de>
1000 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
1001 virtual operands that are not default defs to honor region
1003 (rpo_vn_valueize): Remove ineffective code here.
1005 2018-08-29 Richard Biener <rguenther@suse.de>
1007 PR tree-optimization/87132
1008 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
1009 when skipping defs reachable over backedges.
1011 2018-08-29 Richard Biener <rguenther@suse.de>
1013 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
1014 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
1015 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
1016 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
1017 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
1018 (vn_reference_lookup_3): Remove use of const_parms.
1019 (free_rpo_vn): Do not free const_parms.
1020 (do_rpo_vn): Do not call init_const_parms.
1021 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
1022 SSA_NAME_POINTS_TO_READONLY_MEMORY.
1023 (call_may_clobber_ref_p_1): Likewise.
1025 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
1028 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
1030 (-ftree-scev-cprop): Document.
1032 2018-08-29 Jan Hubicka <jh@suse.cz>
1034 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
1036 (sreal constructor): Update.
1037 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
1038 sreal:operator/): Update.
1040 2018-08-29 Martin Liska <mliska@suse.cz>
1042 * tree-switch-conversion.c (switch_conversion::expand):
1043 Strenghten assumption about gswitch statements.
1045 2018-08-29 Richard Biener <rguenther@suse.de>
1047 PR tree-optimization/87117
1048 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
1049 re-value-number released SSA VDEFs.
1051 2018-08-29 Richard Biener <rguenther@suse.de>
1053 PR tree-optimization/87126
1054 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
1056 2018-08-28 Jim Wilson <jimw@sifive.com>
1058 * config/riscv/pic.md: Rewrite.
1059 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
1061 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
1062 (SOFTF, default_load, softload, softstore): New.
1064 2018-08-28 Jeff Law <law@redhat.com>
1066 * fold-const.c (fold_binary_loc): Remove recently added assert.
1068 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
1070 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
1071 to OP parmeter of generated function.
1073 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
1075 PR tree-optimization/87009
1076 * match.pd: Add boolean optimizations.
1078 2018-08-28 Martin Sebor <msebor@redhat.com>
1081 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
1082 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
1083 (pass_walloca::gate): Use it.
1084 (alloca_call_type): Same.
1085 (pass_walloca::execute): Same.
1086 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
1088 2018-08-28 David Malcolm <dmalcolm@redhat.com>
1090 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
1091 GCC_VERSION for usage of "__gcc_dump_printf__" format from
1094 2018-08-28 Richard Biener <rguenther@suse.de>
1096 PR tree-optimization/87124
1097 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
1098 constants before looking up avail.
1100 2018-08-28 Jakub Jelinek <jakub@redhat.com>
1103 * calls.c (maybe_warn_nonstring_arg): Punt early if
1104 warn_stringop_overflow is zero. Don't call get_range_strlen
1105 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
1106 Swap comparison operands to have constants on rhs. Only use
1107 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
1108 increment lenrng[0].
1110 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
1112 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
1113 use of tree_to_shwi. Remove duplicated test for the size being
1114 a whole number of bytes.
1116 2018-08-28 Richard Biener <rguenther@suse.de>
1118 PR tree-optimization/87117
1119 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
1120 Handle removed stmt without LHS (GIMPLE_NOP).
1122 2018-08-28 Richard Biener <rguenther@suse.de>
1124 PR tree-optimization/87117
1125 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
1126 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
1128 2018-08-28 Richard Biener <rguenther@suse.de>
1130 PR tree-optimization/87117
1131 * tree-ssa-pre.c (compute_avail): Do not make expressions
1132 with predicated values available.
1133 (get_expr_value_id): Assert we do not run into predicated value
1136 2018-08-28 Richard Biener <rguenther@suse.de>
1138 PR tree-optimization/87117
1139 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
1140 get virtual operands.
1141 (get_expr_operands): Handle STRING_CST like other decls.
1143 2018-08-28 Martin Liska <mliska@suse.cz>
1145 * tree.h: Update documentation of fndecl_built_in_p
1149 2018-08-27 Jeff Law <law@redhat.com>
1150 PR tree-optimization/87110
1151 * tree-ssa-dse.c (compute_trims): Handle non-constant
1154 2018-08-27 Martin Sebor <msebor@redhat.com>
1156 PR tree-optimization/86914
1157 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
1159 2018-08-27 Martin Sebor <msebor@redhat.com>
1161 PR tree-optimization/87112
1162 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
1163 the type of the bound argument.
1165 2018-08-27 Jeff Law <law@redhat.com>
1167 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
1168 type does not have a TYPE_SIZE_UNIT.
1170 2018-08-27 Steve Ellcey <sellcey@cavium.com>
1172 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
1173 with include of backend.h.
1175 2018-08-27 Richard Biener <rguenther@suse.de>
1177 PR tree-optimization/86927
1178 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
1179 use const cond reduction code.
1181 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
1183 PR tree-optimization/85758
1184 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
1186 2018-08-27 David Malcolm <dmalcolm@redhat.com>
1189 * diagnostic-show-locus.c (class layout_range): Update for
1190 conversion of show_caret_p to a tri-state.
1191 (layout_range::layout_range): Likewise.
1192 (make_range): Likewise.
1193 (layout::maybe_add_location_range): Likewise.
1194 (layout::should_print_annotation_line_p): Don't show annotation
1195 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
1196 (layout::get_state_at_point): Update for conversion of
1197 show_caret_p to a tri-state. Bail out early for
1198 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
1199 underlining or source colorization.
1200 (gcc_rich_location::add_location_if_nearby): Update for conversion
1201 of show_caret_p to a tri-state.
1202 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
1203 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
1205 (selftest::test_one_liner_labels): Likewise.
1206 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
1207 conversion of show_caret_p to a tri-state.
1208 * pretty-print.c (text_info::set_location): Likewise.
1209 * pretty-print.h (text_info::set_location): Likewise.
1210 * substring-locations.c (format_warning_n_va): Likewise.
1211 * tree-diagnostic.c (default_tree_printer): Likewise.
1212 * tree-pretty-print.c (newline_and_indent): Likewise.
1214 2018-08-27 David Malcolm <dmalcolm@redhat.com>
1217 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
1218 line above for line-insertion fix-it hints.
1219 (selftest::test_fixit_insert_containing_newline): Update the
1220 expected results, and add a test with line-numbering enabled.
1222 2018-08-27 Martin Liska <mliska@suse.cz>
1225 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
1226 params with DECL_HAS_VALUE_EXPR_P.
1228 2018-08-27 Martin Liska <mliska@suse.cz>
1230 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
1231 selected expansion strategy.
1233 2018-08-27 Martin Liska <mliska@suse.cz>
1235 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
1236 * builtins.c (is_builtin_fn): Likewise.
1237 * attribs.c (diag_attr_exclusions): Use new function
1238 fndecl_built_in_p and remove check for FUNCTION_DECL if
1240 (builtin_mathfn_code): Likewise.
1241 (fold_builtin_expect): Likewise.
1242 (fold_call_expr): Likewise.
1243 (fold_builtin_call_array): Likewise.
1244 (fold_call_stmt): Likewise.
1245 (set_builtin_user_assembler_name): Likewise.
1246 (is_simple_builtin): Likewise.
1247 * calls.c (gimple_alloca_call_p): Likewise.
1248 (maybe_warn_nonstring_arg): Likewise.
1249 * cfgexpand.c (expand_call_stmt): Likewise.
1250 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
1251 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
1252 (cgraph_node::verify_node): Likewise.
1253 * cgraphclones.c (build_function_decl_skip_args): Likewise.
1254 (cgraph_node::create_clone): Likewise.
1255 * config/arm/arm.c (arm_insert_attributes): Likewise.
1256 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
1257 * dse.c (scan_insn): Likewise.
1258 * expr.c (expand_expr_real_1): Likewise.
1259 * fold-const.c (operand_equal_p): Likewise.
1260 (fold_binary_loc): Likewise.
1261 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
1262 * gimple-low.c (lower_stmt): Likewise.
1263 * gimple-pretty-print.c (dump_gimple_call): Likewise.
1264 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Likewise.
1265 * gimple.c (gimple_build_call_from_tree): Likewise.
1266 (gimple_call_builtin_p): Likewise.
1267 (gimple_call_combined_fn): Likewise.
1268 * gimplify.c (gimplify_call_expr): Likewise.
1269 (gimple_boolify): Likewise.
1270 (gimplify_modify_expr): Likewise.
1271 (gimplify_addr_expr): Likewise.
1272 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
1273 * ipa-cp.c (determine_versionability): Likewise.
1274 * ipa-fnsummary.c (compute_fn_summary): Likewise.
1275 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
1276 * ipa-split.c (visit_bb): Likewise.
1277 (split_function): Likewise.
1278 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
1279 * lto-cgraph.c (input_node): Likewise.
1280 * lto-streamer-out.c (write_symbol): Likewise.
1281 * omp-low.c (setjmp_or_longjmp_p): Likewise.
1282 (lower_omp_1): Likewise.
1283 * predict.c (strip_predict_hints): Likewise.
1284 * print-tree.c (print_node): Likewise.
1285 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
1286 * trans-mem.c (is_tm_irrevocable): Likewise.
1287 (is_tm_load): Likewise.
1288 (is_tm_simple_load): Likewise.
1289 (is_tm_store): Likewise.
1290 (is_tm_simple_store): Likewise.
1291 (is_tm_abort): Likewise.
1292 (tm_region_init_1): Likewise.
1293 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
1294 * tree-cfg.c (verify_gimple_call): Likewise.
1295 (move_stmt_r): Likewise.
1296 (stmt_can_terminate_bb_p): Likewise.
1297 * tree-eh.c (lower_eh_constructs_2): Likewise.
1298 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
1299 * tree-inline.c (remap_gimple_stmt): Likewise.
1300 (copy_bb): Likewise.
1301 (estimate_num_insns): Likewise.
1302 (fold_marked_statements): Likewise.
1303 * tree-sra.c (scan_function): Likewise.
1304 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
1305 (optimize_stack_restore): Likewise.
1306 (pass_fold_builtins::execute): Likewise.
1307 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
1308 (mark_all_reaching_defs_necessary_1): Likewise.
1309 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
1310 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
1311 (pass_forwprop::execute): Likewise.
1312 * tree-ssa-loop-im.c (stmt_cost): Likewise.
1313 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1314 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
1315 * tree-ssa-strlen.c (get_string_length): Likewise.
1316 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
1317 (find_func_aliases_for_call): Likewise.
1318 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
1319 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
1320 * tree-tailcall.c (find_tail_calls): Likewise.
1321 * tree.c (need_assembler_name_p): Likewise.
1322 (free_lang_data_in_decl): Likewise.
1323 (get_call_combined_fn): Likewise.
1324 * ubsan.c (is_ubsan_builtin_p): Likewise.
1325 * varasm.c (incorporeal_function_p): Likewise.
1326 * tree.h (DECL_BUILT_IN): Remove and replace with
1328 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
1329 (fndecl_built_in_p): New.
1331 2018-08-27 Martin Liska <mliska@suse.cz>
1333 PR tree-optimization/86847
1334 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
1335 Dump also subtree probability.
1336 (switch_decision_tree::do_jump_if_equal): New function.
1337 (switch_decision_tree::emit_case_nodes): Handle special
1338 situations in balanced tree that can be emitted much simpler.
1339 Fix calculation of probabilities that happen in tree expansion.
1340 * tree-switch-conversion.h (struct cluster): Add
1342 (struct simple_cluster): Likewise.
1343 (struct case_tree_node): Add new function has_child.
1344 (do_jump_if_equal): New.
1346 2018-08-27 Martin Liska <mliska@suse.cz>
1348 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
1349 Add new argument to bit_test_cluster constructor.
1350 (bit_test_cluster::emit): Set bits really number of values
1352 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
1353 probability argument.
1354 * tree-switch-conversion.h (struct bit_test_cluster):
1355 Add m_handles_entire_switch.
1357 2018-08-27 Martin Liska <mliska@suse.cz>
1359 PR tree-optimization/86702
1360 * tree-switch-conversion.c (jump_table_cluster::emit):
1361 Make probabilities even for values in jump table
1362 according to number of cases handled.
1363 (switch_decision_tree::compute_cases_per_edge): Pass
1364 argument to reset_out_edges_aux function.
1365 (switch_decision_tree::analyze_switch_statement): Likewise.
1366 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
1369 2018-08-27 Martin Liska <mliska@suse.cz>
1371 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
1372 cfun argument explicitly.
1373 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
1374 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
1375 function gimple_switch_default_bb.
1376 (convert_switch_statements):
1378 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
1379 * stmt.c (label_to_block_fn): Use label_to_block and pass
1380 cfun argument explicitly and use gimple_switch_label_bb.
1381 (expand_case): Likewise.
1382 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
1383 cfun argument explicitly. Likewise.
1384 (make_edges_bb): Likewise.
1385 (make_cond_expr_edges): Likewise.
1386 (get_cases_for_edge): Likewise.
1387 (make_gimple_switch_edges): Likewise.
1388 (label_to_block_fn): Likewise.
1389 (label_to_block): Likewise.
1390 (make_goto_expr_edges): Likewise.
1391 (make_gimple_asm_edges): Likewise.
1392 (main_block_label): Likewise.
1393 (group_case_labels_stmt): Likewise.
1394 (find_taken_edge_computed_goto): Likewise.
1395 (find_taken_edge_switch_expr): Likewise.
1396 (gimple_verify_flow_info): Likewise.
1397 (gimple_redirect_edge_and_branch): Likewise.
1398 (gimple_switch_label_bb): New function.
1399 (gimple_switch_default_bb): Likewise.
1400 (gimple_switch_edge): Likewise.
1401 (gimple_switch_default_edge): Likewise.
1402 * tree-cfg.h (label_to_block_fn): Remove and replace ...
1403 (label_to_block): ... with this.
1404 (gimple_switch_label_bb): New.
1405 (gimple_switch_default_bb): Likewise.
1406 (gimple_switch_edge): Likewise.
1407 (gimple_switch_default_edge): Likewise.
1408 * tree-cfgcleanup.c (convert_single_case_switch): Use
1409 new gimple functions and pass new argument to label_to_block.
1410 (cleanup_control_flow_bb):
1411 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
1412 cfun argument explicitly.
1413 (make_eh_edges): Likewise.
1414 (redirect_eh_dispatch_edge): Likewise.
1415 (lower_resx): Likewise.
1416 (lower_eh_dispatch): Likewise.
1417 (maybe_remove_unreachable_handlers): Likewise.
1418 (unsplit_eh): Likewise.
1419 (cleanup_empty_eh): Likewise.
1420 (verify_eh_edges): Likewise.
1421 (verify_eh_dispatch_edge): Likewise.
1422 * tree-ssa-dom.c (record_edge_info): Likewise.
1423 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
1424 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
1425 (thread_through_normal_block): Likewise.
1426 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
1427 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
1428 * tree-switch-conversion.c (switch_conversion::collect): Use new
1430 (switch_conversion::check_final_bb): Likewise.
1431 (switch_conversion::gather_default_values): Pass new argument
1433 (switch_conversion::build_constructors): Likewise.
1434 (switch_decision_tree::compute_cases_per_edge): Use new
1435 gimple_switch_edge function.
1436 (switch_decision_tree::analyze_switch_statement): Pass new argument
1438 (switch_decision_tree::try_switch_expansion): Use
1439 gimple_switch_default_edge.
1440 * tree-vrp.c (find_switch_asserts): Pass new argument
1442 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
1443 (vr_values::simplify_switch_using_ranges): Likewise.
1445 2018-08-27 Richard Biener <rguenther@suse.de>
1447 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
1448 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
1450 * tree-ssa-sccvn.h (struct vn_pval): New structure.
1451 (struct vn_nary_op_s): Add unwind_to member. Add
1452 predicated_values flag and put result into a union together
1453 with a linked list of vn_pval.
1454 (struct vn_ssa_aux): Add name member to make maintaining
1455 a map of SSA name to vn_ssa_aux possible. Remove no longer
1456 needed info, dfsnum, low, visited, on_sccstack, use_processed
1457 and range_info_anti_range_p members.
1458 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
1459 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
1461 (vn_valueize): New global.
1462 (vn_context_bb): Likewise.
1463 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
1464 VN_INFO_PTR_INFO): Remove.
1465 * tree-ssa-sccvn.c: ... (rewrite)
1466 (pass_fre::execute): For -O2+ initialize loops and run
1467 RPO VN in optimistic mode (iterating). For -O1 and -Og
1468 run RPO VN in non-optimistic mode.
1469 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
1470 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
1471 * doc/invoke.texi (sccvn-max-scc-size): Remove.
1472 (rpo-vn-max-loop-depth): Document.
1473 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
1474 when valuezing the VUSE signals we walked out of the region.
1475 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
1476 (phi_translate): Set VN context block to use for availability
1478 (compute_avail): Likewise.
1479 (pre_valueize): New function.
1480 (pass_pre::execute): Adjust to the RPO VN API.
1482 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
1483 (propagate_constants_for_unrolling): Remove.
1484 (tree_unroll_loops_completely): Perform value-numbering
1485 on the unrolled bodies loop parent.
1487 2018-08-27 Richard Biener <rguenther@suse.de>
1489 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
1490 for partial antic compute.
1492 2018-08-27 Jakub Jelinek <jakub@redhat.com>
1494 PR rtl-optimization/87065
1495 * combine.c (simplify_if_then_else): Formatting fix.
1496 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
1498 (known_cond): Don't return const_true_rtx for vector modes. Use
1499 CONST0_RTX instead of const0_rtx. Formatting fixes.
1501 2018-08-27 Martin Liska <mliska@suse.cz>
1503 PR gcov-profile/87069
1504 * gcov.c (process_file): Record files already processed
1505 and warn about a file being processed multiple times.
1507 2018-08-27 Martin Liska <mliska@suse.cz>
1510 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1511 Set default values for x_aarch64_*_string strings.
1512 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
1513 prefix. For -mabi do not print '=ABI' in help and use
1514 <option_value> format for -msve-vector-bits and -moverride
1517 2018-08-26 Jeff Law <law@redhat.com>
1519 * config/mips/frame-header-opt.c: Include "backend.h" rather than
1522 2018-08-26 Marek Polacek <polacek@redhat.com>
1524 PR c++/87029, Implement -Wredundant-move.
1525 * doc/invoke.texi: Document -Wredundant-move.
1527 2018-08-25 Martin Sebor <msebor@redhat.com>
1529 PR tree-optimization/87059
1530 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
1531 to the same type as the other.
1532 * fold-const.c (fold_binary_loc): Assert expectation.
1534 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
1536 * config/darwin.c (machopic_legitimize_pic_address): Clean up
1537 extraneous parentheses, dead code section and formatting.
1539 2018-08-24 David Malcolm <dmalcolm@redhat.com>
1542 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
1543 wide enough for jumps in the line-numbering to be visible.
1544 (layout::print_gap_in_line_numbering): New member function.
1545 (layout::calculate_line_spans): When using line numbering, merge
1546 line spans that are only 1 line apart.
1547 (diagnostic_show_locus): When printing line numbers, show gaps in
1548 line numbering directly, rather than printing headers.
1549 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
1550 line-numbering with multiple line spans.
1551 (selftest::test_fixit_insert_containing_newline_2): Add test of
1552 line-numbering, in which the spans are close enough to be merged.
1554 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
1556 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
1557 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
1558 * tree-vrp.c (range_is_nonnull): Remove.
1559 (range_includes_zero_p): Accept value_range instead of min/max.
1560 (extract_range_from_binary_expr_1): Do not early bail on
1562 Use range_includes_zero_p instead of range_is_nonnull.
1563 (extract_range_from_unary_expr): Use range_includes_zero_p instead
1564 of range_is_nonnull.
1565 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
1566 special case VR_ANTI_RANGE.
1567 (vrp_finalize): Same.
1568 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
1570 (range_is_nonnull): Remove.
1571 * vr-values.c (vrp_stmt_computes_nonzero): Use
1572 range_includes_zero_p instead of range_is_nonnull.
1573 (extract_range_basic): Pass value_range to range_includes_zero_p
1574 instead of range_is_nonnull.
1576 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
1578 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
1579 * emit-rtl.h (rtl_data): Remove return_bnd.
1580 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
1581 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
1582 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
1583 (POINTER_BOUNDS_MODE): Remove definition.
1584 (make_pointer_bounds_mode): Remove.
1585 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
1586 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
1587 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
1588 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
1589 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
1590 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
1591 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
1593 * config/i386/i386-modes.def (BND32, BND64): Remove.
1594 * config/i386/i386.c (dbx_register_map): Remove bound registers.
1595 (dbx64_register_map): Ditto.
1596 (svr4_dbx_register_map): Ditto.
1597 (indirect_thunk_bnd_needed): Remove.
1598 (indirect_thunks_bnd_used): Ditto.
1599 (indirect_return_bnd_needed): Ditto.
1600 (indirect_return_via_cx_bnd): Ditto.
1601 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
1602 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
1603 (output_indirect_thunk): Ditto. Remove need_prefix argument.
1604 (output_indirect_thunk_function): Remove handling of
1605 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
1606 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
1607 (ix86_save_reg): Remove handling of crtl->return_bnd.
1608 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
1609 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
1610 and UNSPEC_BNDLX_ADDR.
1611 (ix86_output_indirect_branch_via_reg): Remove handling of
1612 indirect_thunk_prefix_bnd.
1613 (ix86_output_indirect_branch_via_push): Ditto.
1614 (ix86_output_function_return): Ditto.
1615 (ix86_output_indirect_function_return): Ditto.
1616 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
1617 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
1618 (CALL_USED_REGISTERS): Ditto.
1619 (REG_ALLOC_ORDER): Update for removal of bound registers.
1620 (HI_REGISTER_NAMES): Ditto.
1621 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
1622 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
1623 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
1624 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
1625 (FIRST_PSEUDO_REG): Update.
1626 (BND): Remove mode iterator.
1627 * config/i386/predicates.md (bnd_mem_operator): Remove.
1629 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
1631 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
1634 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
1636 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
1637 the case in which the permute needs only a single element and
1638 repeats for every vector of the result. Extend that case to
1639 handle variable-length vectors.
1640 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
1642 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
1645 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
1646 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
1648 2018-08-24 Richard Biener <rguenther@suse.de>
1650 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
1651 bb_flags_allocated members.
1652 (auto_flag): New RAII class for allocating flags.
1653 (auto_edge_flag): New RAII class for allocating edge flags.
1654 (auto_bb_flag): New RAII class for allocating bb flags.
1655 * cfgloop.c (verify_loop_structure): Allocate temporary edge
1657 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
1658 in favor of temporarily allocated BB flag.
1659 * hsa-brig.c: Re-order includes.
1660 * hsa-dump.c: Likewise.
1661 * hsa-regalloc.c: Likewise.
1662 * print-rtl.c: Likewise.
1663 * profile-count.c: Likewise.
1665 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
1668 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
1671 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
1674 * wide-int-range.cc (wide_int_range_div): Do not ignore result
1675 from wide_int_range_multiplicative_op.
1677 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1679 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
1682 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
1684 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
1687 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
1689 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
1690 full GENERIC expression used for replacement.
1692 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
1694 * tree-vrp.c (abs_extent_range): Remove.
1695 (extract_range_into_wide_ints): Pass wide ints by reference.
1696 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
1697 Pass wide ints by reference in all calls to
1698 extract_range_into_wide_ints.
1699 * wide-int-range.cc (wide_int_range_div): New.
1700 * wide-int-range.h (wide_int_range_div): New.
1701 (wide_int_range_includes_zero_p): New.
1702 (wide_int_range_zero_p): New.
1704 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
1706 * config/aarch64/aarch64.md (arches): New enum.
1707 (arch): New enum attr.
1708 (arch_enabled): New attr.
1709 (enabled): Now uses arch_enabled only.
1710 (simd, sve, fp16): Removed attribute.
1711 (fp): Attr now defined in terms of 'arch'.
1712 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
1713 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
1714 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
1715 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
1716 attributes into 'arch'.
1717 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
1718 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
1719 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
1720 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
1721 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
1722 'simd' attribute into 'arch'.
1723 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
1724 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
1725 Convert use of 'fp' attribute to 'arch'.
1726 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
1727 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
1729 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
1730 (different modes) Merge 'fp' and 'simd' into 'arch'.
1731 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
1734 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
1736 PR rtl-optimization/87026
1737 * expmed.c (canonicalize_comparison): If we can no longer create
1738 pseudoregisters, don't.
1740 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
1743 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
1745 * config/arm/arm.c (speculation_barrier_libfunc): New static
1747 (arm_init_libfuncs): Initialize it.
1748 (arm_emit_speculation_barrier): New function.
1749 * config/arm/arm.md (speculation_barrier): Call
1750 arm_emit_speculation_barrier for architectures that do not have
1752 (speculation_barrier_insn): Only match on Armv7 or later.
1754 2018-08-23 Richard Biener <rguenther@suse.de>
1757 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
1760 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
1762 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
1763 of single-vector TBLs.
1764 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
1767 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
1770 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
1771 aarch64_evpc_tbl guard.
1773 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
1775 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
1778 2018-08-22 Martin Sebor <msebor@redhat.com>
1781 * tree-pretty-print.c (pretty_print_string): Add argument.
1782 (dump_generic_node): Call to pretty_print_string with string size.
1784 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
1786 PR rtl-optimization/86771
1787 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
1788 of two SETs into those two SETs, one to be placed at i2, if that SETs
1789 destination is modified between i2 and i3.
1791 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1793 PR tree-optimization/86725
1794 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
1796 (vect_analyze_scalar_cycles_1): Check it.
1798 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1800 PR tree-optimization/86725
1801 * tree-vect-loop.c (vect_is_simple_reduction): When treating
1802 an outer loop phi as a double reduction, make sure that the
1803 single user of the phi result is an inner loop phi.
1805 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1807 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
1808 grouped stores with gaps to a strided group.
1810 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1812 * tree-vect-stmts.c (get_group_load_store_type)
1813 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
1814 first statement in a group.
1816 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1818 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
1819 the sequence used in gcc/gcc.c.
1821 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1824 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
1827 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1829 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
1830 Darwin10-specific unwinder-shim.
1831 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
1832 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
1833 New to cater for Darwin10 Rosetta.
1835 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1837 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
1838 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
1840 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1845 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
1846 (dwarf2out_switch_text_section): Generate a local label for the
1847 second function sub-section and apply it as the second FDE start
1849 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
1850 second sub-section start.
1852 2018-08-22 Richard Biener <rguenther@suse.de>
1854 PR tree-optimization/86988
1855 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
1857 2018-08-22 Richard Biener <rguenther@suse.de>
1859 PR tree-optimization/86945
1860 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
1862 2018-08-22 Alexandre Oliva <oliva@adacore.com>
1864 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
1865 a comment about how uses of r2 for .sdata2 come about.
1867 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
1869 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
1871 2018-08-21 Marek Polacek <polacek@redhat.com>
1873 PR c++/86981, Implement -Wpessimizing-move.
1874 * doc/invoke.texi: Document -Wpessimizing-move.
1876 2018-08-21 Jan Hubicka <jh@suse.cz>
1878 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
1879 * tree.h (is_redundant_typedef): Remove.
1880 * dwarf2out.c (is_redundant_typedef): Turn into static function.
1882 2018-08-21 Jan Hubicka <jh@suse.cz>
1884 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
1887 2018-08-21 Tamar Christina <tamar.christina@arm.com>
1889 * expmed.c (extract_low_bits): Reject invalid subregs early.
1891 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
1894 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
1897 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
1899 * config/vxworks.h: Guard vxworks_asm_out_constructor and
1900 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
1901 * config/vxworks.c: Likewise.
1903 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
1905 * config/vxworks.c: Set targetm.have_ctors_dtors
1906 if HAVE_INITFINI_ARRAY_SUPPORT.
1907 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
1908 if HAVE_INITFINI_ARRAY_SUPPORT.
1910 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
1912 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
1913 default search path for VxWorks < 7.
1915 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
1917 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
1918 (get_format_string): Refer to c_getstr.
1920 2018-08-21 Tom de Vries <tdevries@suse.de>
1922 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
1923 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
1924 (debuginfo_early_stop): Declare.
1925 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
1926 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
1927 (debuginfo_early_stop): New function.
1928 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
1929 and debuginfo_early_stop.
1930 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
1931 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
1932 (general_init): Call debuginfo_early_init.
1933 (finalize): Call debuginfo_fini.
1934 (do_compile): Call debuginfo_init.
1935 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
1937 (@item -fdump-debug, @item -fdump-earlydebug): Add.
1939 2018-08-21 Tom de Vries <tdevries@suse.de>
1941 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
1942 flag_dump_noaddr and flag_dump_unnumbered.
1944 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
1946 * wide-int-range.cc (wide_int_range_abs): New.
1947 (wide_int_range_order_set): Rename from wide_int_range_min_max.
1948 * wide-int-range.h (wide_int_range_abs): New.
1949 (wide_int_range_min_max): New.
1950 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
1951 case to call wide_int_range_abs.
1952 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
1953 (extract_range_from_abs_expr): Delete.
1955 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
1958 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
1959 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
1962 2018-08-20 Nathan Sidwell <nathan@acm.org>
1963 Jeff Law <law@redhat.com>
1965 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
1966 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
1968 2018-08-20 David Malcolm <dmalcolm@redhat.com>
1971 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
1972 (decl_attributes): Likewise.
1973 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
1975 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
1976 * diagnostic-core.h (class auto_diagnostic_group): New class.
1977 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
1978 (diagnostic_report_diagnostic): Handle the first diagnostics within
1980 (emit_diagnostic): Add auto_diagnostic_group instance.
1982 (inform_n): Likewise.
1983 (warning): Likewise.
1984 (warning_at): Likewise.
1985 (warning_n): Likewise.
1986 (pedwarn): Likewise.
1987 (permerror): Likewise.
1989 (error_n): Likewise.
1990 (error_at): Likewise.
1992 (fatal_error): Likewise.
1993 (internal_error): Likewise.
1994 (internal_error_no_backtrace): Likewise.
1995 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
1996 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
1997 * diagnostic.h (struct diagnostic_context): Add fields
1998 "diagnostic_group_nesting_depth",
1999 "diagnostic_group_emission_count", "begin_group_cb",
2001 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
2002 Add auto_diagnostic_group instance(s).
2003 (find_explicit_erroneous_behavior): Likewise.
2004 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
2005 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
2006 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
2007 (gimplify_va_arg_expr): Likewise.
2008 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
2009 (HSA_SORRY_AT): Likewise.
2010 * ipa-devirt.c (compare_virtual_tables): Likewise.
2011 (warn_odr): Likewise.
2012 * multiple_target.c (expand_target_clones): Likewise.
2013 * opts-common.c (cmdline_handle_error): Likewise.
2014 * reginfo.c (globalize_reg): Likewise.
2015 * substring-locations.c (format_warning_n_va): Likewise.
2016 * tree-inline.c (expand_call_inline): Likewise.
2017 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
2018 * tree-ssa-loop-niter.c
2019 (do_warn_aggressive_loop_optimizations): Likewise.
2020 * tree-ssa-uninit.c (warn_uninit): Likewise.
2021 * tree.c (warn_deprecated_use): Likewise.
2023 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
2026 * config/i386/i386.md (eh_return): Always update EH return
2027 address in word_mode.
2029 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
2031 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
2032 TARGET_SPLIT_COMPLEX_ARG is defined.
2034 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
2036 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
2038 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
2041 * expr.c (expand_assignment): Assert that bitpos is positive.
2042 (store_field): Likewise
2043 (expand_expr_real_1): Make sure that bitpos is positive.
2044 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
2047 2018-08-20 Nathan Sidwell <nathan@acm.org>
2049 * Makefile.in (CPP_ID_DATA_H): Delete.
2050 (CPP_INTERNAL_H): Don't add it.
2051 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
2052 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
2054 2018-08-20 Richard Biener <rguenther@suse.de>
2056 PR tree-optimization/78655
2057 * tree-vrp.c (extract_range_from_binary_expr_1): Make
2058 pointer + offset nonnull if either operand is nonnull work.
2060 2018-08-20 Tom de Vries <tdevries@suse.de>
2062 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
2063 unless the referenced die describes the added property using
2064 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
2065 Otherwise, add a DW_AT_location to the referenced die.
2067 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
2070 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
2071 register_operand when calling ix86_set_reg_reg_cost.
2072 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
2073 Set *total to 0 for operands that satisfy x86_64_immediate_operand
2074 predicate and to 1 otherwise.
2076 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
2078 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
2079 emit a diagnostic that it is not supported and reset the option.
2080 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
2081 supported and consume the option. (ASM_FINAL_SPEC): New.
2083 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
2085 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
2088 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
2090 C-SKY port: Documentation
2092 * doc/extend.texi (C-SKY Function Attributes): New section.
2093 * doc/invoke.texi (Option Summary): Add C-SKY options.
2094 (C-SKY Options): New section.
2095 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
2097 2018-08-17 Jojo <jijie_rong@c-sky.com>
2098 Huibin Wang <huibin_wang@c-sky.com>
2099 Sandra Loosemore <sandra@codesourcery.com>
2100 Chung-Lin Tang <cltang@codesourcery.com>
2102 C-SKY port: Backend implementation
2104 * config/csky/*: New.
2105 * common/config/csky/*: New.
2107 2018-08-17 Jojo <jijie_rong@c-sky.com>
2108 Huibin Wang <huibin_wang@c-sky.com>
2109 Sandra Loosemore <sandra@codesourcery.com>
2110 Chung-Lin Tang <cltang@codesourcery.com>
2111 Andrew Jenner <andrew@codesourcery.com>
2113 C-SKY port: Configury
2115 * config.gcc (csky-*-*): New.
2116 * configure.ac: Add csky to targets for dwarf2 debug_line support.
2117 * configure: Regenerated.
2119 2018-08-17 David Malcolm <dmalcolm@redhat.com>
2121 * dump-context.h: Include "dumpfile.h".
2122 (dump_context::dump_printf_va): Convert final param from va_list
2123 to va_list *. Convert from ATTRIBUTE_PRINTF to
2124 ATTRIBUTE_GCC_DUMP_PRINTF.
2125 (dump_context::dump_printf_loc_va): Likewise.
2126 * dumpfile.c: Include "stringpool.h".
2127 (make_item_for_dump_printf_va): Delete.
2128 (make_item_for_dump_printf): Delete.
2129 (class dump_pretty_printer): New class.
2130 (dump_pretty_printer::dump_pretty_printer): New ctor.
2131 (dump_pretty_printer::emit_items): New member function.
2132 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
2134 (dump_pretty_printer::emit_item): New member function.
2135 (dump_pretty_printer::stash_item): New member function.
2136 (dump_pretty_printer::format_decoder_cb): New member function.
2137 (dump_pretty_printer::decode_format): New member function.
2138 (dump_context::dump_printf_va): Reimplement in terms of
2139 dump_pretty_printer.
2140 (dump_context::dump_printf_loc_va): Convert final param from va_list
2142 (dump_context::begin_scope): Reimplement call to
2143 make_item_for_dump_printf.
2144 (dump_printf): Update for change to dump_printf_va.
2145 (dump_printf_loc): Likewise.
2146 (selftest::test_capture_of_dump_calls): Convert "stmt" from
2147 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
2148 with %T, %E, and %G.
2149 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
2150 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
2151 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
2152 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
2153 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
2154 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
2155 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
2156 within a dump_printf_loc call to "%wu".
2157 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
2158 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
2159 missing space after "=".
2160 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
2161 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
2162 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
2163 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
2164 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
2165 duplicate "vectorized" from message.
2167 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
2169 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
2170 polyNxK_t element's TYPE_STRING_FLAG.
2172 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
2174 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
2175 (they were unnamed before). Fix comments.
2177 2018-08-17 Nathan Sidwell <nathan@acm.org>
2179 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
2181 2018-08-17 Richard Biener <rguenther@suse.de>
2183 PR tree-optimization/86841
2184 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
2186 2018-08-17 Martin Liska <mliska@suse.cz>
2188 * common.opt: Remove Warn, Init and Report for options with
2189 Ignore/Deprecated flag. Warning is done automatically for
2191 * config/i386/i386.opt: Likewise.
2192 * config/ia64/ia64.opt: Likewise.
2193 * config/rs6000/rs6000.opt: Likewise.
2194 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
2195 Remove usage of flag_check_pointer_bounds.
2196 * lto-wrapper.c (merge_and_complain): Do not handle
2197 OPT_fcheck_pointer_bounds.
2198 (append_compiler_options): Likewise.
2199 * opt-functions.awk: Do not handle Deprecated.
2200 * optc-gen.awk: Check that Var, Report and Init are not
2201 used for an option with Ignore/Deprecated flag.
2202 * opts-common.c (decode_cmdline_option): Do not report
2204 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
2206 * opts.h (struct cl_option): Remove cl_deprecated flag.
2207 (CL_ERR_DEPRECATED): Remove error enum value.
2209 2018-08-17 Richard Biener <rguenther@suse.de>
2212 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
2213 across a va-arg-pack using call adjust its return value accordingly.
2215 2018-08-16 Martin Sebor <msebor@redhat.com>
2217 PR tree-optimization/86853
2218 * gimple-ssa-sprintf.c (struct format_result): Rename member.
2219 (struct fmtresult): Add member and initialize it in ctors.
2220 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
2221 (format_string): Handle %S the same as %ls. Set MAYFAIL.
2222 (format_directive): Set POSUNDER4K when MAYFAIL is set.
2223 (parse_directive): Handle %C same as %c.
2224 (sprintf_dom_walker::compute_format_length): Adjust.
2225 (is_call_safe): Adjust.
2227 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
2229 * builtins.c (c_strlen): Add new parameter eltsize. Use it
2230 for determining how to count the elements.
2231 * builtins.h (c_strlen): Adjust prototype.
2232 * expr.c (string_constant): Add new parameter mem_size.
2233 Set *mem_size appropriately.
2234 * expr.h (string_constant): Adjust protoype.
2235 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
2236 * gimple-fold.h (get_range_strlen): Adjust prototype.
2237 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
2238 (format_string): Call get_string_length with eltsize.
2240 2018-08-16 David Malcolm <dmalcolm@redhat.com>
2242 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
2243 to emit the span, rather than setting it as the prefix.
2245 2018-08-16 David Malcolm <dmalcolm@redhat.com>
2247 * diagnostic-show-locus.c (layout::start_annotation_line): Add
2248 "margin_char" parameter, defaulting to space. Use it in place
2249 of pp_space for the initial part of the margin.
2250 (layout::print_leading_fixits): Use '+' when filling the margin
2251 of line-insertion fix-it hints.
2253 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2255 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
2258 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2260 * config/rs6000/altivec.md: Don't set length attribute to the default
2262 * config/rs6000/darwin.md: Ditto.
2263 * config/rs6000/dfp.md: Ditto.
2264 * config/rs6000/htm.md: Ditto.
2265 * config/rs6000/rs6000.md: Ditto.
2266 * config/rs6000/sync.md: Ditto.
2267 * config/rs6000/vsx.md: Ditto.
2269 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2271 * config/rs6000/altivec.md: Don't set length attribute to the default
2272 value, for branch instructions.
2273 * config/rs6000/darwin.md: Ditto.
2274 * config/rs6000/rs6000.md: Ditto.
2276 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2278 * config/rs6000/rs6000.md (length): Always define as const_int 4.
2279 (unnamed conditional branch define_insn): Set length to 4 or 8
2280 depending on offset.
2281 (<bd>_<mode>): Similar, for alternative 0.
2282 (<bd>tf_<mode>): Ditto.
2284 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2286 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
2288 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
2290 * doc/rtl.texi: Replace old RTX class names with new names.
2293 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
2295 * expmed.h (canonicalize_comparison): New declaration.
2296 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
2297 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
2298 * optabs.c (prepare_cmp_insn): Likewise.
2299 * rtl.h (unsigned_condition_p): New function which checks if a
2300 comparison operator is unsigned.
2302 2018-08-16 Nathan Sidwell <nathan@acm.org>
2304 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
2305 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
2307 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2310 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
2311 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
2312 (mov<mov>): ..this and enable unconditionally.
2314 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2316 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
2318 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
2320 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
2321 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
2322 "Common" with "Target".
2324 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
2326 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
2327 * doc/invoke.texi (mmitigate-rop): Remove.
2328 * config/i386/i386.c: Do not include "regrename.h".
2329 (ix86_rop_should_change_byte_p, reg_encoded_number)
2330 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
2332 (ix86_reorg): Remove call to ix86_mitigate_rop.
2333 * config/i386/i386.md (attr "modrm_class"): Remove.
2334 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
2335 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
2336 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
2338 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
2340 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
2341 allow folding of mergeh() and mergel() for the float and double types.
2342 (fold_mergehl_helper): Rework to handle building a permute tree
2345 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
2347 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
2350 2018-08-15 David Malcolm <dmalcolm@redhat.com>
2352 * common.opt (fdiagnostics-show-labels): New option.
2353 * diagnostic-show-locus.c (class layout_range): Add field
2355 (class layout): Add field "m_show_labels_p".
2356 (layout_range::layout_range): Add param "label" and use it to
2358 (make_range): Pass in NULL for new "label" param of layout_range's
2360 (layout::layout): Initialize m_show_labels_p.
2361 (layout::maybe_add_location_range): Pass in loc_range->m_label
2362 when constructing layout_range instances.
2363 (struct line_label): New struct.
2364 (layout::print_any_labels): New member function.
2365 (layout::print_line): Call it if label-printing is enabled.
2366 (selftest::test_one_liner_labels): New test.
2367 (selftest::test_diagnostic_show_locus_one_liner): Call it.
2368 * diagnostic.c (diagnostic_initialize): Initialize
2369 context->show_labels_p.
2370 * diagnostic.h (struct diagnostic_context): Add field
2372 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
2373 -fno-diagnostics-show-labels.
2374 * dwarf2out.c (gen_producer_string): Add
2375 OPT_fdiagnostics_show_labels to the ignored options.
2376 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
2378 (gcc_rich_location::maybe_add_expr): Likewise.
2379 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
2380 label" param, defaulting to NULL.
2381 (gcc_rich_location::add_expr): Add "label" param.
2382 (gcc_rich_location::maybe_add_expr): Likewise.
2383 (class text_range_label): New class.
2384 (class range_label_for_type_mismatch): New class.
2385 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
2386 of format_warning_va.
2387 (fmtwarn_n): Likewise for new params of format_warning_n_va.
2388 * lto-wrapper.c (merge_and_complain): Add
2389 OPT_fdiagnostics_show_labels to the "pick one setting" options.
2390 (append_compiler_options): Likewise to the dropped options.
2391 (append_diag_options): Likewise to the passed-on options.
2392 * opts.c (common_handle_option): Handle the new option.
2393 * selftest-diagnostic.c
2394 (test_diagnostic_context::test_diagnostic_context): Enable
2396 * substring-locations.c: Include "gcc-rich-location.h".
2397 (format_warning_n_va): Add "fmt_label" and "param_label" params
2398 and use them as appropriate.
2399 (format_warning_va): Add "fmt_label" and "param_label" params,
2400 passing them on to format_warning_n_va.
2401 (format_warning_at_substring): Likewise.
2402 (format_warning_at_substring_n): Likewise.
2403 * substring-locations.h (format_warning_va): Add "fmt_label" and
2404 "param_label" params.
2405 (format_warning_n_va): Likewise.
2406 (format_warning_at_substring): Likewise.
2407 (format_warning_at_substring_n): Likewise.
2408 * toplev.c (general_init): Initialize global_dc->show_labels_p.
2410 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
2413 * builtins.c (expand_builtin_memcmp): Do not expand the call
2414 when overflow is detected.
2416 2018-08-15 Martin Sebor <msebor@redhat.com>
2418 PR tree-optimization/71625
2419 * config/aarch64/aarch64-builtins.c
2420 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
2422 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
2424 * config/s390/s390.c (s390_reorg): Remove loop.
2426 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
2429 (darwin_function_switched_text_sections): Delete.
2430 * gcc/config/darwin.h
2431 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
2433 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
2436 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
2437 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
2438 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
2440 2018-08-15 Martin Liska <mliska@suse.cz>
2442 PR tree-optimization/86925
2443 * predict.c (expr_expected_value_1): When taking
2444 later predictor, assign also probability.
2445 Use fold_build2_initializer_loc in order to fold
2446 the expression in -frounding-math.
2448 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
2450 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
2452 (expand_vec_perm_1): Try the new method.
2454 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
2457 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
2458 Check whether lra_live_max_point is 0 before dividing.
2460 2018-08-14 Martin Sebor <msebor@redhat.com>
2462 PR tree-optimization/86650
2463 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
2464 (vrp_prop::check_mem_ref): Same.
2466 2018-08-13 Liu Hao <lh_mouse@126.com>
2468 * pretty-print.c (eat_esc_sequence): Swap the foreground and
2469 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
2470 and clear it thereafter, as it only works for DBCS.
2472 2018-08-13 Liu Hao <lh_mouse@126.com>
2474 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
2475 handle returned by _get_osf_handle().
2477 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2479 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
2480 for folding vec_perm.
2482 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2484 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
2485 Add support for gimple-folding of vec_pack() and vec_unpack()
2488 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2490 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
2491 vec_xst variants to the list.
2492 (rs6000_gimple_fold_builtin): Add support for folding unaligned
2493 vector loads and stores.
2495 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
2497 * config.gcc (rs6000-ibm-aix4.x): Delete.
2498 (rs6000-ibm-aix5.1): Delete.
2499 (rs6000-ibm-aix5.2): Delete.
2500 (rs6000-ibm-aix5.3): Delete.
2501 * config/rs6000/aix43.h: Delete.
2502 * config/rs6000/aix51.h: Delete.
2503 * config/rs6000/aix52.h: Delete.
2504 * config/rs6000/t-aix43: Delete.
2506 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
2508 * config/s390/s390.c (s390_decompose_constant_pool_ref):
2510 (s390_decompose_address): Factor out constant pool ref
2513 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2515 * config/nds32/nds32-predicates.c
2516 (nds32_can_use_bclr_p): Change return type as bool.
2517 (nds32_can_use_bset_p): Ditto.
2518 (nds32_can_use_btgl_p): Ditto.
2519 (nds32_can_use_bitci_p): Ditto.
2520 * config/nds32/nds32-protos.h
2521 (nds32_can_use_bclr_p): Change declaration.
2522 (nds32_can_use_bset_p): Ditto.
2523 (nds32_can_use_btgl_p): Ditto.
2524 (nds32_can_use_bitci_p): Ditto.
2526 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2528 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
2529 Support -msched-prolog-epilog option.
2530 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
2532 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2534 * common/config/nds32/nds32-common.c
2535 (nds32_option_optimization_table): Enalbe -malways-align.
2537 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2539 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
2541 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
2542 OPT_misr_secure_ case.
2543 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
2544 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
2545 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
2547 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
2548 (nds32_isr_info): New field security_level.
2549 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
2550 * config/nds32/nds32.md (return_internal): Consider critical attribute.
2551 * config/nds32/nds32.opt (misr-secure): New option.
2552 * config/nds32/nds32_init.inc: New file.
2553 * config/nds32/nds32_isr.h: New file.
2555 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
2557 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
2558 Update comment for atomic instructions.
2559 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
2560 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
2562 (atomic_loaddi): Revise fence expansion to only emit fence prior to
2563 load for __ATOMIC_SEQ_CST model.
2564 (atomic_loaddi_1): Remove float register target.
2565 (atomic_storedi): Handle CONST_INT values.
2566 (atomic_storedi_1): Remove float register source. Add special case
2568 (memory_barrier): New expander and insn.
2570 2018-08-11 Jakub Jelinek <jakub@redhat.com>
2572 PR tree-optimization/86835
2573 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
2574 new_stmt after def_gsi, make sure to insert new_square_stmt after
2575 that stmt, not 2 stmts before it.
2577 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
2580 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
2583 2018-08-10 Martin Liska <mliska@suse.cz>
2586 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
2588 * builtins.c (expand_builtin_expect_with_probability):
2590 (expand_builtin_expect_with_probability): New function.
2591 (build_builtin_expect_predicate): Add new argumnet probability
2592 for BUILT_IN_EXPECT_WITH_PROBABILITY.
2593 (fold_builtin_expect):
2596 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
2597 * builtins.h (fold_builtin_expect): Set new argument.
2598 * doc/extend.texi: Document __builtin_expect_with_probability.
2599 * doc/invoke.texi: Likewise.
2600 * gimple-fold.c (gimple_fold_call): Pass new argument.
2601 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
2602 also BUILT_IN_EXPECT_WITH_PROBABILITY.
2603 * predict.c (get_predictor_value): New function.
2604 (expr_expected_value): Add new argument probability. Assume
2605 that predictor and probability are always non-null.
2606 (expr_expected_value_1): Likewise. For __builtin_expect and
2607 __builtin_expect_with_probability set probability. Handle
2608 combination in binary expressions.
2609 (tree_predict_by_opcode): Simplify code by simply calling
2610 get_predictor_value.
2611 (pass_strip_predict_hints::execute): Add handling of
2612 BUILT_IN_EXPECT_WITH_PROBABILITY.
2613 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
2615 * tree.h (DECL_BUILT_IN_P): New function.
2617 2018-08-10 Martin Liska <mliska@suse.cz>
2619 PR tree-optimization/85799
2620 * passes.def: Add argument for pass_strip_predict_hints.
2621 * predict.c (class pass_strip_predict_hints): Add new argument
2623 (strip_predictor_early): New function.
2624 (pass_strip_predict_hints::execute): Call the function to
2626 (strip_predict_hints): New function.
2627 * predict.def: Fix comment.
2629 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
2631 * Makefile.in: Clarify which tm.texi to copy over to assert the
2632 right to grant a GFDL license for all.
2634 2018-08-09 Jeff Law <law@redhat.com>
2636 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
2639 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
2641 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
2644 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
2646 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
2647 reductions for variable-length vectors.
2649 2018-08-09 David Malcolm <dmalcolm@redhat.com>
2652 * common.opt (fdiagnostics-show-line-numbers): New option.
2653 * diagnostic-show-locus.c (class layout): Add fields
2654 "m_show_line_numbers_p" and "m_linenum_width";
2655 (num_digits): New function.
2656 (test_num_digits): New function.
2657 (layout::layout): Initialize new fields. Update m_x_offset
2658 logic to handle any left margin.
2659 (layout::print_source_line): Print line number when requested.
2660 (layout::start_annotation_line): New member function.
2661 (layout::print_annotation_line): Call it.
2662 (layout::print_leading_fixits): Likewise.
2663 (layout::print_trailing_fixits): Likewise. Update calls to
2664 move_to_column for new parameter.
2665 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
2666 it to potentially call start_annotation_line.
2667 (layout::show_ruler): Call start_annotation_line.
2668 (selftest::test_line_numbers_multiline_range): New selftest.
2669 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
2670 and selftest::test_line_numbers_multiline_range.
2671 * diagnostic.c (diagnostic_initialize): Initialize
2672 show_line_numbers_p.
2673 * diagnostic.h (struct diagnostic_context): Add field
2674 "show_line_numbers_p".
2675 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
2676 -fno-diagnostics-show-line-numbers.
2677 * dwarf2out.c (gen_producer_string): Add
2678 OPT_fdiagnostics_show_line_numbers to the ignored options.
2679 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
2680 one setting" options.
2681 (append_compiler_options): Likewise to the dropped options.
2682 (append_diag_options): Likewise to the passed-on options.
2683 * opts.c (common_handle_option): Handle the new option.
2684 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
2686 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
2688 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
2689 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
2690 third argument of type "const signed char" to descriptions of
2691 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
2692 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
2693 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
2694 __builtin_bcdsub_ov functions.
2696 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
2698 PR tree-optimization/86858
2699 * tree-vect-loop.c (vect_is_simple_reduction): Restore
2700 flow_bb_inside_loop_p calls.
2702 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
2704 PR tree-optimization/86871
2705 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
2706 instead of gimple_assign_lhs.
2708 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
2711 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
2712 register constraint to operand 0.
2713 (add<mode>3_carryinC): Likewise.
2714 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
2716 2018-08-09 Martin Liska <mliska@suse.cz>
2719 * common.opt: Remove extra line.
2721 2018-08-09 Martin Liska <mliska@suse.cz>
2723 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
2724 at the end of a line, make first letter capital and end up
2725 a sentence with a dot.
2726 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
2727 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
2728 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
2729 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
2730 (PARAM_MAX_ISL_OPERATIONS): Likewise.
2731 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
2732 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
2733 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
2734 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
2735 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
2736 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
2737 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
2738 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
2739 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
2740 (PARAM_TREE_REASSOC_WIDTH): Likewise.
2741 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
2742 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
2743 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
2745 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
2748 * config/s390/s390.c (s390_option_override_internal): Reduce the
2749 stack-clash-protection-probe-interval param if it would be too big
2752 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
2755 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
2756 * config/m68k/m68k.c (handle_move_double): Don't call
2757 m68k_final_prescan_insn.
2758 (m68k_adjust_decorated_operand): Renamed from
2759 m68k_final_prescan_insn, remove first and third operand and
2761 (print_operand): Call it.
2762 (print_operand_address): Call it.
2764 2018-08-08 Nathan Sidwell <nathan@acm.org>
2766 * diagnostic.c (diagnostic_report_current_module): Use
2767 linemap_included_from & linemap_included_from_linemap.
2769 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
2771 * config/aarch64/aarch64-cores.def: Add phecda core.
2772 * config/aarch64/aarch64-tune.md: Regenerate.
2773 * doc/invoke.texi: Add phecda core.
2775 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
2778 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
2780 * config/s390/s390.md ("movti"): Add more alternatives for
2781 constant to GPR copies.
2783 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
2785 * config/s390/s390.c: Fix whitespace damage throughout the file.
2786 * config/s390/s390.h: Likewise.
2787 * config/s390/tpf.h: Likewise.
2789 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
2791 * config/s390/s390.c (s390_loadrelative_operand_p):
2792 Remove TARGET_CPU_ZARCH usages.
2793 (s390_rtx_costs): Likewise.
2794 (s390_legitimate_constant_p): Likewise.
2795 (s390_cannot_force_const_mem): Likewise.
2796 (legitimate_reload_constant_p): Likewise.
2797 (s390_preferred_reload_class): Likewise.
2798 (legitimize_pic_address): Likewise.
2799 (legitimize_tls_address): Likewise.
2800 (s390_split_branches): Removed.
2801 (s390_add_execute): Removed.
2802 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
2803 (s390_mainpool_start): Likewise.
2804 (s390_mainpool_finish): Likewise.
2805 (s390_mainpool_cancel): Removed.
2806 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
2807 (s390_chunkify_cancel): Likewise.
2808 (s390_return_addr_rtx): Likewise.
2809 (s390_register_info): Remove split_branches_pending_p uages.
2810 (s390_optimize_register_info): Likewise.
2811 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
2812 split_branches_pending_p usages.
2813 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
2814 (s390_load_got): Likewise.
2815 (s390_expand_split_stack_prologue): Likewise.
2816 (output_asm_nops): Likewise.
2817 (s390_function_profiler): Likewise.
2818 (s390_emit_call): Likewise.
2819 (s390_conditional_register_usage): Likewise.
2820 (s390_optimize_prologue): Likewise.
2821 (s390_reorg): Remove TARGET_CPU_ZARCH and
2822 split_branches_pending_p usages.
2823 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
2825 (s390_output_indirect_thunk_function): Likewise.
2826 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
2827 (TARGET_CPU_ZARCH_P): Removed.
2828 (struct machine_function): Remove split_branches_pending_p.
2829 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
2831 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
2833 * common/config/s390/s390-common.c (processor_flags_table):
2835 * config.gcc: Remove with_arch/with_tune support.
2836 * config/s390/2064.md: Remove cpu attribute comparisons.
2837 * config/s390/driver-native.c (s390_host_detect_local_cpu):
2839 * config/s390/linux.h (ASM_SPEC):
2840 Remove -march support.
2841 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
2842 Use a table to get an arch level.
2843 * config/s390/s390-opts.h (enum processor_type):
2845 * config/s390/s390.c
2846 (processor_table): Remove entries, add arch_level values.
2847 (s390_issue_rate): Remove cases.
2848 (s390_option_override): Adjust
2849 s390_option_override_internal() call.
2850 (s390_option_override_internal): Remove deprecation warning.
2851 (s390_valid_target_attribute_tree): Adjust
2852 s390_option_override_internal() call.
2853 * config/s390/s390.h (struct s390_processor):
2854 Share with s390-c.c, add arch_level field.
2855 * config/s390/s390.md:
2856 Remove occurrences in cpu attribute.
2857 * config/s390/s390.opt: Remove -march/-mtune support.
2858 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
2859 * doc/invoke.texi: Remove deprecation warning.
2861 2018-08-08 Luis Machado <luis.machado@linaro.org>
2863 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
2865 (qdf24xx_tunings): Set vector cost structure to
2866 qdf24xx_vector_cost.
2868 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
2869 <register_sextend>: Set to 3.
2871 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
2874 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
2875 * config/aarch64/aarch64-simd.md
2876 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
2877 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
2878 (aarch64_frecpx<mode>): ...this new pattern.
2879 * config/aarch64/aarch64-simd-builtins.def: Remove comment
2880 about aarch64_frecp<FRECP:frecp_suffix><mode>.
2882 2018-08-07 Martin Liska <mliska@suse.cz>
2885 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
2886 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
2887 * predict.def (PRED_MALLOC_NONNULL): New predictor.
2888 * doc/extend.texi: Document that malloc attribute adds
2891 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
2894 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2895 Define to speculation_safe_value_not_needed.
2897 2018-08-06 Jeff Law <law@redhat.com>
2899 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
2900 the vr_values instance to cprop_into_stmt.
2901 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
2902 (cprop_operand): Also query EVRP to determine if OP is a constant.
2904 2018-08-06 Nathan Sidwell <nathan@acm.org>
2906 * diagnostic.c (diagnostic_report_current_module): Reroll
2907 included-at loop. Translate text.
2909 2018-08-06 David Malcolm <dmalcolm@redhat.com>
2911 * function-tests.c (selftest::test_expansion_to_rtl): Call
2912 free_after_compilation.
2914 2018-08-06 Alan Hayward <alan.hayward@arm.com>
2916 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
2918 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
2920 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
2921 loops with memory block operations from getting unrolled.
2923 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
2926 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2927 Define to speculation_safe_value_not_needed.
2929 2018-08-06 Jeff Law <law@redhat.com>
2931 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
2934 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
2937 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
2938 with all enabled __intN types.
2940 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
2942 2018-08-06 Alan Hayward <alan.hayward@arm.com>
2944 * alias.c (record_set): Check for clobber high.
2945 * cfgexpand.c (expand_gimple_stmt): Likewise.
2946 * combine-stack-adj.c (single_set_for_csa): Likewise.
2947 * combine.c (find_single_use_1): Likewise.
2948 (set_nonzero_bits_and_sign_copies): Likewise.
2949 (get_combine_src_dest): Likewise.
2950 (is_parallel_of_n_reg_sets): Likewise.
2951 (try_combine): Likewise.
2952 (record_dead_and_set_regs_1): Likewise.
2953 (reg_dead_at_p_1): Likewise.
2954 (reg_dead_at_p): Likewise.
2955 * dce.c (deletable_insn_p): Likewise.
2956 (mark_nonreg_stores_1): Likewise.
2957 (mark_nonreg_stores_2): Likewise.
2958 * df-scan.c (df_find_hard_reg_defs): Likewise.
2959 (df_uses_record): Likewise.
2960 (df_get_call_refs): Likewise.
2961 * dwarf2out.c (mem_loc_descriptor): Likewise.
2962 * haifa-sched.c (haifa_classify_rtx): Likewise.
2963 * ira-build.c (create_insn_allocnos): Likewise.
2964 * ira-costs.c (scan_one_insn): Likewise.
2965 * ira.c (equiv_init_movable_p): Likewise.
2966 (rtx_moveable_p): Likewise.
2967 (interesting_dest_for_shprep): Likewise.
2968 * jump.c (mark_jump_label_1): Likewise.
2969 * postreload-gcse.c (record_opr_changes): Likewise.
2970 * postreload.c (reload_cse_simplify): Likewise.
2971 (struct reg_use): Add source expr.
2972 (reload_combine): Check for clobber high.
2973 (reload_combine_note_use): Likewise.
2974 (reload_cse_move2add): Likewise.
2975 (move2add_note_store): Likewise.
2976 * print-rtl.c (print_pattern): Likewise.
2977 * recog.c (decode_asm_operands): Likewise.
2978 (store_data_bypass_p): Likewise.
2979 (if_test_bypass_p): Likewise.
2980 * regcprop.c (kill_clobbered_value): Likewise.
2981 (kill_set_value): Likewise.
2982 * reginfo.c (reg_scan_mark_refs): Likewise.
2983 * reload1.c (maybe_fix_stack_asms): Likewise.
2984 (eliminate_regs_1): Likewise.
2985 (elimination_effects): Likewise.
2986 (mark_not_eliminable): Likewise.
2987 (scan_paradoxical_subregs): Likewise.
2988 (forget_old_reloads_1): Likewise.
2989 * reorg.c (find_end_label): Likewise.
2990 (try_merge_delay_insns): Likewise.
2991 (redundant_insn): Likewise.
2992 (own_thread_p): Likewise.
2993 (fill_simple_delay_slots): Likewise.
2994 (fill_slots_from_thread): Likewise.
2995 (dbr_schedule): Likewise.
2996 * resource.c (update_live_status): Likewise.
2997 (mark_referenced_resources): Likewise.
2998 (mark_set_resources): Likewise.
2999 * rtl.c (copy_rtx): Likewise.
3000 * rtlanal.c (reg_referenced_p): Likewise.
3001 (single_set_2): Likewise.
3002 (noop_move_p): Likewise.
3003 (note_stores): Likewise.
3004 * sched-deps.c (sched_analyze_reg): Likewise.
3005 (sched_analyze_insn): Likewise.
3007 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3009 * cse.c (invalidate_reg): New function extracted from...
3010 (invalidate): ...here.
3011 (canonicalize_insn): Check for clobber high.
3012 (invalidate_from_clobbers): invalidate clobber highs.
3013 (invalidate_from_sets_and_clobbers): Likewise.
3014 (count_reg_usage): Check for clobber high.
3015 (insn_live_p): Likewise.
3016 * cselib.c (cselib_expand_value_rtx_1):Likewise.
3017 (cselib_invalidate_regno): Check for clobber in setter.
3018 (cselib_invalidate_rtx): Pass through setter.
3019 (cselib_invalidate_rtx_note_stores):
3020 (cselib_process_insn): Check for clobber high.
3021 * cselib.h (cselib_invalidate_rtx): Add operand.
3023 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3025 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
3026 (mark_not_eliminable): Likewise.
3027 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
3028 * lra-lives.c (process_bb_lives): Check for clobber high.
3029 * lra.c (new_insn_reg): Remember clobber highs.
3030 (collect_non_operand_hard_regs): Check for clobber high.
3031 (lra_set_insn_recog_data): Likewise.
3032 (add_regs_to_insn_regno_info): Likewise.
3033 (lra_update_insn_regno_info): Likewise.
3035 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3037 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
3038 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
3040 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3042 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
3043 (copy_insn_1): Likewise.
3044 (gen_hard_reg_clobber_high): New gen function.
3045 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
3046 * genemit.c (gen_exp): Likewise.
3047 (gen_emit_seq): Pass through info.
3048 (gen_insn): Check for CLOBBER_HIGH.
3049 (gen_expand): Pass through info.
3050 (gen_split): Likewise.
3051 (output_add_clobbers): Likewise.
3052 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
3053 (remove_clobbers): Likewise.
3054 * rtl.h (gen_hard_reg_clobber_high): New declaration.
3056 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3058 * doc/rtl.texi (clobber_high): Add.
3059 (parallel): Add in clobber high
3060 * rtl.c (rtl_check_failed_code3): Add function.
3061 * rtl.def (CLOBBER_HIGH): Add expression.
3062 * rtl.h (RTL_CHECKC3): Add macro.
3063 (rtl_check_failed_code3): Add declaration.
3064 (XC3EXP): Add macro.
3066 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
3069 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
3070 cfun->machine->max_used_stack_alignment if needed.
3072 2018-08-04 Martin Sebor <msebor@redhat.com>
3074 PR tree-optimization/86571
3075 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
3078 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
3081 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3084 2018-08-03 Jeff Law <law@redhat.com>
3087 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3088 Define to speculation_safe_value_not_needed.
3090 2018-08-03 David Malcolm <dmalcolm@redhat.com>
3092 * doc/gcov.texi (-x): Remove duplicate "to".
3093 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
3094 (-Wif-not-aligned): Remove duplicate "is".
3095 (-flto): Remove duplicate "the".
3096 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
3097 duplicate "v5.00.b".
3098 (MSP430 Options): Remove duplicate "and" from the description
3099 of "-mgprel-sec=regexp".
3100 (x86 Options): Remove duplicate copies of "vmldLog102" and
3101 vmlsLog104 from description of "-mveclibabi=type".
3103 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
3105 * internal-fn.h (first_commutative_argument): Declare.
3106 * internal-fn.c (first_commutative_argument): New function.
3107 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
3108 restrictions for pattern statements. Use first_commutative_argument
3109 to look for commutative operands in calls to internal functions.
3111 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
3113 * Makefile.in (wide-int-range.o): New.
3114 * tree-vrp.c: Move all the wide_int_* functions to...
3115 * wide-int-range.cc: ...here.
3116 * tree-vrp.h: Move all the wide_int_* prototypes to...
3117 * wide-int-range.h: ...here.
3119 2018-08-03 Tom de Vries <tdevries@suse.de>
3121 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
3123 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
3124 * except.c (output_function_exception_table): Do early exit if
3125 targetm_common.except_unwind_info (&global_options) == UI_NONE.
3127 2018-08-03 Martin Liska <mliska@suse.cz>
3129 * predict.c (dump_prediction): Change to 2 digits
3132 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
3134 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
3136 (falkor_am_1_gtov_gtov): ... a new insn reservation.
3138 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
3140 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
3141 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
3142 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
3143 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
3144 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
3146 2018-08-02 David Malcolm <dmalcolm@redhat.com>
3148 * diagnostic-show-locus.c (diagnostic_show_locus): Use
3149 pp_take_prefix when saving the existing prefix.
3150 * diagnostic.c (diagnostic_append_note): Likewise.
3151 * langhooks.c (lhd_print_error_function): Likewise.
3152 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
3153 param's type. Free the existing prefix.
3154 (pp_take_prefix): New function.
3155 (pretty_printer::pretty_printer): Drop the prefix parameter.
3156 Rename the length parameter to match the comment.
3157 (pretty_printer::~pretty_printer): Free the prefix.
3158 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
3160 (struct pretty_printer): Drop the "const" from "prefix" field's
3161 type and clarify memory management.
3162 (pp_set_prefix): Drop the "const" from the 2nd param.
3163 (pp_take_prefix): New decl.
3165 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
3167 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
3168 for word_mode_ok here instead of passing as argument.
3169 (expand_block_compare): Change select_block_compare_mode() call.
3170 (expand_strncmp_gpr_sequence): New function.
3171 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
3173 2018-08-02 Jeff Law <law@redhat.com>
3176 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3177 Define to speculation_safe_value_not_needed.
3180 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3181 Define to speculation_safe_value_not_needed.
3183 2018-08-02 Tom de Vries <tdevries@suse.de>
3186 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
3187 function. Return UI_TARGET unconditionally.
3188 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
3189 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
3191 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3193 * genemit.c (print_overload_test): Fix typo.
3195 2018-08-02 Richard Biener <rguenther@suse.de>
3197 PR tree-optimization/86816
3198 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
3199 which checks for value availability before querying it.
3200 (gvn_uses_equal): Use it.
3201 (same_succ_hash): Likewise.
3202 (gimple_equal_p): Likewise.
3204 2018-08-02 Nick Clifton <nickc@redhat.com>
3207 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3208 Define to speculation_safe_value_not_needed.
3211 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3212 Define to speculation_safe_value_not_needed.
3215 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3216 Define to speculation_safe_value_not_needed.
3219 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3220 Define to speculation_safe_value_not_needed.
3223 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3224 Define to speculation_safe_value_not_needed.
3227 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3228 Define to speculation_safe_value_not_needed.
3231 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3232 Define to speculation_safe_value_not_needed.
3235 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3236 Define to speculation_safe_value_not_needed.
3239 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
3240 speculation_safe_value_not_needed.
3243 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
3244 to speculation_safe_value_not_needed.
3246 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3248 * doc/md.texi: Expand the documentation of instruction names
3249 to mention port-local uses. Document '@' in pattern names.
3250 * read-md.h (overloaded_instance, overloaded_name): New structs.
3252 (md_reader::handle_overloaded_name): New member function.
3253 (md_reader::get_overloads): Likewise.
3254 (md_reader::m_first_overload): New member variable.
3255 (md_reader::m_next_overload_ptr): Likewise.
3256 (md_reader::m_overloads_htab): Likewise.
3257 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
3258 m_next_overload_ptr and m_overloads_htab.
3259 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
3260 (get_mode_token, get_code_token, get_int_token): New functions.
3261 (map_attr_string): Add an optional argument that passes back
3262 the associated iterator.
3263 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
3264 (md_reader::handle_overloaded_name, add_overload_instance): New
3266 (apply_iterators): Handle '@' names. Report an error if '@'
3267 is used without iterators.
3268 (initialize_iterators): Initialize the new iterator_group fields.
3269 * genopinit.c (handle_overloaded_code_for)
3270 (handle_overloaded_gen): New functions.
3271 (main): Use them to print declarations of maybe_code_for_* and
3272 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
3273 * genemit.c (print_overload_arguments, print_overload_test)
3274 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
3275 (main): Use it to print definitions of maybe_code_for_* and
3276 maybe_gen_* functions.
3277 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
3278 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
3279 instead of explicit mode checks.
3280 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
3281 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
3282 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
3283 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
3284 (aarch64_expand_compare_and_swap): Likewise
3285 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
3286 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
3287 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
3288 (aarch64_constant_pool_reload_icode): Delete.
3289 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
3290 instead of aarch64_constant_pool_reload_icode. Use
3291 code_for_aarch64_reload_mov instead of explicit mode checks.
3292 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
3293 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
3294 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
3295 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
3296 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
3297 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
3298 (aarch64_atomic_load_op_code): Delete.
3299 (aarch64_emit_atomic_load_op): Likewise.
3300 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
3301 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
3302 instead of aarch64_emit_atomic_load_op.
3303 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
3304 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
3305 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
3306 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
3307 character before the pattern name.
3308 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
3309 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
3310 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
3311 (aarch64_frecps<mode>): Likewise.
3312 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
3313 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
3314 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
3315 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
3316 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
3318 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3320 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
3321 Allow HFmode constants if TARGET_FP_F16INST.
3323 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
3326 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
3327 No longer check last store for clobber of address register.
3329 2018-08-02 Martin Liska <mliska@suse.cz>
3331 PR gcov-profile/86817
3332 * gcov.c (process_all_functions): New function.
3334 (process_file): Move functions processing to
3335 process_all_functions.
3337 2018-08-02 David Malcolm <dmalcolm@redhat.com>
3339 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
3340 "const" to the "gimple *" and "rtx_insn *" parameters.
3341 * dumpfile.h (dump_user_location_t::dump_user_location_t):
3343 (dump_location_t::dump_location_t): Likewise.
3345 2018-08-01 Martin Sebor <msebor@redhat.com>
3347 PR tree-optimization/86650
3348 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
3349 rather than a "gcall *". Directly pass the data of interest
3350 to percent_K_format, rather than building a temporary CALL_EXPR
3352 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
3353 (gimple_fold_builtin_strncat): Adjust.
3354 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
3355 gcall* argument with gimple*.
3356 * gimple-ssa-warn-restrict.c (check_call): Same.
3357 (wrestrict_dom_walker::before_dom_children): Same.
3358 (builtin_access::builtin_access): Same.
3359 (check_bounds_or_overlap): Same
3360 (maybe_diag_overlap): Same.
3361 (maybe_diag_offset_bounds): Same.
3362 * tree-diagnostic.c (default_tree_printer): Move usage of
3363 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
3365 * tree-pretty-print.c (percent_K_format): Add argument.
3366 * tree-pretty-print.h: Add argument.
3367 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
3368 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
3369 (maybe_diag_stxncpy_trunc): Same.
3370 (handle_builtin_stxncpy): Same.
3371 (handle_builtin_strcat): Same.
3373 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3375 * match.pd: Optimise pointer range checks.
3377 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3379 PR tree-optimization/86758
3380 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
3381 to remove pattern statements.
3383 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3385 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
3386 result of dfs_enumerate_from when constructing stmt_vec_infos,
3387 instead of additionally calling get_loop_body.
3389 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3391 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
3393 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
3394 When creating an iv, assert that the step is not known to be zero.
3395 (vect_setup_realignment): Update call accordingly.
3396 * tree-vect-stmts.c (vectorizable_store): Likewise.
3397 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
3399 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3401 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
3402 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
3403 (vectorizable_reduction): Likewise.
3404 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
3405 (vect_detect_hybrid_slp_stmts): Likewise.
3406 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
3408 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
3410 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
3411 (wide_int_set_zero_nonzero_bits): ...this.
3412 (zero_nonzero_bits_from_vr): Rename to...
3413 (vrp_set_zero_nonzero_bits): ...this.
3414 (extract_range_from_multiplicative_op_1): Abstract wide int
3416 (wide_int_range_multiplicative_op): ...here.
3417 (extract_range_from_binary_expr_1): Extract wide int binary
3418 operations into their own functions.
3419 (wide_int_range_lshift): New.
3420 (wide_int_range_can_optimize_bit_op): New.
3421 (wide_int_range_shift_undefined_p): New.
3422 (wide_int_range_bit_xor): New.
3423 (wide_int_range_bit_ior): New.
3424 (wide_int_range_bit_and): New.
3425 (wide_int_range_trunc_mod): New.
3426 (extract_range_into_wide_ints): New.
3427 (vrp_shift_undefined_p): New.
3428 (extract_range_from_multiplicative_op): New.
3429 (vrp_can_optimize_bit_op): New.
3430 * tree-vrp.h (value_range::dump): New.
3431 (wide_int_range_multiplicative_op): New.
3432 (wide_int_range_lshift):New.
3433 (wide_int_range_shift_undefined_p): New.
3434 (wide_int_range_bit_xor): New.
3435 (wide_int_range_bit_ior): New.
3436 (wide_int_range_bit_and): New.
3437 (wide_int_range_trunc_mod): New.
3438 (zero_nonzero_bits_from_bounds): Rename to...
3439 (wide_int_set_zero_nonzero_bits): ...this.
3440 (zero_nonzero_bits_from_vr): Rename to...
3441 (vrp_set_zero_nonzero_bits): ...this.
3442 (range_easy_mask_min_max): Rename to...
3443 (wide_int_range_can_optimize_bit_op): this.
3444 * vr-values.c (simplify_bit_ops_using_ranges): Rename
3445 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
3447 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3449 * tree-vectorizer.h (vect_orig_stmt): New function.
3450 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
3451 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
3452 (vect_create_epilog_for_reduction): Likewise.
3453 (vectorizable_live_operation): Likewise.
3454 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
3455 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
3456 * tree-vect-stmts.c (vectorizable_call): Likewise.
3457 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
3459 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3461 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
3463 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
3464 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
3465 (vect_transform_loop): Likewise.
3466 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3468 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3470 * tree-vectorizer.h (vect_schedule_slp): Return void.
3471 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3472 (vect_schedule_slp): Likewise.
3474 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3476 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
3478 (vect_transform_loop): Update calls accordingly. Schedule SLP
3479 instances before the main loop, if any exist.
3481 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3483 PR tree-optimization/86749
3484 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
3485 If the lhs is used in a COND_EXPR, check that it is being used
3486 as the "then" or "else" value.
3488 2018-08-01 Tom de Vries <tdevries@suse.de>
3491 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
3492 speculation_safe_value_not_needed.
3494 2018-08-01 Richard Biener <rguenther@suse.de>
3496 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
3499 2018-08-01 Martin Liska <mliska@suse.cz>
3501 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
3502 format how successful transformation is dumped.
3503 (gimple_mod_pow2_value_transform): Likewise.
3504 (gimple_mod_subtract_transform): Likewise.
3505 (gimple_stringops_transform): Likewise.
3507 2018-08-01 Martin Liska <mliska@suse.cz>
3510 * value-prof.c (interesting_stringop_to_profile_p):
3511 Simplify the code and add BUILT_IN_MEMMOVE.
3512 (gimple_stringops_transform): Likewise.
3514 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
3516 * config/aarch64/aarch64-simd.md
3517 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
3518 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
3519 use GPI iterator instead of SI mode.
3521 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
3523 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
3524 rs6000_speculation_barrier.
3525 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
3526 new barrier pattern name.
3528 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
3530 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
3531 (speculation_barrier): New insn.
3533 2018-08-01 Richard Biener <rguenther@suse.de>
3536 * graphite.h: Include isl/id.h and isl/space.h to allow build
3539 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
3542 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
3543 mode for COFF targets.
3544 * defaults.h (TARGET_COFF): Define.
3545 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
3546 TARGET_COFF): Define.
3547 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
3548 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
3549 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
3551 2018-07-31 Alexandre Oliva <oliva@adacore.com>
3553 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
3554 * gimple-streamer-out.c (output_bb): Save it.
3555 * lto-streamer-in.c (input_struct_function_base): Restore
3556 instance discriminator if available. Create map on demand.
3557 * lto-streamer-out.c (output_struct_function_base): Save it if
3559 * final.c (decl_to_instance_map): Document LTO strategy.
3561 2018-07-31 Alexandre Oliva <oliva@adacore.com>
3562 Olivier Hainque <hainque@adacore.com>
3564 * debug.h (decl_to_instance_map_t): New type.
3565 (decl_to_instance_map): Declare.
3566 (maybe_create_decl_to_instance_map): New inline function.
3567 * final.c (bb_discriminator, last_bb_discriminator): New statics,
3568 to track basic block discriminators.
3569 (final_start_function_1): Initialize them.
3570 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
3572 (decl_to_instance_map): New variable.
3573 (map_decl_to_instance, maybe_set_discriminator): New functions.
3574 (notice_source_line): Set discriminator.
3576 2018-07-31 Ian Lance Taylor <iant@golang.org>
3578 * targhooks.c (default_have_speculation_safe_value): Add
3581 2018-07-31 David Malcolm <dmalcolm@redhat.com>
3583 * dump-context.h: Include "pretty-print.h".
3584 (dump_context::refresh_dumps_are_enabled): New decl.
3585 (dump_context::emit_item): New decl.
3586 (class dump_context): Add fields "m_test_pp" and
3588 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
3589 (temp_dump_context::get_dumped_text): New decl.
3590 (class temp_dump_context): Add field "m_pp".
3591 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
3592 (dump_context::refresh_dumps_are_enabled): ...and add a test for
3594 (set_dump_file): Update for above change.
3595 (set_alt_dump_file): Likewise.
3596 (dump_loc): New overload, taking a pretty_printer *.
3597 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
3598 to any test pretty-printer.
3599 (make_item_for_dump_gimple_stmt): New function, adapted from
3600 optinfo::add_gimple_stmt.
3601 (dump_context::dump_gimple_stmt): Call it, and use the result,
3602 eliminating the direct usage of dump_file and alt_dump_file in
3603 favor of indirectly using them via emit_item.
3604 (make_item_for_dump_gimple_expr): New function, adapted from
3605 optinfo::add_gimple_expr.
3606 (dump_context::dump_gimple_expr): Call it, and use the result,
3607 eliminating the direct usage of dump_file and alt_dump_file in
3608 favor of indirectly using them via emit_item.
3609 (make_item_for_dump_generic_expr): New function, adapted from
3611 (dump_context::dump_generic_expr): Call it, and use the result,
3612 eliminating the direct usage of dump_file and alt_dump_file in
3613 favor of indirectly using them via emit_item.
3614 (make_item_for_dump_printf_va): New function, adapted from
3615 optinfo::add_printf_va.
3616 (make_item_for_dump_printf): New function.
3617 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
3618 and use the result, eliminating the direct usage of dump_file and
3619 alt_dump_file in favor of indirectly using them via emit_item.
3620 (make_item_for_dump_dec): New function.
3621 (dump_context::dump_dec): Call it, and use the result,
3622 eliminating the direct usage of dump_file and alt_dump_file in
3623 favor of indirectly using them via emit_item.
3624 (make_item_for_dump_symtab_node): New function, adapted from
3625 optinfo::add_symtab_node.
3626 (dump_context::dump_symtab_node): Call it, and use the result,
3627 eliminating the direct usage of dump_file and alt_dump_file in
3628 favor of indirectly using them via emit_item.
3629 (dump_context::begin_scope): Reimplement, avoiding direct usage
3630 of dump_file and alt_dump_file in favor of indirectly using them
3632 (dump_context::emit_item): New member function.
3633 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
3634 Set up test pretty-printer on the underlying context. Call
3635 refresh_dumps_are_enabled.
3636 (temp_dump_context::~temp_dump_context): Call
3637 refresh_dumps_are_enabled.
3638 (temp_dump_context::get_dumped_text): New member function.
3639 (selftest::verify_dumped_text): New function.
3640 (ASSERT_DUMPED_TEXT_EQ): New macro.
3641 (selftest::test_capture_of_dump_calls): Run all tests twice, with
3642 and then without optinfo enabled. Add uses of
3643 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
3644 * dumpfile.h: Update comment for the dump_* API.
3645 * optinfo-emit-json.cc
3646 (selftest::test_building_json_from_dump_calls): Update for new
3647 param for temp_dump_context ctor.
3648 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
3649 and "m_owned" field.
3650 (optinfo_item::~optinfo_item): Likewise.
3651 (optinfo::add_item): New member function.
3652 (optinfo::emit): Update comment.
3653 (optinfo::add_string): Delete.
3654 (optinfo::add_printf): Delete.
3655 (optinfo::add_printf_va): Delete.
3656 (optinfo::add_gimple_stmt): Delete.
3657 (optinfo::add_gimple_expr): Delete.
3658 (optinfo::add_tree): Delete.
3659 (optinfo::add_symtab_node): Delete.
3660 (optinfo::add_dec): Delete.
3661 * optinfo.h (class dump_context): New forward decl.
3662 (optinfo::add_item): New decl.
3663 (optinfo::add_string): Delete.
3664 (optinfo::add_printf): Delete.
3665 (optinfo::add_printf_va): Delete.
3666 (optinfo::add_gimple_stmt): Delete.
3667 (optinfo::add_gimple_expr): Delete.
3668 (optinfo::add_tree): Delete.
3669 (optinfo::add_symtab_node): Delete.
3670 (optinfo::add_dec): Delete.
3671 (optinfo::add_poly_int): Delete.
3672 (optinfo_item::optinfo_item): Remove "owned" param.
3673 (class optinfo_item): Remove field "m_owned".
3675 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3678 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
3679 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
3680 requested variable alignment.
3681 (expand_one_ssa_partition): Likewise.
3682 (expand_one_var): Likewise.
3684 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3686 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
3687 to speculation_safe_value_not_needed.
3689 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3691 * targhooks.h (speculation_safe_value_not_needed): New prototype.
3692 * targhooks.c (speculation_safe_value_not_needed): New function.
3693 * target.def (have_speculation_safe_value): Update documentation.
3694 * doc/tm.texi: Regenerated.
3696 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3698 * config/aarch64/iterators.md (ALLI_TI): New iterator.
3699 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
3701 (despeculate_copy<ALLI:mode>_insn): New insn.
3702 (despeculate_copyti_insn): New insn.
3703 (despeculate_simple<ALLI:mode>): New insn
3704 (despeculate_simpleti): New insn.
3705 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
3707 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
3708 aarch64_speculation_safe_value.
3709 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
3711 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3713 * config/aarch64/aarch64-speculation.cc: New file.
3714 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
3715 before pass_reorder_blocks.
3716 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
3718 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
3719 X14 and X15 when tracking speculation.
3720 * config/aarch64/aarch64.md (register name constants): Add
3721 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
3722 (unspec): Add UNSPEC_SPECULATION_TRACKER.
3723 (speculation_barrier): New insn attribute.
3724 (cmp<mode>): Allow SP in comparisons.
3725 (speculation_tracker): New insn.
3726 (speculation_barrier): Add speculation_barrier attribute.
3727 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
3728 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
3729 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
3731 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3733 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
3734 aarch64_track_speculation is true.
3735 (tb<optab><mode>1): Likewise.
3736 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
3737 generate CB[N]Z when tracking speculation.
3738 (aarch64_split_compare_and_swap): Likewise.
3739 (aarch64_split_atomic_op): Likewise.
3741 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3743 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
3745 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3747 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
3748 (speculation_barrier): New insn.
3750 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3752 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
3753 * config/arm/arm.md (speculation_barrier): New expand.
3754 (speculation_barrier_insn): New pattern.
3756 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3758 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
3759 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
3760 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
3761 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
3763 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
3764 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
3765 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
3766 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
3767 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
3768 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
3769 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
3770 * builtins.c (expand_speculation_safe_value): New function.
3771 (expand_builtin): Call it.
3772 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
3773 * doc/extend.texi: Document __builtin_speculation_safe_value.
3774 * doc/md.texi: Document "speculation_barrier" pattern.
3775 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
3776 TARGET_HAVE_SPECULATION_SAFE_VALUE.
3777 * doc/tm.texi: Regenerated.
3778 * target.def (have_speculation_safe_value, speculation_safe_value): New
3780 * targhooks.c (default_have_speculation_safe_value): New function.
3781 (default_speculation_safe_value): New function.
3782 * targhooks.h (default_have_speculation_safe_value): Add prototype.
3783 (default_speculation_safe_value): Add prototype.
3785 2018-07-31 David Malcolm <dmalcolm@redhat.com>
3787 * dump-context.h (dump_context::dump_loc): New decl.
3788 * dumpfile.c (dump_context::dump_loc): New member function.
3789 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
3790 and dump_gimple_stmt.
3791 (dump_context::dump_gimple_expr_loc): Likewise, using
3793 (dump_context::dump_generic_expr_loc): Likewise, using
3795 (dump_context::dump_printf_loc_va): Likewise, using
3797 (dump_context::begin_scope): Explicitly using the global function
3798 "dump_loc", rather than the member function.
3800 2018-07-31 Martin Sebor <msebor@redhat.com>
3802 PR tree-optimization/86741
3803 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
3805 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
3807 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
3808 depend on whether prefetch instructions will be emitted or not.
3809 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
3810 will be emitted or not.
3811 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
3812 (TARGET_SETMEM_PFD): New macros.
3814 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3816 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
3817 (NULL_STMT_VEC_INFO): Delete.
3818 (stmt_vec_info::operator*): Likewise.
3819 (stmt_vec_info::operator gimple *): Likewise.
3820 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
3821 of NULL_STMT_VEC_INFO.
3822 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
3823 (vect_reassociating_reduction_p): Likewise.
3824 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
3825 (vectorizable_store): Likewise.
3826 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
3827 (vec_info::free_stmt_vec_infos): Likewise.
3829 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3831 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
3832 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
3834 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3836 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
3837 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
3838 (vec_info::free_stmt_vec_info): New private member functions.
3839 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
3840 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
3841 * tree-parloops.c (gather_scalar_reductions): Remove calls to
3842 set_stmt_vec_info_vec and free_stmt_vec_infos.
3843 * tree-vect-loop.c (_loop_vec_info): Remove call to
3844 set_stmt_vec_info_vec.
3845 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
3846 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
3847 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
3848 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
3849 (vec_info::free_stmt_vec_info): ...these new functions. Remove
3850 assignments in {vec_info::,}new_stmt_vec_info that are redundant
3851 with the clearing in the xcalloc.
3852 (stmt_vec_info_vec): Delete.
3853 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
3854 (vectorize_loops): Likewise.
3855 (vec_info::~vec_info): Remove argument from call to
3856 free_stmt_vec_infos.
3857 (vec_info::add_stmt): Remove vinfo argument from call to
3860 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3862 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
3863 rather than a gimple stmt.
3864 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
3865 information for pattern statements when passed the original
3866 statement; instead wait to be passed the pattern statement itself.
3867 Don't call set_vinfo_for_stmt here.
3868 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
3869 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
3870 stmt_vec_infos here.
3871 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
3872 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
3873 stmt_vec_infos entry.
3875 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3877 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
3878 * tree-vectorizer.c (vec_info::replace_stmt): New function.
3879 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
3880 * tree-vect-stmts.c (vectorizable_call): Likewise.
3881 (vectorizable_simd_clone_call): Likewise.
3883 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3885 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
3886 * tree-vectorizer.c (vec_info::remove_stmt): New function.
3887 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
3888 * tree-vect-loop.c (vect_transform_loop): Likewise.
3889 * tree-vect-slp.c (vect_schedule_slp): Likewise.
3890 * tree-vect-stmts.c (vect_remove_stores): Likewise.
3892 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3894 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
3895 (vect_dr_stmt): Delete.
3896 * tree-vectorizer.c (vec_info::lookup_dr): New function.
3897 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
3899 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
3900 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
3901 (vect_verify_datarefs_alignment, vect_peeling_supportable)
3902 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
3903 (vect_analyze_data_refs): Likewise.
3904 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
3906 (vect_find_same_alignment_drs): Likewise.
3907 (vect_slp_analyze_node_dependences): Update calls accordingly.
3908 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
3909 instead of DR_VECT_AUX.
3910 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
3911 of a vector data references. Use vec_info::lookup_dr instead of
3913 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
3914 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
3915 instead of DR_VECT_AUX.
3917 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3919 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
3921 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
3923 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
3924 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
3925 (vect_gen_prolog_loop_niters): Likewise.
3927 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3929 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
3930 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
3931 (vect_known_alignment_in_bytes, vect_dr_behavior)
3932 (vect_get_scalar_dr_size): Take references as dr_vec_infos
3933 instead of data_references. Update calls to other routines for
3934 which the same change has been made.
3935 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
3936 dr_vec_infos instead of stmt_vec_infos.
3937 (vect_analyze_data_ref_dependence): Update call accordingly.
3938 (vect_slp_analyze_data_ref_dependence)
3939 (vect_record_base_alignments): Use DR_VECT_AUX.
3940 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
3941 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
3942 (vector_alignment_reachable_p, vect_get_data_access_cost)
3943 (vect_peeling_supportable, vect_analyze_group_access_1)
3944 (vect_analyze_group_access, vect_analyze_data_ref_access)
3945 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
3946 (vect_compile_time_alias, vect_small_gap_p)
3947 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
3948 (vect_supportable_dr_alignment): Take references as dr_vec_infos
3949 instead of data_references. Update calls to other routines for
3950 which the same change has been made.
3951 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
3952 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
3953 (vect_slp_analyze_and_verify_node_alignment)
3954 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
3955 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
3956 (vect_setup_realignment): Use dr_vec_infos. Update calls after
3958 (_vect_peel_info::dr): Replace with...
3959 (_vect_peel_info::dr_info): ...this new field.
3960 (vect_peeling_hash_get_most_frequent)
3961 (vect_peeling_hash_choose_best_peeling): Update accordingly.
3962 (vect_peeling_hash_get_lowest_cost):
3963 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
3964 routines for which the same change has been made.
3965 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
3967 * tree-vect-loop-manip.c (get_misalign_in_elems)
3968 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
3970 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
3971 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
3972 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
3973 (get_group_load_store_type, get_negative_load_store_type)
3974 (vect_get_data_ptr_increment, vectorizable_store)
3975 (vectorizable_load): Likewise.
3976 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
3977 Update calls to other routines for which the same change has been made.
3979 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3981 * tree-vectorizer.h (vec_info::move_dr): New member function.
3982 (dataref_aux): Rename to...
3983 (dr_vec_info): ...this and add "dr" and "stmt" fields.
3984 (_stmt_vec_info::dr_aux): Update accordingly.
3985 (_stmt_vec_info::data_ref_info): Delete.
3986 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
3987 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
3988 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
3989 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
3991 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
3992 (STMT_VINFO_DR_INFO): New macro.
3993 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
3994 (set_dr_misalignment): Update after rename of dataref_aux.
3995 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
3996 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
3997 initialization of STMT_VINFO_DATA_REF.
3998 * tree-vectorizer.c (vec_info::move_dr): New function.
3999 * tree-vect-patterns.c (vect_recog_bool_pattern)
4000 (vect_recog_mask_conversion_pattern)
4001 (vect_recog_gather_scatter_pattern): Use it.
4002 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
4003 the "dr" and "stmt" fields of dr_vec_info instead of
4004 STMT_VINFO_DATA_REF.
4006 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4008 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
4009 (is_pattern_stmt_p): Use it.
4010 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
4011 on pattern statements.
4013 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4015 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
4016 original stmt as a stmt_vec_info rather than a gimple stmt.
4017 (vect_pattern_recog_1): Take the statement directly as a
4018 stmt_vec_info, rather than via a gimple_stmt_iterator.
4019 Update call to vect_mark_pattern_stmts.
4020 (vect_pattern_recog): Update calls accordingly.
4022 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4024 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
4025 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
4026 a vect_def_type for the first argument.
4027 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
4028 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
4029 operand if it isn't defined by a vectorized statement.
4030 (vect_build_gather_load_calls): Remove the mask_dt argument and
4031 update calls to vect_get_vec_def_for_stmt_copy.
4032 (vectorizable_bswap): Likewise the dt argument.
4033 (vectorizable_call): Update calls to vectorizable_bswap and
4034 vect_get_vec_def_for_stmt_copy.
4035 (vectorizable_simd_clone_call, vectorizable_assignment)
4036 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
4037 (vectorizable_comparison): Update calls to
4038 vect_get_vec_def_for_stmt_copy.
4039 (vectorizable_store): Likewise. Remove now-unnecessary calls to
4041 (vect_get_loop_based_defs): Remove dt argument and update call
4042 to vect_get_vec_def_for_stmt_copy.
4043 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
4044 and vect_get_vec_def_for_stmt_copy.
4045 (vectorizable_load): Update calls to vect_build_gather_load_calls
4046 and vect_get_vec_def_for_stmt_copy.
4047 * tree-vect-loop.c (vect_create_epilog_for_reduction)
4048 (vectorizable_reduction, vectorizable_live_operation): Update calls
4049 to vect_get_vec_def_for_stmt_copy.
4051 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4053 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
4054 and gimple stmt arguments with a stmt_vec_info.
4055 (vect_record_base_alignments): Update calls accordingly.
4056 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
4057 and gimple stmt arguments with a stmt_vec_info.
4058 (vect_build_slp_tree_1): Remove vinfo argument and update call
4059 to vect_record_max_nunits.
4060 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
4061 and vect_record_max_nunits.
4063 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4065 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
4066 file and take a stmt_vec_info instead of a gimple stmt.
4067 (supportable_widening_operation, vect_finish_replace_stmt)
4068 (vect_finish_stmt_generation, vect_get_store_rhs)
4069 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
4070 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
4071 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
4072 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
4073 (vect_create_data_ref_ptr, bump_vector_ptr)
4074 (vect_permute_store_chain, vect_setup_realignment)
4075 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
4076 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
4077 (vectorizable_reduction, vectorizable_induction)
4078 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
4079 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
4080 than gimple stmts as arguments.
4081 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
4082 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
4083 (can_group_stmts_p, vect_check_gather_scatter)
4084 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
4085 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
4086 (vect_permute_load_chain, vect_shift_permute_load_chain)
4087 (vect_transform_grouped_load)
4088 (vect_record_grouped_load_vectors): Likewise.
4089 * tree-vect-loop.c (vect_fixup_reduc_chain)
4090 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
4091 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
4092 (vectorizable_reduction, vectorizable_induction)
4093 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
4094 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
4095 (vect_get_load_store_mask): Likewise.
4096 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
4097 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
4098 * tree-vect-stmts.c (vect_mark_relevant)
4099 (is_simple_and_all_uses_invariant)
4100 (exist_non_indexing_operands_for_use_p, process_use)
4101 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
4102 (vect_get_vec_def_for_operand, vect_get_vec_defs)
4103 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
4104 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
4105 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
4106 (get_negative_load_store_type, get_load_store_type)
4107 (vect_check_load_store_mask, vect_check_store_rhs)
4108 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
4109 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4110 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4111 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4112 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
4113 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
4114 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
4115 (supportable_widening_operation): Likewise.
4117 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4119 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
4120 a stmt_vec_info instead of a gcall.
4121 (vect_check_gather_scatter): Update call accordingly.
4122 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
4124 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
4125 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
4126 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
4127 instead of a gimple stmt.
4128 (vect_transform_loop): Update calls accordingly.
4129 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
4130 stmt_vec_infos instead of gimple stmts.
4131 (vect_analyze_slp_instance): Update use accordingly.
4132 * tree-vect-stmts.c (read_vector_array, write_vector_array)
4133 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
4134 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
4135 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
4136 (vect_gen_widened_results_half, vect_get_loop_based_defs)
4137 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
4138 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
4139 down to subroutines.
4141 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4143 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
4144 of the worklist from a vector of gimple stmts to a vector of
4146 * tree-vect-stmts.c (vect_mark_relevant, process_use)
4147 (vect_mark_stmts_to_be_vectorized): Likewise
4149 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4151 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
4152 statement before passing it to vect_analyze_stmt.
4153 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
4154 the chain of phi vector definitions. Track the exit phi via its
4156 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
4157 STMT_VINFO_REDUC_DEF.
4158 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
4159 stmt_vec_infos to handle the statement chains.
4160 (vect_get_slp_defs): Record the first statement in the node
4161 using a stmt_vec_info.
4162 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
4163 statements here and pass their stmt_vec_info down to subroutines.
4164 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
4165 down to vect_finish_stmt_generation.
4166 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
4167 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
4168 stmt_vec_infos to subroutines.
4169 (vect_remove_stores): Use stmt_vec_infos to handle the statement
4172 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4174 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
4175 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
4176 (vect_permute_store_chain, vect_setup_realignment)
4177 (vect_permute_load_chain, vect_shift_permute_load_chain)
4178 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
4179 stmts internally, and when passing values to other vectorizer routines.
4180 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
4181 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
4182 (vect_analyze_loop_operations, get_initial_def_for_reduction)
4183 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
4184 (vectorizable_reduction, vectorizable_induction)
4185 (vectorizable_live_operation, vect_transform_loop_stmt)
4186 (vect_transform_loop): Likewise.
4187 * tree-vect-patterns.c (vect_reassociating_reduction_p)
4188 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
4189 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
4190 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4191 (vect_slp_analyze_node_operations_1): Likewise.
4192 * tree-vect-stmts.c (vect_mark_relevant, process_use)
4193 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
4194 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
4195 (vect_finish_stmt_generation_1, get_group_load_store_type)
4196 (get_load_store_type, vect_build_gather_load_calls)
4197 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4198 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4199 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4200 (vectorizable_store, vectorizable_load, vectorizable_condition)
4201 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
4202 (supportable_widening_operation): Likewise.
4203 (vect_get_vector_types_for_stmt): Likewise.
4204 * tree-vectorizer.h (vect_dr_behavior): Likewise.
4206 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4208 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
4209 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
4210 (vect_permute_store_chain, vect_permute_load_chain)
4211 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
4212 repeated stmt_vec_info lookups.
4213 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
4214 (vect_update_ivs_after_vectorizer): Likewise.
4215 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
4216 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
4217 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
4218 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4219 (vect_bb_slp_scalar_cost): Likewise.
4220 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
4222 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4224 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
4225 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
4226 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
4227 to get gassigns and gcalls, rather than operating on generc gimple
4229 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
4230 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
4231 (vectorizable_load, vect_analyze_stmt): Likewise.
4232 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
4234 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4236 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
4237 return stmt_vec_infos rather than gimple stmts. Do not accept
4239 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
4241 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
4242 Update use of get_later_stmt.
4243 (vect_get_constant_vectors): Update call accordingly.
4244 (vect_schedule_slp_instance): Likewise
4245 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
4246 (vect_slp_analyze_instance_dependence): Likewise.
4247 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
4249 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4251 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
4252 (stmt_info_for_cost::stmt_info): ...this new field.
4253 (add_stmt_costs): Update accordingly.
4254 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
4255 (vect_get_known_peeling_cost): Likewise.
4256 (vect_estimate_min_profitable_iters): Likewise.
4257 * tree-vect-stmts.c (record_stmt_cost): Likewise.
4259 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4261 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
4262 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4263 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
4265 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
4267 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4269 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
4270 a gimple stmt to a stmt_vec_info.
4271 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
4273 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4275 * tree-vectorizer.h (vec_info::grouped_stores): Change from
4276 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4277 (_loop_vec_info::reduction_chains): Likewise.
4278 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
4280 * tree-vect-slp.c (vect_analyze_slp): Likewise.
4282 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4284 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
4285 a gimple stmt to a stmt_vec_info.
4286 (_stmt_vec_info::next_element): Likewise.
4287 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
4288 (vect_slp_analyze_and_verify_node_alignment)
4289 (vect_analyze_group_access_1, vect_analyze_group_access)
4290 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
4291 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
4292 (vect_supportable_dr_alignment): Update accordingly.
4293 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
4294 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
4295 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
4296 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
4297 * tree-vect-slp.c (vect_build_slp_tree_1)
4298 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
4299 (vect_split_slp_store_group, vect_analyze_slp_instance)
4300 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
4301 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
4302 (get_group_load_store_type, get_load_store_type)
4303 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
4304 (vect_transform_stmt, vect_remove_stores): Likewise.
4306 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4308 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
4310 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
4311 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
4312 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
4313 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
4314 (vector_alignment_reachable_p, vect_get_data_access_cost)
4315 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
4316 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
4317 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
4318 (vect_analyze_group_access_1, vect_analyze_group_access)
4319 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
4320 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
4321 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
4322 result of vect_dr_stmt and use the stmt_vec_info instead of
4323 the associated gimple stmt.
4324 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
4325 (vect_gen_prolog_loop_niters): Likewise.
4326 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
4328 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4330 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
4331 to a vec<stmt_vec_info>.
4332 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
4333 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
4335 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
4336 to a vec<stmt_vec_info>.
4337 (bst_traits::value_type, bst_traits::value_type): Likewise.
4338 (bst_traits::hash): Update accordingly.
4339 (vect_get_and_check_slp_defs): Change the stmts parameter from
4340 a vec<gimple *> to a vec<stmt_vec_info>.
4341 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
4342 (vect_build_slp_tree): Likewise.
4343 (vect_build_slp_tree_2): Likewise. Update uses of
4344 SLP_TREE_SCALAR_STMTS.
4345 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
4346 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
4347 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
4348 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
4349 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
4350 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
4351 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
4352 (vect_get_constant_vectors, vect_get_slp_defs)
4353 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
4354 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
4355 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
4356 instead of gimple stmts.
4357 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
4358 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
4359 (vect_slp_analyze_instance_dependence): Update uses of
4360 SLP_TREE_SCALAR_STMTS.
4361 (vect_slp_analyze_and_verify_node_alignment): Likewise.
4362 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
4363 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
4364 (get_initial_defs_for_reduction): Likewise.
4365 (vect_create_epilog_for_reduction): Likewise.
4366 (vectorize_fold_left_reduction): Likewise.
4367 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
4368 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
4369 (can_vectorize_live_stmts): Likewise.
4371 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4373 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
4374 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4375 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
4377 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
4379 (gather_scalar_reductions): Update after above interface changes.
4380 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
4381 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
4383 (vect_force_simple_reduction): Likewise.
4384 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
4385 LOOP_VINFO_REDUCTIONS.
4386 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4388 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4390 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
4391 a gimple stmt to a stmt_vec_info.
4392 * tree-vect-loop.c (vect_active_double_reduction_p)
4393 (vect_force_simple_reduction, vectorizable_reduction): Update
4396 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4398 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
4399 vec<gimple *> to a vec<stmt_vec_info>.
4400 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
4401 the reduction_phis argument from a vec<gimple *> to a
4403 (vectorizable_reduction): Likewise the phis local variable that
4404 is passed to vect_create_epilog_for_reduction. Update for new type
4405 of SLP_TREE_VEC_STMTS.
4406 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
4407 (vectorizable_live_operation): Likewise.
4408 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
4409 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
4411 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4413 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
4414 a gimple stmt to a stmt_vec_info.
4415 (vectorizable_condition, vectorizable_live_operation)
4416 (vectorizable_reduction, vectorizable_induction): Pass back the
4417 vectorized statement as a stmt_vec_info.
4418 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
4419 use of STMT_VINFO_VEC_STMT.
4420 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
4421 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
4422 as stmt_vec_infos rather than gimple stmts.
4423 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
4425 (vectorizable_live_operation): Likewise.
4426 (vectorizable_reduction, vectorizable_induction): Likewise,
4427 updating use of STMT_VINFO_VEC_STMT.
4428 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
4429 of STMT_VINFO_VEC_STMT.
4430 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
4431 (vectorizable_simd_clone_call, vectorizable_conversion)
4432 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4433 (vectorizable_store, vectorizable_load, vectorizable_condition)
4434 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
4435 from a gimple stmt to a stmt_vec_info.
4436 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
4437 pointer to a stmt_vec_info to the vectorizable_* routines.
4439 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4441 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
4442 a gimple stmt to a stmt_vec_info.
4443 (is_pattern_stmt_p): Update accordingly.
4444 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
4445 (vect_record_grouped_load_vectors): Likewise.
4446 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
4447 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
4448 (vect_model_reduction_cost): Likewise.
4449 (vect_create_epilog_for_reduction): Likewise.
4450 (vectorizable_reduction, vectorizable_induction): Likewise.
4451 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4452 Return the stmt_vec_info for the pattern statement.
4453 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
4454 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
4455 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
4456 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
4457 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
4458 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
4459 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
4460 (free_stmt_vec_info, vect_is_simple_use): Likewise.
4462 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4464 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
4465 (vect_finish_stmt_generation): Likewise.
4466 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
4467 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
4468 (vect_build_gather_load_calls): Use the return value of the above
4469 functions instead of a separate call to vinfo_for_stmt. Use narrow
4470 scopes for the input gimple stmt and wider scopes for the associated
4471 stmt_vec_info. Use vec_info::lookup_def when setting these
4472 stmt_vec_infos from an SSA_NAME definition.
4473 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4474 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4475 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4476 (vectorizable_store, vectorizable_load, vectorizable_condition)
4477 (vectorizable_comparison): Likewise.
4478 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
4479 (vectorizable_reduction): Likewise.
4481 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4483 * tree-vectorizer.h (vect_is_simple_use): Add an optional
4484 stmt_vec_info * parameter before the optional gimple **.
4485 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
4486 (process_use, vect_get_vec_def_for_operand_1): Update callers.
4487 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
4488 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4489 (vectorizable_live_operation): Likewise.
4490 * tree-vect-patterns.c (type_conversion_p): Likewise.
4491 (vect_look_through_possible_promotion): Likewise.
4492 (vect_recog_rotate_pattern): Likewise.
4493 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
4495 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4497 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
4498 a typedef to a wrapper class.
4499 (NULL_STMT_VEC_INFO): New macro.
4500 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
4501 (stmt_vec_info::operator*): New function.
4502 (stmt_vec_info::operator gimple *): Likewise.
4503 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
4504 (add_stmt_costs): Likewise.
4505 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
4506 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
4507 (vect_get_known_peeling_cost): Likewise.
4508 (vect_estimate_min_profitable_iters): Likewise.
4509 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4510 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
4511 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
4512 (vectorizable_store, free_stmt_vec_infos): Likewise.
4513 (new_stmt_vec_info): Change return type of xcalloc to
4516 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4518 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
4519 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
4520 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
4521 a single_imm_use-based sequence.
4522 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
4524 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4526 * tree-vectorizer.h (vec_info::lookup_def): Declare.
4527 * tree-vectorizer.c (vec_info::lookup_def): New function.
4528 * tree-vect-patterns.c (vect_get_internal_def): Use it.
4529 (vect_widened_op_tree): Likewise.
4530 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
4531 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
4532 (vectorizable_reduction): Likewise.
4533 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
4535 (vect_is_slp_reduction): Update calls accordingly. Use
4536 vec_info::lookup_def.
4537 (vect_is_simple_reduction): Likewise
4538 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
4540 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4542 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
4543 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
4544 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
4546 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
4547 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
4548 (vect_update_vf_for_slp, vect_analyze_loop_operations)
4549 (vect_is_slp_reduction, vectorizable_induction)
4550 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
4551 * tree-vect-patterns.c (vect_init_pattern_stmt):
4552 (vect_determine_min_output_precision_1, vect_determine_precisions)
4553 (vect_pattern_recog): Likewise.
4554 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
4555 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
4556 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
4557 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
4558 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
4559 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
4560 info field from a loop to a loop_vec_info.
4562 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4564 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
4565 (vec_info::add_stmt): Declare.
4566 * tree-vectorizer.c (vec_info::add_stmt): New function.
4567 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
4568 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
4569 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
4570 (vectorizable_induction): Likewise.
4571 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
4572 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
4573 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
4574 (vectorizable_load): Likewise.
4575 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4576 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
4577 (vect_recog_gather_scatter_pattern): Likewise.
4578 (append_pattern_def_seq): Likewise. Remove a check that is
4579 performed by add_stmt itself.
4581 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4583 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
4584 which make_ssa_name was called with new_stmt before new_stmt
4587 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4589 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
4591 (vect_is_slp_reduction): ...here...
4592 (vect_is_simple_reduction): ...and here. Remove repetition of tests
4593 that are already known to be false.
4595 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4597 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
4598 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
4599 STMT_VINFO_NUM_SLP_USES when it's true.
4600 (vect_free_slp_instance): Add a final_p parameter and pass it to
4602 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
4603 (vect_analyze_slp_instance): Likewise.
4604 (vect_slp_analyze_operations): Likewise.
4605 (vect_slp_analyze_bb_1): Likewise.
4606 * tree-vectorizer.c (vec_info): Likewise.
4607 * tree-vect-loop.c (vect_transform_loop): Likewise.
4609 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4611 * tree-vect-loop.c (vectorizable_reduction): Assert that the
4612 function is not called for second and subsequent members of
4615 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4617 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
4618 cases for nested loops from here to ...
4619 (vect_create_epilog_for_reduction): ...here. Only call
4620 vect_is_simple_use for inner-loop reductions.
4622 2018-07-31 Martin Liska <mliska@suse.cz>
4624 PR gcov-profile/85338
4625 PR gcov-profile/85350
4626 PR gcov-profile/85372
4627 * profile.c (struct location_triplet): New.
4628 (struct location_triplet_hash): Likewise.
4629 (output_location): Do not output a BB that
4630 is already recorded for a line.
4631 (branch_prob): Use streamed_locations.
4633 2018-07-31 Martin Liska <mliska@suse.cz>
4635 PR gcov-profile/85370
4636 * coverage.c (coverage_begin_function): Do not mark target
4637 clones as artificial functions.
4639 2018-07-31 Martin Liska <mliska@suse.cz>
4641 PR gcov-profile/83813
4642 PR gcov-profile/84758
4643 PR gcov-profile/85217
4644 PR gcov-profile/85332
4645 * profile.c (branch_prob): Do not record GOTO expressions
4646 for GIMPLE statements which locations are already streamed.
4648 2018-07-31 Olivier Hainque <hainque@adacore.com>
4650 * gcc.c (handle_spec_function): Accept a soft_matched_part
4651 argument, as do_spec_1. Pass it down to ...
4652 (eval_spec_function): Accept a soft_matched_part argument,
4653 and pass it down to ...
4654 (do_spec_2): Accept a soft_matched_part argument, and pass
4655 it down to do_spec_1.
4656 (do_spec_1): Pass soft_matched_part to handle_spec_function.
4657 (handle_braces): Update call to handle_spec_function.
4658 (driver::set_up_specs): Update calls to do_spec_2.
4659 (compare_debug_dump_opt_spec_function): Likewise.
4660 (compare_debug_self_opt_spec_function): Likewise.
4662 2018-07-31 Olivier Hainque <hainque@adacore.com>
4664 * common.opt (nolibc): New option.
4665 * doc/invoke.texi (Link Options): Document it.
4666 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
4667 * config/alpha/linux.h: Likewise.
4668 * config/arc/elf.h: Likewise.
4669 * config/arm/uclinux-elf.h: Likewise.
4670 * config/arm/unknown-elf.h: Likewise.
4671 * config/avr/avrlibc.h: Likewise.
4672 * config/bfin/bfin.h: Likewise.
4673 * config/bfin/linux.h: Likewise.
4674 * config/bfin/uclinux.h: Likewise.
4675 * config/darwin.h: Likewise.
4676 * config/darwin10.h: Likewise.
4677 * config/darwin12.h: Likewise.
4678 * config/gnu-user.h: Likewise.
4679 * config/lm32/uclinux-elf.h: Likewise.
4680 * config/pa/pa-hpux11.h: Likewise.
4681 * config/pa/pa64-hpux.h: Likewise.
4682 * config/sparc/sparc.h: Likewise.
4684 2018-07-31 Olivier Hainque <hainque@adacore.com>
4686 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
4687 undefined variables.
4689 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
4692 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
4695 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
4697 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
4698 terminated string literal.
4700 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
4702 PR rtl-optimization/85160
4703 * combine.c (is_just_move): New function.
4704 (try_combine): Allow combining two instructions into two if neither of
4705 the original instructions was a move.
4707 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
4710 * doc/extend.texi (Global Register Variables): Discourage use of type
4712 (Local Register Variables): Likewise.
4714 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
4716 PR tree-optimization/86506
4717 * hwint.h (ceil_log2): Resync with hwint.c implementation.
4719 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
4722 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
4723 hard_regno, make sure no insn between `from` and `to` clobbers it.
4725 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
4726 Tom de Vries <tdevries@suse.de>
4728 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
4729 (PTX_DEFAULT_RUNTIME_DIM): ... this.
4730 (nvptx_goacc_validate_dims): Set default worker and gang dims to
4731 PTX_DEFAULT_RUNTIME_DIM.
4732 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
4734 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
4736 * config/pa/pa.c (pa_output_addr_vec): Align address table.
4737 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
4738 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
4740 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
4742 * config/rs6000/constraints.md (wG constraint): Delete, no longer
4744 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
4745 predicate to reflect toc fusion has been deleted.
4746 (toc_fusion_mem_raw): Delete, no longer used.
4747 (toc_fusion_mem_wrapped): Likewise.
4748 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
4750 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
4751 Delete, no longer used.
4752 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
4753 meant to be used for toc fusion.
4754 (rs6000_debug_print_mode): Delete toc fusion debugging.
4755 (rs6000_debug_reg_global): Likewise.
4756 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
4757 fusion and secondary reload support that were never used.
4758 (rs6000_option_override_internal): Delete TOC fusion, that was only
4759 partially defined, and it did not work unless you also used the
4761 (rs6000_legitimate_address_p): Delete TOC fusion support.
4762 (rs6000_opt_masks): Likewise.
4763 (fusion_wrap_memory_address): Delete function, no longer used.
4764 (fusion_split_address); Delete TOC fusion support.
4765 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
4766 longer used with toc fusion being deleted.
4767 (TARGET_TOC_FUSION_FP): Likewise.
4768 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
4770 (toc fusion spliter): Delete TOC fusion support.
4771 (toc_fusionload_<mode>): Likewise.
4772 (toc_fusionload_di): Likewise.
4773 (fusion_gpr_load_<mode>): Delete generator function, this insn no
4774 longer needs to be named. Rename predicate to delete TOC fusion.
4775 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
4776 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
4777 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
4778 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
4779 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
4781 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
4783 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
4784 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
4785 __int128 in built-in function prototypes.
4786 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
4787 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
4789 2018-07-27 Martin Sebor <msebor@redhat.com>
4791 PR tree-optimization/86696
4792 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
4793 types, including enums.
4794 (handle_char_store): Be prepared for the above function to fail.
4796 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
4798 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
4799 when optimization level is lower than 2 or optimize for size.
4801 2018-07-26 Martin Sebor <msebor@redhat.com>
4803 PR tree-optimization/86043
4804 PR tree-optimization/86042
4805 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
4806 (get_string_cst_length): Rename...
4807 (get_min_string_length): ...to this. Add argument.
4808 (handle_char_store): Extend to handle multi-character stores by
4810 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
4811 * tree.h (initializer_zerop): Add argument.
4813 2018-07-26 Jakub Jelinek <jakub@redhat.com>
4816 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
4817 declare target to variables if they have always,{to,from,tofrom} map
4820 2018-07-26 Martin Liska <mliska@suse.cz>
4823 * lto-wrapper.c: Add linker_output as prefix
4824 for ltrans_output_file.
4826 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
4828 PR rtl-optimization/85805
4829 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
4830 value for hard registers if that was written in the same mode.
4832 2018-07-26 Martin Liska <mliska@suse.cz>
4834 PR gcov-profile/86536
4835 * gcov.c (format_gcov): Use printf format %.*f directly
4836 and do not handle special values.
4838 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4840 * common/config/arc/arc-common.c (arc_option_optimization_table):
4841 Update default optimizations for size.
4843 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4845 * config/arc/arc.md (movsf_insn): Add short instruction selection.
4846 * config/arc/constraints.md (CfZ): New constraint.
4847 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
4848 (subsf3_fpu): Likewise.
4849 (cmpsf_fpu): Likewise.
4850 (cmpsf_fpu_uneq): Likewise.
4852 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4854 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
4856 (arc_is_uncached_mem_p): uncached applies to both the variable and
4859 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4861 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
4864 2018-07-25 David Malcolm <dmalcolm@redhat.com>
4866 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
4867 field "m_scopes" from vec to auto_vec.
4869 2018-07-25 Martin Liska <mliska@suse.cz>
4871 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
4874 2018-07-25 Richard Biener <rguenther@suse.de>
4877 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
4878 special wrt context_die late.
4879 (gen_subprogram_die): Re-use DIEs in local scope.
4881 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
4883 PR tree-optimization/86644
4884 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
4886 2018-07-25 Martin Liska <mliska@suse.cz>
4889 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
4890 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
4892 2018-07-25 Martin Liska <mliska@suse.cz>
4895 * params.def: Explain ASan abbreviation and provide
4896 a documentation link.
4898 2018-07-24 Martin Sebor <msebor@redhat.com>
4900 PR tree-optimization/86622
4901 PR tree-optimization/86532
4902 * builtins.h (string_length): Declare.
4903 * builtins.c (c_strlen): Correct handling of non-constant offsets.
4904 (check_access): Be prepared for non-constant length ranges.
4905 (string_length): Make extern.
4906 * expr.c (string_constant): Only handle the minor non-constant
4907 array index. Use string_constant to compute the length of
4908 a generic string constant.
4910 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
4912 PR tree-optimization/86618
4913 * tree-vect-stmts.c (vectorizable_call): Don't take the address
4914 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
4916 2018-07-24 David Malcolm <dmalcolm@redhat.com>
4918 PR tree-optimization/86636
4919 * json.cc (json::object::set): Fix comment. Add assertions.
4920 (json::array::append): Move here from json.h. Add comment and an
4922 (json::string::string): Likewise.
4923 * json.h (json::array::append): Move to json.cc.
4924 (json::string::string): Likewise.
4925 * optinfo-emit-json.cc
4926 (optrecord_json_writer::impl_location_to_json): Assert that we
4927 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
4928 wrapper around it. Expand the location once, rather than three
4930 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
4931 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
4933 (optrecord_json_writer::optinfo_to_json): Likewise, in four
4934 places. Fix some overlong lines.
4936 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
4938 * config/aarch64/aarch64-simd.md
4939 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
4940 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
4941 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
4942 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
4943 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
4944 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
4945 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
4946 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
4947 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
4949 2018-07-24 Jakub Jelinek <jakub@redhat.com>
4952 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
4953 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
4954 and abs_d == d, do the power of two handling if profitable.
4956 2018-07-24 Richard Biener <rguenther@suse.de>
4958 * match.pd: Add BIT_FIELD_REF canonicalizations.
4960 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4963 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
4965 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4967 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
4968 terminated STRING_CST object.
4970 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4972 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
4974 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
4976 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
4977 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
4978 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
4979 the elements into a register.
4980 (rs6000_split_v4si_init_di_reg): Delete.
4981 (rs6000_split_v4si_init): Delete.
4982 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
4983 (vsx_init_v4si): Rewrite as a define_expand.
4985 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
4987 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
4988 zero_extend argument from memory): New.
4990 2018-07-22 Martin Sebor <msebor@redhat.com>
4993 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
4994 diagnosing calls with unknown arguments unless -Walloca-larger-than
4995 is restricted to less than PTRDIFF_MAX bytes.
4997 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
4999 * doc/gcov.texi (Invoking Gcov): Editorial changes.
5001 2018-07-20 David Malcolm <dmalcolm@redhat.com>
5003 * pretty-print.c (text_info::set_location): Remove redundant
5004 "line_table" parameter from call to rich_location::set_range.
5006 2018-07-20 Martin Sebor <msebor@redhat.com>
5009 * builtins.c (expand_builtin_alloca): Adjust.
5010 * calls.c (alloc_max_size): Simplify.
5011 * cgraphunit.c (cgraph_node::expand): Adjust.
5012 * common.opt (larger_than_size, warn_frame_larger_than): Remove
5014 (frame_larger_than_size): Same.
5015 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
5016 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
5018 * doc/invoke.texi (GCC Command Options): Document option arguments.
5019 Explain byte-size arguments and suffixes.
5020 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
5021 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
5022 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
5023 * doc/options.texi (UInteger): Expand.
5024 (Host_Wide_Int, ByteSize): Document new properties.
5025 * final.c (final_start_function_1): Include sizes in an error message.
5026 * function.c (frame_offset_overflow): Same.
5027 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
5028 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
5029 Diagnose unbounded alloca calls only for limits of less than
5031 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
5032 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
5034 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
5035 only for limits of less than PTRDIFF_MAX.
5036 * langhooks-def.h (lhd_handle_option): Change function argument
5038 * langhooks.c (lhd_handle_option): Same.
5039 * langhooks.h (handle_option): Same.
5040 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
5042 (var_type, var_type_struct): Same.
5043 (var_set): Handle ByteSize flag.
5044 * optc-gen.awk: Add comments to output to ease debugging. Make
5045 use of HOST_WIDE_INT where appropriate.
5046 * opts-gen-save.awk: Use %lx to format unsigned long.
5047 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
5048 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
5049 arguments. Parse bytes-size suffixes.
5050 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
5051 (enum_value_to_arg): Same.
5052 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
5053 (handle_option): Adjust.
5054 (generate_option): Change function argument to HOST_WIDE_INT.
5055 (cmdline_handle_error): Adjust.
5056 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
5057 (set_option): Change function argument to HOST_WIDE_INT.
5058 (option_enabled): Handle cl_host_wide_int.
5059 (get_option_state): Handle CLVC_SIZE.
5060 (control_warning_option): Same.
5061 * opts.c (common_handle_option): Change function argument to
5062 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
5063 OPT_Wvla_larger_than_.
5064 * opts.h (enum cl_var_type): Add an enumerator.
5065 * stor-layout.c (layout_decl): Print a more meaningful warning.
5066 * toplev.c (output_stack_usage): Adjust.
5068 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
5070 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
5071 call to inline_expand_builtin_string_cmp.
5072 (expand_builtin_strcmp): Likewise.
5073 (expand_builtin_strncmp): Likewise.
5074 (inline_string_cmp): Delete the last parameter, change char_type_node
5075 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
5077 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
5078 the inlining expansion on target where the type of the call has same or
5079 narrower precision than unsigned char.
5081 2018-07-20 David Malcolm <dmalcolm@redhat.com>
5083 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
5084 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
5085 * common.opt (fsave-optimization-record): New option.
5086 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
5087 * doc/invoke.texi (-fsave-optimization-record): New option.
5088 * dumpfile.c: Include "optinfo-emit-json.h".
5089 (struct kv_pair): Move to coretypes.h.
5090 (optgroup_options): Make non-static.
5091 (dump_context::end_scope): Call
5092 optimization_records_maybe_pop_dump_scope.
5093 * dumpfile.h (optgroup_options): New decl.
5094 * json.cc: New file.
5096 * optinfo-emit-json.cc: New file.
5097 * optinfo-emit-json.h: New file.
5098 * optinfo.cc: Include "optinfo-emit-json.h".
5099 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
5100 (optinfo_enabled_p): Check optimization_records_enabled_p.
5101 (optinfo_wants_inlining_info_p): Likewise.
5102 * optinfo.h: Update comment.
5103 * profile-count.c (profile_quality_as_string): New function.
5104 * profile-count.h (profile_quality_as_string): New decl.
5105 (profile_count::quality): New accessor.
5106 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
5107 and optinfo_emit_json_cc_tests.
5108 * selftest.h (selftest::json_cc_tests): New decl.
5109 (selftest::optinfo_emit_json_cc_tests): New decl.
5110 * toplev.c: Include "optinfo-emit-json.h".
5111 (compile_file): Call optimization_records_finish.
5112 (do_compile): Call optimization_records_start.
5113 * tree-ssa-live.c: Include optinfo.h.
5114 (remove_unused_scope_block_p): Retain inlining information if
5115 optinfo_wants_inlining_info_p returns true.
5117 2018-07-20 Richard Biener <rguenther@suse.de>
5120 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
5121 to cover -flto-partition=none.
5123 2018-07-20 Martin Liska <mliska@suse.cz>
5125 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
5126 (get_decl_source_range): Remove unused function.
5128 2018-07-20 Richard Biener <rguenther@suse.de>
5130 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
5131 (struct vn_phi_s): Likewise.
5132 (struct vn_reference_s): Likewise.
5133 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
5134 for searching the slot of an entry known to be in the hash itself.
5135 (vn_phi_hasher::equal): Likewise.
5136 (vn_reference_hasher::equal): Likewise.
5137 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
5139 (optimistic_info, current_info): Remove, keeping only valid_info.
5140 (vn_reference_lookup_1): Remove fallback lookup.
5141 (vn_reference_lookup_2): Likewise.
5142 (vn_nary_op_lookup_1): Likewise.
5143 (vn_phi_lookup): Likewise.
5144 (vn_nary_build_or_lookup_1): Make sure to not chain the built
5146 (vn_reference_insert): Adjust, chain the inserted hash element
5147 at last_inserted_ref.
5148 (vn_reference_insert_pieces): Likewise.
5149 (visit_reference_op_call): Likewise.
5150 (vn_nary_op_insert_into): Chain the inserted hash element at
5152 (vn_nary_op_insert_pieces): Adjust.
5153 (vn_nary_op_insert): Likewise.
5154 (vn_nary_op_insert_stmt): Likewise.
5155 (vn_phi_insert): Adjust, chain the inserted hash element at
5157 (process_scc): Remove clearing and copying the optimistic
5158 table. Instead remove elements inserted during an optimistic
5159 iteration from the single table we maintain.
5160 (init_scc_vn): Adjust.
5161 (free_scc_vn): Likewise.
5162 (sccvn_dom_walker::record_cond): Likewise.
5163 (sccvn_dom_walker::after_dom_children): Likewise.
5165 2018-07-19 Martin Sebor <msebor@redhat.com>
5167 PR tree-optimization/84047
5168 PR tree-optimization/83776
5169 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
5170 (check_array_bounds): Call it.
5172 2018-07-19 Martin Sebor <msebor@redhat.com>
5174 * align.h (align_flags): Use member initialization.
5176 2018-07-19 David Malcolm <dmalcolm@redhat.com>
5178 * Makefile.in (OBJS): Add optinfo.o.
5179 * coretypes.h (class symtab_node): New forward decl.
5180 (struct cgraph_node): New forward decl.
5181 (class varpool_node): New forward decl.
5182 * dump-context.h: New file.
5183 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
5185 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
5186 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
5187 (set_alt_dump_file): Likewise.
5188 (dump_context::~dump_context): New dtor.
5189 (dump_gimple_stmt): Move implementation to...
5190 (dump_context::dump_gimple_stmt): ...this new member function.
5191 Add the stmt to any pending optinfo, creating one if need be.
5192 (dump_gimple_stmt_loc): Move implementation to...
5193 (dump_context::dump_gimple_stmt_loc): ...this new member function.
5194 Start a new optinfo and add the stmt to it.
5195 (dump_gimple_expr): Move implementation to...
5196 (dump_context::dump_gimple_expr): ...this new member function.
5197 Add the stmt to any pending optinfo, creating one if need be.
5198 (dump_gimple_expr_loc): Move implementation to...
5199 (dump_context::dump_gimple_expr_loc): ...this new member function.
5200 Start a new optinfo and add the stmt to it.
5201 (dump_generic_expr): Move implementation to...
5202 (dump_context::dump_generic_expr): ...this new member function.
5203 Add the tree to any pending optinfo, creating one if need be.
5204 (dump_generic_expr_loc): Move implementation to...
5205 (dump_context::dump_generic_expr_loc): ...this new member
5206 function. Add the tree to any pending optinfo, creating one if
5208 (dump_printf): Move implementation to...
5209 (dump_context::dump_printf_va): ...this new member function. Add
5210 the text to any pending optinfo, creating one if need be.
5211 (dump_printf_loc): Move implementation to...
5212 (dump_context::dump_printf_loc_va): ...this new member function.
5213 Start a new optinfo and add the stmt to it.
5214 (dump_dec): Move implementation to...
5215 (dump_context::dump_dec): ...this new member function. Add the
5216 value to any pending optinfo, creating one if need be.
5217 (dump_context::dump_symtab_node): New member function.
5218 (dump_context::get_scope_depth): New member function.
5219 (dump_context::begin_scope): New member function.
5220 (dump_context::end_scope): New member function.
5221 (dump_context::ensure_pending_optinfo): New member function.
5222 (dump_context::begin_next_optinfo): New member function.
5223 (dump_context::end_any_optinfo): New member function.
5224 (dump_context::s_current): New global.
5225 (dump_context::s_default): New global.
5226 (dump_scope_depth): Delete global.
5227 (dumpfile_ensure_any_optinfo_are_flushed): New function.
5228 (dump_symtab_node): New function.
5229 (get_dump_scope_depth): Reimplement in terms of dump_context.
5230 (dump_begin_scope): Likewise.
5231 (dump_end_scope): Likewise.
5232 (selftest::temp_dump_context::temp_dump_context): New ctor.
5233 (selftest::temp_dump_context::~temp_dump_context): New dtor.
5234 (selftest::verify_item): New function.
5235 (ASSERT_IS_TEXT): New macro.
5236 (ASSERT_IS_TREE): New macro.
5237 (ASSERT_IS_GIMPLE): New macro.
5238 (selftest::test_capture_of_dump_calls): New test.
5239 (selftest::dumpfile_c_tests): Call it.
5240 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
5241 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
5242 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
5243 descriptive comment.
5244 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
5245 (dump_node, dump_bb): Move these unrelated decls.
5246 (class dump_manager): Add leading comment.
5247 * optinfo.cc: New file.
5248 * optinfo.h: New file.
5250 2018-07-19 Michael Collison <michael.collison@arm.com>
5251 Richard Henderson <rth@redhat.com>
5253 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
5254 (subti): Handle op1 zero.
5255 (subvti4, usub4ti4): New.
5256 (*sub<GPI>3_compare1_imm): New.
5257 (sub<GPI>3_carryinCV): New.
5258 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
5259 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
5261 2018-07-19 Michael Collison <michael.collison@arm.com>
5262 Richard Henderson <rth@redhat.com>
5264 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
5265 (addti3): Create simpler code if low part is already known to be 0.
5266 (addvti4, uaddvti4): New.
5267 (*add<GPI>3_compareC_cconly_imm): New.
5268 (*add<GPI>3_compareC_cconly): New.
5269 (*add<GPI>3_compareC_imm): New.
5270 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
5271 handle constants within this pattern..
5272 (*add<GPI>3_compareV_cconly_imm): New.
5273 (*add<GPI>3_compareV_cconly): New.
5274 (*add<GPI>3_compareV_imm): New.
5275 (add<GPI>3_compareV): New.
5276 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
5277 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
5278 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
5279 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
5280 with 'comparison' operator.
5281 (*add<GPI>3_compareV_cconly_imm): Ditto.
5282 (*add<GPI>3_compareV_cconly): Ditto.
5283 (*add<GPI>3_compareV_imm): Ditto.
5284 (add<GPI>3_compareV): Ditto.
5285 (add<mode>3_carryinC): Ditto.
5286 (*add<mode>3_carryinC_zero): Ditto.
5287 (*add<mode>3_carryinC): Ditto.
5288 (add<mode>3_carryinV): Ditto.
5289 (*add<mode>3_carryinV_zero): Ditto.
5290 (*add<mode>3_carryinV): Ditto.
5292 2018-07-19 Michael Collison <michael.collison@arm.com>
5293 Richard Henderson <rth@redhat.com>
5295 * config/aarch64/aarch64-modes.def (CC_V): New.
5296 * config/aarch64/aarch64-protos.h
5297 (aarch64_addti_scratch_regs): Declare
5298 (aarch64_subvti_scratch_regs): Declare.
5299 (aarch64_expand_subvti): Declare.
5300 (aarch64_gen_unlikely_cbranch): Declare
5301 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
5302 for signed overflow using CC_Vmode.
5303 (aarch64_get_condition_code_1): Handle CC_Vmode.
5304 (aarch64_gen_unlikely_cbranch): New function.
5305 (aarch64_addti_scratch_regs): New function.
5306 (aarch64_subvti_scratch_regs): New function.
5307 (aarch64_expand_subvti): New function.
5309 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5311 * config/aarch64/aarch64-option-extensions.def: New entry for profile
5313 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
5314 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
5317 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5320 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
5321 address check not strict.
5323 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5325 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
5327 (store_pair_lanes<mode>): Likewise.
5328 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
5329 enum value 'ADDR_QUERY_LDP_STP_N'.
5330 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
5331 (aarch64_print_address_internal): Add declaration.
5332 (aarch64_print_ldpstp_address): Remove.
5333 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
5334 (aarch64_print_operand): Change printing of 'y'.
5335 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
5336 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
5337 'true' rather than '1'.
5338 * config/aarch64/constraints.md (Uml): Likewise.
5339 (Uml): Rename to Umn.
5342 2018-07-19 Richard Biener <rguenther@suse.de>
5344 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
5346 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
5347 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
5348 (vn_reference_hasher): Likewise.
5349 (struct vn_tables_s): Remove obstack and alloc-pool members.
5350 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
5351 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
5352 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
5353 (vn_reference_insert_pieces): Likewise.
5354 (alloc_vn_nary_op_noinit): Adjust.
5355 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
5356 (vn_phi_eq): Adjust.
5357 (shared_lookup_phiargs): Remove.
5358 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
5359 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
5360 (visit_reference_op_call): Likewise.
5361 (copy_nary, copy_phi, copy_reference): Remove.
5362 (process_scc): Rewind the obstack when iterating. Do not
5363 copy the elements to valid_info but just move them from one
5364 hashtable to the other.
5365 (allocate_vn_table): Adjust.
5366 (free_vn_table): Likewise.
5367 (init_scc_vn): Likewise.
5368 (free_scc_vn): Likewise.
5370 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
5373 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
5374 indirect_return as function type attribute.
5375 (ix86_attribute_table): Change indirect_return to function
5377 * doc/extend.texi: Update indirect_return attribute.
5379 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
5381 * wide-int.h (widest2_int): New.
5382 * gimple-fold.c (arith_overflowed_p): Use it.
5383 * tree.h (widest2_int_cst): New.
5384 * tree-vrp.c (wide_int_binop_overflow): Rename from
5385 vrp_int_const_binop.
5386 Rewrite to work on trees.
5387 (extract_range_from_multiplicative_op_1): Abstract code to...
5388 (wide_int_range_min_max): ...here.
5389 (wide_int_range_cross_product): ...and here.
5390 (extract_range_from_binary_expr_1): Abstract overflow code to...
5391 (wide_int_range_mult_wrapping): ...here.
5392 * tree-vrp.h (wide_int_range_cross_product): New.
5393 (wide_int_range_mult_wrapping): New.
5395 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
5396 Julia Koval <julia.koval@intel.com>
5398 * config/i386/x86-tune-costs.h (skylake_memcpy,
5399 skylake_memset): Replace rep_prefix with unrolling for size 512.
5401 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
5404 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
5405 comparision with EQ_EXPR in last stmt.
5407 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
5409 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
5410 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
5411 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
5412 previously known as "PowerPC AltiVec Built-in Functions". Move
5413 some material to new subsubsections "PowerPC AltiVec Built-in
5414 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
5416 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
5417 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
5418 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
5419 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
5421 2018-07-18 Richard Biener <rguenther@suse.de>
5423 PR tree-optimization/86557
5424 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
5427 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
5429 * config/s390/s390.c (s390_function_profiler): Generate CFI.
5431 2018-07-17 Jeff Law <law@redhat.com>
5433 * config/arm/arm.c (get_label_padding): Update for recent
5434 changes to label_to_alignment.
5436 PR tree-optimization/86010
5437 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
5439 * config/mips/mips.c (vr4130_align_insns): Update for recent
5440 changes to label_to_alignment.
5442 * config/frv/frv.c (frv_label_align): Update for recent changes
5443 to label_to_alignment.
5445 * config/nios2/nios2.c (nios2_label_align): Update for recent
5446 changes which dropped ALIGN_LABELS_LOG.
5448 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
5450 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
5451 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
5453 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5455 * config/arc/arc.c (arc_label_align): Use align_labels instead of
5456 deprecated align_labels_log.
5458 2018-07-17 Richard Biener <rguenther@suse.de>
5461 * dwarf2out.c (init_sections_and_labels): Always generate
5462 a debug_line_str_section for early LTO debug.
5463 (dwarf2out_finish): Reset debug_line_str_hash output early.
5464 Bump counter for extra dwarf5 .debug_loc labels to not conflict
5466 (dwarf2out_early_finish): Output debug_line_str.
5468 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
5470 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
5471 index register on z196 or later.
5473 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
5475 * config/s390/s390.c (s390_default_align): Set default function
5477 (s390_override_options_after_change): Call s390_default align.
5478 (s390_option_override_internal): Call s390_default_align.
5479 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
5481 2018-07-17 Jakub Jelinek <jakub@redhat.com>
5484 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
5485 to _looptemp_ clauses, other than the first two.
5487 2018-07-17 Martin Liska <mliska@suse.cz>
5489 * opts.c: Do not enable OPT_falign_* for -Os.
5491 2018-07-17 Martin Liska <mliska@suse.cz>
5493 * align.h (MAX_CODE_ALIGN): New.
5494 (MAX_CODE_ALIGN_VALUE): New.
5495 * common/config/i386/i386-common.c (ix86_handle_option):
5496 (MAX_CODE_ALIGN): Moved to align.h.
5497 * final.c (MAX_CODE_ALIGN): Likewise.
5498 * opts.c (parse_and_check_align_values):
5499 (MAX_CODE_ALIGN): Likewise.
5500 (MAX_CODE_ALIGN_VALUE): Likewise.
5502 2018-07-17 Martin Liska <mliska@suse.cz>
5504 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
5505 in order to fulfil coding style.
5506 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
5507 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
5508 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
5509 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
5510 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
5511 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
5512 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
5513 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
5514 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
5516 2018-07-17 Martin Liska <mliska@suse.cz>
5518 * align.h: New file.
5519 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
5521 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
5522 align_flags of label_to_alignment.
5523 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
5525 * config/m68k/m68k.c: Do not use removed align_labels_value and
5527 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
5528 (LOOP_ALIGN): Likewise.
5529 (LABEL_ALIGN): Likewise.
5530 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
5531 Remove not used macro.
5532 (rs6000_loop_align): Change return type to align_flags.
5533 (rs6000_loop_align_max_skip): Remove.
5534 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
5535 Change return type to align_flags.
5536 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
5537 Remove not used macro.
5538 (rs6000_loop_align): Change return type to align_flags.
5539 (rs6000_loop_align_max_skip): Remove.
5540 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
5541 * config/rx/rx-protos.h (rx_align_for_label): Make it
5543 * config/rx/rx.c (rx_align_for_label): Change return type
5545 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
5547 into align_flags class.
5548 (LABEL_ALIGN): Likewise.
5549 (LOOP_ALIGN): Likewise.
5550 * config/s390/s390.c (s390_label_align): Use align_flags
5552 (s390_asm_output_function_label): Likewise.
5553 * config/sh/sh.c (sh_override_options_after_change):
5554 Use align_flags class directly without macros.
5555 (find_barrier): Likewise.
5556 (barrier_align): Likewise.
5557 (sh_loop_align): Likewise.
5558 * config/spu/spu.c (spu_option_override):
5559 Use align_flags_tuple::get_value instead of removed macros.
5560 (spu_sched_init): Likewise.
5561 * config/spu/spu.h (GTY): Likewise.
5562 * config/visium/visium.c (visium_option_override):
5563 Set "8" as default secondary alignment.
5564 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
5565 in order to guarantee secondary alignment of 8.
5566 * coretypes.h: Include align.h header file.
5567 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
5568 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
5569 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
5570 * doc/tm.texi.in: Likewise.
5571 * final.c (struct label_alignment): Remove not used structure.
5572 (LABEL_ALIGN): Change type to align_flags.
5573 (LOOP_ALIGN): Likewise.
5574 (JUMP_ALIGN): Likewise.
5575 (default_loop_align_max_skip): Remove.
5576 (default_label_align_max_skip): Likewise.
5577 (default_jump_align_max_skip): Likewise.
5578 (default_label_align_after_barrier_max_skip):
5579 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
5580 (LABEL_TO_MAX_SKIP): Remove.
5581 (label_to_alignment): Return align_flags type instead of integer.
5582 (label_to_max_skip): Remove.
5583 (align_fuzz): Use align_flags type.
5584 (compute_alignments): Use align_flags type and use align_flags::max
5585 to combine multiple alignments.
5586 (grow_label_align): Grow vec instead of C array.
5587 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
5588 (shorten_branches): Use align_flags type and use align_flags::max
5589 to combine multiple alignments.
5590 (final_scan_insn_1): Remove usage of secondary alignment that comes
5591 from label alignment, but instead use proper secondary alignment
5592 which is computed in grow_label_align.
5593 * flags.h (struct align_flags_tuple): Move to align.h.
5594 (struct align_flags): Likewise.
5595 (state_align_loops): Rename to align_loops.
5596 (state_align_jumps): Rename to align_jumps.
5597 (state_align_labels): Rename to align_labels.
5598 (state_align_functions): Rename to align_functions.
5599 (align_loops_log): Remove.
5600 (align_jumps_log): Remove.
5601 (align_labels_log): Remove.
5602 (align_functions_log): Remove.
5603 (align_loops_max_skip): Remove.
5604 (align_jumps_max_skip): Remove.
5605 (align_labels_max_skip): Remove.
5606 (align_functions_max_skip): Remove.
5607 (align_loops_value): Remove.
5608 (align_jumps_value): Remove.
5609 (align_labels_value): Remove.
5610 (align_functions_value): Remove.
5611 * output.h (label_to_alignment): Change return type to align_flags.
5612 (label_to_max_skip): Remove.
5613 * target.def: Remove loop_align_max_skip, label_align_max_skip,
5614 jump_align_max_skip macros.
5615 * targhooks.h (default_loop_align_max_skip): Remove.
5616 (default_label_align_max_skip): Likewise.
5617 (default_jump_align_max_skip): Likewise.
5618 (default_label_align_after_barrier_max_skip): Remove.
5619 * toplev.c (read_log_maxskip): Use ::normalize function.
5620 (parse_N_M): Remove not used argument and also call ::normalize.
5621 (parse_alignment_opts): Do not pass unused arguments.
5622 * varasm.c (assemble_start_function): Use directly align_functions
5623 instead of removed macros.
5624 * system.h: Do not poison removed macros.
5626 2018-07-17 Jakub Jelinek <jakub@redhat.com>
5629 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
5630 and cond temporaries don't have reference type if iterator has
5631 pointer type. For init use &for_pre_body instead of pre_p if
5632 for_pre_body is non-empty.
5634 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
5636 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
5637 double-double modes to SFmode directly directly.
5638 (trunc<mode>sf2_fprs): Delete.
5640 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
5642 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
5643 for conversions between IFmode and the decimal floating point modes.
5644 (init_float128_ieee): Use the correct names for conversions between
5645 KFmode and the decimal floating point modes.
5647 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
5649 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
5650 for the conversions between TDmode and IFmode.
5651 (init_float128_ieee): Use more correct names for the conversions
5652 between TDmode and KFmode.
5654 2018-07-16 Jakub Jelinek <jakub@redhat.com>
5656 PR tree-optimization/86526
5657 * builtins.c (expand_builtin_memcmp): Formatting fixes.
5658 (inline_expand_builtin_string_cmp): Likewise.
5659 (inline_string_cmp): Likewise. Use c_readstr instead of
5660 builtin_memcpy_read_str. Add unit_mode temporary.
5662 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
5665 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
5666 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
5668 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
5670 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
5671 Alphabetize prototypes of built-in functions, separating out
5672 built-in functions that are listed in this section but should be
5673 described elsewhere.
5675 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
5678 * expmed.c (emit_store_flag): Do not emit setcc followed by a
5679 conditional move when trapping comparison was split to a
5680 non-trapping one (and vice versa).
5682 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
5684 * config/s390/s390.c (s390_function_profiler): Generate nops
5685 instead of profiler call sequences.
5686 * config/s390/s390.opt: Add the new option.
5688 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
5690 * config/s390/s390.c (s390_function_profiler): Generate
5691 __mcount_loc section.
5692 * config/s390/s390.opt: Add the new option.
5694 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
5696 * common.opt: Add the new warning.
5697 * config/s390/s390.c (s390_function_profiler): Emit "brasl
5698 %r0,__fentry__" when -mfentry is specified.
5699 (s390_option_override_internal): Disallow -mfentry for 31-bit
5701 * config/s390/s390.opt: Add the new option.
5703 2018-07-16 Richard Biener <rguenther@suse.de>
5706 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
5707 for function-local FUNCTION_DECL and RESULT_DECL immediately.
5709 2018-07-16 Martin Liska <mliska@suse.cz>
5712 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
5715 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
5717 * config/arc/arcHS.md: Update ARCHS scheduling rules.
5719 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
5721 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
5723 * config/arc/arc-cpus.def (hs4x): New cpu.
5725 * config/arc/arc-tables.opt: Regenerate.
5726 * config/arc/arc.c (arc_sched_issue_rate): New function.
5727 (TARGET_SCHED_ISSUE_RATE): Define.
5728 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
5729 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
5731 (attr tune): Add ARCHS4x tune values.
5732 (attr tune_dspmpy): Define.
5733 (*tst): Correct instruction type.
5734 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
5735 * config/arc/arcHS4x.md: New file.
5736 * config/arc/fpu.md: Update instruction type attributes.
5737 * config/arc/t-multilib: Regenerate.
5739 2018-07-16 Tom de Vries <tdevries@suse.de>
5742 * var-tracking.c (vt_initialize): Fix pre_dec handling.
5744 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5746 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
5749 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
5751 PR tree-optimization/86514
5752 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
5753 conversion to a boolean type from a type with greater precision.
5755 2018-07-16 Tom de Vries <tdevries@suse.de>
5757 * var-tracking.c (vt_initialize): Print adjusted insn slim if
5758 dump_flags request TDF_SLIM.
5760 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
5762 * fold-const.c (int_const_binop_1): Abstract...
5763 (wide_int_binop): ...wide int code here.
5764 (poly_int_binop): ...poly int code here.
5765 Abstract the rest of int_const_binop_1 into int_const_binop.
5766 * fold-const.h (wide_int_binop): New.
5767 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
5768 Remove useless PLUS/MINUS_EXPR case.
5769 (zero_nonzero_bits_from_vr): Move wide int code...
5770 (zero_nonzero_bits_from_bounds): ...here.
5771 (extract_range_from_binary_expr_1): Move mask optimization code...
5772 (range_easy_mask_min_max): ...here.
5773 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
5774 (range_easy_mask_min_max): New.
5776 2018-07-15 Jeff Law <law@redhat.com>
5779 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
5782 2018-07-14 Jim Wilson <jimw@sifive.com>
5784 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
5786 2018-07-14 Paul Koning <ni1d@arrl.net>
5788 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
5790 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
5792 * lto-streamer-out.c (copy_function_or_variable): Dump info about
5795 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
5796 Steve Munroe <munroesj52@gmail.com>
5798 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
5799 (_mm_andnot_si128): Likewise.
5800 (_mm_or_si128): Likewise.
5801 (_mm_xor_si128): Likewise.
5803 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
5806 * builtins.c (expand_builtin_memcmp): Inline the calls first
5807 when result_eq is false.
5808 (expand_builtin_strcmp): Inline the calls first.
5809 (expand_builtin_strncmp): Likewise.
5810 (inline_string_cmp): New routine. Expand a string compare
5811 call by using a sequence of char comparison.
5812 (inline_expand_builtin_string_cmp): New routine. Inline expansion
5813 a call to str(n)cmp/memcmp.
5814 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
5816 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
5818 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
5820 * config/arm/driver-arm.c: Include arm-native.h.
5821 (host_detect_local_cpu): Use auto-generated data tables.
5822 (vendors, arm_cpu_table): Delete. Move part information to ...
5823 * config/arm/arm-cpus.in: ... here.
5824 * config/arm/parsecpu.awk (gen_native): New function.
5825 (vendor, part): New CPU fields.
5826 (END): Add support for building the native CPU detection tables.
5827 * config/arm/t-arm (arm-native.h): Add build rule.
5828 (driver-arm.o): Add dependency on arm-native.h.
5830 2018-07-13 Richard Biener <rguenther@suse.de>
5833 * match.pd (addr1 - addr2): Allow either of the operand to
5836 2018-07-13 Tom de Vries <tdevries@suse.de>
5838 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
5841 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
5843 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
5844 arrays instead of numbered variables.
5846 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
5848 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
5849 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
5851 2018-07-13 Richard Biener <rguenther@suse.de>
5854 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
5855 instead of get_context_die.
5857 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
5858 Richard Biener <rguenther@suse.de>
5861 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
5862 that the loop latch destination where phi is defined.
5864 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
5866 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
5867 (riscv_expand_epilogue): Add assertion to check interrupt mode.
5868 (riscv_set_current_function): Extract getting interrupt type to new
5870 (riscv_get_interrupt_type): New function.
5871 (riscv_merge_decl_attributes): New function, checking interrupt type is
5873 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
5875 2018-07-12 Paul Koning <ni1d@arrl.net>
5877 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
5880 2018-07-12 Paul Koning <ni1d@arrl.net>
5882 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
5883 zero reference, add doloop_end instead.
5884 * doc/md.texi (decrement_and_branch_until_zero): Remove.
5885 (Looping patterns): Remove decrement_and_branch_until_zero. Add
5886 detail for doloop_end.
5888 2018-07-12 Martin Sebor <msebor@redhat.com>
5891 * attribs.c (decl_attributes): Reject conflicting attributes before
5892 calling attribute handlers.
5894 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
5896 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
5898 (gcc::dump_manager::get_dump_file_name): likewise.
5899 (dump_begin): Likewise.
5900 * dumpfile.h (dump_begin): Update prototype.
5901 (gcc::dump_manager::get_dump_file_name,
5902 gcc::dump_manager::get_dump_file_name): Update prototype.
5904 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
5906 * internal-fn.h (vectorizable_internal_fn_p): New function.
5907 * tree-vect-slp.c (compatible_calls_p): Likewise.
5908 (vect_build_slp_tree_1): Remove nops argument. Handle calls
5909 to internal functions.
5910 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
5912 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
5914 * fold-const.h (inverse_conditions_p): Declare.
5915 * fold-const.c (inverse_conditions_p): New function.
5916 * match.pd: Use inverse_conditions_p. Add folds of view_converts
5917 that test the inverse condition of a conditional internal function.
5918 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
5919 * internal-fn.c (internal_fn_mask_index): Handle conditional
5921 (vectorized_internal_fn_supported_p): New function.
5922 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
5923 (any_pred_load_store): Replace with...
5924 (need_to_predicate): ...this new variable.
5925 (redundant_ssa_names): New variable.
5926 (ifcvt_can_use_mask_load_store): Move initial checks to...
5927 (ifcvt_can_predicate): ...this new function. Handle tree codes
5928 for which a conditional internal function exists.
5929 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
5930 instead of ifcvt_can_use_mask_load_store. Update after variable
5932 (predicate_load_or_store): New function, split out from
5933 predicate_mem_writes.
5934 (check_redundant_cond_expr): New function.
5935 (value_available_p): Likewise.
5936 (predicate_rhs_code): Likewise.
5937 (predicate_mem_writes): Rename to...
5938 (predicate_statements): ...this. Use predicate_load_or_store
5939 and predicate_rhs_code.
5940 (combine_blocks, tree_if_conversion): Update after above name changes.
5941 (ifcvt_local_dce): Handle redundant_ssa_names.
5942 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
5943 general conditional functions.
5944 * tree-vect-stmts.c (vectorizable_call): Likewise.
5946 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
5947 Alan Hayward <alan.hayward@arm.com>
5948 David Sherwood <david.sherwood@arm.com>
5950 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
5951 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
5952 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
5953 plus and minus and convert them into IFN_COND_FMA-based sequences.
5954 (convert_mult_to_fma): Handle conditional plus and minus.
5956 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
5958 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
5959 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
5960 (cond_fnms_optab): New optabs.
5961 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
5963 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
5964 * internal-fn.h (get_conditional_internal_fn): Declare.
5965 (get_unconditional_internal_fn): Likewise.
5966 * internal-fn.c (cond_ternary_direct): New macro.
5967 (expand_cond_ternary_optab_fn): Likewise.
5968 (direct_cond_ternary_optab_supported_p): Likewise.
5969 (FOR_EACH_COND_FN_PAIR): Likewise.
5970 (get_conditional_internal_fn): New function.
5971 (get_unconditional_internal_fn): Likewise.
5972 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
5973 (gimple_match_op::gimple_match_op): Add a new overload for 5
5975 (gimple_match_op::set_op): Likewise.
5976 (gimple_resimplify5): Declare.
5977 * genmatch.c (decision_tree::gen): Generate simplifications for
5979 * gimple-match-head.c (gimple_simplify): Define an overload for
5980 5 operands. Handle calls with 5 arguments in the top-level overload.
5981 (convert_conditional_op): Handle conversions from unconditional
5982 internal functions to conditional ones.
5983 (gimple_resimplify5): New function.
5984 (build_call_internal): Pass a fifth operand.
5985 (maybe_push_res_to_seq): Likewise.
5986 (try_conditional_simplification): Try converting conditional
5987 internal functions to unconditional internal functions.
5988 Handle 3-operand unconditional forms.
5989 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
5990 Define ternary equivalents of the current rules for binary conditional
5992 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
5994 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
5995 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
5996 (optab): Handle them.
5997 (SVE_COND_FP_TERNARY): New int iterator.
5998 (sve_fmla_op, sve_fmad_op): New int attributes.
5999 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
6000 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
6001 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
6003 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6005 * target.def (preferred_else_value): New target hook.
6006 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
6007 * doc/tm.texi: Regenerate.
6008 * targhooks.h (default_preferred_else_value): Declare.
6009 * targhooks.c (default_preferred_else_value): New function.
6010 * internal-fn.h (conditional_internal_fn_code): Declare.
6011 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
6012 (get_conditional_internal_fn): Use it.
6013 (conditional_internal_fn_code): New function.
6014 * gimple-match.h (gimple_match_cond): New struct.
6015 (gimple_match_op): Add a cond member function.
6016 (gimple_match_op::gimple_match_op): Update all forms to take a
6018 * genmatch.c (expr::gen_transform): Use the same condition as res_op
6019 for the suboperation, but don't specify a particular else_value.
6020 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
6021 (visit_nary_op, visit_reference_op_load): Pass
6022 gimple_match_cond::UNCOND to the gimple_match_op constructor.
6023 * gimple-match-head.c: Include tree-eh.h
6024 (convert_conditional_op): New function.
6025 (maybe_resimplify_conditional_op): Likewise.
6026 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
6027 (gimple_resimplify2): Likewise.
6028 (gimple_resimplify3): Likewise.
6029 (gimple_resimplify4): Likewise.
6030 (maybe_push_res_to_seq): Return null for conditional operations.
6031 (try_conditional_simplification): New function.
6032 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
6034 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
6036 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
6038 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
6040 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
6042 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
6044 (hash_tree): Do not hash DECL_FCONTEXT
6045 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
6046 Do not stream DECL_FCONTEXT.
6047 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
6048 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
6050 2018-07-12 Richard Biener <rguenther@suse.de>
6053 * dwarf2out.c (gen_block_die): Only output blocks when they have
6054 at least one !DECL_IGNORED_P variable.
6056 2018-07-12 Richard Biener <rguenther@suse.de>
6059 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
6060 Remove -mieee-fp handling.
6062 2018-07-12 Richard Biener <rguenther@suse.de>
6064 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
6065 left-over from last patch.
6067 2018-07-12 Jakub Jelinek <jakub@redhat.com>
6069 PR tree-optimization/86492
6070 * gimple-ssa-store-merging.c
6071 (imm_store_chain_info::coalesce_immediate_stores): Call
6072 check_no_overlap even for the merge_overlapping case. Formatting fix.
6074 2018-07-12 Richard Biener <rguenther@suse.de>
6077 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
6078 move possibly trapping operations into the conditional.
6080 2018-07-12 Richard Biener <rguenther@suse.de>
6082 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
6083 (vn_lookup_simplify_result): Remove recursion limit applied
6085 (vn_nary_build_or_lookup_1): Adjust.
6086 (try_to_simplify): Likewise.
6087 * gimple-match-head.c (gimple_resimplify1): Instead apply one
6089 (gimple_resimplify2): Likewise.
6090 (gimple_resimplify3): Likewise.
6091 (gimple_resimplify4): Likewise.
6093 2018-07-11 Jakub Jelinek <jakub@redhat.com>
6095 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
6096 Use __mmask64 type instead of __mmask8 for __M argument.
6097 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
6098 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
6100 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
6101 __mmask16 for __M argument.
6102 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
6103 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
6104 to __mmask16 instead of __mmask8.
6105 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
6106 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
6107 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
6108 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
6109 instead of __mmask16 for __U argument.
6110 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
6111 __mmask16 instead of __mmask8 for __U argument.
6112 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
6114 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
6116 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
6118 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
6120 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
6122 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
6124 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
6126 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
6128 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
6130 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
6132 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
6133 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
6134 return type as well as __M argument type and all casts from __mmask8
6136 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
6137 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
6138 return type as well as __M argument type and all casts from __mmask8
6140 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
6141 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
6142 return type as well as __M argument type and all casts from __mmask8
6144 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
6145 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
6146 return type as well as __M argument type and all casts from __mmask8
6148 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
6149 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
6152 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
6154 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
6155 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
6158 2018-07-11 Paul Koning <ni1d@arrl.net>
6160 * doc/md.texi (define_subst): Document how multiple occurrences of
6161 the same argument in the replacement pattern are handled.
6163 2018-07-11 Paul Koning <ni1d@arrl.net>
6165 * doc/extend.texi (Common Variable Attributes): Move "mode" into
6167 (Common Type Attributes): Add "mode" attribute.
6169 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
6171 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
6172 stream DECL_ORIGINAL_TYPE.
6173 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
6174 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
6175 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
6176 Do not walk original type.
6177 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
6179 (write_ts_decl_non_common_tree_pointers): Do not stream
6181 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
6182 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
6184 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
6186 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
6187 thread twice from the same starting edge.
6189 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
6191 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
6192 * gimple.c (gimple_call_nonnull_result_p): ...here...
6193 (gimple_call_nonnull_arg): ...and here.
6194 * gimple.h (gimple_call_nonnull_result_p): New.
6195 (gimple_call_nonnull_arg): New.
6197 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
6199 * config/arm/arm-cpus.in: Move information from fpu field of each
6200 cpu definition to the isa field.
6201 * config/arm/parsecpu.awk (fpu): Delete match rule.
6202 (gen_comm_data): Don't add bits from the CPU's FPU entry.
6204 2018-07-11 Richard Biener <rguenther@suse.de>
6207 * dwarf2out.c (init_sections_and_labels): Use
6208 output_asm_line_debug_info consistently.
6209 (dwarf2out_early_finish): Likewise.
6210 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
6213 2018-07-11 Richard Biener <rguenther@suse.de>
6215 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
6216 Rework father_bb setting in a way to avoid propagating constants
6217 multiple times on a loop body.
6219 2018-07-10 Mark Wielaard <mark@klomp.org>
6222 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
6224 2018-07-10 Richard Biener <rguenther@suse.de>
6226 * hash-map.h (hash_map::iterator::operator*): Return
6227 references to key and value.
6229 2018-07-10 Jakub Jelinek <jakub@redhat.com>
6232 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
6233 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
6234 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
6235 (gimplify_omp_for): For composite loops, move outer
6236 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
6237 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
6238 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
6239 TREE_LIST for both the original class iterator and the "last" helper
6240 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
6241 loop, remember has_decl_expr from outer composite loops for the
6242 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
6244 2018-07-09 Martin Sebor <msebor@redhat.com>
6248 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
6249 accessing implicitly initialized array elements.
6250 * expr.c (string_constant): Handle string initializers of
6251 character arrays within aggregates.
6252 * gimple-fold.c (fold_array_ctor_reference): Add argument.
6253 Store element offset. As a special case, handle zero size.
6254 (fold_nonarray_ctor_reference): Same.
6255 (fold_ctor_reference): Add argument. Store subobject offset.
6256 * gimple-fold.h (fold_ctor_reference): Add argument.
6258 2018-07-09 Paul Koning <ni1d@arrl.net>
6260 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
6261 (pdp11_insn_cost): New function.
6262 (pdp11_md_asm_adjust): New function.
6263 (TARGET_INVALID_WITHIN_DOLOOP): Define.
6264 (pdp11_rtx_costs): Update to match machine better.
6265 (output_addr_const_pdp11): Correct format mismatch warnings.
6266 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
6267 * config/pdp11/pdp11.md: General change to add base_cost and/or
6268 length attributes for use by new pdp11_insn_cost function.
6269 (MIN_BRANCH): Correct definition.
6271 (doloop_end): Use standard pattern name for looping pattern.
6272 (doloop_end_nocc): New.
6273 (movsf): Add another constraint alternative.
6274 (zero_extendqihi2): Add constraint alternatives for not in place
6276 (zero_extendhisi2): Remove.
6277 (shift patterns): Add CC handling variants.
6281 (define_peephole2): New peephole to recognize mov that sets CC for
6284 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6286 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
6287 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
6290 2018-07-09 Jakub Jelinek <jakub@redhat.com>
6293 * real.c (real_nextafter): Return true if result is denormal.
6295 2018-07-09 Martin Liska <mliska@suse.cz>
6297 * common.opt: Add back wrongly removed attribute.
6299 2018-07-09 Richard Biener <rguenther@suse.de>
6302 * dwarf2out.c (gen_block_die): For an early generated DIE
6303 always output high/low PC attributes.
6305 2018-07-09 Tom de Vries <tdevries@suse.de>
6307 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
6308 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
6309 onto VAR_DECL with abstract origin.
6311 2018-07-07 Jim Wilson <jimw@sifive.com>
6313 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
6315 2018-07-07 Tom de Vries <tdevries@suse.de>
6317 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
6319 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
6321 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
6323 (combine_bound): Use wide-int overflow calculation instead of
6325 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
6327 * fold-const.c (int_const_binop_2): Same.
6328 (extract_muldiv_1): Same.
6329 (fold_div_compare): Same.
6330 (fold_abs_const): Same.
6332 * poly-int.h (add): Same.
6336 * predict.c (predict_iv_comparison): Same.
6337 * profile-count.c (slow_safe_scale_64bit): Same.
6338 * simplify-rtx.c (simplify_const_binary_operation): Same.
6339 * tree-chrec.c (tree_fold_binomial): Same.
6340 * tree-data-ref.c (split_constant_offset_1): Same.
6341 * tree-if-conv.c (idx_within_array_bound): Same.
6342 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
6343 * tree-ssa-phiopt.c (minmax_replacement): Same.
6344 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
6345 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
6346 * vr-values.c (vr_values::adjust_range_with_scev): Same.
6347 * wide-int.cc (wi::add_large): Same.
6348 (wi::mul_internal): Same.
6349 (wi::sub_large): Same.
6350 (wi::divmod_internal): Same.
6351 * wide-int.h: Change overflow type to overflow_type for neg, add,
6352 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
6353 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
6354 mul_internal, divmod_internal.
6355 (overflow_type): New enum.
6356 (accumulate_overflow): New.
6358 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
6360 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
6361 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
6363 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
6365 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
6366 argument is checked for zero before entering loop, avoid checking again.
6368 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
6370 * gimplify.h (generic_expr_could_trap_p): Set as global function.
6371 * gimplify.h (generic_expr_could_trap_p): Likwise.
6372 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
6374 2018-07-06 Jakub Jelinek <jakub@redhat.com>
6376 PR tree-optimization/86401
6377 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
6378 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
6379 (fold_bit_and_mask): ... here. New helper function for match.pd.
6380 * fold-const.h (fold_bit_and_mask): Declare.
6381 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
6383 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
6386 * target.def (translate_mode_attribute): New hook.
6387 * targhooks.h (default_translate_mode_attribute): Declare.
6388 * targhooks.c (default_translate_mode_attribute): New function.
6389 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
6390 * doc/tm.texi: Regenerate.
6391 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
6392 (rs6000_translate_mode_attribute): New function.
6394 2018-07-06 Paul Koning <ni1d@arrl.net>
6396 * doc/md.texi (define_split): Document DONE and FAIL.
6397 (define_peephole2): Ditto.
6399 2018-07-05 Jeff Law <law@redhat.com>
6401 PR tree-optimization/86010
6402 * tree-ssa-dse.c (compute_trims): More aggressively trim at
6403 both the head and tail of mem* and str* calls.
6405 2018-07-05 Jim Wilson <jimw@sifive.com>
6407 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
6409 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
6411 * config/aarch64/aarch64-simd.md: correct flags text for
6412 MIN_EXPR replacement.
6414 2018-07-05 James Clarke <jrtc27@jrtc27.com>
6416 * configure: Regenerated.
6418 2018-07-05 Carl Love <cel@us.ibm.com>
6420 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
6421 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
6422 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
6423 VSX_BUILTIN_DOUBLEL_V4SF.
6425 2018-07-05 Martin Sebor <msebor@redhat.com>
6428 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
6429 than its domain to compute its the upper bound of a char array.
6431 2018-07-05 Nathan Sidwell <nathan@acm.org>
6433 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
6434 * doc/cpp.texi: Update comment.
6435 * doc/tm.texi: Rebuilt.
6436 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
6437 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
6438 * doc/extend.texi (Backwards Compatibility): Clarify it is system
6439 headers affected by extern "C".
6440 * system.h: Poison NO_IMPLICIT_EXTERN_C.
6441 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
6442 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
6443 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
6444 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
6445 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
6446 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
6447 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
6448 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
6449 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
6450 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
6451 config/sparc/sp64-elf.h, config/spu/spu.h,
6452 config/stormy16/stormy16.h, config/v850/v850.h,
6453 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
6454 define NO_IMPLICIT_EXTERN_C.
6455 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
6457 2018-07-05 Tamar Christina <tamar.christina@arm.com>
6460 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
6461 instead of GET_MODE_SIZE when comparing Units.
6463 2018-07-05 Tamar Christina <tamar.christina@arm.com>
6466 * rtlanal.c (set_noop_p): Constrain on mode change,
6467 include hard-reg-set.h
6469 2018-07-05 Tamar Christina <tamar.christina@arm.com>
6471 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
6473 2018-07-05 Jakub Jelinek <jakub@redhat.com>
6476 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
6479 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
6481 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
6483 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
6486 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
6488 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
6490 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
6493 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
6495 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
6497 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
6498 neg_*_op* variables.
6500 2018-07-04 Martin Liska <mliska@suse.cz>
6502 * tree-switch-conversion.c: Define
6503 max_ratio_for_speed and max_ratio_for_size constants.
6505 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
6506 Martin Liska <mliska@suse.cz>
6511 * common.opt: Rename align options with 'str_' prefix.
6512 * common/config/i386/i386-common.c (set_malign_value): New
6514 (ix86_handle_option): Use it to set -falign-* options/
6515 * config/aarch64/aarch64-protos.h (struct tune_params): Change
6516 type from int to string.
6517 * config/aarch64/aarch64.c: Update default values from int
6519 * config/alpha/alpha.c (alpha_override_options_after_change):
6521 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
6522 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6523 max skip conditionally.
6524 * config/i386/freebsd.h (SUBALIGN_LOG): New.
6525 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6526 max skip conditionally.
6527 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6528 max skip conditionally.
6529 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
6530 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6531 max skip conditionally.
6532 * config/i386/i386.c (struct ptt): Change type from int to
6534 (ix86_default_align): Set default values.
6535 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
6536 max skip conditionally.
6537 * config/i386/iamcu.h (SUBALIGN_LOG): New.
6538 (ASM_OUTPUT_MAX_SKIP_ALIGN):
6539 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
6540 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6541 max skip conditionally.
6542 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
6543 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
6544 * config/i386/x86-64.h (SUBALIGN_LOG): New.
6545 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6546 max skip conditionally.
6547 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
6548 * config/ia64/ia64.c (ia64_option_override): Set default values
6549 for alignment options.
6550 * config/m68k/m68k.c: Handle new str_align_* options.
6551 * config/mips/mips.c (mips_set_compression_mode): Change
6553 (mips_option_override): Set default values for options.
6554 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
6556 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6558 * config/rx/rx.c (rx_option_override): Likewise.
6559 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
6560 (LABEL_ALIGN): Use align_labels_log.
6561 (LOOP_ALIGN): Use align_loops_align.
6562 * config/s390/s390.c (s390_asm_output_function_label): Use new
6564 * config/sh/sh.c (sh_override_options_after_change):
6565 Change type of constants.
6566 * config/spu/spu.c (spu_sched_init): Likewise.
6567 * config/sparc/sparc.c (sparc_option_override): Set default
6569 * config/visium/visium.c (visium_option_override): Likewise.
6570 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
6571 emit p2align format with last argument if it's not needed.
6572 * doc/invoke.texi: Document extended format of -falign-*.
6573 * final.c: Use align_labels alignment.
6574 * flags.h (struct target_flag_state): Change type to use
6576 (struct align_flags_tuple): New.
6577 (struct align_flags): Likewise.
6578 (align_loops_log): Redefine macro to use new types.
6579 (align_loops_max_skip): Redefine macro to use new types.
6580 (align_jumps_log): Redefine macro to use new types.
6581 (align_jumps_max_skip): Redefine macro to use new types.
6582 (align_labels_log): Redefine macro to use new types.
6583 (align_labels_max_skip): Redefine macro to use new types.
6584 (align_functions_log): Redefine macro to use new types.
6585 (align_loops): Redefine macro to use new types.
6586 (align_jumps): Redefine macro to use new types.
6587 (align_labels): Redefine macro to use new types.
6588 (align_functions): Redefine macro to use new types.
6589 (align_functions_max_skip): Redefine macro to use new types.
6590 (align_loops_value): New macro.
6591 (align_jumps_value): New macro.
6592 (align_labels_value): New macro.
6593 (align_functions_value): New macro.
6594 * function.c (invoke_set_current_function_hook): Propagate
6595 alignment values from flags to global variables default in
6597 * ipa-icf.c (sem_function::equals_wpa): Use
6598 cl_optimization_option_eq instead of memcmp.
6599 * lto-streamer.h (cl_optimization_stream_out): Support streaming
6601 (cl_optimization_stream_in): Likewise.
6602 * optc-save-gen.awk: Support strings in cl_optimization.
6603 * opth-gen.awk: Likewise.
6604 * opts.c (finish_options): Remove error checking of invalid
6606 (MAX_CODE_ALIGN): Remove.
6607 (MAX_CODE_ALIGN_VALUE): Likewise.
6608 (parse_and_check_align_values): New function.
6609 (check_alignment_argument): Likewise.
6610 (common_handle_option): Use check_alignment_argument.
6611 * opts.h (parse_and_check_align_values): Declare.
6612 * toplev.c (init_alignments): Remove.
6613 (read_log_maxskip): New.
6614 (parse_N_M): Likewise.
6615 (parse_alignment_opts): Likewise.
6616 (backend_init_target): Remove usage of init_alignments.
6617 * toplev.h (parse_alignment_opts): Declare.
6618 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
6620 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
6621 * tree.c (cl_option_hasher::equal): New.
6622 * varasm.c: Use new global macros.
6624 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
6626 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
6627 Use a simpler align directive also if MAXSKIP = ALIGN-1.
6628 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6629 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6630 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6631 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
6632 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
6633 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
6634 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
6635 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6636 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6637 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6638 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6640 2018-07-04 Martin Liska <mliska@suse.cz>
6641 Jonathan Wakely <jwakely@redhat.com>
6643 * coverage.c: Use correct type.
6644 * doc/invoke.texi: Language correction.
6646 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
6649 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
6650 ENDBRANCH for non-tail call which may return via indirect branch.
6651 * doc/extend.texi: Document indirect_return attribute.
6653 2018-07-03 Martin Sebor <msebor@redhat.com>
6655 PR tree-optimization/86274
6656 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
6658 (format_floating): Correct handling of infinities and NaNs.
6660 2018-07-03 Martin Sebor <msebor@redhat.com>
6662 * print-tree.c (print_real_cst): New function.
6663 (print_node_brief): Call it.
6664 (print_node): Ditto.
6666 2018-07-03 Jeff Law <law@redhat.com>
6668 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
6669 into a single pattern.
6671 * config/h8300/h8300.md (ors code_iterator): New.
6672 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
6673 a single pattern and single splitter.
6674 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
6675 (iorqi3_1, xorqi3_1): Likewise.
6676 (iorqi3, xorqi3 expanders): Similarly.
6678 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
6679 (movmd_internal) into a single pattern using the P mode iterator.
6680 (movmd splitters): Similarly.
6681 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
6682 (movsd splitters): Similarly.
6684 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
6685 ADDB, ADDW and ADDL into a single ADD attribute which selects the
6686 right table based on the size of the operand.
6687 * config/h8300/h8300.md (length_table): Corresponding changes. All
6688 references to "addb", "addw" and "addl" changed to "add".
6689 (btst patterns): Merge two variants into a single pattern.
6690 (tstqi, tsthi): Likewise.
6691 (addhi3_incdec, addsi3_incdec): Likewise.
6692 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
6693 (mulhi3, mulsi3): Likewise.
6694 (udivhi3, udivsi3): Likewise.
6695 (divhi3, divsi3): Likewise.
6696 (andorqi3, andorhi3, andorsi3): Likewise.
6698 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
6701 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
6702 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
6704 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6706 PR tree-optimization/85694
6707 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
6708 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
6710 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
6711 (<u>avg<mode>3_ceil): New patterns.
6713 2018-07-03 David Malcolm <dmalcolm@redhat.com>
6715 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
6716 scan-tree-dump directive.
6717 * gcc.dg/vect/slp-perm-2.c: Likewise.
6718 * gcc.dg/vect/slp-perm-3.c: Likewise.
6719 * gcc.dg/vect/slp-perm-5.c: Likewise.
6720 * gcc.dg/vect/slp-perm-6.c: Likewise.
6721 * gcc.dg/vect/slp-perm-7.c: Likewise.
6722 * gcc.dg/vect/slp-perm-8.c: Likewise.
6724 2018-07-03 Marek Polacek <polacek@redhat.com>
6727 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
6729 2018-07-03 Richard Biener <rguenther@suse.de>
6732 * tree-ssa-structalias.c (find_func_clobbers): Properly
6733 handle indirect calls.
6735 2018-07-03 Jeff Law <law@redhat.com>
6737 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
6738 (shifts): New code iterator.
6739 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
6740 expander. Fix HImode handling on H8/SX.
6741 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
6742 (subqi3, subhi3, subsi3 expanders): Likewise.
6743 (andqi3, andhi3, andsi3 expanders): Likewise.
6744 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
6745 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
6746 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
6747 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
6748 (zero_extendqihi2, zero_extendqisi2): Likewise.
6749 (extendqihi2, extendqisi2): Likewise.
6750 (rotlqi3, rotlhi3, rotlsi3): Likewise.
6751 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
6752 (rotlqi3_1, rotlhi3_1): Likewise.
6753 (logicalhi3_sn, logicalsi3_sn): Likewise.
6754 (logicalhi3, logicalsi3): Likewise.
6756 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6758 * tree-vect-patterns.c (vect_recog_rotate_pattern)
6759 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
6760 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
6761 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
6762 type to append_pattern_def_seq instead of creating a stmt_vec_info
6764 (build_mask_conversion): Likewise. Remove vinfo argument.
6765 (vect_add_conversion_to_patterm): Likewise, renaming to...
6766 (vect_add_conversion_to_pattern): ...this.
6767 (vect_recog_mask_conversion_pattern): Update call to
6768 build_mask_conversion. Pass the vector type to
6769 append_pattern_def_seq here too.
6770 (vect_recog_gather_scatter_pattern): Update call to
6771 vect_add_conversion_to_pattern.
6773 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6775 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
6776 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
6777 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
6778 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
6779 STMT_VINFO_PATTERN_DEF_SEQ to null here.
6780 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
6781 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
6782 append_pattern_def_seq instead of new_pattern_def_seq.
6783 (vect_recog_divmod_pattern): Do both of the above.
6784 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
6787 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6789 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
6790 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
6791 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
6792 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
6793 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
6794 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
6795 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
6796 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
6797 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
6798 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
6799 parameter with a single stmt_vec_info.
6800 (vect_recog_func_ptr): Likewise.
6801 (vect_recog_gather_scatter_pattern): Likewise, folding in...
6802 (vect_try_gather_scatter_pattern): ...this.
6803 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
6804 the stmt_vec_info of the statement to be matched. Don't clear
6805 STMT_VINFO_RELATED_STMT.
6806 (vect_pattern_recog): Update call accordingly.
6808 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6810 PR tree-optimization/85694
6811 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
6812 (uavgM3_ceil): Document new optabs.
6813 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
6814 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
6816 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
6817 (savg_ceil_optab): New optabs.
6818 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
6819 (vect_vect_recog_func_ptrs): Add it.
6820 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
6821 constant directly from the associated lhs.
6823 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6825 * tree-vect-patterns.c (vect_split_statement): New function.
6826 (vect_convert_input): Use it to try to split an existing cast.
6828 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6830 * poly-int.h (print_hex): New function.
6831 * dumpfile.h (dump_dec, dump_hex): Declare.
6832 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
6833 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
6834 min_input_precision, operation_precision and operation_sign.
6835 * tree-vect-patterns.c (vect_get_range_info): New function.
6836 (vect_same_loop_or_bb_p, vect_single_imm_use)
6837 (vect_operation_fits_smaller_type): Delete.
6838 (vect_look_through_possible_promotion): Add an optional
6839 single_use_p parameter.
6840 (vect_recog_over_widening_pattern): Rewrite to use new
6841 stmt_vec_info infomration. Handle one operation at a time.
6842 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
6843 (vect_truncatable_operation_p, vect_set_operation_type)
6844 (vect_set_min_input_precision): New functions.
6845 (vect_determine_min_output_precision_1): Likewise.
6846 (vect_determine_min_output_precision): Likewise.
6847 (vect_determine_precisions_from_range): Likewise.
6848 (vect_determine_precisions_from_users): Likewise.
6849 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
6850 (vect_vect_recog_func_ptrs): Put over_widening first.
6852 (vect_pattern_recog): Call vect_determine_precisions.
6854 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6856 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
6857 statements that have been replaced by further pattern statements.
6858 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
6860 2018-07-03 Richard Biener <rguenther@suse.de>
6862 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
6863 always set *dt. Dump vectype in vectype overload.
6864 * dumpfile.h (dump_gimple_expr): New function.
6865 (dump_gimple_expr_loc): Likewise.
6866 * dumpfile.c (dump_gimple_expr): New function.
6867 (dump_gimple_expr_loc): Likewise.
6869 2018-07-02 Jeff Law <law@redhat.com>
6871 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
6872 the H8/300, H8/300H and H8/S variants into a single pattern.
6873 (movhi_h8300, movqi_h8300hs): Similarly.
6874 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
6875 (QHI mode iterator): New.
6877 * config/h8300/h8300.md: Remove trailing whitespace.
6879 2018-07-02 Jim Wilson <jimw@sifive.com>
6881 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
6882 instead of emit_insn for interrupt returns.
6883 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
6884 (riscv_sret, riscv_uret): Likewise.
6886 2018-07-02 David Malcolm <dmalcolm@redhat.com>
6888 * pretty-print.c (selftest::test_pp_format): Move save and restore
6889 of quotes to class auto_fix_quotes, and add an instance.
6890 * selftest.c: Include "intl.h".
6891 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
6892 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
6893 * selftest.h (selftest::auto_fix_quotes): New class.
6895 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
6897 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
6898 (aarch64_sve_prepare_conditional_op): Remove.
6899 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
6900 Allow aarch64_simd_reg_or_zero as select operand; remove
6901 the aarch64_sve_prepare_conditional_op call.
6902 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
6903 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
6904 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
6905 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
6906 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
6907 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
6908 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
6909 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
6910 and a splitters to match all of the *_any patterns.
6911 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
6913 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
6914 (SVE_COND_FP_BINARY_REV): Remove.
6915 (sve_int_op_rev, sve_fp_op_rev): New.
6916 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
6917 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
6918 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
6919 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
6920 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
6921 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
6922 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
6923 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
6924 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
6926 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
6927 Remove match_dup 1 from the inner unspec.
6928 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
6930 * config/aarch64/aarch64.md (movprfx): New attr.
6931 (length): Default movprfx to 8.
6932 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
6933 (*madd<SVE_I>, *msub<SVE_I): Likewise.
6934 (*<su>mul<SVE_I>3_highpart): Likewise.
6935 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
6936 (*v<ASHIFT><SVE_I>3): Likewise.
6937 (*<su><MAXMIN><SVE_I>3): Likewise.
6938 (*<su><MAXMIN><SVE_F>3): Likewise.
6939 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
6940 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
6941 (*div<SVE_F>4): Likewise.
6943 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
6945 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
6948 2018-07-02 Richard Biener <rguenther@suse.de>
6950 PR tree-optimization/86363
6951 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
6952 memset argument refers to a non-variable address.
6954 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
6956 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
6957 {PLUS,MINUS}_EXPR code to...
6958 (adjust_symbolic_bound): ...here,
6959 (combine_bound): ...here,
6960 (set_value_range_with_overflow): ...and here.
6962 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
6964 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
6966 (extract_range_from_abs_expr): ...here.
6968 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
6970 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
6971 -fno-omit-frame-pointer when not optimizing.
6973 2018-07-02 Martin Liska <mliska@suse.cz>
6976 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
6977 (propagate_nothrow): Likewise.
6979 2018-07-02 Martin Liska <mliska@suse.cz>
6982 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
6984 2018-07-02 David Malcolm <dmalcolm@redhat.com>
6986 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
6987 function in r262149, changing "loc" param from source_location to
6988 const dump_location_t &.
6989 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
6990 declaration, as above.
6992 2018-07-01 Paul Koning <ni1d@arrl.net>
6994 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
6995 -munit-asm, -mgnu-asm, -mdec-asm.
6996 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
6997 (pdp11_output_labelref): New.
6998 (pdp11_output_def): New.
6999 (pdp11_output_addr_vec_elt): New.
7000 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
7001 %# and %@ format codes.
7002 (pdp11_option_override): New.
7003 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
7004 (pdp11_output_ident): New.
7005 (pdp11_asm_named_section): New.
7006 (pdp11_asm_init_sections): New.
7007 (pdp11_file_start): New.
7008 (pdp11_file_end): New.
7009 (output_ascii): Use .ascii/.asciz for -mdec-asm.
7010 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
7011 %o, like %c but octal.
7012 (pdp11_option_override): New.
7013 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
7015 (DATA_SECTION_ASM_OP): Ditto.
7016 (READONLY_DATA_SECTION_ASM_OP): New.
7017 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
7018 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
7019 (ASM_OUTPUT_LABELREF): Ditto.
7020 (ASM_OUTPUT_DEF): Ditto.
7021 (ASM_OUTPUT_EXTERNAL): New.
7022 (ASM_OUTPUT_SOURCE_FILENAME): New.
7023 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
7024 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
7025 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
7026 %# and %@ format codes.
7027 * config/pdp11/pdp11.opt (mgnu-asm): New.
7028 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
7029 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
7030 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
7032 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
7034 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
7035 dereferencing path[] beyond its length.
7037 (debug_all_paths): New.
7038 (rewire_first_differing_edge): New.
7039 (adjust_paths_after_duplication): New.
7040 (duplicate_thread_path): Call adjust_paths_after_duplication.
7042 (thread_through_all_blocks): Add new argument to
7043 duplicate_thread_path.
7045 2018-06-30 Jim Wilson <jimw@sifive.com>
7047 * config/riscv/predicates.md (p2m1_shift_operand): New.
7048 (high_mask_shift_operand): New.
7049 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
7050 pattern using p2m1_shift_operand.
7051 (lshsi3_zero_extend_3+2): New combiner pattern using
7052 high_mask_shift_operand.
7054 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7056 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
7058 (vect_recog_rotate_pattern): ...here.
7059 (vect_convert_input): Try to insert casts of invariants in the
7061 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
7062 preheader to be empty.
7064 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7066 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
7067 vector type. If given, install it in the new statement's
7069 (vect_element_precision): New function.
7070 (vect_unpromoted_value): New struct.
7071 (vect_unpromoted_value::vect_unpromoted_value): New function.
7072 (vect_unpromoted_value::set_op): Likewise.
7073 (vect_look_through_possible_promotion): Likewise.
7074 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
7075 (vect_widened_op_tree, vect_convert_input): Likewise.
7076 (vect_convert_inputs, vect_convert_output): Likewise.
7077 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
7078 to handle the optional cast of the multiplication result and
7079 vect_widened_op_tree to detect the widened multiplication itself.
7080 Do not require the input and output of promotion casts to have
7081 the same sign, but base the signedness of the operation on the
7082 input rather than the result. If the pattern includes two
7083 promotions, check that those promotions have the same sign.
7084 Do not restrict the MULT_EXPR handling to a double-width result;
7085 handle quadruple-width results and wider. Use vect_convert_inputs
7086 to convert the inputs to the common type.
7087 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
7088 to handle the optional cast of the ABS result. Also allow a sign
7089 change or a sign extension between the ABS and MINUS.
7090 Use vect_widened_op_tree to detect the widened subtraction and use
7091 vect_convert_inputs to convert the inputs to the common type.
7092 (vect_handle_widen_op_by_const): Delete.
7093 (vect_recog_widen_op_pattern): New function.
7094 (vect_recog_widen_mult_pattern): Use it.
7095 (vect_recog_widen_shift_pattern): Likewise.
7096 (vect_recog_widen_sum_pattern): Use
7097 vect_look_through_possible_promotion to handle the promoted
7100 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7102 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
7103 the containing gimple_seq *.
7104 * gimple-iterator.h (gsi_for_stmt): Declare it.
7105 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
7106 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
7107 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
7108 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
7109 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
7111 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
7113 (vect_mark_pattern_stmts): ...here. Handle cases in which the
7114 statement being replaced is part of an existing pattern
7115 definition sequence, inserting the new pattern statements before
7117 (vect_pattern_recog_1): Don't return a bool. If the statement
7118 is already part of a pattern, instead apply pattern matching
7119 to the pattern definition statements. Don't clear the
7120 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
7121 (vect_pattern_recog): Don't break after the first match;
7122 continue processing the pattern definition statements instead.
7123 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
7125 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7127 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
7128 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
7129 (vect_recog_widen_sum_pattern): Use it.
7131 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7133 * tree-vect-loop.c (vectorizable_reduction): Assert that the
7134 phi is not a pattern statement and has not been replaced by
7135 a pattern statement.
7136 * tree-vect-patterns.c (type_conversion_p): Don't check
7137 STMT_VINFO_IN_PATTERN_P.
7138 (vect_recog_vector_vector_shift_pattern): Likewise.
7139 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
7140 the pattern statement rather than the original statement; check
7141 directly for a WIDEN_MULT_EXPR here.
7142 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
7143 vect_is_simple_use to return the pattern statement rather
7144 than the original statement; use is_pattern_stmt_p to check
7145 for such a pattern statement.
7146 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
7147 to return the pattern statement rather than the original statement;
7148 don't do the same transformation here.
7149 (vect_is_simple_use): If the defining statement has been replaced
7150 by a pattern statement, return the pattern statement instead.
7151 Remove the corresponding (local) transformation from the vectype
7154 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7156 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
7157 end and default to null.
7158 * tree-vect-loop.c (vect_create_epilog_for_reduction)
7159 (vectorizable_reduction): Update calls accordingly, dropping the
7160 gimple ** argument if the passed-back statement isn't needed.
7161 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
7162 (vect_recog_rotate_pattern): Likewise.
7163 (vect_recog_mask_conversion_pattern): Likewise.
7164 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
7165 (vect_mask_constant_operand_p): Likewise.
7166 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
7167 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
7168 (get_group_load_store_type, get_load_store_type): Likewise.
7169 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
7170 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
7171 (vectorizable_conversion, vectorizable_assignment): Likewise.
7172 (vectorizable_shift, vectorizable_operation): Likewise.
7173 (vectorizable_store, vect_is_simple_cond): Likewise.
7174 (vectorizable_condition, vectorizable_comparison): Likewise.
7175 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
7176 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
7177 and move it to the end. Cope with null def_stmt_outs.
7179 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
7181 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
7183 2018-06-29 Jeff Law <law@redhat.com>
7185 * config/v850/v850.c (v850_legitimate_address_p): Handle large
7186 displacements for TARGET_V850E2V3 and newer.
7187 (TARGET_LRA_P): Remove. Defaults to LRA now.
7188 * config/v850/v850.md (sign23byte_load): Remove.
7189 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
7190 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
7192 2018-06-29 Martin Liska <mliska@suse.cz>
7195 * coverage.c (coverage_init): Mangle full path name.
7196 * doc/invoke.texi: Document the change.
7197 * gcov-io.c (mangle_path): New.
7198 * gcov-io.h (mangle_path): Likewise.
7199 * gcov.c (mangle_name): Use mangle_path for path mangling.
7201 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7203 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
7204 if starting source register is not even.
7206 2018-06-29 Martin Liska <mliska@suse.cz>
7208 PR tree-optimization/86263
7209 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
7210 Make edge redirection.
7212 2018-06-29 David Malcolm <dmalcolm@redhat.com>
7214 * dumpfile.c (dump_loc): Add indentation based on scope depth.
7215 (dump_scope_depth): New variable.
7216 (get_dump_scope_depth): New function.
7217 (dump_begin_scope): New function.
7218 (dump_end_scope): New function.
7219 * dumpfile.h (get_dump_scope_depth): New declaration.
7220 (dump_begin_scope): New declaration.
7221 (dump_end_scope): New declaration.
7222 (class auto_dump_scope): New class.
7223 (AUTO_DUMP_SCOPE): New macro.
7224 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
7227 2018-06-29 Richard Biener <rguenther@suse.de>
7229 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
7230 compute_all_dependences succeeds.
7231 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
7232 exceed --param loop-max-datarefs-for-datadeps.
7234 2018-06-29 Jakub Jelinek <jakub@redhat.com>
7236 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
7238 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
7241 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
7242 alternative 0 in preferred_for_speed attribute.
7244 2018-06-28 Paul Koning <ni1d@arrl.net>
7246 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
7247 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
7248 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
7249 * config/pdp11/pdp11.md: Correct "length" attribute calculation
7250 for shift insn patterns.
7252 2018-06-28 David Malcolm <dmalcolm@redhat.com>
7254 * cgraph.c (cgraph_node::get_body): Replace assignments to
7255 "dump_file" with calls to set_dump_file.
7256 * dumpfile.c (alt_dump_file): Make static, and group with...
7257 (alt_flags): ...this definition.
7258 (dumps_are_enabled): New variable.
7259 (refresh_dumps_are_enabled): New function.
7260 (set_dump_file): New function.
7261 (set_alt_dump_file): New function.
7262 (gcc::dump_manager::dump_start): Replace assignments to
7263 "dump_file" and "alt_dump_file" with calls to set_dump_file and
7265 (gcc::dump_manager::dump_finish): Likewise.
7266 * dumpfile.h (alt_dump_file): Delete decl.
7267 (dumps_are_enabled): New variable decl.
7268 (set_dump_file): New function decl.
7269 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
7271 * tree-nested.c (lower_nested_functions): Replace assignments to
7272 "dump_file" with calls to set_dump_file.
7274 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
7276 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
7277 goto_locus of each outgoing edge of each basic block.
7279 2018-06-28 Richard Biener <rguenther@suse.de>
7281 * dwarf2out.c (decl_scope_table): Remove.
7282 (push_decl_scope): Likewise.
7283 (pop_decl_scope): Likewise.
7284 (gen_type_die_for_member): Do not call push/pop_decl_scope.
7285 (gen_struct_or_union_type_die): Likewise.
7286 (gen_tagged_type_die): Likewise.
7287 (dwarf2out_init): Do not initialize decl_scope_table.
7288 (dwarf2out_c_finalize): Do not free it.
7290 2018-06-28 Richard Biener <rguenther@suse.de>
7292 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
7293 deciding whether to not re-use a DIE.
7295 2018-06-28 Richard Biener <rguenther@suse.de>
7297 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
7298 DW_AT_abstract_origin attribute.
7300 2018-06-28 Martin Liska <mliska@suse.cz>
7302 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
7303 Use newly introduced constants.
7304 * tree-switch-conversion.h (struct jump_table_cluster):
7305 Define max_ratio_for_size and max_ratio_for_speed.
7307 2018-06-28 Martin Liska <mliska@suse.cz>
7309 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
7310 Add new checking assert to catch invalid state.
7311 (jump_table_cluster::can_be_handled): Handle single case
7313 (jump_table_cluster::is_beneficial): Bail out for such case.
7314 (bit_test_cluster::find_bit_tests):
7315 Add new checking assert to catch invalid state.
7316 (bit_test_cluster::can_be_handled): Handle single case
7318 (bit_test_cluster::is_beneficial): Bail out for such case.
7319 (switch_decision_tree::analyze_switch_statement):
7322 2018-06-28 Martin Liska <mliska@suse.cz>
7324 * common.opt: Introduce -completion option.
7325 * gcc.c (driver_handle_option): Handle it.
7326 (driver::main): Print completions if completion
7328 * opt-suggestions.c (option_proposer::get_completions):
7330 (option_proposer::suggest_completion): Likewise.
7331 (option_proposer::find_param_completions): Likewise.
7332 (verify_autocompletions): Likewise.
7333 (test_completion_valid_options): Likewise.
7334 (test_completion_valid_params): Likewise.
7335 (in_completion_p): Likewise.
7336 (empty_completion_p): Likewise.
7337 (test_completion_partial_match): Likewise.
7338 (test_completion_garbage): Likewise.
7339 (opt_proposer_c_tests): Likewise.
7340 * opt-suggestions.h: Declare new functions.
7341 * opts.c (common_handle_option): Handle OPT__completion_.
7342 * selftest-run-tests.c (selftest::run_tests): Add
7343 opt_proposer_c_tests.
7344 * selftest.c (assert_str_startswith): New.
7345 * selftest.h (assert_str_startswith): Likewise.
7346 (opt_proposer_c_tests): New.
7347 (ASSERT_STR_STARTSWITH): Likewise.
7349 2018-06-28 Martin Liska <mliska@suse.cz>
7351 * Makefile.in: Add opt-suggestions.o.
7352 * gcc-main.c: Include opt-suggestions.h.
7353 * gcc.c (driver::driver): Likewise.
7354 (driver::~driver): Remove m_option_suggestions.
7355 (driver::build_option_suggestions): Moved to option_proposer.
7356 (driver::suggest_option): Likewise.
7357 (driver::handle_unrecognized_options): Use option_proposer.
7358 * gcc.h (class driver): Add new memver m_option_proposer.
7359 * opt-suggestions.c: New file.
7360 * opt-suggestions.h: New file.
7362 2018-06-28 Martin Liska <mliska@suse.cz>
7364 * vec.h (class auto_string_vec): New (moved from auto_argvec).
7365 (auto_string_vec::~auto_string_vec): Likewise.
7367 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
7369 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
7370 prevent_decl_creation_for_types fields up and add reset_location field.
7371 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
7372 statement if id->reset_location is true.
7373 (copy_edges_for_bb): Do not set goto_locus on the new edges if
7374 id->reset_location is true.
7375 (copy_phis_for_bb): Force input_location on the arguments if
7376 id->reset_location is true.
7377 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
7378 is set on the function to be inlined.
7380 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
7382 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
7384 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
7386 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
7387 registers for Pmode.
7388 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
7389 hard registers for the clobbered pseudo.
7391 2018-06-27 Paul Koning <ni1d@arrl.net>
7393 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
7394 mutually exclusive options.
7395 * config/pdp11/constraints.md (h): New constraint.
7396 (O): Update definition to match shift code generation.
7397 (D): New constraint.
7398 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
7400 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
7402 (output_jump): Change arguments.
7403 (pdp11_fixed_cc_regs): New function.
7404 (pdp11_cc_mode): Ditto.
7405 (pdp11_expand_shift): Ditto.
7406 (pdp11_assemble_shift): Ditto.
7407 (pdp11_small_shift): Ditto.
7408 (pdp11_branch_cost): Remove.
7409 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
7411 (pdp11_register_move_cost): Update for CC registers.
7412 (pdp11_rtx_costs): Add case for LSHIFTRT.
7413 (pdp11_output_jump): Add CCNZ mode conditional branches.
7414 (notice_update_cc_on_set): Remove.
7415 (pdp11_cc_mode): New function.
7416 (simple_memory_operand): Correct pre/post decrement case.
7417 (no_side_effect_operand): New function.
7418 (pdp11_regno_reg_class): Add CC_REGS class.
7419 (pdp11_fixed_cc_regs): New function.
7420 (pdp11_small_shift): New function.
7421 (pdp11_expand_shift): New function to expand shift insns.
7422 (pdp11_assemble_shift): New function to output shifts.
7423 (pdp11_branch_cost): Remove.
7424 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
7425 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
7426 (WCHAR_TYPE): Ditto.
7427 (PTRDIFF_TYPE): Ditto.
7428 (ADJUST_INSN_LENGTH): New macro.
7429 (FIXED_REGISTERS): Add CC registers.
7430 (CALL_USED_REGISTERS): Ditto.
7432 (REG_CLASS_NAMES): Ditto.
7433 (REG_CLASS_CONTENTS): Ditto.
7434 (SELECT_CC_MODE): Use new function.
7435 (TARGET_FLAGS_REGNUM): New macro.
7436 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
7437 (cc0_reg_rtx): Remove.
7438 (CC_STATUS_MDEP): Remove.
7439 (CC_STATUS_MDEFP_INIT): Remove.
7440 (CC_IN_FPU): Remove.
7441 (NOTICE_UPDATE_CC): Remove.
7442 (REGISTER_NAMES): Add CC registers.
7443 (BRANCH_COST): Change to constant 1.
7444 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
7446 * config/pdp11/pdp11.opt (mbcopy): Remove.
7447 (mbcopy-builtin): Remove.
7448 (mbranch-cheap): Remove.
7449 (mbranch-expensive): Remove.
7450 * config/pdp11/predicates.md (expand_shift_operand): Update to
7451 match shift code generation.
7452 (ccnz_operator): New predicate.
7453 * doc/invoke.texi (PDP-11 Options): Remove deleted options
7454 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
7455 Remove non-existent option -mabshi, -mno-abshi. Document mutually
7457 * doc/md.texi (PDP-11): Document new D and h constraints. Update
7458 description of O constraint.
7460 2018-06-27 Jeff Law <law@redhat.com>
7461 Austin Law <austinklaw@gmail.com>
7463 * config/v850/v850.md (addsi3_set_flags): New pattern.
7464 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
7465 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
7466 (zero_extendhisi2_v850_set_flags): Likewise.
7467 (zero_extendqisi2_v850_set_flags): Likewise.
7468 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
7469 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
7470 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
7472 * config/v850/v850-protos.h (notice_update_cc): Remove.
7473 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
7474 (v850_print_operand): Handle 'D' and "d".
7475 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
7476 Add handling of arithmetic/logical operations compared against zero.
7477 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
7478 Do not look at v850_compare_op, instead get mode from last argument.
7479 (v850_gen_compare): Remove
7480 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
7481 after reload for prologue insns.
7482 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
7484 (construct_save_jarl): Likewise.
7485 (TARGET_FLAGS_REGNUM): Define.
7486 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
7487 (NOTICE_UPDATE_CC): Remove.
7488 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
7489 than cc0. Conditionalize on reload_completed.
7490 (cmpsi_insn, setfcc_insn): Likewise.
7491 (tst1 splitter): Turn into define_and_split which sets the flags
7493 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
7494 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
7495 (cstoresf4, cstoredf4): Clobber the flags.
7496 (cmpsi, cmpsf, cmpdf): Remove expanders.
7497 (setf_insn): Remove pattern.
7498 (addsi3): Turn into define_and_split which clobbers the flags after
7499 reload and a suitable pattern (addsi3_clobber_flags) for use after
7501 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
7502 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
7503 (ashrsi3, ashrsi3_v850e2): Likewise.
7504 (bins): Clobber the flags.
7505 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
7506 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
7507 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
7508 (call_value_internal_short, call_value_internal_long): Likewise.
7509 (callt_save_interrupt, callt_return_interrupt): Likewise.
7510 (save_interrupt, return_interrupt): Likewise.
7511 (callt_save_all_interrupt, save_all_interrupt): Likewise.
7512 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
7513 (restore_all_interrupt, _restore_all_interrupt): Likewise.
7514 (All FP comparisons): Only allow after reload has completed.
7516 (divh, divhu): Tweak output template.
7517 (branch_z_normal, branch_z_invert): Remove
7518 (branch_nz_normal, branch_nz_invert): Likewise.
7519 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
7521 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
7522 * config/v850/v850.c (notice_update_cc): Remove.
7523 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
7524 (CC_NO_CARRY): Likewise.
7525 (NOTICE_UPDATE_CC): Define to nothing.
7526 * config/v850/v850.md: Remove block comment on cc0 handling
7527 Remove "cc" attribute from all patterns. Remove cc_status handling
7528 from all patterns. Minor formatting fixes.
7530 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7532 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
7533 (cortex-a76.cortex-a55): Likewise.
7534 * config/aarch64/aarch64-tune.md: Regenerate.
7535 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
7536 cortex-a76.cortex-a55.
7538 2018-06-27 Jeff Law <law@redhat.com>
7540 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
7541 (MULTILIB_DIRNAMES): Similarly.
7543 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
7545 * gimple.h (gimple_return_retbnd): Delete.
7546 (gimple_return_set_retbnd): Likewise.
7547 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
7548 gimple_return_set_retbnd.
7549 * gimple-pretty-print.c (dump_gimple_return): Remove call to
7550 gimple_return_retbnd and adjust.
7551 * tree-inline.h (struct copy_body_data): Remove retbnd field.
7552 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
7553 Explicitly return NULL in a couple more cases. Move assertion
7554 on debug statements and remove unreachable code.
7555 (reset_debug_binding): Do not test id->retbnd.
7556 (expand_call_inline): Do not set it.
7558 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
7560 * configure.ac: Add --disable-gcov option.
7561 * configure: Regenerate.
7562 * Makefile.in: Honour @enable_gcov@.
7563 * doc/install.texi: Document --disable-gcov.
7565 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7567 * config/arm/arm-cpus.in (cortex-a76): New entry.
7568 (cortex-a76.cortex-a55): Likewise.
7569 * config/arm/arm-tables.opt: Regenerate.
7570 * config/arm/arm-tune.md: Likewise.
7571 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
7572 * doc/invoke.texi (ARM Options): Document cortex-a76 and
7573 cortex-a76.cortex-a55.
7575 2018-06-27 Tamar Christina <tamar.christina@arm.com>
7578 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
7580 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
7582 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
7584 (EPILOGUE_USES): Likewise.
7586 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
7588 * tree-inline.c (remap_location): New function extracted from...
7589 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
7590 (copy_phis_for_bb): ...here. Call remap_location.
7591 (copy_cfg_body): Adjust call to copy_edges_for_bb.
7593 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
7595 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
7596 unaligned vsx for 16B memset.
7598 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
7601 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
7602 ieee128_float_type_node to long_double_type_node unless
7603 TARGET_LONG_DOUBLE_128 is set.
7605 2018-06-26 David Malcolm <dmalcolm@redhat.com>
7607 * cfgloop.c (get_loop_location): Convert return type from
7608 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
7609 by implicit construction from rtx_insn *, and using
7610 dump_user_location_t::from_function_decl for the fallback case.
7611 * cfgloop.h (get_loop_location): Convert return type from
7612 location_t to dump_user_location_t.
7613 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
7614 dump_printf_loc to pass in a dump_location_t rather than a
7615 location_t, via the gimple stmt.
7616 * coverage.c (get_coverage_counts): Update calls to
7617 dump_printf_loc to pass in dump_location_t rather than a
7619 * doc/optinfo.texi (Dump types): Convert example of
7620 dump_printf_loc from taking "locus" to taking "insn". Update
7621 description of the "_loc" calls to cover dump_location_t.
7622 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
7624 (dump_user_location_t::dump_user_location_t): New constructors,
7625 from gimple *stmt and rtx_insn *.
7626 (dump_user_location_t::from_function_decl): New function.
7627 (dump_loc): Make static.
7628 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
7629 const dump_location_t &.
7630 (dump_generic_expr_loc): Delete.
7631 (dump_printf_loc): Convert param "loc" from location_t to
7632 const dump_location_t &.
7633 (selftest::test_impl_location): New function.
7634 (selftest::dumpfile_c_tests): New function.
7635 * dumpfile.h: Include "profile-count.h".
7636 (class dump_user_location_t): New class.
7637 (struct dump_impl_location_t): New struct.
7638 (class dump_location_t): New class.
7639 (dump_printf_loc): Convert 2nd param from source_location to
7640 const dump_location_t &.
7641 (dump_generic_expr_loc): Delete.
7642 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
7643 const dump_location_t &.
7644 * gimple-fold.c (fold_gimple_assign): Update call to
7645 dump_printf_loc to pass in a dump_location_t rather than a
7646 location_t, via the gimple stmt.
7647 (gimple_fold_call): Likewise.
7648 * gimple-loop-interchange.cc
7649 (loop_cand::analyze_iloop_reduction_var): Update for change to
7650 check_reduction_path.
7651 (tree_loop_interchange::interchange): Update for change to
7653 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
7654 change in return-type of find_loop_location.
7655 (graphite_regenerate_ast_isl): Likewise.
7656 * graphite-optimize-isl.c (optimize_isl): Likewise.
7657 * graphite.c (graphite_transform_loops): Likewise.
7658 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
7659 pass in a dump_location_t rather than a location_t, via the
7661 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
7662 * ipa.c (walk_polymorphic_call_targets): Likewise.
7663 * loop-unroll.c (report_unroll): Convert "locus" param from
7664 location_t to dump_location_t.
7665 (decide_unrolling): Update for change to get_loop_location's
7667 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
7668 location_t to dump_user_location_t.
7669 (grid_find_single_omp_among_assignments_1): Updates calls to
7670 dump_printf_loc to pass in a dump_location_t rather than a
7671 location_t, via the gimple stmt.
7672 (grid_parallel_clauses_gridifiable): Convert "tloc" from
7673 location_t to dump_location_t. Updates calls to dump_printf_loc
7674 to pass in a dump_location_t rather than a location_t, via the
7676 (grid_inner_loop_gridifiable_p): Likewise.
7677 (grid_dist_follows_simple_pattern): Likewise.
7678 (grid_gfor_follows_tiling_pattern): Likewise.
7679 (grid_target_follows_gridifiable_pattern): Likewise.
7680 (grid_attempt_target_gridification): Convert initialization
7681 of local "grid" from memset to zero-initialization; FIXME: does
7682 this require C++11? Update call to dump_printf_loc to pass in a
7683 optinfo_location rather than a location_t, via the gimple stmt.
7684 * profile.c (read_profile_edge_counts): Updates call to
7685 dump_printf_loc to pass in a dump_location_t rather than a
7687 (compute_branch_probabilities): Likewise.
7688 * selftest-run-tests.c (selftest::run_tests): Call
7690 * selftest.h (dumpfile_c_tests): New decl.
7691 * tree-loop-distribution.c (pass_loop_distribution::execute):
7692 Update for change in return type of find_loop_location.
7693 * tree-parloops.c (parallelize_loops): Likewise.
7694 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
7695 "locus" from location_t to dump_user_location_t.
7696 (canonicalize_loop_induction_variables): Likewise.
7697 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
7698 for change in return type of find_loop_location.
7699 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
7700 to dump_printf_loc to pass in a dump_location_t rather than a
7701 location_t, via the stmt.
7702 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
7704 * tree-vect-loop-manip.c (find_loop_location): Convert return
7705 type from source_location to dump_user_location_t.
7706 (vect_do_peeling): Update for above change.
7707 (vect_loop_versioning): Update for change in type of
7709 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
7710 from location_t to dump_user_location_t.
7711 (vect_estimate_min_profitable_iters): Update for change in type
7713 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
7714 location_t to dump_location_t.
7715 (vect_slp_bb): Update for change in type of vect_location.
7716 * tree-vectorizer.c (vect_location): Convert from source_location
7717 to dump_user_location_t.
7718 (try_vectorize_loop_1): Update for change in vect_location's type.
7719 (vectorize_loops): Likewise.
7720 (increase_alignment): Likewise.
7721 * tree-vectorizer.h (vect_location): Convert from source_location
7722 to dump_user_location_t.
7723 (find_loop_location): Convert return type from source_location to
7724 dump_user_location_t.
7725 (check_reduction_path): Convert 1st param from location_t to
7726 dump_user_location_t.
7727 * value-prof.c (check_counter): Update call to dump_printf_loc to
7728 pass in a dump_user_location_t rather than a location_t; update
7729 call to error_at for change in type of "locus".
7730 (check_ic_target): Update call to dump_printf_loc to
7731 pass in a dump_user_location_t rather than a location_t, via the
7734 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
7736 * config/s390/s390.h (enum processor_flags): Do not use
7737 default tune parameter when -march was specified.
7739 2018-06-26 Jakub Jelinek <jakub@redhat.com>
7742 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
7743 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
7746 2018-06-26 Richard Biener <rguenther@suse.de>
7748 PR tree-optimization/86287
7750 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
7751 (vect_analyze_loop): Initialize n_stmts.
7753 2018-06-26 Richard Biener <rguenther@suse.de>
7756 * fold-const.c (fold_convertible_p): Pointer extension
7759 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
7762 * dwarf2out.c (loc_list_has_views): Adjust comments.
7763 (dw_loc_list): Split single cross-partition range with
7766 2018-06-25 Jeff Law <law@redhat.com>
7768 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
7769 on -mbig-switch by default.
7771 * config/v850/predicates.md (const_float_1_operand): Fix match_code
7773 (const_float_0_operand): Remove unused predicate.
7774 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
7775 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
7776 (recipsf2): New expander. Original pattern now called
7778 (recipdf2, recipdf2_insn): Similarly.
7779 (rsqrtsf2, rsqrtsf2_insn): Similarly
7780 (rsqrtdf2, rsqrtdf2_insn): Similarly
7782 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
7784 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
7785 Simplify logic for FreeBSD (twice).
7787 2018-06-25 Martin Sebor <msebor@redhat.com>
7789 PR tree-optimization/86204
7790 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
7791 a strnlen result if it's less than the length of the string.
7793 2018-06-25 Martin Sebor <msebor@redhat.com>
7795 PR tree-optimization/85700
7796 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
7797 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
7798 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
7800 2018-06-25 Martin Sebor <msebor@redhat.com>
7802 * doc/extend.texi (Zero-length arrays): Update and clarify.
7804 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
7806 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
7807 added IEEE/IBM long double multilib support on PowerPC little
7808 endian Linux systems.
7809 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
7810 (MULTILIB_DEFAULTS): Likewise.
7811 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7813 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
7814 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
7815 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
7817 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
7820 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
7821 (REORDER_45): Likewise.
7823 2018-06-25 Jeff Law <law@redhat.com>
7825 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
7826 dividend to 32 bits. Adjust length.
7827 (udivmodhi4): Cleanup output template. Fix length.
7829 2018-06-25 Carl Love <cel@us.ibm.com>
7831 * config/rs6000/vsx.md: Change word selector to prefered location.
7833 2018-06-25 Richard Biener <rguenther@suse.de>
7835 PR tree-optimization/86304
7836 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
7837 epilogue-if-converted loops as well.
7839 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
7841 * lto-section-out.c (lto_begin_section): Do not print section
7842 name for noaddr and unnumbered dumps.
7844 2018-06-25 Richard Biener <rguenther@suse.de>
7846 * tree-vectorizer.h (struct vec_info_shared): New structure
7847 with parts split out from struct vec_info and loop_nest from
7848 struct _loop_vec_info.
7849 (struct vec_info): Adjust accordingly.
7850 (struct _loop_vec_info): Likewise.
7851 (LOOP_VINFO_LOOP_NEST): Adjust.
7852 (LOOP_VINFO_DATAREFS): Likewise.
7853 (LOOP_VINFO_DDRS): Likewise.
7854 (struct _bb_vec_info): Likewise.
7855 (BB_VINFO_DATAREFS): Likewise.
7856 (BB_VINFO_DDRS): Likewise.
7857 (struct _stmt_vec_info): Add dr_aux member.
7858 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
7859 (DR_MISALIGNMENT_UNINITIALIZED): New.
7860 (set_dr_misalignment): Adjust.
7861 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
7862 (vect_analyze_loop): Adjust prototype.
7863 (vect_analyze_loop_form): Likewise.
7864 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
7865 Compute dependences lazily.
7866 (vect_record_base_alignments): Use shared datarefs/ddrs.
7867 (vect_verify_datarefs_alignment): Likewise.
7868 (vect_analyze_data_refs_alignment): Likewise.
7869 (vect_analyze_data_ref_accesses): Likewise.
7870 (vect_analyze_data_refs): Likewise.
7871 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
7872 constructor parameter for shared part.
7873 (vect_analyze_loop_form): Pass in shared part and adjust.
7874 (vect_analyze_loop_2): Pass in storage for the number of
7875 stmts. Move loop nest finding to the caller. Compute
7877 (vect_analyze_loop): Pass in shared part.
7878 (vect_transform_loop): Verify shared datarefs are unchanged.
7879 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
7880 constructor parameter for shared part.
7881 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
7882 (vect_slp_bb): Verify shared datarefs are unchanged before
7884 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
7886 (new_stmt_vec_info): Initialize DR_AUX misalignment to
7887 DR_MISALIGNMENT_UNINITIALIZED.
7888 * tree-vectorizer.c (vec_info::vec_info): Add constructor
7889 parameter for shared part.
7890 (vec_info::~vec_info): Adjust.
7891 (vec_info_shared::vec_info_shared): New.
7892 (vec_info_shared::~vec_info_shared): Likewise.
7893 (vec_info_shared::save_datarefs): Likewise.
7894 (vec_info_shared::check_datarefs): Likewise.
7895 (try_vectorize_loop_1): Construct shared part live for analyses
7896 of a single loop for multiple vector sizes.
7897 * tree-parloops.c (gather_scalar_reductions): Adjust.
7899 2018-06-25 Richard Biener <rguenther@suse.de>
7901 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
7902 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
7903 (vect_analyze_data_refs): Remove similar code from here and
7904 simplify accordingly.
7906 2018-06-25 Richard Biener <rguenther@suse.de>
7908 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
7909 for reverse storage order accesses rather than asserting
7910 they cannot happen here.
7912 2018-06-25 Tom de Vries <tdevries@suse.de>
7915 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
7916 Use data16 instead of .byte for insn prefix.
7918 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
7921 * parser.c (make_char_string_pack): Pass this literal chars
7922 through cpp_interpret_string.
7923 (cp_parser_userdef_numeric_literal): Check the result of
7924 make_char_string_pack.
7926 2018-06-24 Maya Rashish <coypu@sdf.org>
7928 * ginclude/stddef.h: Simplify conditions around avoiding
7929 re-definition of __size_t.
7931 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
7933 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
7934 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
7936 2018-06-22 Maya Rashish <coypu@sdf.org>
7938 * doc/invoke.texi (mno-fancy-math-387): Update for changes
7939 made to OpenBSD and NetBSD through the years.
7941 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
7943 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
7944 behavior of vec_pack (vector double, vector double) to match
7945 behavior of vec_float2 (vector double, vector double).
7947 2018-06-22 Olivier Hainque <hainque@adacore.com>
7949 * gimplify.c (gimplify_function_tree): Prevent creation
7950 of a trampoline for the address of the current function
7951 passed to entry/exit instrumentation hooks.
7953 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
7956 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
7959 2018-06-22 Martin Liska <mliska@suse.cz>
7961 PR tree-optimization/86263
7962 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
7963 Bail out if is_enabled is false.
7964 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
7966 (jump_table_cluster::is_enabled): New function.
7968 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
7970 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
7971 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
7972 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
7973 (lto_input_ts_binfo_tree_pointers): Likewise.
7974 * tree-streamer-out.c (streamer_write_tree_bitfields,
7975 write_ts_binfo_tree_pointers): Likewise.
7976 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
7978 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
7980 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
7982 2018-06-22 Martin Liska <mliska@suse.cz>
7984 * symbol-summary.h (get): Make it pure and inline move
7985 functionality from ::get function.
7986 (get): Remove and inline into ::get and ::get_create.
7987 (get_create): Move code from ::get function.
7989 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7992 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
7993 -x assembler-with-cpp.
7995 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7997 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
7998 _FILE_OFFSET_BITS=64 for C++.
8000 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
8002 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
8003 conversion insn that shows up when pr85657-3.c is compiled using
8004 IEEE 128-bit long double.
8005 (neg<mode>2_internal): Use the correct mode to check whether the
8006 mode is IBM extended.
8007 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
8008 multiply and divide external functions from being created more
8011 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
8013 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
8015 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
8016 the edge can be forwarded.
8017 (cfg_layout_merge_blocks): Likewise.
8019 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
8021 * except.c (finish_eh_generation): Commit edge insertions only after
8022 the EH edges have been redirected from post-landing to landing pads.
8024 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
8026 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
8027 create_tmp_var_for to create the FRAME decl.
8028 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
8030 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
8032 * tree-inline.c (copy_edges_for_bb): Minor tweak.
8033 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
8034 debug statement when resetting its value.
8035 (expand_call_inline): Copy the locus of the call onto the assignment
8036 of the return value, if any. Use local variable in more cases.
8038 2018-06-21 Martin Liska <mliska@suse.cz>
8040 * ipa-pure-const.c (propagate_nothrow): Use
8041 funct_state_summaries->get.
8042 (dump_malloc_lattice): Likewise.
8043 (propagate_malloc): Likewise.
8045 2018-06-21 Richard Biener <rguenther@suse.de>
8047 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
8048 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
8049 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
8051 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
8052 BLOCK_ABSTRACT_ORIGIN unconditionally.
8054 2018-06-21 David Malcolm <dmalcolm@redhat.com>
8056 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
8058 * ipa-reference.c (ipa_reference_c_finalize): Delete
8059 ipa_ref_opt_sum_summaries and set it to NULL.
8061 2018-06-21 Tom de Vries <tdevries@suse.de>
8063 PR tree-optimization/85859
8064 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
8065 test with comment from bb_no_side_effects_p.
8067 2018-06-21 Richard Biener <rguenther@suse.de>
8069 PR tree-optimization/86232
8070 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
8071 max for constant niter.
8073 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
8075 * config/aarch64/aarch64-simd.md
8076 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
8078 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
8080 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
8081 Make opernads of the unspec commutative.
8083 2018-06-21 Richard Biener <rguenther@suse.de>
8085 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
8086 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
8087 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
8088 (vect_analyze_data_ref_dependence): Re-order checks to deal with
8090 (vect_record_base_alignments): Do not record base alignment
8091 for gathers or scatters.
8092 (vect_compute_data_ref_alignment): Drop return value that is always
8093 true. Bail out early for gathers or scatters.
8094 (vect_enhance_data_refs_alignment): Bail out early for gathers
8096 (vect_find_same_alignment_drs): Likewise.
8097 (vect_analyze_data_refs_alignment): Remove dead code.
8098 (vect_slp_analyze_and_verify_node_alignment): Likewise.
8099 (vect_analyze_data_refs): For possible gathers or scatters do
8100 not create an alternate DR, just check their possible validity
8101 and mark them. Adjust DECL_NONALIASED handling to not rely
8103 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
8104 update inits of gathers or scatters.
8105 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
8106 Also copy gather/scatter flag to pattern vinfo.
8108 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
8110 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
8111 behavior of vec_packsu (vector unsigned long long, vector unsigned
8112 long long) to match behavior of vec_packs with same signature.
8114 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
8115 Thomas Schwinge <thomas@codesourcery.com>
8116 Cesar Philippidis <cesar@codesourcery.com>
8118 * gimplify.c (gimplify_scan_omp_clauses): Add support for
8119 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
8120 (gimplify_adjust_omp_clauses): Likewise.
8121 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
8122 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
8123 (gimplify_omp_target_update): Update handling of acc update and
8125 * omp-low.c (install_var_field): Remove unused parameter
8126 base_pointers_restrict.
8127 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
8128 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
8130 (omp_target_base_pointers_restrict_p): Delete.
8131 (scan_omp_target): Update call to scan_sharing_clauses.
8132 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
8134 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
8135 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
8136 (convert_local_omp_clauses): Likewise.
8137 * tree-pretty-print.c (dump_omp_clause): Likewise.
8138 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
8140 (omp_clause_code_name): Likewise.
8142 2018-06-20 Jakub Jelinek <jakub@redhat.com>
8145 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
8148 PR tree-optimization/86231
8149 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
8150 anti-range don't overwrite *vr0min before using it to compute *vr0max.
8152 2018-06-20 Tom de Vries <tdevries@suse.de>
8154 PR tree-optimization/86097
8155 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
8156 iv type if signedness of iv type is not the same as that of *nit.
8158 2018-06-20 Jakub Jelinek <jakub@redhat.com>
8160 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
8161 EDGE_EH edges, verify they are all EDGE_EH.
8163 2018-06-20 Maya Rashish <coypu@sdf.org>
8165 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
8167 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8169 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
8170 * config/aarch64/aarch64.c (xgene1_tunings): Add
8171 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
8172 (aarch64_mode_valid_for_sched_fusion_p):
8173 Allow 16-byte modes.
8174 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
8175 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
8177 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
8179 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
8180 * config/aarch64/iterators.md (VQ2): New mode iterator.
8182 2018-06-20 Martin Liska <mliska@suse.cz>
8184 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
8185 Change default ratio from 10 to 8.
8187 2018-06-20 Martin Liska <mliska@suse.cz>
8189 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
8191 (bit_test_cluster::find_bit_tests): Likewise.
8192 (switch_decision_tree::analyze_switch_statement): Find clusters.
8193 * tree-switch-conversion.h (struct jump_table_cluster): Document
8196 2018-06-20 Martin Liska <mliska@suse.cz>
8198 * tree-switch-conversion.c (switch_conversion::collect):
8199 Record m_uniq property.
8200 (switch_conversion::expand): Bail out for special conditions.
8201 (group_cluster::~group_cluster): New.
8202 (group_cluster::group_cluster): Likewise.
8203 (group_cluster::dump): Likewise.
8204 (jump_table_cluster::emit): New.
8205 (switch_decision_tree::fix_phi_operands_for_edges): New.
8206 (struct case_node): Remove struct.
8207 (jump_table_cluster::can_be_handled): New.
8208 (case_values_threshold): Moved to header.
8209 (reset_out_edges_aux): Likewise.
8210 (jump_table_cluster::is_beneficial): New.
8211 (bit_test_cluster::can_be_handled): Likewise.
8212 (add_case_node): Remove.
8213 (bit_test_cluster::is_beneficial): New.
8214 (case_bit_test::cmp): New.
8215 (bit_test_cluster::emit): New.
8216 (expand_switch_as_decision_tree_p): Remove.
8217 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
8218 (fix_phi_operands_for_edge): Likewise.
8219 (switch_decision_tree::analyze_switch_statement): New.
8220 (compute_cases_per_edge): Move ...
8221 (switch_decision_tree::compute_cases_per_edge): ... here.
8222 (try_switch_expansion): Likewise.
8223 (switch_decision_tree::try_switch_expansion): Likewise.
8224 (record_phi_operand_mapping): Likewise.
8225 (switch_decision_tree::record_phi_operand_mapping): Likewise.
8226 (emit_case_decision_tree): Likewise.
8227 (switch_decision_tree::emit): Likewise.
8228 (balance_case_nodes): Likewise.
8229 (switch_decision_tree::balance_case_nodes): Likewise.
8230 (dump_case_nodes): Likewise.
8231 (switch_decision_tree::dump_case_nodes): Likewise.
8232 (emit_jump): Likewise.
8233 (switch_decision_tree::emit_jump): Likewise.
8234 (emit_cmp_and_jump_insns): Likewise.
8235 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
8236 (emit_case_nodes): Likewise.
8237 (switch_decision_tree::emit_case_nodes): Likewise.
8238 (conditional_probability): Remove.
8239 * tree-switch-conversion.h (enum cluster_type): New.
8241 (struct cluster): Likewise.
8242 (cluster::cluster): Likewise.
8243 (struct simple_cluster): Likewise.
8244 (simple_cluster::simple_cluster): Likewise.
8245 (struct group_cluster): Likewise.
8246 (struct jump_table_cluster): Likewise.
8247 (struct bit_test_cluster): Likewise.
8248 (struct min_cluster_item): Likewise.
8249 (struct case_tree_node): Likewise.
8250 (case_tree_node::case_tree_node): Likewise.
8251 (jump_table_cluster::case_values_threshold): Likewise.
8252 (struct case_bit_test): Likewise.
8253 (struct switch_decision_tree): Likewise.
8254 (struct switch_conversion): Likewise.
8255 (switch_decision_tree::reset_out_edges_aux): Likewise.
8257 2018-06-20 Martin Liska <mliska@suse.cz>
8259 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
8260 (hoist_edge_and_branch_if_true): Likewise.
8261 (expand_switch_using_bit_tests_p): Likewise.
8262 (struct case_bit_test): Likewise.
8263 (case_bit_test_cmp): Likewise.
8264 (emit_case_bit_tests): Likewise.
8265 (switch_conversion::switch_conversion): New class.
8266 (struct switch_conv_info): Remove old struct.
8267 (collect_switch_conv_info): More to ...
8268 (switch_conversion::collect): ... this.
8269 (check_range): Likewise.
8270 (switch_conversion::check_range): Likewise.
8271 (check_all_empty_except_final): Likewise.
8272 (switch_conversion::check_all_empty_except_final): Likewise.
8273 (check_final_bb): Likewise.
8274 (switch_conversion::check_final_bb): Likewise.
8275 (create_temp_arrays): Likewise.
8276 (switch_conversion::create_temp_arrays): Likewise.
8277 (free_temp_arrays): Likewise.
8278 (gather_default_values): Likewise.
8279 (switch_conversion::gather_default_values): Likewise.
8280 (build_constructors): Likewise.
8281 (switch_conversion::build_constructors): Likewise.
8282 (constructor_contains_same_values_p): Likewise.
8283 (switch_conversion::contains_same_values_p): Likewise.
8284 (array_value_type): Likewise.
8285 (switch_conversion::array_value_type): Likewise.
8286 (build_one_array): Likewise.
8287 (switch_conversion::build_one_array): Likewise.
8288 (build_arrays): Likewise.
8289 (switch_conversion::build_arrays): Likewise.
8290 (gen_def_assigns): Likewise.
8291 (switch_conversion::gen_def_assigns): Likewise.
8292 (prune_bbs): Likewise.
8293 (switch_conversion::prune_bbs): Likewise.
8294 (fix_phi_nodes): Likewise.
8295 (switch_conversion::fix_phi_nodes): Likewise.
8296 (gen_inbound_check): Likewise.
8297 (switch_conversion::gen_inbound_check): Likewise.
8298 (process_switch): Use the newly created class.
8299 (switch_conversion::expand): New.
8300 (switch_conversion::~switch_conversion): New.
8301 * tree-switch-conversion.h: New file.
8303 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8305 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
8306 tree-vect-patterns.c.
8307 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
8308 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
8309 (vect_recog_sad_pattern): Likewise.
8310 (vect_recog_widen_sum_pattern): Likewise.
8311 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
8312 (vect_recog_widen_shift_pattern): Remove the type_in argument.
8313 (vect_recog_rotate_pattern): Likewise.
8314 (vect_recog_mult_pattern): Likewise.
8315 (vect_recog_vector_vector_shift_pattern): Likewise.
8316 (vect_recog_divmod_pattern): Likewise.
8317 (vect_recog_mixed_size_cond_pattern): Likewise.
8318 (vect_recog_bool_pattern): Likewise.
8319 (vect_recog_mask_conversion_pattern): Likewise.
8320 (vect_try_gather_scatter_pattern): Likewise.
8321 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
8322 (vect_recog_over_widening_pattern): Likewise.
8323 (vect_recog_gather_scatter_pattern): Likewise.
8324 (vect_recog_func_ptr): Move from tree-vectorizer.h
8325 (vect_vect_recog_func_ptrs): Move further down the file.
8326 (vect_recog_func): Likewise. Remove the third argument.
8327 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
8328 (vect_pattern_recog_1): Expect the pattern function to do any
8329 necessary target tests. Also expect it to provide a vector type.
8330 Remove the type_in handling.
8332 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8334 * tree-vect-patterns.c (vect_pattern_detected): New function.
8335 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
8336 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
8337 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
8338 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
8339 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
8340 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
8341 (vect_recog_mask_conversion_pattern)
8342 (vect_try_gather_scatter_pattern): Likewise.
8344 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8346 * tree-vect-patterns.c (vect_get_internal_def): New function.
8347 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
8348 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
8349 (search_type_for_mask_1): Use it.
8351 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8353 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
8354 redundant WIDEN_SUM_EXPR handling.
8355 (vect_recog_sad_pattern): Likewise.
8357 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8359 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
8360 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
8361 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
8362 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
8363 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
8365 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8367 * tree-vect-stmts.c (vectorizable_call): Make sure that we
8368 use the stmt_vec_info of the original bb statement for the
8369 new zero assignment, even if the call is part of a pattern.
8371 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8373 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
8374 that the sequence is attached to the original statement rather
8375 than the pattern statement.
8376 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
8377 PATTERN_DEF_SEQ from the original statement rather than
8378 the main pattern statement.
8379 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
8380 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
8381 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
8383 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8385 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
8386 definition statements before the early exit for statements that aren't
8388 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
8390 (vect_transform_loop): ...here. Process pattern definition
8391 statements without first checking whether the main pattern
8392 statement is live or relevant.
8394 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
8396 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
8397 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
8399 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
8401 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
8402 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
8403 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
8404 (expand_block_compare): Change select_block_compare_mode call.
8405 (expand_strncmp_align_check): Use new functions, fix comment.
8406 (emit_final_str_compare_gpr): New function.
8407 (expand_strn_compare): Refactor and clean up code.
8408 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
8410 2018-06-19 Tony Reix <tony.reix@atos.com>
8411 Damien Bergamini <damien.bergamini@atos.com>
8412 David Edelsohn <dje.gcc@gmail.com>
8414 * collect2.c (static_obj): New variable.
8415 (static_libs): New variable.
8416 (is_in_list): Uncomment declaration.
8417 (main): Track AIX libraries linked statically.
8418 (is_in_list): Uncomment definition.
8419 (scan_prog_file): Don't add AIX shared libraries initializer
8420 to constructor list if linking statically.
8422 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
8424 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
8426 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
8428 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
8430 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
8433 2018-06-19 Martin Liska <mliska@suse.cz>
8435 * config/i386/i386.c (ix86_can_inline_p): Do not use
8436 ipa_fn_summaries::get_create.
8437 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
8439 (devirtualization_time_bonus): Likewise.
8440 (ipcp_propagate_stage): Likewise.
8441 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
8442 (edge_set_predicate): Likewise.
8443 (evaluate_conditions_for_known_args): Likewise.
8444 (evaluate_properties_for_edge): Likewise.
8445 (ipa_call_summary::reset): Tranform to ...
8446 (ipa_call_summary::~ipa_call_summary): ... this.
8447 (ipa_fn_summary::reset): Transform to ...
8448 (ipa_fn_summary::~ipa_fn_summary): ... this.
8449 (ipa_fn_summary_t::remove): Rename to ...
8450 (ipa_fn_summary_t::remove_callees): ... this.
8451 (ipa_fn_summary_t::duplicate): Use placement new
8452 instead of memory copy.
8453 (ipa_call_summary_t::duplicate): Likewise.
8454 (ipa_call_summary_t::remove): Remove.
8455 (dump_ipa_call_summary): Change get_create to get.
8456 (ipa_dump_fn_summary): Dump only when summary exists.
8457 (analyze_function_body): Use symbol_summary::get instead
8459 (compute_fn_summary): Likewise.
8460 (estimate_edge_devirt_benefit): Likewise.
8461 (estimate_edge_size_and_time): Likewise.
8462 (inline_update_callee_summaries): Likewise.
8463 (remap_edge_change_prob): Likewise.
8464 (remap_edge_summaries): Likewise.
8465 (ipa_merge_fn_summary_after_inlining): Likewise.
8466 (write_ipa_call_summary): Likewise.
8467 (ipa_fn_summary_write): Likewise.
8468 (ipa_free_fn_summary): Likewise.
8469 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
8470 (struct ipa_call_summary): Likewise.
8471 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
8473 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
8474 (estimate_size_after_inlining): Likewise.
8475 (estimate_growth): Likewise.
8476 (growth_likely_positive): Likewise.
8477 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
8478 (inline_call): Likewise.
8479 * ipa-inline.c (caller_growth_limits): Likewise.
8480 (can_inline_edge_p): Likewise.
8481 (can_inline_edge_by_limits_p): Likewise.
8482 (compute_uninlined_call_time): Likewise.
8483 (compute_inlined_call_time): Likewise.
8484 (want_inline_small_function_p): Likewise.
8485 (edge_badness): Likewise.
8486 (update_caller_keys): Likewise.
8487 (update_callee_keys): Likewise.
8488 (inline_small_functions): Likewise.
8489 (inline_to_all_callers_1): Likewise.
8490 (dump_overall_stats): Likewise.
8491 (early_inline_small_functions): Likewise.
8492 (early_inliner): Likewise.
8493 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
8494 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
8495 * ipa-pure-const.c (malloc_candidate_p): Likewise.
8496 * ipa-split.c (execute_split_functions): Likewise.
8497 * symbol-summary.h: Likewise.
8498 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
8500 2018-06-19 Richard Biener <rguenther@suse.de>
8502 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
8503 (vectorize_loops): ... here. Fix dbgcnt handling.
8504 (try_vectorize_loop): Wrap try_vectorize_loop_1.
8506 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
8509 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
8510 ieee128 argument takes up only one (vector) register, not two (floating
8513 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
8515 * gimplify.c (gimplify_init_constructor): Really never clear for an
8516 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
8518 2018-06-19 Richard Biener <rguenther@suse.de>
8520 PR tree-optimization/86179
8521 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
8522 after failed recognition.
8524 2018-06-18 Martin Sebor <msebor@redhat.com>
8527 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
8528 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
8529 Handle integer subtraction.
8530 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
8531 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
8533 2018-06-18 David Malcolm <dmalcolm@redhat.com>
8535 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
8536 param from rtx to rtx_insn *.
8537 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
8539 (frv_ifcvt_modify_insn): Likwise.
8540 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
8541 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
8542 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
8543 as_a <rtx_insn *> cast to local "unprotected_region" once
8544 it's been established that it's not NULL or pc_rtx.
8545 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
8546 param "sethi" from rtx to rtx_insn *.
8547 (nds32_group_float_insns): Likewise for param "insn".
8548 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
8550 (vax_output_int_subtract): Likewise.
8551 * config/vax/vax.c (vax_output_int_add): Likewise for param
8553 (vax_output_int_subtract): Likewise.
8554 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
8555 (emit_pattern_after): Likewise for param "after".
8556 (emit_insn_after): Likewise.
8557 (emit_jump_insn_after): Likewise.
8558 (emit_call_insn_after): Likewise.
8559 (emit_debug_insn_after): Likewise.
8560 (emit_pattern_before): Likewise for param "before".
8561 (emit_insn_before): Likewise.
8562 (emit_jump_insn_before): Likewise.
8563 * final.c (get_insn_template): Likewise for param "insn", removing
8565 * output.h (get_insn_template): Likewise for 2nd param.
8566 * rtl.h (emit_insn_before): Likewise.
8567 (emit_jump_insn_before): Likewise.
8568 (emit_debug_insn_before_noloc): Likewise.
8569 (emit_insn_after): Likewise.
8570 (emit_jump_insn_after): Likewise.
8571 (emit_call_insn_after): Likewise.
8572 (emit_debug_insn_after): Likewise.
8573 (set_insn_deleted): Likewise for param.
8575 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
8578 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
8579 floating point modes, so that IFmode is numerically greater than
8580 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
8581 to declare the ordering. This prevents IFmode from being
8582 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
8583 machine. Include rs6000-modes.h to share the fractional values
8584 between genmodes* and the rest of the compiler.
8588 * config/rs6000/rs6000-modes.h: New file.
8589 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
8590 meaning of rs6000_long_double_size so that 126..128 selects an
8591 appropriate 128-bit floating point type.
8592 (rs6000_option_override_internal): Likewise.
8593 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
8594 (TARGET_LONG_DOUBLE_128): Change the meaning of
8595 rs6000_long_double_size so that 126..128 selects an appropriate
8596 128-bit floating point type.
8597 (LONG_DOUBLE_TYPE_SIZE): Update comment.
8598 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
8599 source and destination to match the standard usage.
8600 (truncifkf2): Likewise.
8601 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
8602 ISA 2.07 to use an explicit clobber, instead of passing in a
8604 (copysign<mode>3_soft): Likewise.
8606 2018-06-18 David Malcolm <dmalcolm@redhat.com>
8608 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
8609 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
8610 (vect_slp_analyze_instance_dependence): Likewise.
8611 (vect_enhance_data_refs_alignment): Likewise.
8612 (vect_analyze_data_refs_alignment): Likewise.
8613 (vect_slp_analyze_and_verify_instance_alignment
8614 (vect_analyze_data_ref_accesses): Likewise.
8615 (vect_prune_runtime_alias_test_list): Likewise.
8616 (vect_analyze_data_refs): Likewise.
8617 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
8618 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
8619 (vect_analyze_scalar_cycles_1): Likewise.
8620 (vect_get_loop_niters): Likewise.
8621 (vect_analyze_loop_form_1): Likewise.
8622 (vect_update_vf_for_slp): Likewise.
8623 (vect_analyze_loop_operations): Likewise.
8624 (vect_analyze_loop): Likewise.
8625 (vectorizable_induction): Likewise.
8626 (vect_transform_loop): Likewise.
8627 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
8628 * tree-vect-slp.c (vect_analyze_slp): Likewise.
8629 (vect_make_slp_decision): Likewise.
8630 (vect_detect_hybrid_slp): Likewise.
8631 (vect_slp_analyze_operations): Likewise.
8632 (vect_slp_bb): Likewise.
8633 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
8634 (vectorizable_bswap): Likewise.
8635 (vectorizable_call): Likewise.
8636 (vectorizable_simd_clone_call): Likewise.
8637 (vectorizable_conversion): Likewise.
8638 (vectorizable_assignment): Likewise.
8639 (vectorizable_shift): Likewise.
8640 (vectorizable_operation): Likewise.
8641 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
8643 2018-06-18 Martin Sebor <msebor@redhat.com>
8645 PR tree-optimization/81384
8646 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
8647 * builtins.c (expand_builtin_strnlen): New function.
8648 (expand_builtin): Call it.
8649 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
8650 * builtins.def (BUILT_IN_STRNLEN): New.
8651 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
8652 Warn for bounds in excess of maximum object size.
8653 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
8654 single-value ranges. Handle strnlen.
8655 (handle_builtin_strlen): Handle strnlen.
8656 (strlen_check_and_optimize_stmt): Same.
8657 * doc/extend.texi (Other Builtins): Document strnlen.
8659 2018-06-18 Maya Rashish <coypu@sdf.org>
8661 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
8662 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
8663 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
8665 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
8667 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
8669 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8671 * tree.c (escaped_string::escape): Replace cast to char * by
8672 const_cast<char *> (unescaped).
8674 2018-06-18 Nick Clifton <nickc@redhat.com>
8677 * tree.c (escaped_string): New class. Converts an unescaped
8678 string into its escaped equivalent.
8679 (warn_deprecated_use): Use the new class to convert the
8680 deprecation message, if present.
8681 (test_escaped_strings): New self test.
8682 (test_c_tests): Add test_escaped_strings.
8683 * doc/extend.texi (deprecated): Add a note that the
8684 deprecation message is affected by the -fmessage-length
8685 option, and that control characters will be escaped.
8686 (#pragma GCC error): Document this pragma.
8687 (#pragma GCC warning): Likewise.
8688 * doc/invoke.texi (-fmessage-length): Document this option's
8689 effect on the #warning and #error preprocessor directives and
8690 the deprecated attribute.
8692 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
8694 * tree.c (decl_value_expr_lookup): Revert latest change.
8695 (decl_value_expr_insert): Likewise.
8697 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
8699 * gimplify.c (nonlocal_vlas): Delete.
8700 (nonlocal_vla_vars): Likewise.
8701 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
8703 (gimplify_body): Do not create and destroy nonlocal_vlas.
8704 * tree-nested.c: Include diagnostic.h.
8705 (use_pointer_in_frame): Tweak.
8706 (lookup_field_for_decl): Add assertion and declare the transformation.
8707 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
8708 internal error when the reference is in a wrong context. Do not
8709 create a debug decl by default.
8710 (note_nonlocal_block_vlas): Delete.
8711 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
8712 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
8713 create a debug decl by default.
8714 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
8715 call is in a wrong context.
8716 (fixup_vla_decls): New function.
8717 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
8718 debug variables were created.
8719 * tree.c (decl_value_expr_lookup): Add checking assertion.
8720 (decl_value_expr_insert): Likewise.
8722 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
8725 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
8726 * tree-scalar-evolution.c (interpret_expr): Likewise.
8727 (expression_expensive_p): Likewise.
8728 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
8729 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
8730 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
8731 (ssa_defined_by_minus_one_stmt_p): New.
8733 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
8736 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
8737 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
8738 * dojump.c (do_jump): Likewise.
8739 * expr.c (expand_expr_real_2): Check operand type's sign.
8740 * fold-const.c (const_unop): Handle ABSU_EXPR.
8741 (fold_abs_const): Likewise.
8742 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
8743 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
8744 (strip_sign_op_1): Likesise.
8745 * match.pd: Add new pattern to generate ABSU_EXPR.
8746 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
8747 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
8748 * tree-eh.c (operation_could_trap_helper_p): Likewise.
8749 * tree-inline.c (estimate_operator_cost): Likewise.
8750 * tree-pretty-print.c (dump_generic_node): Likewise.
8751 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
8752 * tree.def (ABSU_EXPR): New.
8754 2018-06-16 Jakub Jelinek <jakub@redhat.com>
8757 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
8758 documented as preserved for backward compatibility only.
8759 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
8761 PR rtl-optimization/86108
8762 * bb-reorder.c (create_forwarder_block): Renamed to ...
8763 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
8764 jump from new landing pad to the second part.
8765 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
8768 2018-06-15 Jakub Jelinek <jakub@redhat.com>
8771 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
8772 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
8773 Only call store_expr for halves if the mode is the same.
8776 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
8777 Fix up comment formatting.
8779 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8781 * typed-splay-tree.h (typed_splay_tree::remove): New function.
8782 (typed_splay_tree::closure,
8783 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
8784 (typed_splay_tree::typed_splay_tree,
8785 typed_splay_tree::operator =): Declared private.
8786 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
8787 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
8788 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
8789 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
8790 typed_splay_tree::splay_tree_splay,
8791 typed_splay_tree::splay_tree_foreach_helper,
8792 typed_splay_tree::splay_tree_insert,
8793 typed_splay_tree::splay_tree_remove,
8794 typed_splay_tree::splay_tree_lookup,
8795 typed_splay_tree::splay_tree_predecessor,
8796 typed_splay_tree::splay_tree_successor,
8797 typed_splay_tree::splay_tree_min,
8798 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
8799 (typed_splay_tree::root, typed_splay_tree::comp,
8800 typed_splay_tree::delete_key,
8801 typed_splay_tree::delete_value): New data members.
8802 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
8803 typed_splay_tree::remove.
8805 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
8807 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
8808 -mginv and -mno-ginv to the assembler.
8809 * config/mips/mips.opt (-mcrc): New option.
8811 * doc/invoke.text (-mcrc): Document.
8814 2018-06-15 Nick Clifton <nickc@redhat.com>
8817 * tree.c (escaped_string): New class. Converts an unescaped
8818 string into its escaped equivalent.
8819 (warn_deprecated_use): Use the new class to convert the
8820 deprecation message, if present.
8821 (test_escaped_strings): New self test.
8822 (test_c_tests): Add test_escaped_strings.
8823 * doc/extend.texi (deprecated): Add a note that the
8824 deprecation message is affected by the -fmessage-length
8825 option, and that control characters will be escaped.
8826 (#pragma GCC error): Document this pragma.
8827 (#pragma GCC warning): Likewise.
8828 * doc/invoke.texi (-fmessage-length): Document this option's
8829 effect on the #warning and #error preprocessor directives and
8830 the deprecated attribute.
8832 2018-06-15 Richard Biener <rguenther@suse.de>
8834 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
8835 here, also noting vector size used.
8836 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
8837 size used in MSG_OPTIMIZED_LOCATIONS dump.
8838 (pass_slp_vectorize::execute): Adjust.
8840 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
8843 * config/arc/arc.c (arc_return_address_register): Fix
8846 2018-06-15 Richard Biener <rguenther@suse.de>
8849 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
8850 leave useless conversion stripping to force_gimple_operand_gsi.
8851 (gimplify_build2): Likewise.
8852 (gimplify_build1): Likewise.
8854 2018-06-15 Richard Biener <rguenther@suse.de>
8857 * tree-cfg.c (move_stmt_op): unshare invariant addresses
8858 before adjusting their block.
8860 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
8862 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
8863 multilibs for *-*-rtems*.
8864 * config/riscv/t-rtems: New file.
8866 2018-06-14 Jakub Jelinek <jakub@redhat.com>
8869 * match.pd ((A +- CST1) +- CST2): Punt if last resort
8870 unsigned_type_for returns NULL.
8873 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
8874 subregs of multi-word pseudos unless the float mode has word size.
8876 2018-06-14 Richard Biener <rguenther@suse.de>
8879 * tree-vect-generic.c (build_word_mode_vector_type): Remove
8880 duplicate and harmful type_hash_canon.
8881 * tree.c (type_hash_canon): Assert we didn't find ourselves.
8883 2018-06-14 Richard Biener <rguenther@suse.de>
8886 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
8889 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
8891 * config/rtems.h (STDINT_LONG32): Define.
8893 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
8894 Prachi Godbole <prachi.godbole@imgtec.com>
8896 * config/mips/mips-cpus.def: Define P6600.
8897 * config/mips/mips-tables.opt: Regenerate.
8898 * config/mips/mips.c (mips_ucbranch_type): New enum.
8899 (mips_rtx_cost_data): Add support for P6600.
8900 (mips_issue_rate): Likewise.
8901 (mips_multipass_dfa_lookahead): Likewise.
8902 (mips_avoid_hazard): Likewise.
8903 (mips_reorg_process_insns): Likewise.
8904 (mips_classify_branch_p6600): New function.
8905 * config/mips/mips.h (TUNE_P6600): New define.
8906 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
8907 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
8908 * config/mips/mips.md: Include p6600.md.
8909 (processor): Add p6600.
8910 * config/mips/p6600.md: New file.
8911 * doc/invoke.texi: Add p6600 to supported architectures.
8913 2018-06-13 Martin Sebor <msebor@redhat.com>
8915 PR tree-optimization/86114
8916 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
8918 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
8920 2018-06-13 Richard Biener <rguenther@suse.de>
8922 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
8923 Properly set vector type of the intermediate stmt.
8924 * tree-vect-stmts.c (vectorizable_operation): The destination
8925 var always has vectype_out type.
8927 2018-06-13 Jeff Law <law@redhat.com>
8929 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
8930 integer 0 for argument to print_rtl_with_bb.
8931 (rl78_reorg): Likewise.
8933 2018-06-13 David Malcolm <dmalcolm@redhat.com>
8935 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
8936 from rtx to rtx_insn *.
8937 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
8939 (add_sched_insns_for_speculation): Likewise for local "target",
8940 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
8941 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
8942 from rtx_insn ** to rtx_code_label **.
8943 (reorg_emit_nops): Likewise.
8944 (c6x_reorg): Likewise for local "call_labels".
8945 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
8947 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
8948 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
8949 the loops over LABEL_REFS.
8950 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
8952 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
8953 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
8954 (split_branches): Strengthen local "olabel" from rtx to
8955 rtx_insn *, adding a safe_as_a cast.
8956 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
8958 (add_insn_after): Likewise for first two params.
8959 (add_insn_before): Likewise.
8960 (remove_insn): Likewise for param.
8961 (emit_pattern_before_noloc): Likewise for second and third params.
8962 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
8963 (emit_call_insn_before_noloc): Likewise.
8964 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
8966 (emit_barrier_before): Likewise.
8967 (emit_label_before): Strengthen "label" param from "rtx" to
8968 "rtx_code_label *". Strengthen "before" param from "rtx" to
8970 (emit_insn_after_1): Strengthen "after" param from "rtx" to
8972 (emit_pattern_after_noloc): Likewise.
8973 (emit_insn_after_noloc): Likewise.
8974 (emit_jump_insn_after_noloc): Likewise.
8975 (emit_call_insn_after_noloc): Likewise.
8976 (emit_debug_insn_after_noloc): Likewise.
8977 (emit_barrier_after): Likewise.
8978 (emit_label_after): Likewise for both params.
8979 (emit_pattern_after_setloc): Likewise for "after" param. Convert
8980 "loc" param from "int" to "location_t".
8981 (emit_insn_after_setloc): Likewise.
8982 (emit_jump_insn_after_setloc): Likewise.
8983 (emit_call_insn_after_setloc): Likewise.
8984 (emit_debug_insn_after_setloc): Likewise.
8985 (emit_pattern_before_setloc): Likewise for "before" param. Convert
8986 "loc" param from "int" to "location_t".
8987 (emit_pattern_before): Convert NULL_RTX to NULL.
8988 (emit_insn_before_setloc): Convert "loc" param from "int" to
8990 (emit_jump_insn_before_setloc): Likewise.
8991 (emit_call_insn_before_setloc): Likewise.
8992 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
8993 rtx_insn *. Convert "loc" param from "int" to "location_t".
8994 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
8995 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
8996 Convert 3rd param from "int" to "location_t".
8997 (emit_barrier_before, emit_barrier_after, next_real_insn):
8998 Strengthen param from rtx to rtx_insn *.
8999 (emit_label_before): Strengthen 1st param from "rtx" to
9000 "rtx_code_label *". Strengthen 2nd param from "rtx" to
9002 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
9003 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
9004 Strengthen 2nd param from "rtx" to "rtx_insn *".
9005 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
9006 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
9007 Likewise. Convert 3rd param from "int" to "location_t".
9008 (emit_label_after): Strengthen 1st param from "rtx" to
9010 (next_real_insn, remove_insn): Strengthen param from "rtx" to
9012 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
9013 from "rtx" to "rtx_insn *".
9015 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
9017 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
9018 bodies streamed in with -Q.
9019 * dumpfile.c (dump_files): Add lto-stream-out dump file.
9020 * dumpfile.h (tree_dump_index): Add lto_stream_out.
9021 * gimple-streamer-out.c: Include gimple-pretty-print.h
9022 (output_bb): Dump stmts streamed.
9023 * lto-section-out.c: Include print-tree.h
9024 (lto_begin_section): Dump sections created.
9025 (lto_output_decl_index): Dump decl encoded.
9026 * lto-streamer-out.c: Include print-tree.h
9027 (create_output_block): Dump output block created.
9028 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
9029 (output_function): Dump function output.
9030 (output_constructor): Dump constructor streamed.
9031 (write_global_stream): Output indexes encoded.
9032 (produce_asm_for_decls): Dump streams encoded.
9033 * lto-streamer.c (streamer_dump_file): New global var.
9034 * lto-streamer.h (streamer_dump_file): Declare.
9035 * passes.c (ipa_write_summaries): Initialize streamer dump.
9036 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
9039 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
9042 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
9043 offsets for register save directives. Emit a second batch of save
9044 directives, if need be, when the function accesses prior frames.
9046 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9048 * config/arc/fpu.md (fmasf4): Force operand to register.
9049 (fnmasf4): Likewise.
9051 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9053 * config/arc/arc-protos.h (arc_pad_return): Remove.
9054 * config/arc/arc.c (machine_function): Remove force_short_suffix
9056 (arc_print_operand): Adjust printing of '&'.
9057 (arc_verify_short): Remove conditional printing of short suffix.
9058 (arc_final_prescan_insn): Remove reference to size_reason.
9059 (pad_return): New function.
9060 (arc_reorg): Call pad_return.
9061 (arc_pad_return): Remove.
9062 (arc_init_machine_status): Remove reference to force_short_suffix.
9063 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
9064 (attr length): When attribute iscompact is true force to 2
9065 regardless; in the case of maybe check if we want to force the
9066 instruction to have 4 bytes length.
9067 (nopv): Change it to generate 4 byte long nop as well.
9068 (blockage): New pattern.
9069 (simple_return): Remove call to arc_pad_return.
9070 (p_return_i): Likewise.
9072 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9074 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
9076 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9078 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
9081 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9083 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
9084 for ARC700 and ARCv2.
9086 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
9089 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
9090 operands[2] instead of operands[1].
9093 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9095 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
9096 case, check whether the outer register overlaps an unallocatable
9097 register, not just whether it fits the required class.
9099 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9101 * poly-int.h (can_div_trunc_p): Add new overload in which all values
9103 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
9104 (memrefs_conflict_p): Likewise.
9105 (init_alias_analysis): Likewise.
9106 * cfgexpand.c (expand_debug_expr): Likewise.
9107 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
9108 * cse.c (fold_rtx): Likewise.
9109 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
9110 * expr.c (emit_block_move_hints): Likewise.
9111 (clear_storage_hints, push_block, emit_push_insn): Likewise.
9112 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
9113 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
9114 (emit_group_store): Likewise.
9115 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
9116 to read the PRE/POST_MODIFY increment.
9117 * calls.c (store_one_arg): Use strip_offset.
9118 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
9120 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
9122 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
9123 (simplify_binary_operation_1): Extend CONST_INT handling to
9125 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
9126 than a HOST_WIDE_INT.
9127 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
9129 (adjust_mems, add_stores): Update accodingly.
9130 (vt_canonicalize_addr): Track polynomial offsets.
9131 (emit_note_insn_var_location): Likewise.
9132 (vt_add_function_parameter): Likewise.
9133 (vt_initialize): Likewise.
9135 2018-06-12 Jeff Law <law@redhat.com>
9137 * config.gcc (alpha*-*-freebsd*): Remove.
9138 * config/alpha/freebsd.h: Remove.
9140 2018-06-12 David Malcolm <dmalcolm@redhat.com>
9143 * spellcheck-tree.c (levenshtein_distance): Rename to...
9144 (get_edit_distance): ...this, and update for underlying renaming.
9145 * spellcheck-tree.h (levenshtein_distance): Rename to...
9146 (get_edit_distance): ...this.
9147 * spellcheck.c (levenshtein_distance): Rename to...
9148 (get_edit_distance): ...this. Convert from Levenshtein distance
9149 to Damerau-Levenshtein distance by supporting transpositions of
9150 adjacent characters. Rename "v1" to "v_next" and "v0" to
9152 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
9153 (selftest::test_edit_distance_unit_test_oneway): ...this, and
9154 update for underlying renaming.
9155 (selftest::levenshtein_distance_unit_test): Rename to...
9156 (selftest::test_get_edit_distance_unit): ...this, and update for
9157 underlying renaming.
9158 (selftest::test_find_closest_string): Add example from PR 69968
9159 where transposition helps
9160 (selftest::test_metric_conditions): Update for renaming.
9161 (selftest::test_metric_conditions): Likewise.
9162 (selftest::spellcheck_c_tests): Likewise.
9163 * spellcheck.h (levenshtein_distance): Rename both overloads to...
9164 (get_edit_distance): ...this.
9165 (best_match::consider): Update for renaming.
9167 2018-06-12 Martin Sebor <msebor@redhat.com>
9169 PR tree-optimization/85259
9170 * builtins.c (compute_objsize): Handle constant offsets.
9171 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
9172 true iff a warning has been issued.
9173 * gimple.h (gimple_nonartificial_location): New function.
9174 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
9175 gimple_nonartificial_location and handle -Wno-system-headers.
9176 (handle_builtin_stxncpy): Same.
9178 2018-06-12 Martin Sebor <msebor@redhat.com>
9181 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
9183 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9185 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9186 BUILTIN_VEC_XST entries for pointer to double and long long.
9188 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
9191 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
9193 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
9196 2018-06-12 Martin Liska <mliska@suse.cz>
9198 * doc/options.texi: Document IntegerRange.
9200 2018-06-12 Martin Liska <mliska@suse.cz>
9202 * config/i386/i386.opt: Make MPX-related options as Deprecated.
9203 * opt-functions.awk: Handle Deprecated flag.
9204 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
9206 (read_cmdline_option): Report warning for a deprecated option.
9207 * opts.h (struct cl_option): Add new field cl_deprecated.
9208 (CL_ERR_DEPRECATED): New.
9210 2018-06-12 Martin Liska <mliska@suse.cz>
9212 * doc/options.texi: Document Deprecated option flag.
9214 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9216 * config/arc/arc-arch.h (arc_extras): New enum.
9217 (arc_cpu_t):Add field extra.
9218 (arc_cpu_types): Consider the extras.
9219 * config/arc/arc-cpus.def: Add extras info.
9220 * config/arc/arc-opts.h (processor_type): Consider extra field.
9221 * config/arc/arc.c (arc_override_options): Handle extra field.
9223 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9225 * config/arc/arc-arch.h: Update ARC_OPTX macro.
9226 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
9228 * config/arc/arc.c (arc_init): Update pic warning.
9229 (irq_range): Update irq range parsing warnings.
9230 (arc_override_options): Update various warning messages.
9231 (arc_handle_aux_attribute): Likewise.
9233 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
9235 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
9237 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
9239 * doc/sourcebuild.texi: Document usage of line number 0 in verify
9240 compiler messages directives.
9242 2018-06-12 Matthew Fortune <mfortune@gmail.com>
9244 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
9245 * config/mips/mips-tables.opt: Regenerate.
9246 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
9248 * doc/invoke.texi: Document -march=i6500.
9250 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
9252 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
9253 (i6400_gpmul): Add cpu_unit.
9254 (i6400_gpdiv): Likewise.
9255 (i6400_msa_add_d): Update reservations.
9256 (i6400_msa_int_add) Likewise.
9257 (i6400_msa_short_logic3) Likewise.
9258 (i6400_msa_short_logic2) Likewise.
9259 (i6400_msa_short_logic) Likewise.
9260 (i6400_msa_move) Likewise.
9261 (i6400_msa_cmp) Likewise.
9262 (i6400_msa_short_float2) Likewise.
9263 (i6400_msa_div_d) Likewise.
9264 (i6400_msa_long_logic1) Likewise.
9265 (i6400_msa_long_logic2) Likewise.
9266 (i6400_msa_mult) Likewise.
9267 (i6400_msa_long_float2) Likewise.
9268 (i6400_msa_long_float4) Likewise.
9269 (i6400_msa_long_float5) Likewise.
9270 (i6400_msa_long_float8) Likewise.
9271 (i6400_fpu_fadd): Include frint type.
9272 (i6400_fpu_store): New define_insn_reservation.
9273 (i6400_fpu_load): Likewise.
9274 (i6400_fpu_move): Likewise.
9275 (i6400_fpu_fcmp): Likewise.
9276 (i6400_fpu_fmadd): Likewise.
9277 (i6400_int_mult): Include imul3nc type and update reservation.
9278 (i6400_int_div): Include idiv3 type and update reservation.
9279 (i6400_int_load): Update to check type not move_type.
9280 (i6400_int_store): Likewise.
9281 (i6400_int_prefetch): Set zero latency.
9283 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
9285 * gcc.c: Document new %@{...} sequence.
9286 (LINK_COMMAND_SPEC): Use it for the -L switches.
9287 (cpp_unique_options): Use it for the -I switches.
9288 (at_file_argbuf): New global variable.
9289 (in_at_file): Likewise.
9290 (alloc_args): Create at_file_argbuf.
9291 (clear_args): Truncate at_file_argbuf.
9292 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
9293 (open_at_file): New function.
9294 (close_at_file): Likewise.
9295 (create_at_file): Delete.
9296 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
9299 (validate_switches_from_spec): Deal with %@{...} sequence.
9300 (validate_switches): Likewise.
9301 (driver::finalize): Call clear_args.
9303 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
9305 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
9307 2018-06-11 Martin Sebor <msebor@redhat.com>
9309 * doc/invoke.texi (-Wall): List -Wc++17-compat.
9310 (Wno-class-memaccess): Add @opindex.
9311 (Wno-templates, Wno-multiple-inheritance): Same.
9312 (Wno-virtual-inheritance, Wno-namespaces): Same.
9313 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
9314 (Wno-format-overflow, Wno-format-truncation): Same.
9315 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
9316 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
9317 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
9318 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
9319 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
9320 (Wno-misspelled-isr): Same.
9322 2018-06-11 Martin Sebor <msebor@redhat.com>
9324 * PR tree-optimization/86083
9325 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
9327 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
9329 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
9331 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
9334 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
9335 on the correct operand.
9336 (*movdi_internal64): Ditto.
9338 2018-06-11 Martin Liska <mliska@suse.cz>
9340 PR tree-optimization/86089
9341 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
9343 2018-06-11 Julia Koval <julia.koval@intel.com>
9345 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
9346 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
9347 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
9349 2018-06-11 Olivier Hainque <hainque@adacore.com>
9351 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
9352 for Ada with strict dwarf2.
9354 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
9357 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
9360 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
9362 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
9364 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
9366 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
9369 2018-06-08 Martin Liska <mliska@suse.cz>
9371 * tree-cfg.h (debug_function): Fix argument type to match
9374 2018-06-08 Martin Liska <mliska@suse.cz>
9376 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
9377 Remove usage of MPX-related (and removed) fields.
9378 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
9380 2018-06-08 David Malcolm <dmalcolm@redhat.com>
9382 * cfg.c (debug): Use TDF_NONE rather than 0.
9383 * cfghooks.c (debug): Likewise.
9384 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
9385 (struct dump_option_value_info): Convert to...
9386 (struct kv_pair): ...this template type.
9387 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
9389 (optinfo_verbosity_options): Likewise.
9390 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
9392 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
9393 than int for "optgroup_flags" param.
9394 (dump_generic_expr_loc): Use dump_flags_t rather than int for
9396 (dump_dec): Likewise.
9397 (dump_finish): Use TDF_NONE rather than 0.
9398 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
9399 rather than int for "optgroup_flags" param. Use TDF_NONE rather
9400 than 0. Update for change to option_ptr.
9401 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
9402 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
9403 0. Update for changes to optinfo_verbosity_options and
9405 (opt_info_switch_p): Convert optgroup_flags from int to
9407 (dump_basic_block): Use dump_flags_t rather than int
9408 for "dump_kind" param.
9409 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
9410 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
9411 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
9412 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
9413 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
9414 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
9415 TDF_NONE): Convert from macros to...
9416 (enum dump_flag): ...this new enum.
9417 (dump_flags_t): Update to use enum.
9418 (operator|, operator&, operator~, operator|=, operator&=):
9419 Implement for dump_flags_t.
9420 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
9421 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
9422 Convert from macros to...
9423 (enum optgroup_flag): ...this new enum.
9424 (optgroup_flags_t): New typedef.
9425 (operator|, operator|=): Implement for optgroup_flags_t.
9426 (struct dump_file_info): Convert field "alt_flags" to
9427 dump_flags_t. Convert field "optgroup_flags" to
9429 (dump_basic_block): Use dump_flags_t rather than int for param.
9430 (dump_generic_expr_loc): Likewise.
9431 (dump_dec): Likewise.
9432 (dump_register): Convert param "optgroup_flags" to
9434 (opt_info_enable_passes): Likewise.
9435 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
9437 * gimple-pretty-print.c (debug): Likewise.
9438 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
9439 (merged_store_group::apply_stores): Likewise.
9440 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
9441 * gimple.c (verify_gimple_pp): Likewise.
9442 * graphite-poly.c (print_pbb_body): Likewise.
9443 * passes.c (pass_manager::register_one_dump_file): Convert
9444 local "optgroup_flags" to optgroup_flags_t.
9445 * print-tree.c (print_node): Use TDF_NONE rather than 0.
9447 (debug_body): Likewise.
9448 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
9449 to optgroup_flags_t.
9450 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
9452 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
9453 (convert_mult_to_fma): Likewise.
9454 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
9455 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
9456 * tree-vect-data-refs.c (dump_lower_bound): Convert param
9457 "dump_kind" to dump_flags_t.
9459 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
9461 * config/rs6000/rs6000.c (min, max): Delete.
9463 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
9465 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
9466 -mabi=spe and -mabi=no-spe.
9468 2018-06-08 Martin Liska <mliska@suse.cz>
9470 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
9471 where we expect an existing summary.
9473 2018-06-08 Martin Liska <mliska@suse.cz>
9475 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
9476 * ipa-inline.h (estimate_edge_growth): Likewise.
9478 2018-06-08 Martin Liska <mliska@suse.cz>
9480 * cgraph.c (function_version_hasher::hash): Use
9481 cgraph_node::get_uid ().
9482 (function_version_hasher::equal):
9483 * cgraph.h (cgraph_node::get_uid): New method.
9484 * ipa-inline.c (update_caller_keys): Use
9485 cgraph_node::get_uid ().
9486 (update_callee_keys): Likewise.
9487 * ipa-utils.c (searchc): Likewise.
9488 (ipa_reduced_postorder): Likewise.
9489 * lto-cgraph.c (input_node): Likewise.
9490 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
9491 * symbol-summary.h (symtab_insertion): Likewise.
9492 (symtab_removal): Likewise.
9493 (symtab_duplication): Likewise.
9494 * tree-pretty-print.c (dump_function_header): Likewise.
9495 * tree-sra.c (convert_callers_for_node): Likewise.
9497 2018-06-08 Martin Liska <mliska@suse.cz>
9499 * cgraph.c (symbol_table::create_edge): Always assign a new
9501 (symbol_table::free_edge): Do not recycle numbers.
9502 * cgraph.h (cgraph_edge::get): New method.
9503 * symbol-summary.h (symtab_removal): Use it.
9504 (symtab_duplication): Likewise.
9505 (call_summary::hashable_uid): Remove.
9507 2018-06-08 Martin Liska <mliska@suse.cz>
9509 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
9510 (initialize_growth_caches): Remove.
9511 (free_growth_caches): Likewise.
9512 (do_estimate_edge_time): Use edge_growth_cache.
9513 (do_estimate_edge_size): Likewise.
9514 (do_estimate_edge_hints): Likewise.
9515 * ipa-inline.c (reset_edge_caches): Likewise.
9516 (recursive_inlining): Likewise.
9517 (inline_small_functions): Likewise.
9518 * ipa-inline.h (initialize_growth_caches): Remove.
9519 (estimate_edge_size): Likewise.
9520 (estimate_edge_time): Likewise.
9521 (estimate_edge_hints): Likewise.
9522 (reset_edge_growth_cache): Likewise.
9523 * symbol-summary.h (call_summary::remove): New method.
9525 2018-06-08 Martin Liska <mliska@suse.cz>
9527 * ipa-cp.c (class edge_clone_summary): New summary.
9528 (grow_edge_clone_vectors): Remove.
9529 (ipcp_edge_duplication_hook): Remove.
9530 (class edge_clone_summary_t): New call_summary class.
9531 (ipcp_edge_removal_hook): Remove.
9532 (edge_clone_summary_t::duplicate): New function.
9533 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
9534 (create_specialized_node): Likewise.
9535 (ipcp_driver): Initialize edge_clone_summaries and do not
9538 2018-06-08 Martin Liska <mliska@suse.cz>
9540 * symbol-summary.h (get): New function.
9541 (call_summary::m_initialize_when_cloning): New class member.
9543 2018-06-08 Martin Liska <mliska@suse.cz>
9545 * cgraph.c (cgraph_node::remove): Do not recycle uid.
9546 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
9547 (symbol_table::allocate_cgraph_symbol): Do not set uid.
9548 * passes.c (uid_hash_t): Record removed_nodes by their uids.
9549 (remove_cgraph_node_from_order): Use the removed_nodes set.
9550 (do_per_function_toporder): Likwise.
9551 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
9552 instead of summary_uid.
9553 (symtab_removal): Likewise.
9554 (symtab_duplication): Likewise.
9556 2018-06-08 Martin Liska <mliska@suse.cz>
9558 * ipa-cp.c (ipcp_store_bits_results): Use
9559 ipcp_transformation_sum.
9560 (ipcp_store_vr_results): Likewise.
9561 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
9563 (ipcp_transformation_initialize): ... this.
9564 (ipa_set_node_agg_value_chain):
9565 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
9566 (write_ipcp_transformation_info): Likewise.
9567 (read_ipcp_transformation_info): Likewise.
9568 (ipcp_update_bits): Likewise.
9569 (ipcp_update_vr): Likewise.
9570 (ipcp_transform_function): Likewise.
9571 * ipa-prop.h: Rename ipcp_transformation_summary to
9572 ipcp_transformation.
9573 (class ipcp_transformation_t): New function summary.
9574 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
9575 (ipa_get_agg_replacements_for_node): Likewise.
9577 2018-06-08 Martin Liska <mliska@suse.cz>
9579 * ipa-pure-const.c (struct funct_state_d): Do it class instead
9581 (class funct_state_summary_t): New function_summary class.
9582 (has_function_state): Remove.
9583 (get_function_state): Likewise.
9584 (set_function_state): Likewise.
9585 (add_new_function): Likewise.
9586 (funct_state_summary_t::insert): New function.
9587 (duplicate_node_data): Remove.
9588 (remove_node_data): Remove.
9589 (funct_state_summary_t::duplicate): New function.
9590 (register_hooks): Create new funct_state_summaries.
9591 (pure_const_generate_summary): Use it.
9592 (pure_const_write_summary): Likewise.
9593 (pure_const_read_summary): Likewise.
9594 (propagate_pure_const): Likewise.
9595 (propagate_nothrow): Likewise.
9596 (dump_malloc_lattice): Likewise.
9597 (propagate_malloc): Likewise.
9598 (execute): Do not register hooks, just remove summary
9600 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
9603 2018-06-08 Martin Liska <mliska@suse.cz>
9605 * ipa-reference.c (remove_node_data): Remove.
9606 (duplicate_node_data): Likewise.
9607 (class ipa_ref_var_info_summary_t): New class.
9608 (class ipa_ref_opt_summary_t): Likewise.
9609 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
9610 (get_reference_optimization_summary): Use
9611 ipa_ref_opt_sum_summaries.
9612 (set_reference_vars_info): Remove.
9613 (set_reference_optimization_summary): Likewise.
9614 (ipa_init): Create summaries.
9615 (init_function_info): Use function summary.
9616 (ipa_ref_opt_summary_t::duplicate): New function.
9617 (ipa_ref_opt_summary_t::remove): New function.
9618 (get_read_write_all_from_node): Fix GNU coding style.
9619 (propagate): Use function summary.
9620 (write_node_summary_p): Fix GNU coding style.
9621 (stream_out_bitmap): Likewise.
9622 (ipa_reference_read_optimization_summary): Use function summary.
9623 (ipa_reference_c_finalize): Do not release hooks.
9625 2018-06-08 Martin Liska <mliska@suse.cz>
9627 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
9628 (analyze_function_body): Extract multiple calls of get_create.
9629 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
9630 * ipa-inline.c (recursive_inlining): Use ::get method.
9631 * ipa-inline.h (estimate_edge_growth): Likewise.
9633 2018-06-08 Martin Liska <mliska@suse.cz>
9635 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
9637 (hsa_function_summary::hsa_function_summary): Use the new enum
9639 (hsa_gpu_implementation_p): Use hsa_summaries::get.
9640 * hsa-gen.c (hsa_get_host_function): Likewise.
9641 (get_brig_function_name): Likewise.
9642 * ipa-hsa.c (process_hsa_functions): Likewise.
9643 (ipa_hsa_write_summary): Likewise.
9644 * symbol-summary.h (symtab_duplication): Use ::get function/
9645 (get): New function.
9647 2018-06-08 Martin Liska <mliska@suse.cz>
9649 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
9651 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
9652 (hsa_register_kernel): Likewise.
9653 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
9654 * hsa-gen.c (hsa_get_host_function): Likewise.
9655 (get_brig_function_name): Likewise.
9656 (generate_hsa): Likewise.
9657 (pass_gen_hsail::execute): Likewise.
9658 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
9659 (devirtualization_time_bonus): Likewise.
9660 (ipcp_propagate_stage): Likewise.
9661 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
9662 (edge_set_predicate): Likewise.
9663 (evaluate_conditions_for_known_args): Likewise.
9664 (evaluate_properties_for_edge): Likewise.
9665 (ipa_fn_summary::reset): Likewise.
9666 (ipa_fn_summary_t::duplicate): Likewise.
9667 (dump_ipa_call_summary): Likewise.
9668 (ipa_dump_fn_summary): Likewise.
9669 (analyze_function_body): Likewise.
9670 (compute_fn_summary): Likewise.
9671 (estimate_edge_devirt_benefit): Likewise.
9672 (estimate_edge_size_and_time): Likewise.
9673 (estimate_calls_size_and_time): Likewise.
9674 (estimate_node_size_and_time): Likewise.
9675 (inline_update_callee_summaries): Likewise.
9676 (remap_edge_change_prob): Likewise.
9677 (remap_edge_summaries): Likewise.
9678 (ipa_merge_fn_summary_after_inlining): Likewise.
9679 (ipa_update_overall_fn_summary): Likewise.
9680 (read_ipa_call_summary): Likewise.
9681 (inline_read_section): Likewise.
9682 (write_ipa_call_summary): Likewise.
9683 (ipa_fn_summary_write): Likewise.
9684 (ipa_free_fn_summary): Likewise.
9685 * ipa-hsa.c (process_hsa_functions): Likewise.
9686 (ipa_hsa_write_summary): Likewise.
9687 (ipa_hsa_read_section): Likewise.
9688 * ipa-icf.c (sem_function::merge): Likewise.
9689 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
9690 (do_estimate_edge_time): Likewise.
9691 (estimate_size_after_inlining): Likewise.
9692 (estimate_growth): Likewise.
9693 (growth_likely_positive): Likewise.
9694 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
9695 (inline_call): Likewise.
9696 * ipa-inline.c (caller_growth_limits): Likewise.
9697 (can_inline_edge_p): Likewise.
9698 (can_inline_edge_by_limits_p): Likewise.
9699 (compute_uninlined_call_time): Likewise.
9700 (compute_inlined_call_time): Likewise.
9701 (want_inline_small_function_p): Likewise.
9702 (edge_badness): Likewise.
9703 (update_caller_keys): Likewise.
9704 (update_callee_keys): Likewise.
9705 (recursive_inlining): Likewise.
9706 (inline_small_functions): Likewise.
9707 (inline_to_all_callers_1): Likewise.
9708 (dump_overall_stats): Likewise.
9709 (early_inline_small_functions): Likewise.
9710 (early_inliner): Likewise.
9711 * ipa-inline.h (estimate_edge_growth): Likewise.
9712 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
9713 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
9714 * ipa-prop.h (IPA_NODE_REF): Likewise.
9715 (IPA_EDGE_REF): Likewise.
9716 * ipa-pure-const.c (malloc_candidate_p): Likewise.
9717 (propagate_malloc): Likewise.
9718 * ipa-split.c (execute_split_functions): Likewise.
9719 * symbol-summary.h: Rename get to get_create.
9721 (get_create): Likewise.
9722 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
9724 2018-06-08 Martin Liska <mliska@suse.cz>
9726 * symbol-summary.h (release): Move definition out of class
9728 (symtab_removal): Likewise.
9729 (symtab_duplication): Likewise.
9731 2018-06-08 Martin Liska <mliska@suse.cz>
9733 * symbol-summary.h (function_summary): Move constructor
9734 implementation out of class declaration.
9735 (release): Likewise.
9736 (symtab_insertion): Likewise.
9737 (symtab_removal): Likewise.
9738 (symtab_duplication): Likewise.
9741 2018-06-08 Martin Liska <mliska@suse.cz>
9743 * Makefile.in: Remove support for MPX (macros, related functions,
9744 fields in cgraph_node, ...).
9745 * builtin-types.def (BT_BND): Likewise.
9746 (BT_FN_BND_CONST_PTR): Likewise.
9747 (BT_FN_CONST_PTR_BND): Likewise.
9748 (BT_FN_VOID_PTR_BND): Likewise.
9749 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
9750 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
9751 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
9752 (expand_builtin_mempcpy_with_bounds): Likewise.
9753 (expand_builtin_memset_with_bounds): Likewise.
9754 (expand_builtin_memset_args): Likewise.
9755 (std_expand_builtin_va_start): Likewise.
9756 (expand_builtin): Likewise.
9757 (expand_builtin_with_bounds): Likewise.
9758 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
9759 (DEF_LIB_BUILTIN_CHKP): Likewise.
9760 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
9761 (DEF_CHKP_BUILTIN): Likewise.
9762 (BUILT_IN_MEMCPY): Likewise.
9763 (BUILT_IN_MEMMOVE): Likewise.
9764 (BUILT_IN_MEMPCPY): Likewise.
9765 (BUILT_IN_MEMSET): Likewise.
9766 (BUILT_IN_STPCPY): Likewise.
9767 (BUILT_IN_STRCAT): Likewise.
9768 (BUILT_IN_STRCHR): Likewise.
9769 (BUILT_IN_STRCPY): Likewise.
9770 (BUILT_IN_STRLEN): Likewise.
9771 (BUILT_IN_MEMCPY_CHK): Likewise.
9772 (BUILT_IN_MEMMOVE_CHK): Likewise.
9773 (BUILT_IN_MEMPCPY_CHK): Likewise.
9774 (BUILT_IN_MEMSET_CHK): Likewise.
9775 (BUILT_IN_STPCPY_CHK): Likewise.
9776 (BUILT_IN_STRCAT_CHK): Likewise.
9777 (BUILT_IN_STRCPY_CHK): Likewise.
9778 * calls.c (store_bounds): Likewise.
9779 (emit_call_1): Likewise.
9780 (special_function_p): Likewise.
9781 (maybe_warn_nonstring_arg): Likewise.
9782 (initialize_argument_information): Likewise.
9783 (finalize_must_preallocate): Likewise.
9784 (compute_argument_addresses): Likewise.
9785 (expand_call): Likewise.
9786 * cfgexpand.c (expand_call_stmt): Likewise.
9787 (expand_return): Likewise.
9788 (expand_gimple_stmt_1): Likewise.
9789 (pass_expand::execute): Likewise.
9790 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
9791 (cgraph_node::remove): Likewise.
9792 (cgraph_node::dump): Likewise.
9793 (cgraph_node::verify_node): Likewise.
9794 * cgraph.h (chkp_function_instrumented_p): Likewise.
9795 (symtab_node::get_alias_target): Likewise.
9796 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
9797 (cgraph_local_p): Likewise.
9798 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
9799 (cgraph_edge::rebuild_references): Likewise.
9800 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
9801 (walk_polymorphic_call_targets): Likewise.
9802 (cgraph_node::expand_thunk): Likewise.
9803 (symbol_table::output_weakrefs): Likewise.
9804 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
9805 (ix86_handle_option): Likewise.
9806 * config/i386/constraints.md: Likewise.
9807 * config/i386/i386-builtin-types.def (BND): Likewise.
9811 * config/i386/i386-builtin.def (BDESC_END): Likewise.
9812 (BDESC_FIRST): Likewise.
9814 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
9815 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
9816 * config/i386/i386.c (enum reg_class): Likewise.
9817 (ix86_target_string): Likewise.
9818 (ix86_option_override_internal): Likewise.
9819 (ix86_conditional_register_usage): Likewise.
9820 (ix86_valid_target_attribute_inner_p): Likewise.
9821 (ix86_set_indirect_branch_type): Likewise.
9822 (ix86_set_current_function): Likewise.
9823 (ix86_function_arg_regno_p): Likewise.
9824 (init_cumulative_args): Likewise.
9825 (ix86_function_arg_advance): Likewise.
9826 (ix86_function_arg): Likewise.
9827 (ix86_pass_by_reference): Likewise.
9828 (ix86_function_value_regno_p): Likewise.
9829 (ix86_function_value_1): Likewise.
9830 (ix86_function_value_bounds): Likewise.
9831 (ix86_return_in_memory): Likewise.
9832 (ix86_setup_incoming_vararg_bounds): Likewise.
9833 (ix86_va_start): Likewise.
9834 (indirect_thunk_need_prefix): Likewise.
9835 (print_reg): Likewise.
9836 (ix86_print_operand): Likewise.
9837 (ix86_expand_call): Likewise.
9838 (ix86_output_function_return): Likewise.
9839 (reg_encoded_number): Likewise.
9840 (BDESC_VERIFYS): Likewise.
9841 (ix86_init_mpx_builtins): Likewise.
9842 (ix86_init_builtins): Likewise.
9843 (ix86_emit_cmove): Likewise.
9844 (ix86_emit_move_max): Likewise.
9845 (ix86_expand_builtin): Likewise.
9846 (ix86_builtin_mpx_function): Likewise.
9847 (ix86_get_arg_address_for_bt): Likewise.
9848 (ix86_load_bounds): Likewise.
9849 (ix86_store_bounds): Likewise.
9850 (ix86_load_returned_bounds): Likewise.
9851 (ix86_store_returned_bounds): Likewise.
9852 (ix86_class_likely_spilled_p): Likewise.
9853 (ix86_hard_regno_mode_ok): Likewise.
9854 (x86_order_regs_for_local_alloc): Likewise.
9855 (ix86_mitigate_rop): Likewise.
9856 (ix86_bnd_prefixed_insn_p): Likewise.
9857 (ix86_mpx_bound_mode): Likewise.
9858 (ix86_make_bounds_constant): Likewise.
9859 (ix86_initialize_bounds): Likewise.
9860 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
9861 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
9862 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
9863 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
9864 (TARGET_CHKP_BOUND_MODE): Likewise.
9865 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
9866 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
9867 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
9868 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
9869 * config/i386/i386.h (TARGET_MPX): Likewise.
9870 (TARGET_MPX_P): Likewise.
9871 (VALID_BND_REG_MODE): Likewise.
9872 (FIRST_BND_REG): Likewise.
9873 (LAST_BND_REG): Likewise.
9874 (enum reg_class): Likewise.
9875 (BND_REG_P): Likewise.
9876 (BND_REGNO_P): Likewise.
9877 (BNDmode): Likewise.
9878 (ADJUST_INSN_LENGTH): Likewise.
9879 * config/i386/i386.md: Likewise.
9880 * config/i386/i386.opt: Likewise.
9881 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
9882 (defined): Likewise.
9883 (LINK_MPX): Likewise.
9884 (MPX_SPEC): Likewise.
9885 (LIBMPX_SPEC): Likewise.
9886 (LIBMPXWRAPPERS_SPEC): Likewise.
9887 (CHKP_SPEC): Likewise.
9888 * config/i386/predicates.md: Likewise.
9889 * dbxout.c (dbxout_type): Likewise.
9890 * doc/extend.texi: Likewise.
9891 * doc/invoke.texi: Likewise.
9892 * doc/md.texi: Likewise.
9893 * doc/tm.texi: Likewise.
9894 * doc/tm.texi.in: Likewise.
9895 * dwarf2out.c (is_base_type): Likewise.
9896 (gen_formal_types_die): Likewise.
9897 (gen_subprogram_die): Likewise.
9898 (gen_type_die_with_usage): Likewise.
9899 (gen_decl_die): Likewise.
9900 (dwarf2out_late_global_decl): Likewise.
9901 * expr.c (expand_assignment): Likewise.
9902 (emit_storent_insn): Likewise.
9903 (store_expr_with_bounds): Likewise.
9904 (store_expr): Likewise.
9905 (expand_expr_real_1): Likewise.
9906 * expr.h (store_expr_with_bounds): Likewise.
9907 * function.c (use_register_for_decl): Likewise.
9908 (struct bounds_parm_data): Likewise.
9909 (assign_parms_augmented_arg_list): Likewise.
9910 (assign_parm_find_entry_rtl): Likewise.
9911 (assign_parm_is_stack_parm): Likewise.
9912 (assign_parm_load_bounds): Likewise.
9913 (assign_bounds): Likewise.
9914 (assign_parms): Likewise.
9915 (expand_function_start): Likewise.
9916 * gcc.c (CHKP_SPEC): Likewise.
9917 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9918 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
9919 (wrestrict_dom_walker::check_call): Likewise.
9920 * gimple.c (gimple_build_call_from_tree): Likewise.
9921 * gimple.h (enum gf_mask): Likewise.
9922 (gimple_call_with_bounds_p): Likewise.
9923 (gimple_call_set_with_bounds): Likewise.
9924 * gimplify.c (gimplify_init_constructor): Likewise.
9925 * ipa-cp.c (initialize_node_lattices): Likewise.
9926 (propagate_constants_across_call): Likewise.
9927 (find_more_scalar_values_for_callers_subset): Likewise.
9928 * ipa-hsa.c (process_hsa_functions): Likewise.
9929 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
9930 * ipa-icf.c (sem_function::merge): Likewise.
9931 * ipa-inline.c (early_inliner): Likewise.
9932 * ipa-pure-const.c (warn_function_noreturn): Likewise.
9933 (warn_function_cold): Likewise.
9934 (propagate_pure_const): Likewise.
9935 * ipa-ref.h (enum GTY): Likewise.
9936 * ipa-split.c (find_retbnd): Likewise.
9937 (consider_split): Likewise.
9938 (split_function): Likewise.
9939 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
9940 * ipa.c (walk_polymorphic_call_targets): Likewise.
9941 (symbol_table::remove_unreachable_nodes): Likewise.
9942 (process_references): Likewise.
9943 (cgraph_build_static_cdtor_1): Likewise.
9944 * lto-cgraph.c (lto_output_node): Likewise.
9945 (output_refs): Likewise.
9946 (compute_ltrans_boundary): Likewise.
9947 (input_overwrite_node): Likewise.
9948 (input_node): Likewise.
9949 (input_cgraph_1): Likewise.
9950 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
9951 * passes.c (pass_manager::execute_early_local_passes): Likewise.
9952 (class pass_chkp_instrumentation_passes): Likewise.
9953 (make_pass_chkp_instrumentation_passes): Likewise.
9954 * passes.def: Likewise.
9955 * rtl.h (struct GTY): Likewise.
9956 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
9957 * stor-layout.c (layout_type): Likewise.
9958 * symtab.c: Likewise.
9959 * target.def: Likewise.
9960 * targhooks.c (default_chkp_bound_type): Likewise.
9961 (default_chkp_bound_mode): Likewise.
9962 (default_builtin_chkp_function): Likewise.
9963 (default_chkp_function_value_bounds): Likewise.
9964 (default_chkp_make_bounds_constant): Likewise.
9965 (default_chkp_initialize_bounds): Likewise.
9966 * targhooks.h (default_chkp_bound_type): Likewise.
9967 (default_chkp_bound_mode): Likewise.
9968 (default_builtin_chkp_function): Likewise.
9969 (default_chkp_function_value_bounds): Likewise.
9970 (default_chkp_make_bounds_constant): Likewise.
9971 (default_chkp_initialize_bounds): Likewise.
9972 * toplev.c (compile_file): Likewise.
9973 (process_options): Likewise.
9974 * tree-core.h (DEF_BUILTIN): Likewise.
9975 (DEF_BUILTIN_CHKP): Likewise.
9976 * tree-inline.c (declare_return_variable): Likewise.
9977 (remap_gimple_stmt): Likewise.
9978 (copy_bb): Likewise.
9979 (initialize_inlined_parameters): Likewise.
9980 (expand_call_inline): Likewise.
9981 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
9982 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
9983 (make_pass_ipa_chkp_produce_thunks): Likewise.
9984 (make_pass_chkp): Likewise.
9985 (make_pass_chkp_opt): Likewise.
9986 (make_pass_chkp_instrumentation_passes): Likewise.
9987 * tree-pretty-print.c (dump_generic_node): Likewise.
9988 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
9989 * tree-ssa-dce.c (propagate_necessity): Likewise.
9990 (eliminate_unnecessary_stmts): Likewise.
9991 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
9992 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
9993 * tree-ssa-sccvn.h: Likewise.
9994 * tree-ssa-strlen.c (get_string_length): Likewise.
9995 (valid_builtin_call): Likewise.
9996 (adjust_last_stmt): Likewise.
9997 (handle_builtin_strchr): Likewise.
9998 (handle_builtin_strcpy): Likewise.
9999 (handle_builtin_stxncpy): Likewise.
10000 (handle_builtin_memcpy): Likewise.
10001 (handle_builtin_strcat): Likewise.
10002 (strlen_check_and_optimize_stmt): Likewise.
10003 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
10004 * tree-streamer-in.c: Likewise.
10005 * tree-streamer.c (record_common_node): Likewise.
10006 * tree.c (tree_code_size): Likewise.
10007 (wide_int_to_tree_1): Likewise.
10008 (type_contains_placeholder_1): Likewise.
10009 (build_common_tree_nodes): Likewise.
10010 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
10011 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
10012 (POINTER_BOUNDS_P): Likewise.
10013 (BOUNDED_TYPE_P): Likewise.
10014 (BOUNDED_P): Likewise.
10015 (CALL_WITH_BOUNDS_P): Likewise.
10016 (pointer_bounds_type_node): Likewise.
10017 * value-prof.c (gimple_ic): Likewise.
10018 * var-tracking.c (vt_add_function_parameters): Likewise.
10019 * varasm.c (make_decl_rtl): Likewise.
10020 (assemble_start_function): Likewise.
10021 (output_constant): Likewise.
10022 (maybe_assemble_visibility): Likewise.
10023 * varpool.c (ctor_for_folding): Likewise.
10024 * chkp-builtins.def: Remove.
10025 * ipa-chkp.c: Remove.
10026 * ipa-chkp.h: Remove.
10027 * rtl-chkp.c: Remove.
10028 * rtl-chkp.h: Remove.
10029 * tree-chkp-opt.c: Remove.
10030 * tree-chkp.c: Remove.
10031 * tree-chkp.h: Remove.
10033 2018-06-07 Carl Love <cel@us.ibm.com>
10035 * config/rs6000/vsx.md (vextract_fp_from_shorth,
10036 vextract_fp_from_shortl): Add BE support.
10038 2018-06-07 Paul Koning <ni1d@arrl.net>
10040 * compare-elim.c (try_merge_compare): Don't merge compare if
10041 address contains a side effect.
10042 (try_eliminate_compare): Likewise.
10044 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
10046 * config.gcc: Support "tremont".
10047 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
10048 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
10050 * config/i386/i386.c (m_TREMONT): Define.
10051 (processor_target_table): Add "tremont".
10052 (PTA_TREMONT): Define.
10053 (ix86_lea_outperforms): Add TARGET_TREMONT.
10054 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
10055 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
10056 and M_INTEL_GOLDMONT_PLUS.
10057 (fold_builtin_cpu): Add "tremont".
10058 (ix86_add_stmt_cost): Add TARGET_TREMONT.
10059 (ix86_option_override_internal): Add "tremont".
10060 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
10061 (processor_type): Add PROCESSOR_TREMONT.
10062 * config/i386/x86-tune.def: Add m_TREMONT.
10063 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
10065 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10067 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
10068 symbol defined for msp430i* devices to be lower case.
10070 2018-06-07 Richard Biener <rguenther@suse.de>
10072 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
10073 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
10074 Properly wrap signed arithmetic if overflow wraps.
10076 2018-06-07 Jakub Jelinek <jakub@redhat.com>
10078 PR tree-optimization/69615
10079 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
10080 of a cast from a same precision integral SSA_NAME in a bb dominated
10081 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
10082 cast to utype if rhs2 has already a compatible type.
10084 2018-06-07 Richard Biener <rguenther@suse.de>
10086 PR tree-optimization/85935
10087 * graphite-scop-detection.c (find_params_in_bb): Analyze
10088 condition operands with respect to the correct loop. Assert
10089 the analysis doesn't fail.
10091 2018-06-04 Carl Love <cel@us.ibm.com>
10093 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
10094 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
10095 as it is slightly cheaper.
10096 (first_match_or_eos_index_<mode>):
10097 Calculate index using natural element order.
10098 (first_match_index_<mode>):
10099 Calculate index using natural element order.
10100 (first_match_or_eos_index_<mode>):
10101 Calculate index using natural order.
10102 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
10103 for BE and LE modes.
10104 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
10105 P9V_BUILTIN_VCLZLSBB_V16QI.
10106 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
10109 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
10111 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
10112 indentation and line wrap for many prototypes. Add missing
10113 @smallexample directives around block of prototypes for vec_xl and
10116 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
10118 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
10119 track if we pass or return IEEE 128-bit floating point.
10120 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
10121 C++ mangling that is compatible with GCC 8.1.
10122 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
10123 (init_cumulative_args): Note if we pass or return IEEE 128-bit
10124 floating point types.
10125 (rs6000_function_arg_advance_1): Likewise.
10126 (rs6000_mangle_type): Optionally generate mangled names that match
10127 what GCC 8.1 generated for IEEE 128-bit floating point types.
10128 (rs6000_globalize_decl_name): If we have an external function that
10129 passes or returns IEEE 128-bit types, generate a weak reference
10130 from the mangled name used in GCC 8.1 to the current mangled
10132 (rs6000_init_builtins): Make __ibm128 use the long double type if
10133 long double is IBM extended double. Make __float128 use the long
10134 double type if long double is IEEE 128-bit.
10137 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
10138 macro for __ibm128 built-in functions.
10139 (PACK_IF): Add __ibm128 pack/unpack functions.
10140 (UNPACK_IF): Likewise.
10141 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
10142 enable long double built-in functions if long double is IEEE
10143 128-bit floating point.
10144 (rs6000_invalid_builtin): Update long double built-in function
10146 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
10147 functions, adjust the built-in function to use the long double
10148 built-in function if __ibm128 and long double are the same type.
10149 * doc/extend.texi (PowerPC builtins): Update documention for
10150 __builtin_{,un}pack_longdouble. Add documentation for
10151 __builtin_{,un}pack_ibm128.
10153 2018-06-06 Jim Wilson <jimw@sifive.com>
10155 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
10156 (struct machine_function): New field interrupt_mode.
10157 (riscv_handle_type_attribute): New function. Add forward declaration.
10158 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
10159 (riscv_expand_epilogue): Check interrupt_mode field.
10160 (riscv_set_current_function): Check interrupt attribute args and
10161 set interrupt_mode field.
10162 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
10163 (riscv_sret, riscv_uret): New.
10164 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
10165 new arguments to interrupt attribute.
10167 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
10170 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
10171 Don't handle -mcpu=power8 if -mpower9-vector is also used.
10173 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
10175 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
10176 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
10177 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
10178 several redundant entries.
10180 2018-06-06 David Malcolm <dmalcolm@redhat.com>
10182 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
10183 type from "rtx" to "rtx_insn *".
10184 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
10185 for local "call_insn", removing cast.
10186 (ix86_expand_call): Likewise, introducing a "call_insn" local.
10188 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
10190 PR tree-optimization/86066
10191 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
10192 for BIT_INSERT_EXPR stores.
10194 2018-06-06 Richard Biener <rguenther@suse.de>
10196 PR tree-optimization/86062
10197 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
10198 component refs ontop
10199 of to be offsetted base.
10201 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10203 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
10204 to be static and remove check on interrupt attribute name.
10206 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
10208 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
10209 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
10211 2018-06-05 Steve Ellcey <sellcey@cavium.com>
10214 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
10216 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
10217 Remove second argument, change how error is called.
10218 (aarch64_layout_arg): Remove second argument from
10219 aarch64_err_no_fpadvsimd call.
10220 (aarch64_init_cumulative_args): Ditto.
10221 (aarch64_gimplify_va_arg_expr): Ditto.
10222 * config/aarch64/aarch64.md (mov<mode>): Ditto.
10224 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
10226 * config/i386/i386.md (simple_return_indirect_internal): New expander.
10227 (*simple_return_indirect_internal<mode>): Rename from
10228 simple_return_indirect_internal. Use W mode iterator.
10229 (rstorssp): New expander.
10230 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
10231 (clrssbsy): New expander.
10232 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
10234 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
10236 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
10238 (cmse_check_pointed_object): Likewise.
10240 2018-06-05 Martin Liska <mliska@suse.cz>
10242 PR gcov-profile/47618
10243 * doc/invoke.texi: Document how -fprofile-dir format
10246 2018-06-05 Richard Biener <rguenther@suse.de>
10248 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
10249 removal pretend DOM info isn't available so we do not update
10250 it and only remove edges, not dominated blocks. Actually free
10251 DOM info in case we removed something. Remove unreachable blocks.
10252 (mfb_keep_latches): Work with either DOM info or marked backedges.
10253 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
10254 first. Mark backedges if DOM info isn't available.
10255 (Re-)compute DOM info after cleanup_control_flow_pre.
10257 2018-06-05 Richard Biener <rguenther@suse.de>
10259 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
10260 (locus_discrim_hasher::hash): Adjust.
10261 (locus_discrim_hasher::equal): Likewise.
10262 (next_discriminator_for_locus): Work on line directly.
10263 (same_line_p): Pass in expanded locus1 as well.
10264 (assign_discriminators): Avoid redundant location expansions.
10266 2018-06-05 Richard Biener <rguenther@suse.de>
10268 PR tree-optimization/86046
10269 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
10270 if required after clearing TREE_ADDRESSABLE.
10272 2018-06-05 Richard Biener <rguenther@suse.de>
10274 PR tree-optimization/86047
10275 * tree-ssa-loop.c (for_each_index): Glob handling of all
10276 decls and constants and really handle all of them.
10278 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10281 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
10282 qualifier_void_pointer and qualifier_const_void_pointer.
10283 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
10284 (arm_init_builtins): Handle the above.
10285 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
10286 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
10287 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
10290 2018-06-05 Martin Liska <mliska@suse.cz>
10292 * auto-profile.c (read_autofdo_file): Do not use
10293 gcov_ctr_summary struct.
10294 (afdo_callsite_hot_enough_for_early_inline): Likewise.
10295 * coverage.c (struct counts_entry): Likewise.
10296 (read_counts_file): Read just single summary entry.
10297 (get_coverage_counts): Use gcov_summary struct.
10298 * coverage.h (get_coverage_counts): Likewise.
10299 * gcov-dump.c (dump_working_sets): Likewise.
10300 (tag_summary): Dump just single summary.
10301 * gcov-io.c (gcov_write_summary): Write just histogram
10303 (gcov_read_summary): Read just single summary.
10304 (compute_working_sets): Use gcov_summary struct.
10305 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
10306 of GCOV_COUNTERS_SUMMABLE.
10307 (GCOV_COUNTERS_SUMMABLE): Remove.
10308 (GCOV_FIRST_VALUE_COUNTER): Replace with
10309 GCOV_COUNTER_V_INTERVAL.
10310 (struct gcov_ctr_summary): Remove.
10311 (struct gcov_summary): Directly use fields of former
10313 (compute_working_sets): Use gcov_summary struct.
10314 * gcov.c (read_count_file): Do not use ctrs fields.
10315 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
10317 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
10319 * profile.c: Likewise.
10320 * profile.h: Likewise.
10322 2018-06-05 Martin Liska <mliska@suse.cz>
10324 PR gcov-profile/84846
10325 * gcov.c (output_lines): Print working directory only
10326 in intermediate format.
10328 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
10330 * config/s390/s390-builtin-types.def: Add void function type.
10331 * config/s390/s390-builtins.def: Use the function type for the
10334 2018-06-04 Jim Wilson <jimw@sifive.com>
10336 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
10338 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
10339 and maybe_eh_return. Change regno to unsigned int. Use new args to
10340 handle EH_RETURN_DATA_REGNO registers properly.
10341 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
10342 (riscv_expand_epilogue): Update comment. Change argument name and
10343 type. Update code to use new name and type. Pass new args to
10344 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
10346 * config/riscv/riscv.md (NORMAL_RETURN): New.
10347 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
10348 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
10349 (eh_return): Call gen_eh_return_internal and emit barrier.
10350 (eh_return_internal): Call riscv_expand_epilogue.
10352 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
10354 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
10355 bit_insertion field and declare can_be_merged_into method.
10356 (merged_store_group::can_be_merged_into): New method.
10357 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
10358 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
10359 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
10361 2018-06-04 Richard Biener <rguenther@suse.de>
10363 PR tree-optimization/85955
10364 * builtins.c (fold_builtin_sincos): Convert pointers to
10365 destination to appropriate type before dereferencing.
10367 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
10369 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
10371 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
10373 * expr.c (expand_expr_real_1): Force the operand into memory if
10374 its TYPE_MODE is BLKmode and if there is no integer mode for
10375 the number of bits being extracted.
10377 2018-06-04 Jakub Jelinek <jakub@redhat.com>
10381 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
10382 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
10384 2018-06-04 Richard Biener <rguenther@suse.de>
10386 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
10387 (cleanup_tree_cfg_noloop): ... single caller. Do
10388 start_recording_case_labels later.
10390 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
10392 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
10393 to _IMMINTRIN_H_INCLUDED.
10394 * config/i386/pconfigintrin.h: Ditto.
10395 * config/i386/waitpkgintrin.h: Ditto.
10396 * config/i386/immintrin.h: Add includes for sgxintrin.h,
10397 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
10398 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
10399 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
10400 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
10401 waitpkgintrin.h and cldemoteintrin.h.
10403 2018-06-04 Richard Biener <rguenther@suse.de>
10405 PR tree-optimization/86038
10406 * tracer.c (find_best_successor): Check probability for
10407 being initialized, bail out if not.
10409 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
10412 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
10413 of bits to ignore when comparing architectures.
10415 2018-06-04 Jakub Jelinek <jakub@redhat.com>
10417 PR tree-optimization/69615
10418 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
10419 maximum or minimum of the type, try to merge it also as if
10420 range1 is + [-, x - 1] or + [x + 1, -].
10423 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
10425 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
10427 PR tree-optimization/86034
10428 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
10429 the unsigned bitfield type in a bit insertion sequence if it does not
10430 have a larger precision than the bitfield size.
10431 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
10433 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
10435 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
10437 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
10439 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
10440 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
10441 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
10442 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
10444 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
10446 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
10447 Disable -fdelete-null-pointer-checks for ELF toolchain.
10449 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
10450 Kito Cheng <kito.cheng@gmail.com>
10452 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
10453 (nds32le-*-*, nds32be-*-*): Integrate checking process.
10454 (nds32*-*-*): Add glibc and uclibc conditions.
10455 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
10456 (TARGET_EXCEPT_UNWIND_INFO): Define.
10457 * config/nds32/elf.h: New file.
10458 * config/nds32/linux.h: New file.
10459 * config/nds32/nds32-elf.opt: New file.
10460 * config/nds32/nds32-linux.opt: New file.
10461 * config/nds32/nds32-fp-as-gp.c
10462 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
10463 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
10465 (nds32_asm_file_end): Ditto.
10466 (nds32_print_operand): Ditto.
10467 (nds32_insert_attributes): Ditto.
10468 (nds32_init_libfuncs): New function.
10469 (TARGET_HAVE_TLS): Define.
10470 (TARGET_INIT_LIBFUNCS): Define.
10471 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
10473 (TARGET_ELF): Apply different mcmodel setting.
10474 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
10475 been migrated into elf.h and linux.h files.
10476 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
10477 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
10478 (mcmodel): The content has been migrated into nds32-elf.opt and
10479 nds32-linux.opt files.
10480 * config/nds32/t-elf: New file.
10481 * config/nds32/t-linux: New file.
10483 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
10484 Shiva Chen <shiva0217@gmail.com>
10486 * config/nds32/constants.md (unspec_volatile_element): Add
10487 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
10488 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
10490 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
10491 (make_pass_nds32_fp_as_gp): Declare.
10492 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
10494 (nds32_asm_function_end_prologue): Remove unused asm output.
10495 (nds32_asm_function_begin_epilogue): Remove unused asm output.
10496 (nds32_asm_file_start): Output necessary fp_as_gp information.
10497 (nds32_option_override): Adjust register usage.
10498 (nds32_expand_prologue): Consider fp_as_gp situation.
10499 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
10500 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
10503 (simple_return): Ditto.
10504 (omit_fp_begin): Output special directive for fp_as_gp.
10505 (omit_fp_end): Output special directive for fp_as_gp.
10506 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
10507 mforbid-fp-as-gp): New options.
10509 2018-06-01 Mark Wielaard <mark@klomp.org>
10511 * dwarf2out.c (dwarf2out_finish): Remove generation of
10512 DW_AT_loclists_base.
10514 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
10516 * gimple-ssa-store-merging.c: Include gimple-fold.h.
10517 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
10518 (struct merged_store_group): Add bit_insertion field.
10519 (dump_char_array): Use standard hexadecimal format.
10520 (merged_store_group::merged_store_group): Set bit_insertion to false.
10521 (merged_store_group::apply_stores): Use optimal buffer size. Deal
10522 with BIT_INSERT_EXPR stores. Move up code updating the mask and
10523 also print the mask in the dump file.
10524 (pass_store_merging::gate): Minor tweak.
10525 (imm_store_chain_info::coalesce_immediate): Fix wrong association
10526 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
10527 stores with INTEGER_CST stores.
10528 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
10529 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
10530 and use it throughout. Generate bit insertion sequences if need be.
10531 (pass_store_merging::process_store): Remove redundant condition.
10532 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
10534 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
10536 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
10537 the 128-bit floating point types. Fix function comment.
10539 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10541 * config/aarch64/aarch64-simd.md
10542 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
10544 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
10547 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
10549 PR tree-optimization/85989
10550 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
10552 (backprop::intersect_uses): Check it when deciding whether this
10553 is a backedge reference.
10554 (backprop::process_block): Add each phi to m_visited_phis
10555 after visiting it, then clear it at the end.
10557 2018-06-01 Richard Biener <rguenther@suse.de>
10559 * tree-vectorizer.h (vect_dr_stmt): New function.
10560 (vect_get_load_cost): Adjust.
10561 (vect_get_store_cost): Likewise.
10562 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
10563 Use vect_dr_stmt instead of DR_SMTT.
10564 (vect_record_base_alignments): Likewise.
10565 (vect_calculate_target_alignment): Likewise.
10566 (vect_compute_data_ref_alignment): Likewise and make static.
10567 (vect_update_misalignment_for_peel): Likewise.
10568 (vect_verify_datarefs_alignment): Likewise.
10569 (vector_alignment_reachable_p): Likewise.
10570 (vect_get_data_access_cost): Likewise. Pass down
10571 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
10572 (vect_get_peeling_costs_all_drs): Likewise.
10573 (vect_peeling_hash_get_lowest_cost): Likewise.
10574 (vect_enhance_data_refs_alignment): Likewise.
10575 (vect_find_same_alignment_drs): Likewise.
10576 (vect_analyze_data_refs_alignment): Likewise.
10577 (vect_analyze_group_access_1): Likewise.
10578 (vect_analyze_group_access): Likewise.
10579 (vect_analyze_data_ref_access): Likewise.
10580 (vect_analyze_data_ref_accesses): Likewise.
10581 (vect_vfa_segment_size): Likewise.
10582 (vect_small_gap_p): Likewise.
10583 (vectorizable_with_step_bound_p): Likewise.
10584 (vect_prune_runtime_alias_test_list): Likewise.
10585 (vect_analyze_data_refs): Likewise.
10586 (vect_supportable_dr_alignment): Likewise.
10587 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
10588 (vect_gen_prolog_loop_niters): Likewise.
10589 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
10590 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
10592 (vect_recog_mask_conversion_pattern): Likewise.
10593 (vect_try_gather_scatter_pattern): Likewise.
10594 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
10595 to vect_get_store_cost.
10596 (vect_get_store_cost): Get stmt_info instead of DR.
10597 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
10598 (vect_get_load_cost): Get stmt_info instead of DR.
10600 2018-06-01 Richard Biener <rguenther@suse.de>
10602 PR middle-end/86017
10603 * gimple-fold.c (var_decl_component_p): Also allow offsetted
10604 vars wrapped in MEM_REFs.
10606 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
10608 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10609 Fix subreg tests so that we only return a choice between
10610 GENERAL_REGS and FP_REGS if the original classes included both.
10612 2018-06-01 Richard Biener <rguenther@suse.de>
10615 * tree-ssa-structalias.c (get_function_part_constraint):
10616 Handle NULL fi->decl.
10617 (find_func_aliases_for_call): Properly handle indirect
10618 fi from direct call.
10619 (find_func_clobbers): Likewise.
10620 (ipa_pta_execute): Likewise.
10621 (create_variable_info_for): For functions that are ifunc_resolver
10622 resolve to a varinfo that contains the result of the resolver call.
10623 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
10626 2018-05-31 Michael Collison <michael.collison@arm.com>
10628 * config/aarch64/aarch64.md:
10629 (*fix_to_zero_extenddfdi2): New pattern.
10630 * gcc.target/aarch64/fix_extend1.c: New testcase.
10632 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
10634 PR middle-end/78809
10635 PR middle-end/83026
10636 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
10637 and BUILT_IN_STRNCMP_EQ.
10638 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
10639 BUILT_IN_STRNCMP_EQ.
10640 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
10641 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
10642 (gimple_fold_builtin): Likewise.
10643 * tree-ssa-strlen.c (compute_string_length): New function.
10644 (determine_min_obsize): New function.
10645 (handle_builtin_string_cmp): New function to handle calls to
10646 string compare functions.
10647 (strlen_optimize_stmt): Add handling to builtin string compare
10649 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
10650 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
10651 * tree.c (build_common_builtin_nodes): Add new defines of
10652 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
10654 2018-05-31 Jakub Jelinek <jakub@redhat.com>
10657 * bb-reorder.c (pass_partition_blocks::gate): Return false for
10658 functions with naked attribute.
10660 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
10662 * config/i386/sse.md (avx_vec_concat<mode>):
10663 Substitute concat_tg_mode mode attribute with xtg_mode.
10664 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
10665 (concat_tg_mode): Remove mode attribute.
10667 2018-05-31 Martin Sebor <msebor@redhat.com>
10670 * calls.c (alloc_max_size): Correct a logic error/typo.
10671 Treat excessive arguments as infinite. Warn for invalid arguments.
10672 * doc/invoke.texi (-Walloc-size-larger-than): Update.
10674 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
10677 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
10678 and movx for Haswell.
10680 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
10681 Cesar Philippidis <cesar@codesourcery.com>
10683 PR middle-end/85879
10684 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
10685 when emitting error on private/firstprivate reductions.
10686 * omp-low.c (lower_omp_target): Avoid reference-type processing
10687 on pointers for firstprivate clause.
10689 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
10691 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
10694 * config/aarch64/aarch64-simd.md
10695 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
10696 (aarch64_ld1_x3_<mode>): Likewise
10697 (aarch64_st1x2<VALLDIF:mode>): Likewise
10698 (aarch64_st1_x2_<mode>): Likewise
10699 (aarch64_st1x3<VALLDIF:mode>): Likewise
10700 (aarch64_st1_x3_<mode>): Likewise
10701 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
10702 (vld1_s8_x3): Likewise.
10703 (vld1_u16_x3): Likewise.
10704 (vld1_s16_x3): Likewise.
10705 (vld1_u32_x3): Likewise.
10706 (vld1_s32_x3): Likewise.
10707 (vld1_u64_x3): Likewise.
10708 (vld1_s64_x3): Likewise.
10709 (vld1_f16_x3): Likewise.
10710 (vld1_f32_x3): Likewise.
10711 (vld1_f64_x3): Likewise.
10712 (vld1_p8_x3): Likewise.
10713 (vld1_p16_x3): Likewise.
10714 (vld1_p64_x3): Likewise.
10715 (vld1q_u8_x3): Likewise.
10716 (vld1q_s8_x3): Likewise.
10717 (vld1q_u16_x3): Likewise.
10718 (vld1q_s16_x3): Likewise.
10719 (vld1q_u32_x3): Likewise.
10720 (vld1q_s32_x3): Likewise.
10721 (vld1q_u64_x3): Likewise.
10722 (vld1q_s64_x3): Likewise.
10723 (vld1q_f16_x3): Likewise.
10724 (vld1q_f32_x3): Likewise.
10725 (vld1q_f64_x3): Likewise.
10726 (vld1q_p8_x3): Likewise.
10727 (vld1q_p16_x3): Likewise.
10728 (vld1q_p64_x3): Likewise.
10729 (vst1_s64_x2): Likewise.
10730 (vst1_u64_x2): Likewise.
10731 (vst1_f64_x2): Likewise.
10732 (vst1_s8_x2): Likewise.
10733 (vst1_p8_x2): Likewise.
10734 (vst1_s16_x2): Likewise.
10735 (vst1_p16_x2): Likewise.
10736 (vst1_s32_x2): Likewise.
10737 (vst1_u8_x2): Likewise.
10738 (vst1_u16_x2): Likewise.
10739 (vst1_u32_x2): Likewise.
10740 (vst1_f16_x2): Likewise.
10741 (vst1_f32_x2): Likewise.
10742 (vst1_p64_x2): Likewise.
10743 (vst1q_s8_x2): Likewise.
10744 (vst1q_p8_x2): Likewise.
10745 (vst1q_s16_x2): Likewise.
10746 (vst1q_p16_x2): Likewise.
10747 (vst1q_s32_x2): Likewise.
10748 (vst1q_s64_x2): Likewise.
10749 (vst1q_u8_x2): Likewise.
10750 (vst1q_u16_x2): Likewise.
10751 (vst1q_u32_x2): Likewise.
10752 (vst1q_u64_x2): Likewise.
10753 (vst1q_f16_x2): Likewise.
10754 (vst1q_f32_x2): Likewise.
10755 (vst1q_f64_x2): Likewise.
10756 (vst1q_p64_x2): Likewise.
10757 (vst1_s64_x3): Likewise.
10758 (vst1_u64_x3): Likewise.
10759 (vst1_f64_x3): Likewise.
10760 (vst1_s8_x3): Likewise.
10761 (vst1_p8_x3): Likewise.
10762 (vst1_s16_x3): Likewise.
10763 (vst1_p16_x3): Likewise.
10764 (vst1_s32_x3): Likewise.
10765 (vst1_u8_x3): Likewise.
10766 (vst1_u16_x3): Likewise.
10767 (vst1_u32_x3): Likewise.
10768 (vst1_f16_x3): Likewise.
10769 (vst1_f32_x3): Likewise.
10770 (vst1_p64_x3): Likewise.
10771 (vst1q_s8_x3): Likewise.
10772 (vst1q_p8_x3): Likewise.
10773 (vst1q_s16_x3): Likewise.
10774 (vst1q_p16_x3): Likewise.
10775 (vst1q_s32_x3): Likewise.
10776 (vst1q_s64_x3): Likewise.
10777 (vst1q_u8_x3): Likewise.
10778 (vst1q_u16_x3): Likewise.
10779 (vst1q_u32_x3): Likewise.
10780 (vst1q_u64_x3): Likewise.
10781 (vst1q_f16_x3): Likewise.
10782 (vst1q_f32_x3): Likewise.
10783 (vst1q_f64_x3): Likewise.
10784 (vst1q_p64_x3): Likewise.
10786 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10788 * config/msp430/msp430.c (msp430_output_labelref): Prepend
10789 user_label_prefix to name.
10791 * tree-core.h: Update comment about the format of NAME string
10792 passed to handler in attribute_spec.
10794 * config/msp430/msp430.md: Remove erroneous subreg expression from
10795 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
10796 zero_extend{q,h}isi2.
10798 2018-05-30 Borislav Petkov <bp@suse.de>
10800 * doc/extend.texi: Document some architecture specific
10801 constraints and sort entries.
10803 2018-05-30 Martin Sebor <msebor@redhat.com>
10805 PR middle-end/85369
10806 * builtins.c (expand_builtin_stpcpy_1): New function.
10807 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
10808 only if the former succeeds.
10810 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
10812 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
10815 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10817 * doc/invoke.texi (-flinker-output): Document
10819 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10821 * passes.c (ipa_write_summaries): Only modify statements if body
10823 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
10824 incrementally linking.
10825 (ipa_passes): Likewise.
10826 * lto-cgraph.c (lto_output_node): When incrementally linking do not
10827 pass down resolution info.
10828 * common.opt (flag_incremental_link): Update info.
10829 * gcc.c (plugin specs): Turn flinker-output=* to
10830 -plugin-opt=-linker-output-known
10831 * toplev.c (compile_file): Also cut compilation when doing incremental
10833 * flag-types. (enum lto_partition_model): Add
10834 LTO_LINKER_OUTPUT_NOLTOREL.
10835 (invoke.texi): Add -flinker-output docs.
10836 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
10837 link same way as WPA; do not stream in dead initializers.
10839 * dwarf2out.c (dwarf2out_die_ref_for_decl,
10840 darf2out_register_external_decl): Support incremental link.
10842 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10844 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
10846 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10848 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
10849 it down to simple_object_copy_lto_debug_sections.
10850 (run_gcc): Determine incremental LTO link time and configure
10851 lto1 into non-wpa mode, disable renaming of debug sections.
10853 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
10855 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
10856 descriptions of various incorrectly documented functions.
10858 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
10861 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
10862 address check not strict.
10864 2018-05-30 Richard Biener <rguenther@suse.de>
10866 PR tree-optimization/85964
10867 * tracer.c (better_p): Drop initialized count check, we only
10868 call the function with initialized counts now.
10869 (find_best_successor): Do find a best edge if one
10870 has uninitialized count.
10871 (find_best_predecessor): Likewise. Do BB frequency check only
10872 if count is initialized.
10874 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
10876 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
10877 (aarch64_ldrstr_offset_compare): New.
10878 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
10879 load/store orderings.
10880 (aarch64_gen_adjusted_ldpstp): Likewise.
10882 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
10884 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10885 Check for subset of GENERAL_REGS and FP_REGS.
10886 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
10889 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
10891 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
10892 and wi::to_poly_offset. Add the current offset and then check
10893 whether the sum fits, rather than using an unchecked addition of
10894 a checked term. Check for a shwi rather than a uhwi.
10895 * expr.c (get_bit_range): Use tree_to_poly_uint64.
10896 (store_constructor): Use poly_int_tree_p.
10897 (expand_expr_real_1): Likewise.
10898 * function.c (assign_temp): Likewise.
10899 * fold-const.c (const_binop): Use poly_int_tree_p and
10900 wi::to_poly_offset.
10901 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
10903 * ipa-icf-gimple.c (func_checker::compare_operand): Use
10904 to_poly_offset for MEM offsets.
10905 * ipa-icf.c (sem_variable::equals): Likewise.
10906 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
10907 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
10908 wi::to_poly_offset for BIT_FIELD_REF offsets.
10909 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
10910 wi::to_poly_offset.
10911 * var-tracking.c (emit_note_insn_var_location): Use
10912 tree_to_poly_uint64.
10914 2018-05-29 Jim Wilson <jimw@sifive.com>
10916 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
10918 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
10921 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
10922 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
10924 (sse4_1_round<mode>2): Use nonimmediate_operand
10925 for operand 1 predicate.
10927 2018-05-29 Martin Sebor <msebor@redhat.com>
10928 Richard Biener <rguenther@suse.de>
10931 * calls.c (get_size_range): Call determine_value_range instead
10932 of get_value_range..
10933 * tree-vrp.h (determine_value_range): Declared new function.
10934 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
10936 2018-05-29 Richard Biener <rguenther@suse.de>
10938 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
10939 sure to use non-pattern stmts for get_earlier_stmt arguments.
10940 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
10941 called on pattern stmts.
10942 (get_later_stmt): Likewise.
10944 2018-05-29 Martin Liska <mliska@suse.cz>
10946 PR gcov-profile/85759
10947 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
10950 2018-05-29 Jakub Jelinek <jakub@redhat.com>
10952 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
10954 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
10958 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
10959 VEC_PACK_FLOAT_EXPR): New tree codes.
10960 * tree-pretty-print.c (op_code_prio): Handle
10961 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
10962 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
10963 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
10964 * tree-inline.c (estimate_operator_cost): Likewise.
10965 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
10966 * fold-const.c (const_binop): Likewise.
10967 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
10968 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
10969 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
10970 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
10971 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
10972 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
10973 * expr.c (expand_expr_real_2): Likewise.
10974 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
10975 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
10976 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
10978 * optabs.c (expand_widen_pattern_expr): For
10979 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
10980 sign from result type rather than operand's type.
10981 (expand_binop_directly): For vec_packu_float_optab and
10982 vec_packs_float_optab allow result type to be different from operand's
10984 * optabs-tree.c (optab_for_tree_code): Handle
10985 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
10986 VEC_PACK_FLOAT_EXPR. Formatting fixes.
10987 * tree-vect-generic.c (expand_vector_operations_1): Handle
10988 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
10989 VEC_PACK_FLOAT_EXPR.
10990 * tree-vect-stmts.c (supportable_widening_operation): Handle
10992 (supportable_narrowing_operation): Handle FLOAT_EXPR.
10993 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
10994 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
10995 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
10996 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
10998 (vec_pack<floatprefix>_float_<mode>): New expander.
10999 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
11001 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
11002 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
11003 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
11004 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
11005 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
11007 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
11008 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
11009 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
11010 VEC_PACK_FLOAT_EXPR): Document.
11012 2018-05-29 Richard Biener <rguenther@suse.de>
11014 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
11016 (stmt_vec_info_vec): Make pointer.
11017 (init_stmt_vec_info_vec): Remove.
11018 (free_stmt_vec_info_vec): Likewise.
11019 (set_stmt_vec_info_vec): New function.
11020 (free_stmt_vec_infos): Likewise.
11021 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
11022 (set_vinfo_for_stmt): Likewise.
11023 (get_earlier_stmt): Likewise.
11024 (get_later_stmt): Likewise.
11025 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
11026 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
11027 (vec_info::~vec_info): Free stmt_vec_infos.
11028 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
11029 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
11030 (pass_slp_vectorize::execute): Likewise.
11031 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
11032 (free_stmt_vec_info_vec): Likewise.
11033 (set_stmt_vec_info_vec): New function.
11034 (free_stmt_vec_infos): Likewise.
11035 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
11036 the global stmt_vec_info_vec.
11037 * tree-parloops.c (gather_scalar_reductions): Use
11038 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
11041 2018-05-29 Richard Biener <rguenther@suse.de>
11043 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
11045 2018-05-29 Martin Liska <mliska@suse.cz>
11046 David Malcolm <dmalcolm@redhat.com>
11048 * vec.c (test_reverse): New.
11049 (vec_c_tests): Add new test.
11050 * vec.h (vl_ptr>::reverse): New function.
11052 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
11054 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
11056 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
11059 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
11061 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
11063 2018-05-28 Richard Biener <rguenther@suse.de>
11065 PR tree-optimization/85933
11066 * tree-vect-data-refs.c (vect_record_base_alignments): Only
11067 look at stmts marked as vectorizable.
11069 2018-05-28 Richard Biener <rguenther@suse.de>
11071 PR tree-optimization/85934
11072 * tree-vect-generic.c (expand_vector_operations_1): Hoist
11073 vector boolean check before scalar optimization.
11075 2018-05-28 Jakub Jelinek <jakub@redhat.com>
11077 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
11080 2018-05-28 Mark Wielaard <mark@klomp.org>
11082 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
11083 if it is an expression containing a minus sign.
11085 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
11087 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
11089 2018-05-27 Paul Koning <ni1d@arrl.net>
11091 * config/pdp11/pdp11.md (truncsihi2): Remove.
11093 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
11094 Chung-Ju Wu <jasonwucj@gmail.com>
11096 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
11098 (unaligned_store_dw): Ditto.
11099 * config/nds32/nds32-memory-manipulation.c
11100 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
11101 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
11102 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
11103 (emit_setmem_word_loop): Rename to ...
11104 (emit_setmem_doubleword_loop): ... this.
11105 (nds32_gen_dup_4_byte_to_word_value): New function.
11106 (nds32_gen_dup_8_byte_to_double_word_value): New function.
11107 (nds32_expand_setmem_loop): Refine implementation.
11108 (nds32_expand_setmem_loop_v3m): Ditto.
11109 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
11112 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
11114 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
11116 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
11118 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
11119 (nds32_init_machine_status): Initialize machine->attr_naked_p and
11120 machine->attr_no_prologue_p.
11121 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
11122 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
11123 (nds32_expand_epilogue): Consider attr_naked_p.
11124 (nds32_expand_epilogue_v3pop): Likewise.
11125 (nds32_can_use_return_insn): Likewise.
11126 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
11127 attr_no_prologue_p fields.
11128 * config/nds32/nds32.opt (mret-in-naked-func): New option.
11130 2018-05-27 Jakub Jelinek <jakub@redhat.com>
11133 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
11135 * config/i386/sse.md
11136 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
11138 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
11140 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
11142 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
11144 (*<floatsuffix>floatv2div2sf2): Rename to ...
11145 (*float<floatunssuffix>v2div2sf2): ... this.
11146 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
11147 (float<floatunssuffix>v2div2sf2_mask): ... this.
11148 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
11149 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
11150 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
11152 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
11154 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
11156 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
11158 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
11160 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
11162 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
11163 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
11164 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
11165 gen_ufix_truncv8dfv8si2.
11166 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
11167 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
11168 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
11169 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
11170 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
11171 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
11172 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
11173 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
11175 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
11179 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
11181 2018-05-25 Jim Wilson <jimw@sifive.com>
11183 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
11184 * config/riscv/riscv.c (struct machine_function): Add
11185 interrupt_handler_p and attribute_checked_p fields.
11186 (riscv_attribute_table): Add interrupt.
11187 (riscv_interrupt_type_p): New.
11188 (riscv_save_reg_p): Save extra regs for interrupt handler.
11189 (riscv_use_save_libcall): Return false for interrupt handler.
11190 (riscv_first_stack_step): Add forward declaration.
11191 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
11192 for interrupt handler with large frame. Use it for saved reg list.
11193 (riscv_expand_prologue): Move flag_stack_usage_info support to
11194 eliminate duplication.
11195 (riscv_expand_epilogue): Generate mret for interrupt handler.
11196 (riscv_epilogue_uses): New.
11197 (riscv_can_use_return_insn): Return false for interrupt handler.
11198 (riscv_function_ok_for_sibcall): Likewise.
11199 (riscv_set_current_function): Add interrupt handler support.
11200 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
11201 * config/riscv/riscv.md (UNSPECV_MRET): New.
11203 (riscv_frflags, riscv_fsflags): Use tab after opcode.
11205 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
11207 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
11209 PR tree-optimization/85712
11210 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
11211 this candidate has already been replaced in-situ by a copy.
11213 2018-05-25 Jason Merrill <jason@redhat.com>
11215 PR c++/80485 - inline function non-zero address.
11216 * symtab.c (nonzero_address): Check DECL_COMDAT.
11218 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
11221 * config/alpha/alpha.md (ashlsi3): New insn pattern.
11222 (*ashlsi_se): Rename from *ashldi_se. Define as sign
11223 extension of SImode operation. Use const123_operand predicate.
11224 (*saddsi_1): Remove.
11225 (*saddl_se_1): Ditto.
11226 (*ssubsi_1): Ditto.
11227 (*ssubl_se_1): Ditto.
11228 * config/alpha/predicates.md (const123_operand): New predicate.
11229 * config/alpha/constraints.md (P): Use IN_RANGE.
11231 2018-05-25 Richard Biener <rguenther@suse.de>
11233 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
11235 (ref_maybe_used_by_stmt_p): Likewise.
11236 (stmt_may_clobber_ref_p): Likewise.
11237 (stmt_may_clobber_ref_p_1): Likewise.
11238 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
11240 (ref_maybe_used_by_stmt_p): Likewise.
11241 (stmt_may_clobber_ref_p): Likewise.
11242 (stmt_may_clobber_ref_p_1): Likewise.
11243 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
11244 the alias oracle to disambiguate DRs with stmts DR analysis
11246 (vect_analyze_data_refs): Do not give up on not analyzable
11247 DRs for BB vectorization. Remove code truncating the dataref
11250 2018-05-25 Jakub Jelinek <jakub@redhat.com>
11253 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
11254 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
11255 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
11257 2018-05-25 Richard Biener <rguenther@suse.de>
11259 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
11260 function, combining stmt data ref gathering and fatal analysis
11262 (vect_analyze_data_refs): Remove now redudnant code and simplify.
11263 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
11264 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
11265 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
11266 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
11268 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11270 PR tree-optimization/85720
11271 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
11272 SCC if all partitions are builtins.
11273 (version_loop_by_alias_check): New parameter. Generate cancelable
11274 runtime alias check if all partitions are builtins.
11275 (distribute_loop): Update call to above function.
11277 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11279 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
11280 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
11281 (parm_default_def_partition_arg): Ditto.
11282 (set_parm_default_def_partition): Ditto.
11283 (get_parm_default_def_partitions): Ditto and make it static.
11284 (get_undefined_value_partitions): Ditto and make it static.
11285 (remove_ssa_form): Refactor call to init_var_map here.
11286 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
11287 computation for loop region.
11288 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
11289 (register_default_def): Delete.
11290 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
11291 (parm_default_def_partition_arg): Ditto.
11292 (set_parm_default_def_partition): Ditto.
11293 (get_parm_default_def_partitions): Ditto and make it static.
11294 (get_undefined_value_partitions): Ditto and make it static.
11295 (coalesce_with_default, coalesce_with_default): Update comment.
11296 (create_coalesce_list_for_region): New func factored out from
11297 create_outofssa_var_map.
11298 (populate_coalesce_list_for_outofssa): New func factored out from
11299 create_outofssa_var_map and coalesce_ssa_name.
11300 (create_outofssa_var_map): Delete.
11301 (coalesce_ssa_name): Refactor to support live range computation.
11302 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
11303 (get_parm_default_def_partitions): Delete.
11304 (get_undefined_value_partitions): Ditto.
11305 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
11306 computation for loop region.
11307 (new_tree_live_info, loe_visit_block): Ditto.
11308 (live_worklist, set_var_live_on_entry): Ditto.
11309 (calculate_live_on_exit, verify_live_on_entry): Ditto.
11310 * tree-ssa-live.h (struct _var_map): New fields.
11311 (init_var_map): Change decl.
11312 (region_contains_p): New.
11314 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11316 * tree-ssa-live.h (live_merge_and_clear): Delete.
11318 2018-05-25 Richard Biener <rguenther@suse.de>
11321 * tree-dump.c (dequeue_and_dump): Remove access to removed
11322 operand 2 of a SWITCH_EXPR.
11324 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11326 * doc/sourcebuild.texi (vect_double_cond_arith): Include
11327 multiplication and division.
11328 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
11329 (cond_udiv@var{m}, cond_umod@var{m}): Document.
11330 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
11331 (cond_udiv_optab, cond_umod_optab): New optabs.
11332 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
11333 (IFN_COND_RDIV): New internal functions.
11334 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
11335 TRUNC_MOD_EXPR and RDIV_EXPR.
11336 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
11337 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
11339 (SVE_INT_BINARY): Include mult.
11340 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
11341 (optab, sve_int_op): Handle mult.
11342 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
11344 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
11345 for SVE_INT_BINARY_SD.
11347 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11349 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
11350 (optab, sve_int_op): Handle div and udiv.
11351 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
11352 for SVE_INT_BINARY_SD.
11353 (*<optab><mode>3): New insn for the same.
11355 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11357 * tree-vect-patterns.c: Include predict.h.
11358 (vect_recog_divmod_pattern): Restrict check for division support
11359 to when optimizing for size.
11361 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11363 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
11364 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
11365 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
11366 (gimple_match_op::set_op): Likewise.
11367 (gimple_resimplify4): Declare.
11368 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
11369 (expr::gen_transform): Likewise.
11370 (decision_tree::gen): Generate a simplification routine for 4 operands.
11371 * gimple-match-head.c (gimple_simplify): Add an overload for
11372 4 operands. In the top-level function, handle up to 4 call
11373 arguments and call gimple_resimplify4.
11374 (gimple_resimplify4): New function.
11375 (build_call_internal): Pass a fourth operand.
11376 (maybe_push_to_seq): Likewise.
11377 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
11378 Fold VEC_COND_EXPRs of an operation and a default value into
11379 an IFN_COND_* function if possible.
11380 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
11382 (SVE_COND_FP_BINARY): Include them.
11383 (optab, sve_fp_op): Handle them.
11384 (SVE_INT_BINARY_REV): New code iterator.
11385 (SVE_COND_FP_BINARY_REV): New int iterator.
11386 (commutative): New int attribute.
11387 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
11389 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
11391 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
11392 (*cond_<optab><mode>): New patterns for reversed operands.
11394 2018-05-25 Richard Biener <rguenther@suse.de>
11396 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
11397 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
11398 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
11399 (STMT_VINFO_GROUPED_ACCESS): Adjust.
11400 * tree-vect-data-refs.c (everywhere): Adjust users.
11401 * tree-vect-loop.c (everywhere): Likewise.
11402 * tree-vect-slp.c (everywhere): Likewise.
11403 * tree-vect-stmts.c (everywhere): Likewise.
11404 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
11406 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11408 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
11410 (gcc_cv_as_section_exclude): ... this.
11411 Try Solaris as #exclude syntax.
11412 * configure: Regenerate.
11413 * config.in: Regenerate.
11414 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
11416 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
11417 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
11419 * varasm.c (default_elf_asm_named_section): Don't check if
11420 HAVE_GAS_SECTION_EXCLUDE is defined.
11422 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11424 * doc/md.texi: Update the documentation of the cond_* optabs
11425 to mention the new final operand. Fix GET_MODE_NUNITS call.
11426 Describe the scalar case too.
11427 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
11428 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
11430 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
11431 (get_conditional_internal_fn): Update comment.
11432 * tree-vect-loop.c (vectorizable_reduction): Pass the original
11433 accumulator value as a final argument to conditional functions.
11434 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
11435 a define_expand and add an "else" operand. Assert for now that
11436 the else operand is equal to operand 2. Use SVE_INT_BINARY and
11437 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
11438 (*cond_<optab><mode>): New patterns.
11439 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
11440 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
11441 (UNSPEC_COND_EOR): Delete.
11442 (optab): Remove associated mappings.
11443 (SVE_INT_BINARY): New code iterator.
11444 (sve_int_op): Remove int attribute and add "minus" to the code
11446 (SVE_COND_INT_OP): Delete.
11447 (SVE_COND_FP_OP): Rename to...
11448 (SVE_COND_FP_BINARY): ...this.
11450 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11452 * optabs.c (can_reuse_operands_p): New function.
11453 (maybe_legitimize_operands): Try to reuse the results for
11456 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
11458 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
11459 Add {q} suffix to insn mnemonic.
11461 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
11463 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
11464 (msp430_warn_func_return): New.
11466 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
11468 * fold-const.c (tree_nonzero_bits): New function.
11469 * fold-const.h (tree_nonzero_bits): Likewise.
11470 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
11471 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
11473 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
11477 * varasm.c (assemble_alias): Check ifunc_resolver only on
11480 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
11483 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
11484 when memory input operand is handled.
11486 2018-05-24 Luis Machado <luis.machado@linaro.org>
11488 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
11490 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
11492 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
11494 * match.pd: Delay FMA folds until after vectorization.
11496 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
11499 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
11500 address check not strict.
11502 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
11504 * gimple-match.h (gimple_match_op): New class.
11505 (mprts_hook): Replace parameters with a gimple_match_op *.
11506 (maybe_build_generic_op): Likewise.
11507 (gimple_simplified_result_is_gimple_val): Replace parameters with
11508 a const gimple_match_op *.
11509 (gimple_simplify): Replace code_helper * and tree * parameters with
11510 a gimple_match_op * parameter.
11511 (gimple_resimplify1): Replace code_helper *, tree and tree *
11512 parameters with a gimple_match_op * parameter.
11513 (gimple_resimplify2): Likewise.
11514 (gimple_resimplify3): Likewise.
11515 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
11516 parameters with a gimple_match_op * parameter.
11517 * gimple-match-head.c (gimple_simplify): Change prototypes of
11518 auto-generated functions to take a gimple_match_op * instead of
11519 separate code_helper * and tree * parameters. Make the same
11520 change in the top-level overload and update calls to the
11521 gimple_resimplify routines. Update calls to the auto-generated
11522 functions and to maybe_push_res_to_seq in the publicly-facing
11523 operation-specific gimple_simplify overloads.
11524 (gimple_match_op::MAX_NUM_OPS): Define.
11525 (gimple_resimplify1): Replace rcode and ops with a single res_op
11526 parameter. Update call to gimple_simplify.
11527 (gimple_resimplify2): Likewise.
11528 (gimple_resimplify3): Likewise.
11529 (mprts_hook): Replace parameters with a gimple_match_op *.
11530 (maybe_build_generic_op): Likewise.
11531 (build_call_internal): Replace type, nargs and ops with
11532 a gimple_match_op *.
11533 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
11534 with a single gimple_match_op *. Update calls to mprts_hook,
11535 build_call_internal and gimple_simplified_result_is_gimple_val.
11536 Factor out code that is common to the tree_code and combined_fn cases.
11537 * genmatch.c (expr::gen_transform): Replace tem_code and
11538 tem_ops with a gimple_match_op called tem_op. Update calls
11539 to the gimple_resimplify functions and maybe_push_res_to_seq.
11540 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
11541 res_ops. Update call to the gimple_resimplify functions.
11542 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
11543 (decision_tree::gen): Make the functions take a gimple_match_op *
11544 called res_op instead of separate res_code and res_ops parameters.
11545 Update call accordingly.
11546 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
11547 and ops with a single res_op parameter. Update calls to
11548 maybe_build_generic_op and maybe_push_res_to_seq.
11549 (fold_stmt_1): Update calls to gimple_simplify and
11550 replace_stmt_with_simplification.
11551 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
11552 and gimple_simplified_result_is_gimple_val.
11553 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
11555 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
11556 with a gimple_match_op *.
11557 (vn_nary_build_or_lookup): Likewise. Update call to
11558 vn_nary_build_or_lookup_1.
11559 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
11560 gimple_match_op *. Update calls to the gimple_resimplify routines
11561 and to gimple_simplified_result_is_gimple_val.
11562 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
11563 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
11564 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
11565 (visit_nary_op): Likewise.
11566 (visit_reference_op_load): Likewise.
11568 2018-05-23 Luis Machado <luis.machado@linaro.org>
11570 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
11571 modifier for printing the step amount.
11573 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11576 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
11579 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
11581 * doc/sourcebuild.texi (Endianness): New subsubsection.
11583 2018-05-23 Luis Machado <luis.machado@linaro.org>
11585 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
11586 <prefetch_dynamic_strides>: New const bool field.
11587 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
11588 prefetch_dynamic_strides.
11589 (exynosm1_prefetch_tune): Likewise.
11590 (thunderxt88_prefetch_tune): Likewise.
11591 (thunderx_prefetch_tune): Likewise.
11592 (thunderx2t99_prefetch_tune): Likewise.
11593 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
11595 (aarch64_override_options_internal): Update to set
11596 PARAM_PREFETCH_DYNAMIC_STRIDES.
11597 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
11598 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
11599 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
11600 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
11601 prefetch-dynamic-strides setting.
11603 2018-05-23 Luis Machado <luis.machado@linaro.org>
11605 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
11606 <minimum_stride>: New const int field.
11607 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
11608 minimum_stride field defaulting to -1.
11609 (exynosm1_prefetch_tune): Likewise.
11610 (thunderxt88_prefetch_tune): Likewise.
11611 (thunderx_prefetch_tune): Likewise.
11612 (thunderx2t99_prefetch_tune): Likewise.
11613 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
11614 <default_opt_level>: Set to 3.
11615 (aarch64_override_options_internal): Update to set
11616 PARAM_PREFETCH_MINIMUM_STRIDE.
11617 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
11618 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
11619 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
11620 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
11621 stride is constant and is below the minimum stride threshold.
11623 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11625 * config/arm/arm-cpus.in (mode26): Delete.
11626 (armv4): Delete mode26 reference.
11627 * config/arm/arm.c (arm_configure_build_target): Delete use of
11630 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
11632 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
11634 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
11635 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
11637 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
11638 (floatunsdidf2): Ditto.
11640 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
11642 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
11643 (fixuns_trunc<mode>si2_avx512f): Ditto.
11644 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
11645 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
11646 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
11648 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
11650 PR rtl-optimization/79985
11651 * df-scan.c (df_insn_refs_collect): Remove special case for
11652 global registers and asm statements.
11654 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
11656 * extend.texi (Global Register Variables): Rewrite the bullet list.
11657 Note that the register is available for allocation. Note that access
11658 via inline asm must use constraints. Add note about async-signal
11659 handlers. Remove paragraph about automagic register selection.
11661 2018-05-23 Richard Biener <rguenther@suse.de>
11663 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
11664 of fixed offset from memset VN.
11666 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
11668 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
11669 first_interp field.
11670 (alloc_cand_and_find_basis): Initialize first_interp field.
11671 (slsr_process_mul): Modify first_interp field.
11672 (slsr_process_add): Likewise.
11673 (slsr_process_cast): Modify first_interp field for each new
11675 (slsr_process_copy): Likewise.
11676 (dump_candidate): Dump first_interp field.
11677 (replace_mult_candidate): Process all interpretations, not just
11679 (replace_rhs_if_not_dup): Likewise.
11680 (replace_one_candidate): Likewise.
11682 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
11684 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
11686 (aarch64_needs_frame_chain): New function.
11687 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
11689 2018-05-23 Sudakshina Das <sudi.das@arm.com>
11692 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
11693 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
11694 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
11695 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
11696 as true for strict-align.
11697 (aarch64_can_inline_p): Perform checks even when callee has no
11698 attributes to check for strict alignment.
11699 * doc/extend.texi (AArch64 Function Attributes): Document
11701 * doc/invoke.texi: (AArch64 Options): Likewise.
11703 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
11705 PR tree-optimization/85853
11706 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
11707 the handling of the root of the node to...
11708 (vect_slp_analyze_node_operations_1): ...this new function,
11709 and run the whole thing with the child nodes' def types
11710 set according to their SLP node's def type.
11712 2018-05-23 Richard Biener <rguenther@suse.de>
11714 PR middle-end/85874
11715 * tree-data-ref.c (create_runtime_alias_checks): Defer
11716 and ignore overflow warnings.
11718 2018-05-23 Yury Gribov <tetra2005@gmail.com>
11720 PR tree-optimization/85822
11721 * tree-vrp.c (is_masked_range_test): Fix handling of negative
11724 2018-05-23 Richard Biener <rguenther@suse.de>
11726 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
11727 memset constants via native_interpret_expr.
11729 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
11732 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
11734 (cgraph_node::create_alias): Likewise.
11735 (cgraph_node::get_availability): Check ifunc_resolver instead
11736 of looking up ifunc attribute.
11737 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
11738 * varasm.c (do_assemble_alias): Likewise.
11739 (assemble_alias): Likewise.
11740 (default_binds_local_p_3): Likewise.
11741 * cgraph.h (cgraph_node): Add ifunc_resolver.
11742 (cgraph_node::only_called_directly_or_aliased_p): Return false
11743 for IFUNC resolver.
11744 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
11746 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
11747 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
11748 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
11749 instead of looking up ifunc attribute.
11751 2018-05-22 Luis Machado <luis.machado@linaro.org>
11753 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
11755 2018-05-22 Martin Sebor <msebor@redhat.com>
11757 PR middle-end/85359
11758 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
11759 only when expasion succeeds.
11760 (expand_builtin_strcmp): Same.
11761 (expand_builtin_strncmp): Same.
11763 2018-05-22 Martin Sebor <msebor@redhat.com>
11765 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
11767 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
11768 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11770 * config/aarch64/aarch64-ldpstp.md: Replace uses of
11771 aarch64_mem_pair_operand with memory_operand and delete operand swapping
11773 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
11774 Add check for legitimate_address.
11775 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
11776 (aarch64_swap_ldrstr_operands): New.
11777 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
11780 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
11781 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11783 * config/aarch64/aarch64.md: New patterns to generate stp
11785 (store_pair_sw, store_pair_dw): New patterns to generate stp for
11786 single words and double words.
11787 (load_pair_sw, load_pair_dw): Likewise.
11788 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
11790 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
11792 * config/aarch64/aarch64-ldpstp.md: Modify peephole
11793 for different mode ldpstp and add peephole for merged zero stores.
11794 Likewise for loads.
11795 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
11797 (aarch64_gen_store_pair): Rename calls to match new patterns.
11798 (aarch64_gen_load_pair): Rename calls to match new patterns.
11799 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
11800 (load_pair<DREG:mode><DREG2:mode>): ... This.
11801 (store_pair<mode>): Rename to...
11802 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
11803 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
11804 New mode iterators.
11805 (V_INT_EQUIV): Handle SImode.
11806 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
11809 2018-05-22 Martin Sebor <msebor@redhat.com>
11812 * calls.c (maybe_warn_nonstring_arg): Use string length to set
11813 or ajust the presumed bound on an operation to avoid unnecessary
11816 2018-05-22 Martin Sebor <msebor@redhat.com>
11818 PR tree-optimization/85826
11819 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
11820 assuming that a DECL necesarily has a constant size.
11822 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
11824 PR middle-end/85862
11825 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
11827 2018-05-22 Richard Biener <rguenther@suse.de>
11829 PR tree-optimization/85834
11830 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
11831 non-constant and non-zero memset arguments.
11833 2018-05-22 Martin Liska <mliska@suse.cz>
11836 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
11838 2018-05-22 Richard Biener <rguenther@suse.de>
11840 PR tree-optimization/85863
11841 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
11842 comparisons when vectype is specified.
11843 (vectorizable_condition): Do not specify vectype for
11844 vect_is_simple_cond when SLP vectorizing.
11846 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
11849 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
11850 define __ibm128 as long double.
11851 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
11852 as a distinct type when IEEE 128-bit support is enabled.
11853 (init_float128_ieee): Fix up conversions between IFmode and IEEE
11854 128-bit types to use the correct functions.
11855 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
11856 convert between 128-bit floating point types that have different
11857 modes but the same representation, instead of using gen_lowpart to
11859 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
11861 (IFKF_reg): New attributes to give the register constraints for
11863 (extend<mode>tf2_internal): New insns to mark an explicit
11864 conversion between 128-bit floating point types that have a
11865 different mode but share the same representation.
11867 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
11869 PR tree-optimization/85814
11870 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
11871 a null return from get_strinfo when unsharing the next
11872 strinfo in the chain.
11874 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
11877 * varasm.c (weak_finish): Clean up weak_decls.
11879 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11881 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
11882 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
11883 UNSPEC_UADALP values.
11884 * config/aarch64/iterators.md (ABAL): New int iterator.
11887 (sur): Add mappings for the above.
11888 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
11890 (aarch64_<sur>abal<mode>_4): Likewise.
11891 (aarch64_<sur>adalp<mode>_3): Likewise.
11892 (<sur>sadv16qi): New define_expand.
11894 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
11896 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
11897 (*movdf_internal): Ditto.
11898 (*rcpsf2_sse): Ditto.
11899 (*rsqrtsf2_sse): Ditto.
11900 (*sqrt<mode>2_sse): Ditto.
11902 2018-05-21 Tamar Christina <tamar.christina@arm.com>
11904 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
11906 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
11907 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
11908 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
11909 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
11911 * config/aarch64/arm_neon.h: Likewise.
11912 * config/aarch64/iterators.md (VQ_I): New.
11914 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
11916 * config.gcc: Add arc/t-multilib-linux to tmake_file for
11918 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
11921 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
11923 * config/nds32/constraints.md (S): New constraint.
11924 * config/nds32/nds32.md (call_internal): Use constraint S.
11925 (call_value_internal): Likewise.
11926 (sibcall_internal): Likewise.
11927 (sibcall_value_internal): Likewise.
11929 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
11930 Chung-Ju Wu <jasonwucj@gmail.com>
11932 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
11933 into consideration.
11935 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
11936 Chung-Ju Wu <jasonwucj@gmail.com>
11938 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
11939 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
11940 (nds32_rtx_costs_impl): Simplify.
11941 (nds32_address_cost_impl): Simplify.
11942 (nds32_init_rtx_costs): New function.
11943 (nds32_rtx_costs_speed_prefer): Likewise.
11944 (nds32_rtx_costs_size_prefer): Likewise.
11945 (nds32_address_cost_speed_prefer): Likewise.
11946 (nds32_address_cost_speed_fwprop): Likewise.
11947 (nds32_address_cost_size_prefer): Likewise.
11948 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
11949 * config/nds32/nds32.c (nds32_option_override): Use
11950 nds32_init_rtx_costs function.
11952 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
11954 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
11955 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
11956 (TARGET_PIPELINE_N8): Likewise.
11957 (TARGET_PIPELINE_N10): Likewise.
11958 (TARGET_PIPELINE_N13): Likewise.
11959 (TARGET_PIPELINE_GRAYWOLF): Likewise.
11961 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
11963 * config/nds32/nds32-fpu.md: Update copyright year.
11965 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
11967 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
11969 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
11971 * config/nds32/nds32.c
11972 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
11973 * config/nds32/nds32.opt (minline-asm-r15): New option.
11975 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
11977 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
11979 * config/nds32/nds32.md (abssi2): New pattern.
11981 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
11983 * config/i386/i386.md (rex64namesuffix): New mode attribute.
11984 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
11985 Merge insn pattern from sse_cvtsi2ss<round_name> and
11986 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
11987 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
11988 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
11989 using SWI48 mode iterator.
11990 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
11991 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
11992 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
11993 pattern from sse_cvttss2si<round_saeonly_name>
11994 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
11995 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
11996 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
11997 using SWI48 mode iterator.
11998 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
11999 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
12000 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
12001 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
12002 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
12003 using SWI48 mode iterator.
12004 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
12005 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
12006 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
12007 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
12008 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
12009 SWI48 mode iterator.
12010 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
12011 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
12012 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
12013 pattern from sse_cvttsd2si<round_saeonly_name>
12014 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
12016 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12018 * config/nds32/nds32-md-auxiliary.c
12019 (nds32_valid_smw_lwm_base_p): Refine.
12020 (nds32_output_smw_single_word): Refine.
12021 (nds32_output_smw_double_word): New.
12022 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
12024 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12026 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
12027 (nds32_output_stack_pop): Refine.
12028 (nds32_expand_unaligned_load): Refine.
12029 (nds32_expand_unaligned_store): Refine.
12031 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
12032 Chung-Ju Wu <jasonwucj@gmail.com>
12034 * config/nds32/constants.md: Add TP_REGNUM constant.
12035 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
12036 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
12038 * config/nds32/nds32-doubleword.md: Consider flag_pic.
12039 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
12040 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
12041 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
12042 and PIC code generation.
12043 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
12045 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
12047 * config/nds32/nds32.md: Support TLS and PIC.
12048 * config/nds32/nds32.c: Support TLS and PIC.
12049 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
12050 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
12053 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12055 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
12056 mode with E_ prefix.
12058 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
12059 Chung-Ju Wu <jasonwucj@gmail.com>
12061 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
12062 * config/nds32/nds32-md-auxiliary.c
12063 (symbolic_reference_mentioned_p): New.
12064 (nds32_legitimize_ict_address): New.
12065 (nds32_expand_ict_move): New.
12066 (nds32_indirect_call_referenced_p): New.
12067 (nds32_symbol_binds_local_p): Delete.
12068 (nds32_long_call_p): Modify.
12069 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
12070 * config/nds32/nds32-protos.h
12071 (symbolic_reference_mentioned_p): Declare.
12072 (nds32_legitimize_ict_address): Declare.
12073 (nds32_expand_ict_move): Declare.
12074 (nds32_indirect_call_referenced_p): Declare.
12075 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
12076 (nds32_relax_group): Use nds32_ict_const_p as condition.
12077 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
12078 (nds32_asm_file_start): Output ict_model directive in asm code.
12079 (nds32_legitimate_address_p): Consider indirect call.
12080 (nds32_print_operand): Consider indirect call.
12081 (nds32_print_operand_address): Consider indirect call.
12082 (nds32_insert_attributes): Handle "indirect_call" attribute.
12083 (TARGET_LEGITIMATE_ADDRESS_P): Define.
12084 (TARGET_LEGITIMATE_CONSTANT_P): Define.
12085 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
12086 (TARGET_DELEGITIMIZE_ADDRESS): Define.
12087 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
12088 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
12089 (TARGET_ICT_MODEL_SMALL): Define.
12090 (TARGET_ICT_MODEL_LARGE): Define.
12091 * config/nds32/nds32.md (movsi): Consider ict model.
12092 (call, call_value): Consider ict model.
12093 (sibcall, sibcall_value): Consider ict model.
12094 * config/nds32/nds32.opt (mict-model): New option.
12095 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
12098 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
12099 Monk Chiang <sh.chiang04@gmail.com>
12100 Jim Wilson <jimw@sifive.com>
12102 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
12103 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
12104 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
12105 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
12106 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
12107 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
12108 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
12109 compute save_libcall_adjustment properly.
12110 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
12111 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
12112 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
12113 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
12114 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
12115 (ABI_SPEC): Handle mabi=ilp32e.
12116 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
12117 (RVE): Add RVE mask.
12118 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
12119 <-march>: Add rv32e as an example.
12121 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
12124 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
12125 (intra_create_variable_infos): Handle C++ constructors.
12127 2018-05-18 Martin Liska <mliska@suse.cz>
12129 * passes.def: Remove a redundant pass.
12131 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
12134 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
12136 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12138 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
12140 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
12142 (armv2, armv2a, armv3, armv3m): Delete architectures.
12143 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
12144 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
12145 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
12147 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
12148 (*mulsidi3adddi): Likewise.
12149 (mulsidi3): Likewise.
12150 (*mulsidi3_nov6): Likewise.
12151 (umulsidi3): Likewise.
12152 (umulsidi3_nov6): Likewise.
12153 (umaddsidi4): Likewise.
12154 (*umulsidi3adddi): Likewise.
12155 (smulsi3_highpart): Likewise.
12156 (*smulsi3_highpart_nov6): Likewise.
12157 (umulsi3_highpart): Likewise.
12158 (*umulsi3_highpart_nov6): Likewise.
12159 * config/arm/arm.h (arm_arch3m): Delete.
12160 * config/arm/arm.c (arm_arch3m): Delete.
12161 (arm_option_override_internal): Update armv3-related comment.
12162 (arm_configure_build_target): Delete use of isa_bit_mode32.
12163 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
12164 (arm_rtx_costs_internal): Delete check of arm_arch3m.
12165 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
12166 (mulsa3): Likewise.
12167 (mulusa3): Likewise.
12168 * config/arm/arm-protos.h (arm_arch3m): Delete.
12169 * config/arm/arm-tables.opt: Regenerate.
12170 * config/arm/arm-tune.md: Likewise.
12171 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
12172 deleted architectures.
12174 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12176 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
12177 (armv5t, armv5te): New features.
12178 (ARMv5, ARMv5e): Delete fgroups.
12179 (ARMv5t, ARMv5te): Adjust for above changes.
12180 (ARMv6m): Likewise.
12181 (armv5, armv5e): Delete arches.
12182 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
12184 (*call_reg_arm): Likewise.
12185 (*call_value_reg_armv5): Likewise.
12186 (*call_value_reg_arm): Likewise.
12187 (*call_symbol): Likewise.
12188 (*call_value_symbol): Likewise.
12189 (*sibcall_insn): Likewise.
12190 (*sibcall_value_insn): Likewise.
12191 (clzsi2): Likewise.
12192 (prefetch): Likewise.
12193 (define_split and define_peephole2 dependent on arm_arch5):
12195 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
12197 (TARGET_ARM_QBIT): Likewise.
12198 (TARGET_DSP_MULTIPLY): Likewise.
12199 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
12200 (arm_arch5, arm_arch5e): Delete.
12201 (arm_arch5t, arm_arch5te): Declare.
12202 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
12203 (arm_arch5t): Declare.
12204 (arm_option_reconfigure_globals): Update for the above.
12205 (arm_options_perform_arch_sanity_checks): Update comment, replace
12206 use of arm_arch5 with arm_arch5t.
12207 (use_return_insn): Likewise.
12208 (arm_emit_call_insn): Likewise.
12209 (output_return_instruction): Likewise.
12210 (arm_final_prescan_insn): Likewise.
12211 (arm_coproc_builtin_available): Likewise.
12212 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
12213 arm_arch5e with arm_arch5t and arm_arch5te.
12214 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
12215 (arm_arch5t, arm_arch5te): Declare.
12216 * config/arm/arm-tables.opt: Regenerate.
12217 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
12218 * config/arm/t-multilib: Likewise.
12219 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
12220 instead of arm_arch5.
12221 (*call_reg_thumb1): Likewise.
12222 (*call_value_reg_thumb1_v5): Likewise.
12223 (*call_value_reg_thumb1): Likewise.
12224 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
12226 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
12228 2018-05-18 Martin Liska <mliska@suse.cz>
12230 PR gcov-profile/84846
12231 * doc/gcov.texi: Document -t option of gcov tool.
12233 2018-05-18 Martin Liska <mliska@suse.cz>
12235 PR gcov-profile/84846
12236 * gcov.c (print_usage): Add new -t option.
12237 (process_args): Handle the option.
12238 (generate_results): Use stdout as output when requested by
12241 2018-05-18 Martin Liska <mliska@suse.cz>
12243 PR gcov-profile/84846
12244 * coverage.c (coverage_init): Write PWD to .gcno file.
12245 * doc/gcov.texi: Document how working directory is printed.
12246 * gcov-dump.c (dump_gcov_file): Print PWD.
12247 * gcov.c (output_intermediate_file): Likewise.
12248 (read_graph_file): Read PWD string.
12249 (output_lines): Print PWD.
12251 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12253 PR middle-end/85817
12254 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
12255 for retval and return false if all args to phi are zero.
12257 2018-05-18 Richard Biener <rguenther@suse.de>
12259 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
12261 (evrp_dom_walker::before_dom_children): Call it.
12263 2018-05-18 Richard Biener <rguenther@suse.de>
12265 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
12266 results when processing array refs with variable index.
12268 2018-05-18 Toon Moene <toon@moene.org>
12270 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
12271 directly after that of -floop-interchange. Indicate that both
12272 options are enabled by default when specifying -O3.
12274 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12276 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
12277 iterator. Delete separate integer-mode vec_set<mode> expander.
12278 (aarch64_simd_vec_setv2di): Delete.
12279 (vec_setv2di): Delete.
12280 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
12281 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
12282 the "w, r" alternative.
12284 2018-05-18 Martin Liska <mliska@suse.cz>
12286 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
12287 * tree-pass.h (make_pass_lower_switch_O0): New function.
12288 * tree-switch-conversion.c (node_has_low_bound): Remove.
12289 (node_has_high_bound): Likewise.
12290 (node_is_bounded): Likewise.
12291 (class pass_lower_switch): Make it a template type and create
12293 (pass_lower_switch::execute): Add template argument.
12294 (make_pass_lower_switch): New function.
12295 (make_pass_lower_switch_O0): New function.
12296 (do_jump_if_equal): Remove.
12297 (emit_case_nodes): Simplify to just handle all 3 cases and leave
12298 all the hard work to tree optimization passes.
12300 2018-05-18 Martin Liska <mliska@suse.cz>
12302 * dbgcnt.c (limit_low): Renamed from limit.
12303 (limit_high): New variable.
12304 (dbg_cnt_is_enabled): Check for upper limit.
12305 (dbg_cnt): Adjust dumping.
12306 (dbg_cnt_set_limit_by_index): Add new argument for high
12308 (dbg_cnt_set_limit_by_name): Likewise.
12309 (dbg_cnt_process_single_pair): Parse new format.
12310 (dbg_cnt_process_opt): Use strtok.
12311 (dbg_cnt_list_all_counters): Remove 'value' and add
12313 * doc/invoke.texi: Document changes.
12315 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
12317 * doc/sourcebuild.texi (scalar_all_fma): Document.
12318 * tree.def (FMA_EXPR): Delete.
12319 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
12320 * internal-fn.c (ternary_direct): New macro.
12321 (expand_ternary_optab_fn): Likewise.
12322 (direct_ternary_optab_supported_p): Likewise.
12323 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
12324 * builtins.c (fold_builtin_fma): Delete.
12325 (fold_builtin_3): Don't call it.
12326 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
12327 * expr.c (expand_expr_real_2): Likewise.
12328 * fold-const.c (operand_equal_p): Likewise.
12329 (fold_ternary_loc): Likewise.
12330 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
12331 * gimple.c (DEFTREECODE): Likewise.
12332 * gimplify.c (gimplify_expr): Likewise.
12333 * optabs-tree.c (optab_for_tree_code): Likewise.
12334 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12335 * tree-eh.c (operation_could_trap_p): Likewise.
12336 (stmt_could_throw_1_p): Likewise.
12337 * tree-inline.c (estimate_operator_cost): Likewise.
12338 * tree-pretty-print.c (dump_generic_node): Likewise.
12339 (op_code_prio): Likewise.
12340 * tree-ssa-loop-im.c (stmt_cost): Likewise.
12341 * tree-ssa-operands.c (get_expr_operands): Likewise.
12342 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
12343 * fold-const-call.h (fold_fma): Delete.
12344 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
12345 CFN_FNMA and CFN_FNMS.
12346 (fold_fma): Delete.
12347 * genmatch.c (combined_fn): New enum.
12348 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
12349 (commutative_op): New function.
12350 (commutate): Use it. Handle more than 2 operands.
12351 (dt_operand::gen_gimple_expr): Use commutative_op.
12352 (parser::parse_expr): Allow :c to be used with non-binary
12353 operators if the commutative operand is known.
12354 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
12355 CFN_FMS, CFN_FNMA and CFN_FNMS.
12356 (backprop::process_assign_use): Remove FMA_EXPR handling.
12357 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
12358 (gen_hsa_fma): New function.
12359 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
12360 IFN_FNMA and IFN_FNMS.
12361 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
12362 * gimple-fold.h (follow_all_ssa_edges): Declare.
12363 * gimple-fold.c (follow_all_ssa_edges): New function.
12364 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
12365 gimple_build interface and use follow_all_ssa_edges to fold the result.
12366 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
12367 instead of checking for optabs directly.
12368 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
12369 rather than FMA_EXPRs.
12370 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
12371 call to IFN_FMA instead of an FMA_EXPR.
12373 2018-05-17 Jim Wilson <jimw@sifive.com>
12375 * expr.c (do_tablejump): When converting index to Pmode, if we have a
12376 sign extended promoted subreg, and the range does not have the sign bit
12377 set, then do a sign extend.
12379 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
12380 test, check for sign extended subreg and/or constant operands, and
12381 do a sign extend in that case.
12383 2018-05-17 Steve Ellcey <sellcey@cavium.com>
12385 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
12386 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
12388 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
12389 Change logics_shift_reg to logics_shift_imm.
12390 (thunderx2t99_fp_loadpair_basic): Delete.
12391 (thunderx2t99_fp_storepair_basic): Delete.
12392 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
12393 (thunderx2t99_asimd_polynomial): Delete.
12394 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
12395 and neon_fp_mul_d_scalar_q.
12396 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
12397 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
12398 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
12399 (thunderx2t99_asimd_lut): Add missing tbl types.
12400 (thunderx2t99_asimd_ext): Delete.
12401 (thunderx2t99_asimd_load1_1_mult): Delete.
12402 (thunderx2t99_asimd_load1_2_mult): Delete.
12403 (thunderx2t99_asimd_load1_ldp): New.
12404 (thunderx2t99_asimd_load1): New.
12405 (thunderx2t99_asimd_load2): Add missing *load2* types.
12406 (thunderx2t99_asimd_load3): New.
12407 (thunderx2t99_asimd_load4): New.
12408 (thunderx2t99_asimd_store1_1_mult): Delete.
12409 (thunderx2t99_asimd_store1_2_mult): Delete.
12410 (thunderx2t99_asimd_store2_mult): Delete.
12411 (thunderx2t99_asimd_store2_onelane): Delete.
12412 (thunderx2t99_asimd_store_stp): New.
12413 (thunderx2t99_asimd_store1): New.
12414 (thunderx2t99_asimd_store2): New.
12415 (thunderx2t99_asimd_store3): New.
12416 (thunderx2t99_asimd_store4): New.
12418 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
12420 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
12421 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
12423 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
12424 Segher Boessenkool <segher@kernel.crashing.org>
12427 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
12430 2018-05-17 Richard Biener <rguenther@suse.de>
12432 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
12433 for pruning loop and prune defs feeding only already visited PHIs.
12435 2018-05-17 Richard Biener <rguenther@suse.de>
12437 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
12439 2018-05-17 Bin Cheng <bin.cheng@arm.com>
12440 Richard Biener <rguenther@suse.de>
12442 PR tree-optimization/85793
12443 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
12444 for VMAT_ELEMENTWISE.
12446 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
12448 * internal-fn.h (lookup_internal_fn): Declare
12449 * internal-fn.c (lookup_internal_fn): New function.
12450 * gimple.c (gimple_build_call_from_tree): Handle calls to
12451 internal functions.
12452 * gimple-pretty-print.c (dump_gimple_call): Print "." before
12453 internal function names.
12454 * tree-pretty-print.c (dump_generic_node): Likewise.
12455 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
12457 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
12459 * gimple-fold.h (gimple_build): Make the function forms take
12460 combined_fn rather than built_in_function.
12461 (gimple_simplify): Likewise.
12462 * gimple-match-head.c (gimple_simplify): Likewise.
12463 * gimple-fold.c (gimple_build): Likewise.
12464 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
12465 rather than gimple_build_call_internal.
12466 (get_initial_defs_for_reduction): Likewise.
12467 (vect_create_epilog_for_reduction): Likewise.
12468 (vectorizable_live_operation): Likewise.
12470 2018-05-17 Martin Liska <mliska@suse.cz>
12472 * gimple-ssa-sprintf.c (format_directive): Do not use
12473 space in between 'G_' and '('.
12475 2018-05-17 Jakub Jelinek <jakub@redhat.com>
12478 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
12479 even if the mask is not all ones.
12482 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
12484 (ix86_gimple_fold_builtin): Likewise.
12487 * config/i386/i386.c: Include tree-vector-builder.h.
12488 (ix86_vector_shift_count): New function.
12489 (ix86_fold_builtin): Fold shift builtins by scalar count.
12490 (ix86_gimple_fold_builtin): Likewise.
12492 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
12493 _mm512_setzero): New intrinsics.
12495 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
12496 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12498 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
12499 code generation for cases where splatting a value is not useful.
12500 * simplify-rtx.c (simplify_ternary_operation): Simplify
12501 vec_merge across a vec_duplicate and a paradoxical subreg forming
12502 a vector mode to a vec_concat.
12504 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
12506 * config.gcc: Support "goldmont-plus".
12507 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
12509 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12510 PROCESSOR_GOLDMONT_PLUS.
12511 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
12512 (processor_target_table): Add "goldmont-plus".
12513 (PTA_GOLDMONT_PLUS): Define.
12514 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
12515 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
12516 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
12517 (fold_builtin_cpu): Add "goldmont-plus".
12518 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
12519 (ix86_option_override_internal): Add "goldmont-plus".
12520 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
12521 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
12522 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
12523 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
12525 2018-05-17 Richard Biener <rguenther@suse.de>
12527 PR tree-optimization/85757
12528 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
12529 remove defs that only feed that PHI from further processing.
12531 2018-05-16 Jim Wilson <jimw@sifive.com>
12533 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
12535 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
12537 2018-05-16 Mark Wielaard <mark@klomp.org>
12539 * dwarf2out.c (count_index_strings): New function.
12540 (output_indirect_strings): Call count_index_strings and generate
12541 header for dwarf_version >= 5.
12543 2018-05-16 Mark Wielaard <mark@klomp.org>
12545 * dwarf2out.c (dwarf_FORM): New function.
12546 (set_indirect_string): Use dwarf_FORM.
12547 (reset_indirect_string): Likewise.
12548 (size_of_die): Likewise.
12549 (value_format): Likewise.
12550 (output_die): Likewise.
12551 (add_skeleton_AT_string): Likewise.
12552 (output_macinfo_op): Likewise.
12553 (index_string): Likewise.
12554 (output_index_string_offset): Likewise.
12555 (output_index_string): Likewise.
12556 (count_index_strings): Likewise.
12558 2018-05-16 Carl Love <cel@us.ibm.com>
12560 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
12561 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
12563 2018-05-16 Martin Jambor <mjambor@suse.cz>
12565 * ipa-prop.c (ipa_free_all_edge_args): Remove.
12566 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
12568 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
12570 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
12571 (fnma<mode>4): Likewise.
12572 (fms<mode>4): Likewise.
12573 (fnms<mode>4): Likewise.
12574 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
12575 (aarch64_fnma<mode>4): Likewise.
12576 (aarch64_fms<mode>4): Likewise.
12577 (aarch64_fnms<mode>4): Likewise.
12578 (aarch64_fnmadd<mode>4): Likewise.
12580 2018-05-16 Jason Merrill <jason@redhat.com>
12582 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
12584 2018-05-16 Richard Biener <rguenther@suse.de>
12586 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
12587 (dump_stmt_cost): Declare.
12588 (add_stmt_cost): Dump cost we add.
12589 (add_stmt_costs): New function.
12590 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
12591 No longer exported.
12592 (vect_analyze_stmt): Adjust prototype.
12593 (vectorizable_condition): Likewise.
12594 (vectorizable_live_operation): Likewise.
12595 (vectorizable_reduction): Likewise.
12596 (vectorizable_induction): Likewise.
12597 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
12598 cost vector to pass to vectorizable_ and record afterwards.
12599 (vect_model_reduction_cost): Take cost vector argument and adjust.
12600 (vect_model_induction_cost): Likewise.
12601 (vectorizable_reduction): Likewise.
12602 (vectorizable_induction): Likewise.
12603 (vectorizable_live_operation): Likewise.
12604 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
12605 SLP_TREE_NUMBER_OF_VEC_STMTS.
12606 (vect_analyze_slp_cost_1): Remove.
12607 (vect_analyze_slp_cost): Likewise.
12608 (vect_slp_analyze_node_operations): Take visited args and
12609 a target cost vector. Avoid processing already visited stmt sets.
12610 (vect_slp_analyze_operations): Use a local cost vector to gather
12611 costs and register those of non-discarded instances.
12612 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
12613 (vect_schedule_slp_instance): Remove copying of
12614 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
12616 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
12617 adding cost. Record cost entry location.
12618 (vect_prologue_cost_for_slp_op): Function to compute cost of
12619 a constant or invariant generated for SLP vect in the prologue,
12620 split out from vect_analyze_slp_cost_1.
12621 (vect_model_simple_cost): Make static. Adjust for SLP costing.
12622 (vect_model_promotion_demotion_cost): Likewise.
12623 (vect_model_store_cost): Likewise, make static.
12624 (vect_model_load_cost): Likewise.
12625 (vectorizable_bswap): Add cost vector arg and adjust.
12626 (vectorizable_call): Likewise.
12627 (vectorizable_simd_clone_call): Likewise.
12628 (vectorizable_conversion): Likewise.
12629 (vectorizable_assignment): Likewise.
12630 (vectorizable_shift): Likewise.
12631 (vectorizable_operation): Likewise.
12632 (vectorizable_store): Likewise.
12633 (vectorizable_load): Likewise.
12634 (vectorizable_condition): Likewise.
12635 (vectorizable_comparison): Likewise.
12636 (can_vectorize_live_stmts): Likewise.
12637 (vect_analyze_stmt): Likewise.
12638 (vect_transform_stmt): Adjust calls to vectorizable_*.
12639 * tree-vectorizer.c: Include gimple-pretty-print.h.
12640 (dump_stmt_cost): New function.
12642 2018-05-16 Richard Biener <rguenther@suse.de>
12644 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
12645 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
12646 * tree-ssa-dse.c: Include tree-ssa-loop.h.
12647 (check_name): New callback.
12648 (dse_classify_store): Track cycles via a visited bitmap of PHI
12649 defs and simplify handling of in-loop and across loop dead stores
12650 and properly fail for loop-variant refs. Handle byte-tracking with
12651 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
12654 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
12656 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
12657 (vect_get_mask_type_for_stmt): Likewise.
12658 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
12660 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
12661 to determine the statement's vector type and the vector type that
12662 should be used for calculating nunits. Deal with cases in which
12663 the type has to be deferred.
12664 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
12665 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
12666 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
12667 (vect_determine_vf_for_stmt): New functions, split out from...
12668 (vect_determine_vectorization_factor): ...here.
12669 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
12670 (vect_get_mask_type_for_stmt): New functions, split out from
12671 vect_determine_vectorization_factor.
12673 2018-05-16 Richard Biener <rguenther@suse.de>
12675 * tree-cfg.c (verify_gimple_assign_ternary): Properly
12676 verify the [VEC_]COND_EXPR embedded comparison.
12678 2018-05-15 Martin Sebor <msebor@redhat.com>
12680 PR tree-optimization/85753
12681 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
12682 RECORD_TYPE in addition to ARRAY_TYPE.
12684 2018-05-15 Martin Sebor <msebor@redhat.com>
12686 PR middle-end/85643
12687 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
12689 2018-05-15 Richard Biener <rguenther@suse.de>
12691 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
12692 add by_clobber_p one. Change algorithm to collect all defs
12693 representing uses we need to walk and try reducing them to
12694 a single one before failing.
12695 (dse_dom_walker::dse_optimize_stmt): Adjust.
12697 2018-05-13 Mark Wielaard <mark@klomp.org>
12699 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
12700 (size_of_loc_descr): Likewise.
12701 (output_loc_operands): Likewise.
12702 (output_loc_operands_raw): Likewise.
12703 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
12704 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
12705 (hash_loc_operands): Likewise.
12706 (compare_loc_operands): Likewise.
12708 2018-05-14 Mark Wielaard <mark@klomp.org>
12710 * dwarf2out.c (count_index_addrs): New function.
12711 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
12713 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12715 PR tree-optimization/83648
12716 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
12717 return value as malloc candidate.
12719 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12722 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
12723 param as true in call to suggest_attribute.
12725 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
12727 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
12728 -mreadonly-in-sdata.
12730 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12732 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
12734 (aarch64_crypto_aesd_fused): Likewise.
12736 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
12738 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
12739 (movsi_aarch64): Likewise.
12740 (load_pairsi): Likewise.
12741 (load_pairdi): Likewise.
12742 (store_pairsi): Likewise.
12743 (store_pairdi): Likewise.
12744 (load_pairsf): Likewise.
12745 (load_pairdf): Likewise.
12746 (store_pairsf): Likewise.
12747 (store_pairdf): Likewise.
12748 (zero_extend): Likewise.
12749 (trunc): Swap alternatives.
12750 (fcvt_target): Add '?' to prefer w over r.
12752 2018-05-14 Jakub Jelinek <jakub@redhat.com>
12755 * config/i386/i386.md: Disallow non-commutative arithmetics in
12756 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
12757 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
12758 in the peephole2 before it.
12760 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
12762 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
12763 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
12764 (ix86_handle_option): Handle -mcldemote.
12765 * config.gcc: New header.
12766 * config/i386/cldemoteintrin.h: New file.
12767 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
12768 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
12770 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12771 OPTION_MASK_ISA_CLDEMOTE.
12772 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
12773 (ix86_valid_target_attribute_inner_p): Ditto.
12774 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
12775 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
12776 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
12777 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
12778 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
12780 * config/i386/i386.opt: Add -mcldemote.
12781 * config/i386/x86intrin.h: New header.
12782 * doc/invoke.texi: Add -mcldemote.
12784 2018-05-14 Richard Biener <rguenther@suse.de>
12786 * doc/match-and-simplify.texi: Adjust :s documentation.
12788 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
12790 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
12791 intended memcpy size.
12792 (REORDER_45): Likewise.
12794 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
12796 * sort.cc: New file.
12797 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
12798 * vec.c (qsort_chk): Use gcc_qsort.
12799 * Makefile.in (OBJS-libcommon): Add sort.o.
12800 (build/sort.o): New target. Use it...
12801 (BUILD_RTL): ... here, and...
12802 (build/gencfn-macros): ... here, and...
12803 (build/genmatch): ... here.
12805 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
12806 Chung-Ju Wu <jasonwucj@gmail.com>
12808 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
12809 * config/nds32/nds32-graywolf.md: New file.
12810 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
12811 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
12813 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
12814 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
12815 * config/nds32/nds32.md (pipeline_model): Add graywolf.
12816 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
12817 * config/nds32/pipelines.md: Include n15 settings.
12819 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
12820 Chung-Ju Wu <jasonwucj@gmail.com>
12822 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
12823 * config/nds32/nds32-n13.md: New file.
12824 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
12825 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
12827 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
12828 * config/nds32/nds32.md (pipeline_model): Add n13.
12829 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
12830 * config/nds32/pipelines.md: Include n13 settings.
12832 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
12833 Chung-Ju Wu <jasonwucj@gmail.com>
12835 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
12836 * config/nds32/nds32-n10.md: New file.
12837 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
12838 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
12840 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
12841 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
12842 * config/nds32/nds32.md (pipeline_model): Add n10.
12843 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
12844 * config/nds32/pipelines.md: Include n10 settings.
12846 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
12847 Kito Cheng <kito.cheng@gmail.com>
12848 Chung-Ju Wu <jasonwucj@gmail.com>
12850 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
12851 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
12852 Add enum values for DSP extension instructions.
12853 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
12855 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
12856 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
12857 New code iterators.
12858 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
12859 * config/nds32/nds32-dspext.md: New file for DSP implementation.
12860 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
12861 * config/nds32/nds32-intrinsic.md: Likewise.
12862 * config/nds32/nds32_intrinsic.h: Likewise.
12863 * config/nds32/nds32-md-auxiliary.c: Likewise.
12864 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
12865 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
12866 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
12867 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
12868 * config/nds32/nds32-protos.h: New declarations for DSP extension.
12869 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
12870 TYPE_DMAC in switch statement.
12871 * config/nds32/nds32.c: New checking and implementation for DSP
12872 extension instructions.
12873 * config/nds32/nds32.h: Likewise.
12874 * config/nds32/nds32.md: Likewise.
12875 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
12876 * config/nds32/predicates.md: Implement new predicates for DSP
12879 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
12881 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
12882 Reformat alternatives and attributes so it is easier to identify
12883 which constraints/attributes go with which instruction.
12884 (mov<mode>_hardfloat32, FMOVE64): Likewise.
12885 (mov<mode>_softfloat32, FMOVE64): Likewise.
12886 (mov<mode>_hardfloat64, FMOVE64): Likewise.
12887 (mov<mode>_softfloat64, FMOVE64): Likewise.
12889 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
12891 * doc/extend.texi (PowerPC Built-in Functions): Rename this
12893 (Basic PowerPC Built-in Functions): The new name of the
12894 subsection previously known as "PowerPC Built-in Functions".
12895 (Basic PowerPC Built-in Functions Available on all Configurations):
12897 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
12898 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
12899 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
12900 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
12902 2018-05-11 Martin Jambor <mjambor@suse.cz>
12905 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
12908 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
12911 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
12913 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
12915 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
12916 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
12917 (ix86_handle_option): Handle -mwaitpkg.
12918 * config.gcc: New header.
12919 * config/i386/cpuid.h (bit_WAITPKG): New bit.
12920 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
12921 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
12923 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12924 OPTION_MASK_ISA_WAITPKG.
12925 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
12926 (ix86_option_override_internal): Add PTA_WAITPKG.
12927 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
12928 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
12929 IX86_BUILTIN_TPAUSE.
12930 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
12931 __builtin_ia32_umwait and __builtin_ia32_tpause.
12932 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
12933 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
12934 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
12935 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
12936 UNSPECV_TPAUSE): New.
12937 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
12938 * config/i386/i386.opt: Add -mwaitpkg.
12939 * config/i386/waitpkgintrin.h: New file.
12940 * config/i386/x86intrin.h: New header.
12941 * doc/invoke.texi: Add -mwaitpkg.
12943 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
12946 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
12948 (cortex-m0): Use armv6s-m isa.
12949 (cortex-m0plus): Likewise.
12950 (cortex-m1): Likewise.
12951 (cortex-m0.small-multiply): Likewise.
12952 (cortex-m0plus.small-multiply): Likewise.
12953 (cortex-m1.small-multiply): Likewise.
12955 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
12956 Jakub Jelinek <jakub@redhat.com>
12958 PR tree-optimization/85692
12959 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
12960 source permute as well.
12962 2018-05-11 Martin Liska <mliska@suse.cz>
12965 * doc/extend.texi: Document LLVM style format for no_sanitize
12968 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
12970 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
12971 mode_supports_vsx_dform_quad to mode_supports_dq_form.
12972 (mode_supports_vsx_dform_quad): Likewise.
12973 (mode_supports_vmx_dform): Move these functions to be next to the
12974 other mode_supports functions.
12975 (mode_supports_dq_form): Likewise.
12976 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
12977 mode_supports_dq_form.
12978 (reg_offset_addressing_ok_p): Likewise.
12979 (offsettable_ok_by_alignment): Likewise.
12980 (rs6000_legitimate_offset_address_p): Likewise.
12981 (legitimate_lo_sum_address_p): Likewise.
12982 (rs6000_legitimize_address): Likewise.
12983 (rs6000_legitimize_reload_address): Likewise.
12984 (rs6000_secondary_reload_inner): Likewise.
12985 (rs6000_preferred_reload_class): Likewise.
12986 (rs6000_output_move_128bit): Likewise.
12988 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
12990 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
12991 Generate SImode target register for null target.
12992 <case IX86_BUILTIN_XGETBV>: Ditto.
12993 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
12994 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
12996 2018-05-10 Carl Love <cel@us.ibm.com>
12998 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
12999 dcbtt and dcbtstt if operands[2] is 0.
13001 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
13004 * config/i386/sse.md (usadv64qi): New expander.
13006 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
13008 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
13009 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
13010 -maltivec=be support.
13011 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
13012 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
13013 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13014 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
13015 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
13016 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
13017 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
13018 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
13019 altivec_vsumsws): Adjust.
13020 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
13021 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
13022 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
13024 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
13025 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
13026 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
13027 (altivec_lve<VI_char>x): Delete expand.
13028 (*altivec_lve<VI_char>x_internal): Rename to...
13029 (altivec_lve<VI_char>x): ... this.
13030 (altivec_lvxl_<mode>): Delete expand.
13031 (*altivec_lvxl_<mode>_internal): Rename to ...
13032 (altivec_lvxl_<mode>): ... this.
13033 (altivec_stvxl_<mode>): Delete expand.
13034 (*altivec_stvxl_<mode>_internal): Rename to ...
13035 (altivec_stvxl_<mode>): ... this.
13036 (altivec_stve<VI_char>x): Delete expand.
13037 (*altivec_stve<VI_char>x_internal): Rename to ...
13038 (altivec_stve<VI_char>x): ... this.
13039 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
13040 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
13041 reduc_plus_scal_<mode>): Adjust.
13042 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
13044 (rs6000_cpu_cpp_builtins): Adjust.
13045 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
13046 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
13047 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
13048 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
13049 -maltivec=be support.
13050 (rs6000_split_vec_extract_var): Adjust.
13051 (rs6000_split_v4si_init): Adjust.
13052 (swap_selector_for_mode): Delete.
13053 (altivec_expand_lvx_be, altivec_expand_stvx_be,
13054 altivec_expand_stvex_be): Delete.
13055 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
13056 -maltivec=be support.
13057 (rs6000_gimple_fold_builtin): Ditto.
13058 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
13060 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
13061 (TARGET_DIRECT_MOVE_64BIT): Adjust.
13062 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
13063 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
13064 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
13065 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
13066 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
13067 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
13068 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
13069 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
13070 anonymous split): Adjust.
13071 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
13072 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
13074 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
13076 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
13077 when --with-gxx-include-dir is also specified.
13078 * configure: Regenerate.
13080 2018-05-09 Jim Wilson <jimw@sifive.com>
13083 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
13084 * config/riscv/t-withmultilib: New.
13085 * config/riscv/withmultilib.h: New.
13086 * doc/install.texi: Document RISC-V --with-multilib-list support.
13088 2018-05-09 Richard Biener <rguenther@suse.de>
13090 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
13092 (vect_bb_vectorization_profitable_p): Adjust. Compute
13093 actual scalar cost using the cost vector and the add_stmt_cost
13096 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13098 PR rtl-optimization/85645
13099 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
13100 in the REG_CFA_REGISTER note for LR, don't leave it empty.
13102 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13104 PR rtl-optimization/85645
13105 * shrink-wrap.c (spread_components): Return a boolean saying if
13106 anything was changed.
13107 (try_shrink_wrapping_separate): Iterate spread_components until
13108 nothing changes anymore.
13110 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13112 PR rtl-optimization/85645
13113 * regrename.c (build_def_use): Also kill the chains that include the
13114 destination of a REG_CFA_REGISTER note.
13116 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13118 PR rtl-optimization/85645
13119 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
13120 insn that has a REG_CFA_REGISTER note.
13122 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
13124 * cfgexpand.c (expand_clobber): New function.
13125 (expand_gimple_stmt_1): Use it.
13126 * tree-vect-stmts.c (vect_clobber_variable): New function,
13128 (vectorizable_simd_clone_call): ...here.
13129 (vectorizable_store): Emit a clobber either side of an
13130 IFN_STORE_LANES sequence.
13131 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
13133 2018-05-09 Tom de Vries <tom@codesourcery.com>
13136 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
13137 (define_insn "trap_if_false"): Add exit after trap.
13139 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
13141 PR rtl-optimization/85638
13142 * bb-reorder.c: Include common/common-target.h.
13143 (create_forwarder_block): New function extracted from...
13144 (fix_up_crossing_landing_pad): ...here. Rename into...
13145 (dw2_fix_up_crossing_landing_pad): ...this.
13146 (sjlj_fix_up_crossing_landing_pad): New function.
13147 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
13148 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
13149 from both partitions and exit the loop after one iteration.
13151 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13154 * doc/extend.texi (PowerPC Built-in Functions): Rename this
13156 (Basic PowerPC Built-in Functions): The new name of the
13157 subsection previously known as "PowerPC Built-in Functions".
13158 (Basic PowerPC Built-in Functions Available on all Configurations):
13160 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
13162 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
13164 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
13166 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
13169 2018-05-08 Jim Wilson <jimw@sifive.com>
13171 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
13172 (LD_EMUL_SUFFIX): New.
13173 (LINK_SPEC): Use it.
13175 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13177 * doc/extend.texi (PowerPC Built-in Functions): Rename this
13179 (Basic PowerPC Built-in Functions): The new name of the
13180 subsection previously known as "PowerPC Built-in Functions".
13181 (Basic PowerPC Built-in Functions Available on all Configurations):
13183 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
13185 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
13187 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
13189 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
13192 2018-05-08 Jakub Jelinek <jakub@redhat.com>
13195 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
13196 after cmpelim optimization.
13198 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
13200 * config.gcc: Support "goldmont".
13201 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
13202 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13203 PROCESSOR_GOLDMONT.
13204 * config/i386/i386.c (m_GOLDMONT): Define.
13205 (processor_target_table): Add "goldmont".
13206 (PTA_GOLDMONT): Define.
13207 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
13208 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
13209 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
13210 (fold_builtin_cpu): Add "goldmont".
13211 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
13212 (ix86_option_override_internal): Add "goldmont".
13213 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
13214 (processor_type): Add PROCESSOR_GOLDMONT.
13215 * config/i386/i386.md: Add CPU "glm".
13216 * config/i386/glm.md: New file.
13217 * config/i386/x86-tune.def: Add m_GOLDMONT.
13218 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
13220 2018-05-08 Jakub Jelinek <jakub@redhat.com>
13223 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
13225 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
13226 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
13227 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
13230 * config/i386/i386.c (ix86_fold_builtin): Handle
13231 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
13234 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
13235 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
13237 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
13240 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
13241 (check_arch): Likewise.
13242 (check_fpu): Return the result rather than printing it.
13243 (end arch): Fix operator precedence.
13244 (end cpu): Likewise.
13245 (END): Print the result from check_fpu.
13247 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
13248 Alan Hayward <alan.hayward@arm.com>
13249 David Sherwood <david.sherwood@arm.com>
13251 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
13252 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
13253 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
13254 (*fcmuo<mode>_and): New patterns.
13256 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
13258 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
13259 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
13260 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
13261 (cmp_op, sve_imm_con): New code attributes.
13262 (SVE_COND_INT_CMP, imm_con): Delete.
13263 (cmp_op): Remove above unspecs from int attribute.
13264 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
13266 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
13267 comparison-specific unspecs.
13268 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
13269 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
13270 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
13271 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
13272 (*vec_fcm<cmp_op><mode>): Rename to...
13273 (*fcm<cmp_op><mode>): ...this and adjust likewise.
13274 (*vec_fcmuo<mode>): Rename to...
13275 (*fcmuo<mode>): ...this and adjust likewise.
13276 (*pred_fcm<cmp_op><mode>): New pattern.
13277 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
13278 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
13280 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
13282 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
13283 (aarch64_emit_sve_predicated_cond): New function.
13284 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
13285 (aarch64_emit_unspec_cond_or): Replace with...
13286 (aarch64_emit_sve_or_conds): ...this new function. Use
13287 aarch64_emit_sve_ptrue_op for the individual comparisons and
13288 aarch64_emit_binop to OR them together.
13289 (aarch64_emit_inverted_unspec_cond): Replace with...
13290 (aarch64_emit_sve_inverted_cond): ...this new function. Use
13291 aarch64_emit_sve_ptrue_op for the comparison and
13292 aarch64_emit_unop to invert the result.
13293 (aarch64_expand_sve_vec_cmp_float): Update after the above
13294 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
13296 2018-05-07 Nathan Sidwell <nathan@acm.org>
13298 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
13299 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
13300 (Backwards Compatibility): Likewise.
13302 2018-05-07 Luis Machado <luis.machado@linaro.org>
13306 2018-05-07 Luis Machado <luis.machado@linaro.org>
13308 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13309 <prefetch_dynamic_strides>: New const bool field.
13310 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13311 prefetch_dynamic_strides.
13312 (exynosm1_prefetch_tune): Likewise.
13313 (thunderxt88_prefetch_tune): Likewise.
13314 (thunderx_prefetch_tune): Likewise.
13315 (thunderx2t99_prefetch_tune): Likewise.
13316 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
13318 (aarch64_override_options_internal): Update to set
13319 PARAM_PREFETCH_DYNAMIC_STRIDES.
13320 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
13321 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
13322 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
13323 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
13324 prefetch-dynamic-strides setting.
13326 2018-05-07 Luis Machado <luis.machado@linaro.org>
13328 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13329 <minimum_stride>: New const int field.
13330 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13331 minimum_stride field.
13332 (exynosm1_prefetch_tune): Likewise.
13333 (thunderxt88_prefetch_tune): Likewise.
13334 (thunderx_prefetch_tune): Likewise.
13335 (thunderx2t99_prefetch_tune): Likewise.
13336 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
13337 (aarch64_override_options_internal): Update to set
13338 PARAM_PREFETCH_MINIMUM_STRIDE.
13339 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
13340 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
13341 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
13342 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
13343 stride is constant and is below the minimum stride threshold.
13345 2018-05-07 Luis Machado <luis.machado@linaro.org>
13347 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
13350 2018-05-07 Luis Machado <luis.machado@linaro.org>
13352 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13353 <prefetch_dynamic_strides>: New const bool field.
13354 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13355 prefetch_dynamic_strides.
13356 (exynosm1_prefetch_tune): Likewise.
13357 (thunderxt88_prefetch_tune): Likewise.
13358 (thunderx_prefetch_tune): Likewise.
13359 (thunderx2t99_prefetch_tune): Likewise.
13360 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
13362 (aarch64_override_options_internal): Update to set
13363 PARAM_PREFETCH_DYNAMIC_STRIDES.
13364 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
13365 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
13366 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
13367 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
13368 prefetch-dynamic-strides setting.
13370 2018-05-07 Luis Machado <luis.machado@linaro.org>
13372 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13373 <minimum_stride>: New const int field.
13374 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13375 minimum_stride field.
13376 (exynosm1_prefetch_tune): Likewise.
13377 (thunderxt88_prefetch_tune): Likewise.
13378 (thunderx_prefetch_tune): Likewise.
13379 (thunderx2t99_prefetch_tune): Likewise.
13380 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
13381 (aarch64_override_options_internal): Update to set
13382 PARAM_PREFETCH_MINIMUM_STRIDE.
13383 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
13384 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
13385 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
13386 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
13387 stride is constant and is below the minimum stride threshold.
13389 2018-05-06 Jakub Jelinek <jakub@redhat.com>
13392 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
13393 the type is addressable. Don't force op into register if it has
13396 2018-05-05 Roland McGrath <mcgrathr@google.com>
13399 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
13400 any section for which we don't know a specific type it should have,
13401 regardless of name. Previously this was done only for the exact
13402 names ".init_array", ".fini_array", and ".preinit_array".
13403 (default_elf_asm_named_section): Add comment about
13404 relationship with default_section_type_flags and SECTION_NOTYPE.
13405 (get_section): Don't consider it a type conflict if one side has
13406 SECTION_NOTYPE and the other doesn't, as long as neither has the
13407 SECTION_BSS et al used in the default_section_type_flags logic.
13409 2018-05-05 Tom de Vries <tom@codesourcery.com>
13412 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
13413 (workaround_barsyncs): New function.
13414 (nvptx_reorg): Use workaround_barsyncs.
13415 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
13416 (define_expand "nvptx_membar_cta"): New define_expand.
13417 (define_insn "*nvptx_membar_cta"): New insn.
13419 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
13421 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
13422 To improve optimization opportunities.
13423 * builtin-types.def: The new needed builtin types for the above.
13425 2018-05-04 Richard Biener <rguenther@suse.de>
13427 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
13428 * gimple-ssa-store-merging.c
13429 (imm_store_chain_info::output_merged_store): Remove redundant create,
13430 release split_store vector contents on failure.
13431 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
13432 scalar stmt vector on cache hit.
13434 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
13436 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
13438 * config.gcc (powerpc-xilinx-eabi*): Remove.
13439 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
13441 (fusion_addis_mem_combo_load): Ditto.
13442 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
13444 (rs6000_cpu_cpp_builtins): Ditto.
13445 * config/rs6000/rs6000-linux.c
13446 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
13447 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
13448 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
13450 (rs6000_setup_reg_addr_masks): Ditto.
13451 (rs6000_init_hard_regno_mode_ok): Ditto.
13452 (rs6000_option_override_internal): Ditto.
13453 (legitimate_lo_sum_address_p): Ditto.
13454 (rs6000_legitimize_address): Ditto.
13455 (rs6000_legitimize_reload_address): Ditto.
13456 (rs6000_legitimate_address_p): Ditto.
13457 (abi_v4_pass_in_fpr): Ditto.
13458 (setup_incoming_varargs): Ditto.
13459 (rs6000_gimplify_va_arg): Ditto.
13460 (rs6000_split_multireg_move): Ditto.
13461 (rs6000_savres_strategy): Ditto.
13462 (rs6000_emit_prologue_components): Ditto.
13463 (rs6000_emit_epilogue_components): Ditto.
13464 (rs6000_emit_prologue): Ditto.
13465 (rs6000_emit_epilogue): Ditto.
13466 (rs6000_elf_file_end): Ditto.
13467 (rs6000_function_value): Ditto.
13468 (rs6000_libcall_value): Ditto.
13469 * config/rs6000/rs6000.h: Ditto.
13470 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
13471 (TARGET_MINMAX): ... this. New.
13472 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
13473 * config/rs6000/rs6000.md: Remove Xilinx FP support.
13474 (*movsi_internal1_single): Delete.
13475 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
13476 mfpu=, mxilinx-fpu): Delete.
13477 * config/rs6000/singlefp.h: Delete.
13478 * config/rs6000/sysv4.h: Remove Xilinx FP support.
13479 * config/rs6000/t-rs6000: Ditto.
13480 * config/rs6000/t-xilinx: Delete.
13481 * config/rs6000/titan.md: Adjust for fp_type removal.
13482 * config/rs6000/vsx.md: Remove Xilinx FP support.
13483 (VStype_simple): Delete.
13484 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
13485 * config/rs6000/xfpu.h: Delete.
13486 * config/rs6000/xfpu.md: Delete.
13487 * config/rs6000/xilinx.h: Delete.
13488 * config/rs6000/xilinx.opt: Delete.
13489 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
13490 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
13492 2018-05-04 Tom de Vries <tom@codesourcery.com>
13495 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
13498 2018-05-04 Richard Biener <rguenther@suse.de>
13500 PR middle-end/85627
13501 * tree-complex.c (update_complex_assignment): We are always in SSA form.
13502 (expand_complex_div_wide): Likewise.
13503 (expand_complex_operations_1): Likewise.
13504 (expand_complex_libcall): Preserve EH info of the original stmt.
13505 (tree_lower_complex): Handle removed blocks.
13506 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
13507 on complex multiplication and division libcall builtins.
13509 2018-05-04 Richard Biener <rguenther@suse.de>
13511 PR middle-end/85574
13512 * fold-const.c (negate_expr_p): Restrict negation of operand
13513 zero of a division to when we know that can happen without
13515 (fold_negate_expr_1): Likewise.
13517 2018-05-04 Jakub Jelinek <jakub@redhat.com>
13520 * real.h (real_nextafter): Declare.
13521 * real.c (real_nextafter): New function.
13522 * fold-const-call.c (fold_const_nextafter): New function.
13523 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
13524 CASE_CFN_NEXTTOWARD.
13525 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
13526 even when arg1_mode is different from arg0_mode.
13528 2018-05-03 Nathan Sidwell <nathan@acm.org>
13530 * doc/extend.texi (Deprecated Features): Remove
13531 -ffriend-injection.
13532 (Backwards Compatibility): Likewise.
13533 * doc/invoke.texi (C++ Language Options): Likewise.
13534 (C++ Dialect Options): Likewise.
13536 2018-05-03 Jakub Jelinek <jakub@redhat.com>
13539 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
13540 _mm512_mask_mullox_epi64): New intrinsics.
13542 2018-05-03 Tom de Vries <tom@codesourcery.com>
13545 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
13546 dump files): Add offload-tree.
13548 2018-05-03 Richard Biener <rguenther@suse.de>
13550 PR tree-optimization/85615
13551 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
13552 to loops not nested in BBs loop father to avoid creating multi-entry
13555 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13557 PR tree-optimization/70291
13558 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
13559 arguments. Change return type to tree. Emit libcall as a new
13560 statement rather than replacing existing one when inplace_p is true.
13561 (expand_complex_multiplication_components): New function.
13562 (expand_complex_multiplication): Expand floating-point complex
13563 multiplication using the above.
13564 (expand_complex_division): Rename inner_type parameter to type.
13565 Update expand_complex_libcall call-site.
13566 (expand_complex_operations_1): Update expand_complex_multiplication
13567 and expand_complex_division call-sites.
13569 2018-05-02 Jakub Jelinek <jakub@redhat.com>
13572 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
13573 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
13574 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
13575 the highest significant bit of the shift count mask is clear. In
13576 check whether and[sq]i3 is needed verify that all significant bits
13577 of the shift count other than the highest are set.
13579 2018-05-02 Tom de Vries <tom@codesourcery.com>
13582 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
13583 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
13584 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
13585 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
13586 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
13587 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
13588 __builtin_goacc_parlevel_size.
13590 2018-05-02 Richard Biener <rguenther@suse.de>
13592 PR tree-optimization/85597
13593 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
13594 do not use split vect_get_vec_defs call but call vect_get_slp_defs
13597 2018-05-02 Tom de Vries <tom@codesourcery.com>
13600 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
13601 dump files): Add ltrans-tree.
13603 2018-05-02 Tom de Vries <tom@codesourcery.com>
13606 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
13607 dump files): Add wpa-ipa.
13609 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
13611 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
13612 powerpc*-*-linux*paired* target.
13613 * config/rs6000/750cl.h: Delete.
13614 * config/rs6000/paired.h: Delete.
13615 * config/rs6000/paired.md: Delete.
13616 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
13618 * config/rs6000/rs6000-builtin.def: Remove paired float support.
13619 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13620 comment. Remove paired float support.
13621 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
13622 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
13624 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
13625 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
13627 * config/rs6000/rs6000.c: Remove paired float support.
13628 (paired_expand_vector_init, paired_expand_vector_move,
13629 paired_emit_vector_compare, paired_emit_vector_cond_expr,
13630 (paired_expand_lv_builtin, paired_expand_stv_builtin,
13631 paired_expand_builtin, paired_expand_predicate_builtin,
13632 paired_init_builtins): Delete.
13633 * config/rs6000/rs6000.h: Remove paired float support.
13634 * config/rs6000/rs6000.md: Remove paired float support.
13635 (move_from_CR_ov_bit): Delete.
13636 * config/rs6000/rs6000.opt (mpaired): Delete.
13637 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
13638 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
13640 2018-05-02 Richard Biener <rguenther@suse.de>
13642 PR middle-end/85567
13643 * gimplify.c (gimplify_save_expr): When in SSA form allow
13644 SAVE_EXPRs to compute to SSA vars.
13646 2018-05-02 Jakub Jelinek <jakub@redhat.com>
13649 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
13650 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
13651 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
13652 clobber operands[2], instead use a new pseudo. Formatting fixes.
13654 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
13656 PR tree-optimization/85586
13657 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
13658 exit early for statements in the same group if the accesses are
13661 2018-05-02 Tom de Vries <tom@codesourcery.com>
13664 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
13667 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
13669 PR tree-optimization/85143
13670 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
13672 2018-05-01 Tom de Vries <tom@codesourcery.com>
13675 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
13676 not found" error message.
13678 2018-05-01 Tom de Vries <tom@codesourcery.com>
13681 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
13682 * vec.c (test_ordered_remove_if): New function.
13683 (vec_c_tests): Call test_ordered_remove_if.
13684 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
13685 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
13686 * tree-vect-patterns.c (vect_pattern_recog_1): Use
13687 VEC_ORDERED_REMOVE_IF.
13689 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13691 PR tree-optimization/82665
13692 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
13693 pointer subtraction where arguments come from a memchr call.
13695 2018-05-01 Jakub Jelinek <jakub@redhat.com>
13697 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
13698 --push-state --as-needed and --pop-state instead of --as-needed and
13699 --no-as-needed if ld supports it.
13700 * configure: Regenerated.
13703 * doc/install.texi2html: Replace _002d with - and _002a with * in
13704 generated html files using sed.
13706 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13709 * gcc-rich-location.c (blank_line_before_p): New function.
13710 (use_new_line): New function.
13711 (gcc_rich_location::add_fixit_insert_formatted): New function.
13712 * gcc-rich-location.h
13713 (gcc_rich_location::add_fixit_insert_formatted): New function.
13715 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13717 * selftest.c (assert_streq): Rename "expected" and "actual" to
13718 "val1" and "val2". Extend NULL-handling to cover both inputs
13719 symmetrically, while still requiring both to be non-NULL for a pass.
13720 * selftest.h (assert_streq): Rename "expected" and "actual" to
13722 (ASSERT_EQ): Likewise.
13723 (ASSERT_EQ_AT): Likewise.
13724 (ASSERT_KNOWN_EQ): Likewise.
13725 (ASSERT_KNOWN_EQ_AT): Likewise.
13726 (ASSERT_NE): Likewise.
13727 (ASSERT_MAYBE_NE): Likewise.
13728 (ASSERT_MAYBE_NE_AT): Likewise.
13729 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
13730 the assertion to pass.
13731 (ASSERT_STREQ_AT): Likewise.
13733 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
13735 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
13736 interaction with -pie.
13738 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13740 * selftest.h: Fix alphabetization of per-source-file selftest
13743 2018-04-30 Jason Merrill <jason@redhat.com>
13745 PR c++/61982 - dead stores to destroyed objects.
13746 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
13749 2018-04-30 Jason Merrill <jason@redhat.com>
13751 * tree.c (build_clobber): New.
13752 * tree.h: Declare it.
13753 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
13755 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13757 * diagnostic-show-locus.c (layout::layout): Update for
13758 location_get_source_line returning a char_span.
13759 (struct char_span): Move to input.h.
13760 (struct correction): Update for fields in char_span becoming
13762 (struct source_line): Update for location_get_source_line
13763 returning a char_span.
13764 (layout::print_line): Likewise.
13765 * edit-context.c (edited_file::print_content): Likewise.
13766 (edited_file::print_diff_hunk): Likewise.
13767 (edited_file::print_run_of_changed_lines): Likewise.
13768 (edited_file::get_num_lines): Likewise.
13769 (edited_line::edited_line): Likewise.
13770 * final.c (asm_show_source): Likewise.
13771 * input.c (location_get_source_line): Convert return type
13772 from const char * to char_span, losing the final "line_len"
13774 (dump_location_info): Update for the above.
13775 (get_substring_ranges_for_loc): Likewise. Use a char_span
13776 when handling the literal within the line.
13777 (test_reading_source_line): Update for location_get_source_line
13778 returning a char_span.
13779 * input.h (class char_span): Move here from
13780 diagnostic-show-locus.c, converting from a struct to a class.
13781 Make data members private.
13782 (char_span::operator bool): New.
13783 (char_span::length): New.
13784 (char_span::get_buffer): New.
13785 (char_span::operator[]): New.
13786 (char_span::subspan): Make const.
13787 (char_span::xstrdup): New.
13788 (location_get_source_line): Convert return type from const char *
13789 to char_span, losing the final "line_size" param.
13791 2018-04-30 Jan Hubicka <jh@suse.cz>
13793 * lto-wrapper.c (ltrans_priorities): New static var.
13794 (cmp_priority): New.
13795 (run_gcc): Read priorities and if doing parallel build order
13796 the Makefile by them.
13798 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13800 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
13802 2018-04-30 Richard Biener <rguenther@suse.de>
13804 * tree-cfg.c (verify_address): Remove base argument, add
13805 flag whether to check TREE_ADDRESSABLE and do that.
13806 (verify_expr): Remove.
13807 (verify_types_in_gimple_reference): Add pieces from verify_expr.
13808 (verify_gimple_assign_single): Likewise.
13809 (verify_gimple_switch): Likewise.
13810 (verify_expr_location_1): Dereference tp once. Add (disabled)
13811 piece from verify_expr.
13812 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
13814 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
13816 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
13818 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
13820 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
13821 (small_data_pattern): Likewise.
13822 (arc_rewrite_small_data): Likewise.
13823 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
13824 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
13825 (get_symbol_alignment): New function.
13826 (legitimate_small_data_address_p): Likewise.
13827 (legitimate_scaled_address): Update, call
13828 legitimate_small_data_address_p.
13829 (output_sdata): New static variable.
13830 (arc_print_operand): Update how we handle small data operands.
13831 (arc_print_operand_address): Likewise.
13832 (arc_legitimate_address_p): Update, use
13833 legitimate_small_data_address_p.
13834 (arc_rewrite_small_data_p): Remove.
13835 (arc_rewrite_small_data_1): Likewise.
13836 (arc_rewrite_small_data): Likewise.
13837 (small_data_pattern): Likewise.
13838 (compact_sda_memory_operand): Update to use
13839 legitimate_small_data_address_p and get_symbol_alignment.
13840 (prepare_move_operands): Don't rewite sdata pattern.
13841 (prepare_extend_operands): Remove.
13842 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
13844 (zero_extendqisi2): Likewise.
13845 (zero_extendhisi2): Likewise.
13846 (extendqihi2): Likewise.
13847 (extendqisi2): Likewise.
13848 (extendhisi2): Likewise.
13849 (addsi3): Likewise.
13850 (subsi3): Likewise.
13851 (andsi3): Likewise.
13852 * config/arc/constraints.md (Usd): Change it to memory constraint.
13854 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
13856 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
13857 as source of std instructions.
13858 * config/arc/arc.md (movsi_insn): Update pattern predicate to
13859 allow 6-bit constants as source for store instructions.
13860 (movdi_insn): Update instruction pattern to allow 6-bit constants
13861 as source for store instructions.
13863 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
13865 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
13867 2018-04-30 Nathan Sidwell <nathan@acm.org>
13868 Sandra Loosemore <sandra@codesourcery.com>
13870 * dumpfile.c (dump_open): Allow '-' for stdout.
13871 * doc/invoke.texi (Developer Options): Document dump filename
13872 determination early. Document stdin/stdout selection.
13874 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
13876 Microblaze Target: PIC data text relative
13878 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
13879 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
13881 * config/microblaze/microblaze.h (microblaze_constant_address_p):
13882 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
13883 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
13884 New addressing mode for data-text relative position indepenedent code.
13885 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
13886 'ADDRESS_SYMBOLIC_TXT_REL'.
13887 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
13888 (microblaze_legitimate_pic_operand): Exclude function calls from
13889 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
13890 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
13892 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
13893 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
13894 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
13895 for 'address + offset'.
13896 (microblaze_expand_prologue): Add new function prologue call for
13898 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
13899 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
13900 table in case of TARGET_PIC_DATA_TEXT_REL.
13901 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
13902 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
13903 Add new macros 'UNSPEC_TEXT',
13904 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
13905 + exclude function calls from 'UNSPEC_PLT' in case of data text
13907 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
13908 new target hook for generating address diff vector tables in case of
13910 * doc/tm.texi : Regenerate.
13911 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
13912 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
13913 of addr diff vector generation.
13914 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
13915 target hook definition.
13916 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
13917 Add default function for generate_pic_addr_diff_vec -> flag_pic.
13918 * doc/invoke.texi (Add new pic option): Add new microblaze pic
13919 option for data text relative.
13921 2018-04-30 Richard Biener <rguenther@suse.de>
13923 * tree-chrec.h (evolution_function_is_constant_p): Remove
13925 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
13927 2018-04-30 Richard Biener <rguenther@suse.de>
13930 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
13932 2018-04-30 Richard Biener <rguenther@suse.de>
13934 PR tree-optimization/28364
13935 PR tree-optimization/85275
13936 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
13937 copying first exit test.
13939 2018-04-28 Mark Wielaard <mark@klomp.org>
13941 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
13942 dwarf_version >= 5.
13943 (dwarf_AT): Handle DW_AT_addr_base.
13944 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
13946 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
13949 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
13950 (*ashl<dwi>3_doubleword_mask_1): Ditto.
13951 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
13952 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
13954 2018-04-28 Richard Biener <rguenther@suse.de>
13956 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
13957 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
13958 to reflect use. Only add interesting stmts.
13960 2018-04-27 Martin Jambor <mjambor@suse.cz>
13963 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
13964 the jump function allows for passing through aggregate values.
13966 2018-04-27 David Malcolm <dmalcolm@redhat.com>
13968 * input.h (in_system_header_at): Convert from macro to inline
13970 (from_macro_expansion_at): Likewise.
13971 (from_macro_definition_at): Likewise.
13973 2018-04-27 Jeff Law <law@redhat.com>
13975 * config.gcc: Mark tile* targets as deprecated/obsolete.
13977 2018-04-27 Richard Biener <rguenther@suse.de>
13979 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
13982 2018-04-27 Richard Biener <rguenther@suse.de>
13984 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
13986 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
13988 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
13989 with Yd constraint. Set "preferred_for_speed" attribute from
13990 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
13991 with Yd constraint.
13992 (*movdi_internal): Ditto.
13993 (movti_interunit splitters): Remove
13994 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
13995 (movdi_interunit splitters): Ditto.
13996 * config/i386/constraints.md (Ye): Remove.
13997 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
13999 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14002 * config/aarch64/constraints.md (Usg): Limit to 31.
14003 (Usj): Limit to 63.
14005 2018-04-27 Jakub Jelinek <jakub@redhat.com>
14007 PR tree-optimization/85529
14008 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
14009 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
14010 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
14011 zero extension or masking of the MSB bit.
14012 (optimize_range_tests): Add FIRST_BB argument, pass it through
14013 to optimize_range_tests_var_bound.
14014 (maybe_optimize_range_tests, reassociate_bb): Adjust
14015 optimize_range_tests callers.
14017 2018-04-26 Richard Biener <rguenther@suse.de>
14018 Jakub Jelinek <jakub@redhat.com>
14020 * cgraph.h (symbol_table): Just declare debug method here.
14021 * symtab.c (symbol_table::debug): Define.
14023 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
14025 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
14027 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
14029 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
14030 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
14031 (*movdi_internal): Substitute Yi and Yj constraint with x
14032 and Ym and Yn constraint with y constraint. Update "isa"
14033 attribute and set "preferred_for_speed" attribute from
14034 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
14035 (*movsi_internal): Ditto.
14036 (*movdf_internal): Ditto.
14037 (*movsf_internal): Ditto.
14038 (*zero_extendsidi2): Ditto.
14039 * config/i386/sse.md (vec_set<mode>_0): Ditto.
14040 (sse2_loadld): Ditto.
14041 (*vec_extract<ssevecmodelower>_0): Ditto.
14042 (*vec_extractv4si_0_zext_sse4): Ditto.
14043 (vec_concatv2di): Ditto.
14044 (*vec_dup<mode>): Ditto.
14045 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
14046 * config/i386/constraints.md (Yi): Remove.
14051 2018-04-26 Nathan Sidwell <nathan@acm.org>
14053 * dumpfile.c (dump_open): New.
14054 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
14055 (dump_finish): Detect stdio/stderr by value not name.
14057 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
14059 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
14061 2018-04-26 Tom de Vries <tom@codesourcery.com>
14064 * config/nvptx/nvptx.c (verify_neutering_jumps)
14065 (verify_neutering_labels): New function
14066 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
14068 2018-04-26 Tom de Vries <tom@codesourcery.com>
14071 * config/nvptx/nvptx.c (needs_neutering_p): New function.
14072 (nvptx_single): Use needs_neutering_p to skip over insns that do not
14075 2018-04-26 Richard Biener <rguenther@suse.de>
14076 Tom de Vries <tom@codesourcery.com>
14079 * lto-streamer-out.c (output_function): Fixup loops if required to match
14080 discovery done in the reader.
14082 2018-04-26 Richard Biener <rguenther@suse.de>
14084 PR tree-optimization/85116
14085 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
14086 have a loop exit from the single latch predecessor. Remove
14087 case of header with just condition.
14088 (ch_base::copy_headers): Exclude infinite loops from any
14090 (pass_ch::execute): Record exits.
14092 2018-04-26 Richard Biener <rguenther@suse.de>
14094 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
14095 prologue cost vector and pass it to vect_get_load_cost.
14096 (vect_get_peeling_costs_all_drs): Likewise.
14097 (vect_peeling_hash_get_lowest_cost): Likewise.
14098 (vect_enhance_data_refs_alignment): Likewise.
14100 2018-04-26 Richard Biener <rguenther@suse.de>
14102 PR middle-end/85450
14103 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
14104 checking of integer<->pointer conversions.
14105 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
14106 sign-/zero-extending pointer types.
14107 (expand_omp_for_static_chunk): Likewise.
14109 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
14110 Jean Lee <xiaoyur347@gmail.com>
14112 * config/mips/mips.c (mips_asan_shadow_offset): New function.
14113 (TARGET_ASAN_SHADOW_OFFSET): Define.
14114 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
14115 true for -fsanitize=address.
14117 2018-04-25 Mark Wielaard <mark@klomp.org>
14119 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
14122 2018-04-25 Jakub Jelinek <jakub@redhat.com>
14124 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
14125 than "alu", remove explicit "memory" and "imm_disp" attributes.
14126 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
14128 PR middle-end/85414
14129 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
14130 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
14131 gen_lowpart_no_emit.
14133 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
14136 * config/i386/i386.c (ix86_expand_builtin): Change memory
14137 operand to XI, extend p0 to Pmode.
14138 * config/i386/i386.md: Change unspec volatile and operand
14139 1 mode to XI, change operand 0 mode to P.
14141 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14143 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
14144 GET_MODE_MASK before any checking.
14145 (nds32_can_use_bset_p): Likewise.
14146 (nds32_can_use_btgl_p): Likewise.
14148 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14150 * config/nds32/nds32-doubleword.md: New define_split pattern for
14151 illegal register number.
14153 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14155 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
14157 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14159 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
14161 2018-04-25 Richard Biener <rguenther@suse.de>
14163 * lto-streamer.h (LTO_major_version): Bump to 8.
14165 2018-04-25 Jakub Jelinek <jakub@redhat.com>
14167 * BASE-VER: Set to 9.0.0.
14169 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
14171 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
14172 in __abskf2 and __powikf2.
14174 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14177 * config/aarch64/constraints.md (Usg, Usj): New constraints.
14178 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
14179 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
14180 Use the above on operand 2. Reindent.
14181 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
14183 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
14186 * common/config/i386/i386-common.c (ix86_handle_option): Don't
14188 * config/i386/i386.opt (mcet): Removed.
14189 * doc/install.texi: Remove -mcet documentation.
14190 * doc/invoke.texi: Likewise.
14192 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
14195 * doc/install.texi: Remove -mcet from bootstrap-cet.
14197 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14200 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
14201 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
14205 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
14206 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
14207 containing a CONST_VECTOR.
14209 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
14211 * doc/install.texi: Update newlib dependency for nvptx.
14213 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14216 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
14217 instead of INTVAL when shifting x left.
14219 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
14221 PR tree-optimization/85478
14222 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
14223 vect_grouped_store_supported for single element vectors.
14225 2018-04-24 Richard Biener <rguenther@suse.de>
14228 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
14229 load cost increase to the case of non-constant step.
14231 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14234 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
14235 destination if any_malformed_asm.
14237 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
14239 PR middle-end/85496
14240 * expr.c (store_field): In the bitfield case, if the value comes from
14241 a function call and is returned in registers by means of a PARALLEL,
14242 do not change the mode of the temporary unless BLKmode and VOIDmode.
14244 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
14246 PR rtl-optimization/85423
14247 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
14248 dependencies to debug insns when the previous insn is non-debug.
14250 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
14252 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
14253 enums into a single definition.
14254 (fls): Fix predicates and printing.
14257 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
14259 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
14260 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
14261 and short u6 immediate.
14262 (check_if_valid_sleep_operand): Remove.
14263 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
14265 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14267 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
14268 flag_always_save_lp condition.
14269 * config/nds32/nds32.opt (malways-save-lp): New option.
14271 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14273 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
14274 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
14275 * config/nds32/nds32.h
14276 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
14277 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
14279 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14281 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
14282 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
14284 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14285 Chung-Ju Wu <jasonwucj@gmail.com>
14287 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
14289 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
14290 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
14292 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14294 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
14296 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14298 * config/nds32/nds32-protos.h (nds32_data_alignment,
14299 nds32_local_alignment): Declare.
14300 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
14301 nds32_local_alignment): New functions.
14302 (TARGET_CONSTANT_ALIGNMENT): Define.
14303 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
14305 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14307 * config/nds32/nds32.c
14308 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
14309 (TARGET_MODES_TIEABLE_P): Likewise.
14311 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14313 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
14314 level Ofast and Og.
14316 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
14317 Chung-Ju Wu <jasonwucj@gmail.com>
14319 * config/nds32/constants.md (unspec_volatile_element): Add enum values
14320 for unaligned access.
14321 * config/nds32/nds32-intrinsic.c: Implementation of expanding
14323 * config/nds32/nds32-intrinsic.md: Likewise.
14324 * config/nds32/nds32_intrinsic.h: Likewise.
14325 * config/nds32/nds32.h (nds32_builtins): Likewise.
14326 * config/nds32/nds32.opt (munaligned-access): New option.
14327 * config/nds32/nds32.c (nds32_asm_file_start): Display
14328 flag_unaligned_access status.
14330 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
14332 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
14333 -mno-relax is present.
14334 * config/riscv/linux.h (LINK_SPEC): Ditto.
14336 2018-04-20 Martin Sebor <msebor@redhat.com>
14339 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
14341 (gimple_fold_builtin_stxcpy_chk): Same.
14342 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
14344 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
14347 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
14348 __powikf2 when long double is IEEE 128-bit.
14350 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
14352 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
14353 step to make sure stack always aligned.
14355 2018-04-20 Carl Love <cel@us.ibm.com>
14358 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
14359 size check for arg0.
14361 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
14362 Tom de Vries <tom@codesourcery.com>
14365 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
14366 Emit insns for calls too.
14367 (nvptx_find_par): Always look for worker-level predecessor insn.
14368 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
14370 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
14371 (nvptx_process_pars): Propagate frames for calls.
14373 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
14376 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
14378 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
14379 (ix86_handle_option): Don't handle OPT_mibt.
14380 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
14382 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
14384 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
14385 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
14386 (ix86_target_macros): Define __CET__ with flag_cf_protection
14387 for -fcf-protection.
14388 * config/i386/i386.c (isa2_opts): Remove -mibt.
14389 * config/i386/i386.h (TARGET_IBT): Removed.
14390 (TARGET_IBT_P): Likewise.
14391 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
14392 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
14393 * config/i386/i386.opt (mcet): Update help message.
14394 (mshstk): Likewise.
14396 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
14397 -mcet as an alias for -mshstk.
14399 2018-04-20 Richard Biener <rguenther@suse.de>
14401 PR middle-end/85475
14402 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
14403 complexity by forcing a single use of the multiply operand.
14405 2018-04-20 Martin Jambor <mjambor@suse.cz>
14408 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
14409 recursion dependency to only apply to non-clones.
14411 2018-04-20 Martin Jambor <mjambor@suse.cz>
14414 * ipa-cp.c (create_specialized_node): Check that clones of
14415 self-recursive edges exist during IPA-CP.
14417 2018-04-19 Toon Moene <toon@moene.org>
14419 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
14422 2018-04-19 Jakub Jelinek <jakub@redhat.com>
14424 PR tree-optimization/85467
14425 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
14426 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
14427 VECTOR_CST element to type.
14429 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
14432 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
14433 * config/i386/i386.md (builtin_setjmp_setup): Removed.
14434 (builtin_longjmp): Likewise.
14435 (save_stack_nonlocal): New pattern.
14436 (restore_stack_nonlocal): Likewise.
14438 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
14441 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
14442 Replace ASM_OUTPUT_LABEL with fprintf.
14444 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
14447 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
14448 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
14449 * config/i386/i386-c.c (ix86_target_macros_internal): Also
14450 define __IBT__ and __SHSTK__ for -fcf-protection.
14451 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
14453 (ix86_trampoline_init): Likewise.
14454 (x86_output_mi_thunk): Likewise.
14455 (ix86_notrack_prefixed_insn_p): Likewise.
14456 (ix86_option_override_internal): Don't disallow -fcf-protection.
14457 * config/i386/i386.md (rdssp<mode>): Also enable for
14459 (incssp<mode>): Likewise.
14460 (nop_endbr): Likewise.
14461 * config/i386/i386.opt (mcet): Change help message to built-in
14464 (mshstk): Likewise.
14465 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
14466 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
14467 enable CET built-in functions.
14469 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
14471 * common/config/i386/i386-common.c
14472 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
14473 OPTION_MASK_ISA_MOVDIRI_UNSET,
14474 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
14475 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
14476 * config.gcc (movdirintrin.h): New header.
14477 * config/i386/cpuid.h (bit_MOVDIRI,
14478 bit_MOVDIR64B): New bits.
14479 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
14481 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
14482 (VOID, PVOID, PCVOID)): New function types.
14483 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
14484 __builtin_ia32_directstoreu_u64,
14485 __builtin_ia32_movdir64b): New builtins.
14486 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
14487 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
14489 (ix86_valid_target_attribute_inner_p): Ditto.
14490 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
14491 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
14492 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
14493 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
14494 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
14495 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
14496 (movdiri<mode>, movdir64b_<mode>): New.
14497 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
14498 * config/i386/immintrin.h: Include movdirintrin.h.
14499 * config/i386/movdirintrin.h: New file.
14500 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
14502 2018-04-19 Richard Biener <rguenther@suse.de>
14504 PR middle-end/85455
14505 * cfg.c (clear_bb_flags): When loop state says we have
14506 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
14508 2018-04-19 Richard Biener <rguenther@suse.de>
14510 PR tree-optimization/84737
14511 * tree-vect-data-refs.c (vect_copy_ref_info): New function
14512 copying restrict info.
14513 (vect_setup_realignment): Use it.
14514 * tree-vectorizer.h (vect_copy_ref_info): Declare.
14515 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
14516 the first DR to all generated stores.
14517 (vectorizable_load): Likewise for loads.
14519 2018-04-19 Jakub Jelinek <jakub@redhat.com>
14521 PR tree-optimization/85446
14522 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
14523 the integral and pointer types to have the same precision.
14525 * doc/install.texi: Document --disable-cet being the default and
14528 2018-04-18 Martin Liska <mliska@suse.cz>
14530 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
14533 2018-04-18 Martin Liska <mliska@suse.cz>
14536 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
14539 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
14540 arguments if they are comparable.
14542 2018-04-18 Martin Liska <mliska@suse.cz>
14545 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
14548 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
14551 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
14554 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
14555 ENDBR after calling __morestack.
14557 2018-04-18 David Malcolm <dmalcolm@redhat.com>
14560 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
14561 by using gcc_base_ver to generate a gcc_driver_version, and use
14562 it when generating GCC_DRIVER_NAME.
14563 * configure: Regenerate.
14565 2018-04-18 Jakub Jelinek <jakub@redhat.com>
14568 * config.gcc: Obsolete powerpc*-*-*spe*.
14570 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14573 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
14574 (stabstr_D): Change type of unum from unsigned int to
14575 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
14578 2018-04-17 Jim Wilson <jimw@sifive.com>
14581 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
14582 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
14583 Set arg_pointer_offset after using pretend_args_size.
14585 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14587 PR rtl-optimization/85431
14588 * dse.c (record_store): Ignore zero width stores.
14591 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
14592 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
14593 __builtin_stack_restore rather than after it.
14594 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
14595 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
14596 argument instead of virtual_dynamic_stack_rtx.
14598 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
14600 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
14602 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14603 Add note to error message to explain internal mapping of overloaded
14604 built-in function name to non-overloaded built-in function name.
14605 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
14608 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
14611 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
14612 where the inputs overlap with the output.
14614 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14616 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
14617 (=v, v) alternative and explicit "memory" attribute.
14618 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
14619 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
14621 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
14622 "sselog1" type instead of "sselog".
14623 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
14624 "sselog". Remove explicit "memory" attribute.
14625 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
14626 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
14628 (vec_extract_hi_v32hi): Merge all alternatives into one, use
14629 "sselog1" type instead of "sselog". Remove explicit "memory"
14631 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
14632 use "sselog1" type instead of "sselog". Remove explicit "memory"
14634 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
14635 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
14637 (vec_extract_hi_v64qi): Merge all alternatives into one, use
14638 "sselog1" type instead of "sselog". Remove explicit "memory"
14640 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
14641 use "sselog1" type instead of "sselog". Remove explicit "memory"
14645 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
14647 PR middle-end/85414
14648 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
14651 2018-04-17 Martin Jambor <mjambor@suse.cz>
14654 * ipa-cp.c (create_specialized_node): Call
14655 expand_all_artificial_thunks if necessary.
14657 2018-04-17 Martin Liska <mliska@suse.cz>
14660 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
14661 in message, remote space in between '_G' and '('.
14663 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14666 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
14667 avx512f_vmcmp<mode>3<round_saeonly_name>,
14668 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
14669 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
14670 avx512f_rndscale<mode><round_saeonly_name>,
14671 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
14672 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
14673 Use %<iptr>2 instead of %2 for -masm=intel.
14674 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
14675 avx512f_vcvttss2usi<round_saeonly_name>,
14676 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
14678 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
14679 avx512f_vcvttsd2usi<round_saeonly_name>,
14680 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
14681 Use %q1 instead of %1 for -masm=intel.
14682 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
14683 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
14684 of %3 for -masm=intel.
14685 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
14687 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
14689 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
14691 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
14692 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
14694 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
14696 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
14697 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
14698 %g1 and one with %0 and %1.
14699 (avx512er_vmrcp28<mode><round_saeonly_name>,
14700 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
14701 %1 for -masm=intel.
14702 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
14703 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
14704 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
14705 of %0 and %{%4%} for -masm=intel.
14706 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
14707 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
14708 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
14709 order of %0 and %{%5%}%{z%} for -masm=intel.
14711 2018-04-17 Jan Hubicka <jh@suse.cz>
14714 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
14716 2018-04-17 Martin Liska <mliska@suse.cz>
14719 * multiple_target.c (create_dispatcher_calls): Set apostrophes
14720 for target_clone error message. Make default implementation
14721 clone to be a local declaration.
14722 (separate_attrs): Add new argument and check for an empty
14724 (expand_target_clones): Handle it.
14725 (ipa_target_clone): Make redirection just for target_clones
14728 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
14729 Tom de Vries <tom@codesourcery.com>
14731 PR middle-end/84955
14732 * omp-expand.c (expand_oacc_for): Add dummy false branch for
14733 tiled basic blocks without omp continue statements.
14735 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
14738 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
14739 vec_extract expression as having side effects to make sure it gets
14742 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
14745 * config/i386/i386.c (get_builtin_code_for_version): Check
14748 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
14751 * config.gcc: Support "skylake".
14752 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14754 * config/i386/i386.c (m_SKYLAKE): Define.
14755 (processor_target_table): Add "skylake".
14756 (ix86_option_override_internal): Add "skylake".
14757 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
14758 PROCESSOR_CANNONLAKE.
14759 (get_builtin_code_for_version): Fix priority for
14760 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
14761 PROCESSOR_SKYLAKE-AVX512.
14762 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
14763 (processor_type): Add PROCESSOR_SKYLAKE.
14765 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
14766 Jason Merrill <jason@redhat.com>
14769 * convert.c (convert_to_integer_1): Use direct recursion for
14770 enumeral types and types with a precision less than the number
14771 of bits in their mode.
14773 2018-04-16 Julia Koval <julia.koval@intel.com>
14776 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
14777 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
14779 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
14782 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
14783 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
14784 and -mno-direct-move.
14786 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
14789 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
14790 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
14791 Ensure negative shifts result in {0}.
14793 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
14795 PR rtl-optimization/79916
14796 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
14797 regs (if any) to define how to gnerate SD moves when LRA is in
14800 2018-04-13 Jakub Jelinek <jakub@redhat.com>
14802 PR rtl-optimization/85393
14803 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
14804 * except.c (expand_dw2_landing_pad_for_region): Make static.
14805 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
14806 a label and unconditional jump to old_bb, rather than
14807 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
14810 PR rtl-optimization/85376
14811 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
14812 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
14813 instead of a specific value.
14815 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
14816 Bin Cheng <bin.cheng@arm.com>
14818 PR tree-optimization/82965
14819 PR tree-optimization/83991
14820 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
14821 by_profile_only parameter.
14822 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
14823 information if the loop was predicted to iterate too many times.
14824 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
14826 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
14829 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
14832 2018-04-13 Martin Liska <mliska@suse.cz>
14833 Jakub Jelinek <jakub@redhat.com>
14835 PR middle-end/81657
14836 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
14837 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
14838 * builtins.c (expand_builtin_memory_copy_args): Use
14839 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
14840 handle dest_addr == pc_rtx.
14842 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
14845 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
14846 asked to not generate direct moves.
14847 (fix_trunc<mode>si2_stfiwx): Similar.
14848 (fix_trunc<mode>si2_internal): Similar.
14850 2018-04-12 Jakub Jelinek <jakub@redhat.com>
14853 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
14854 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
14855 lookup if dest in some wider mode is known to be const0_rtx and
14856 if so, record permanent equivalence for it to be ZERO_EXTEND of
14857 the narrower mode destination.
14859 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
14861 * lto-streamer-out.c (output_function): Revert 259346.
14862 * omp-expand.c (expand_oacc_for): Likewise.
14864 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
14866 PR rtl-optimization/85354
14867 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
14868 * sel-sched.c (sel_global_init): ... here.
14870 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
14873 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
14874 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
14875 mode for PE-COFF targets.
14876 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
14877 (i386_pe_asm_lto_end): Likewise.
14878 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
14879 (TARGET_ASM_LTO_END): Likewise.
14880 * config/i386/winnt.c (saved_debug_info_level): New static variable.
14881 (i386_pe_asm_lto_start): New function.
14882 (i386_pe_asm_lto_end): Likewise.
14884 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
14885 Richard Biener <rguenther@suse.de>
14887 PR middle-end/84955
14888 * lto-streamer-out.c (output_function): Fix CFG loop state before
14890 * omp-expand.c (expand_oacc_for): Handle calls to internal
14891 functions like regular functions.
14893 2018-04-12 Richard Biener <rguenther@suse.de>
14896 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
14897 for the early LTO debug to properly generate references to it
14898 during DIE emission. Do not re-use that for the skeleton for
14900 (dwarf2out_early_finish): Likewise.
14902 2018-04-12 Jakub Jelinek <jakub@redhat.com>
14905 * config/i386/sse.md
14906 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
14907 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
14908 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
14909 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
14910 and output is a reg, avoid creating invalid lowpart subreg, but
14911 instead split into a 512-bit move. Don't split if not AVX512VL,
14912 input is xmm16+ reg and output is a mem.
14913 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
14914 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
14915 xmm16+ reg and output is a mem.
14917 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14919 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
14920 also for flag_dwarf2_cfi_asm.
14922 2018-04-12 Jakub Jelinek <jakub@redhat.com>
14924 PR rtl-optimization/85342
14925 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
14926 a bool scalar var inside of the loop instead. Don't try to update
14927 recog_data.operand after failed apply_change_group.
14929 2018-04-12 Tom de Vries <tom@codesourcery.com>
14932 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
14933 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
14934 array with flexible array member as array without given dimension.
14935 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
14936 argument for undefined param to true.
14938 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
14941 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
14942 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
14943 from PowerPC section.
14944 * config/rs6000/sysv4.opt (mcall-): Improve help text.
14945 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
14946 help text that is too long.
14947 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
14948 help text that is too long.
14949 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
14950 help text that is too long.
14952 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
14954 * config/alpha/alpha.md (stack_probe_internal): Rename
14955 from "probe_stack". Update all callers.
14957 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
14959 PR rtl-optimization/84566
14960 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
14961 sched_macro_fuse_insns.
14963 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
14966 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
14967 (compute_block_dependences): ... from here.
14969 2018-04-11 Jakub Jelinek <jakub@redhat.com>
14971 PR tree-optimization/85331
14972 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
14973 from int to HOST_WIDE_INT.
14975 2018-04-11 Martin Jambor <mjambor@suse.cz>
14978 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
14979 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
14980 not the same as the source val.
14981 (cgraph_edge_brings_value_p): New parameter.
14982 (gather_edges_for_value): Pass destination value to
14983 cgraph_edge_brings_value_p.
14984 (perhaps_add_new_callers): Likewise.
14985 (get_info_about_necessary_edges): Likewise and exclude values brought
14986 only by self-recursive edges.
14987 (create_specialized_node): Redirect only clones of self-calling edges.
14988 (+self_recursive_pass_through_p): New function.
14989 (find_more_scalar_values_for_callers_subset): Use it.
14990 (find_aggregate_values_for_callers_subset): Likewise.
14991 (known_aggs_to_agg_replacement_list): Removed.
14992 (decide_whether_version_node): Re-calculate known constants for all
14993 remaining context clones.
14995 2018-04-11 Richard Biener <rguenther@suse.de>
14998 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
14999 from early DWARF output.
15000 (dwarf2out_early_finish): Output line info unconditionally into
15001 early DWARF and add reference to it.
15003 2018-04-11 Jakub Jelinek <jakub@redhat.com>
15006 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
15007 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
15008 other than V2DFmode using iptr mode attribute.
15009 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
15011 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
15013 PR rtl-optimization/84659
15014 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
15016 2018-04-11 Jakub Jelinek <jakub@redhat.com>
15019 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
15021 (output_loc_list): Pass address of a dummy size variable even in the
15022 locview handling loop.
15023 (index_location_lists): Add comment on why skip_loc_list_entry can't
15026 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
15029 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
15032 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
15035 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
15036 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
15037 and -mstring-compare-inline-limit.
15039 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
15042 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
15043 for stack clash protection in a register whenever we need it to be in
15046 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
15048 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
15049 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
15051 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
15054 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
15056 (mlong-double-): Ditto.
15057 * config/rs6000/sysv4.opt (msdata=): Ditto.
15058 (mtls-size=): Ditto.
15060 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
15062 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
15063 erroneous entries for
15064 "vector int vec_ldl (int, long int *)", and
15065 "vector unsigned int vec_ldl (int, unsigned long int *)".
15066 Add comments and entries for
15067 "vector bool char vec_ldl (int, bool char *)",
15068 "vector bool short vec_ldl (int, bool short *)",
15069 "vector bool int vec_ldl (int, bool int *)",
15070 "vector bool long long vec_ldl (int, bool long long *)",
15071 "vector pixel vec_ldl (int, pixel *)",
15072 "vector long long vec_ldl (int, long long *)",
15073 "vector unsigned long long vec_ldl (int, unsigned long long *)".
15074 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
15075 type tree bool_long_long_type_node and correct definition of
15076 bool_V2DI_type_node to make reference to this new type tree.
15077 (rs6000_mangle_type): Replace erroneous reference to
15078 bool_long_type_node with bool_long_long_type_node.
15079 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
15080 comments to emphasize sign distinctions for char and int types and
15081 replace RS6000_BTI_bool_long constant with
15082 RS6000_BTI_bool_long_long constant. Also add comment to restrict
15083 use of RS6000_BTI_pixel.
15084 (bool_long_type_node): Remove this macro definition.
15085 (bool_long_long_type_node): New macro definition
15087 2018-04-10 Jakub Jelinek <jakub@redhat.com>
15089 PR rtl-optimization/85300
15090 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
15091 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
15092 simplify_unary_operation fails.
15094 2018-04-10 Martin Liska <mliska@suse.cz>
15096 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
15097 cgraph_edge and ipa_ref.
15099 2018-04-10 Jakub Jelinek <jakub@redhat.com>
15103 * config/i386/sse.md
15104 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
15105 computation of the VEC_MERGE selector from mask.
15106 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
15107 Fix decoding of the VEC_MERGE selector into mask.
15109 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
15111 PR tree-optimization/85286
15112 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
15114 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
15116 * final.c (final_1): Set insn_last_address as well as
15117 insn_current_address.
15119 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15122 * explow.c (emit_stack_probe): Call validize_mem on memory location
15123 before passing it to gen_probe_stack. Create address operand and
15124 legitimize it for the probe_stack_address case.
15126 2018-04-09 Jan Hubicka <jh@suse.cz>
15129 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
15130 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
15131 * tree.c (free_lang_data_in_type): Fix handling of binfos;
15133 (free_lang_data): Rebuild type inheritance graph.
15135 2018-04-09 Martin Sebor <msebor@redhat.com>
15137 * invoke.texi (-finline-small-functions): Mention other optimization
15139 (-findirect-inlining, -fpartial-inlining): Same.
15140 (-finline-functions-called-once): Same.
15141 (-freorder-blocks-and-partition): Same.
15143 2018-04-09 Jan Hubicka <jh@suse.cz>
15146 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
15147 jumps; choose last target that matches the criteria (i.e.
15148 no partition changes for non-crossing jumps).
15149 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
15150 support for redirecting crossing jumps to non-crossing.
15152 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
15154 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
15155 also for naked functions.
15157 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
15159 * config/arc/arc.md (add_shift): New pattern.
15160 (add_shift2): Likewise.
15161 (sub_shift): Likewise.
15162 (sub_shift_cmp0_noout): Likewise.
15163 (compare_si_ashiftsi): Likewise.
15164 (xbfu_cmp0_noout): New combine pattern.
15165 (xbfu_cmp0"): Likewise.
15166 (movsi_set_cc_insn): Place the predicable variant first.
15167 (commutative_binary_cmp0_noout): Remove clobber.
15168 (commutative_binary_cmp0): New pattern.
15169 (noncommutative_binary_cmp0): Likewise.
15170 (noncommutative_binary_cmp0_noout): Likewise.
15171 (noncommutative_binary_comparison_result_used): Removed.
15172 (rsub_cmp0): New pattern.
15173 (rsub_cmp0_noout): Likewise.
15174 (extzvsi): Changed, keep only meaningful variants.
15175 (SQH, SEZ): New iterators.
15176 (SQH_postfix): New mode attribute.
15177 (SEZ_prefix): New code attribute.
15178 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
15179 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
15180 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
15181 of numerical value.
15182 (noncommutative_operator): Check the availability of barrel
15185 2018-04-09 Richard Biener <rguenther@suse.de>
15187 PR tree-optimization/85284
15188 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
15189 Only use the niter constraining form of simple_iv when the exit
15190 is always executed.
15192 2018-04-09 Tom de Vries <tom@codesourcery.com>
15195 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
15196 (define_expand "*memory_barrier"): New define_expand.
15197 (define_insn "memory_barrier"): New insn.
15199 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15201 PR rtl-optimization/80463
15202 PR rtl-optimization/83972
15203 PR rtl-optimization/83480
15205 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
15206 correct producer for the insn.
15207 (tidy_control_flow): Fixup seqnos in case of debug insns.
15209 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15211 PR rtl-optimization/83913
15213 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
15214 different sched-times when merging exprs.
15216 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15218 PR rtl-optimization/83962
15220 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
15221 tidy_fallthru_edge and tidy_control_flow.
15223 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15225 PR rtl-optimization/83530
15227 * sel-sched.c (force_next_insn): New global variable.
15228 (remove_insn_for_debug): When force_next_insn is true, also leave only
15229 next insn in the ready list.
15230 (sel_sched_region): When the region wasn't scheduled, make another pass
15231 over it with force_next_insn set to 1.
15233 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
15235 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
15237 * config/nds32/constants.md (unspec_volatile_element): Add enum values
15238 for interrupt control.
15239 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
15240 functions for interrupt control.
15241 * config/nds32/nds32-intrinsic.md: Likewise.
15242 * config/nds32/nds32_intrinsic.h: Likewise.
15243 * config/nds32/nds32.h (nds32_builtins): Likewise.
15245 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
15247 * config/nds32/nds32.c (nds32_init_machine_status,
15248 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
15249 strict_aligned_p field.
15250 (nds32_expand_to_rtl_hook): New function.
15251 (TARGET_EXPAND_TO_RTL_HOOK): Define.
15252 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
15254 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15255 Chung-Ju Wu <jasonwucj@gmail.com>
15257 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
15258 * config/nds32/nds32-n7.md: New file.
15259 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
15260 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
15262 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
15263 * config/nds32/nds32.md (pipeline_model): Add n7.
15264 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
15265 * config/nds32/pipelines.md: Include n7 settings.
15267 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15268 Chung-Ju Wu <jasonwucj@gmail.com>
15270 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
15271 * config/nds32/nds32-e8.md: New file.
15272 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
15273 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
15275 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
15276 * config/nds32/nds32.md (pipeline_model): Add e8.
15277 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
15278 * config/nds32/pipelines.md: Include e8 settings.
15280 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15281 Chung-Ju Wu <jasonwucj@gmail.com>
15283 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
15284 * config/nds32/nds32-n8.md: New file.
15285 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
15286 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
15288 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
15289 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
15290 * config/nds32/nds32.md (pipeline_model): Add n8.
15291 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
15292 * config/nds32/pipelines.md: Include n8 settings.
15294 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15295 Chung-Ju Wu <jasonwucj@gmail.com>
15297 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
15298 * config/nds32/nds32-n9-2r1w.md: New file.
15299 * config/nds32/nds32-n9-3r2w.md: New file.
15300 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
15301 nds32_register_ports): New or modify for cpu n9.
15302 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
15304 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
15305 * config/nds32/nds32-utils.c: New file.
15306 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
15307 TARGET_MUL_SLOW): Define.
15308 * config/nds32/nds32.md (pipeline_model): New attribute.
15309 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
15310 New options that support cpu n9.
15311 * config/nds32/pipelines.md: Include n9 settings.
15312 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
15314 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
15316 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
15317 information if necessary.
15318 (output_cond_branch_compare_zero): Likewise.
15319 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
15320 (nds32_target_alignment): Refine for alignment.
15321 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
15322 (FUNCTION_BOUNDARY): Modify.
15323 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
15325 * config/nds32/nds32.opt (malways-align, malign-functions): New.
15327 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
15329 * config/nds32/constants.md (unspec_volatile_element): Add values for
15330 TLB operation and data prefetch.
15331 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
15332 functions for TLB operation and data prefetch.
15333 * config/nds32/nds32-intrinsic.md: Likewise.
15334 * config/nds32/nds32_intrinsic.h: Likewise.
15335 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
15336 (nds32_print_operand): Likewise.
15337 * config/nds32/nds32.h (nds32_builtins): Likewise.
15339 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15340 Andrew Pinski <pinsika@gcc.gnu.org>
15342 PR middle-end/82976
15343 * match.pd: Use constant_boolean_node of correct type instead of
15344 boolean_true_node or boolean_false_node for simplifying
15345 pointer comparisons to zero.
15347 2018-04-07 Jakub Jelinek <jakub@redhat.com>
15349 PR tree-optimization/80021
15350 * tree.c (verify_type_variant): Make error call in verify_variant_match
15351 translatable and remove final full stop.
15353 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15355 * config/nds32/constants.md (unspec_volatile_element): Add
15356 UNSPEC_VOLATILE_EH_RETURN.
15357 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
15358 nds32_output_stack_pop): Support dwarf exception handling process.
15359 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
15360 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
15361 exception handling process.
15362 (nds32_compute_stack_frame): Likewise.
15363 (nds32_return_addr_rtx): Likewise.
15364 (nds32_initial_elimination_offset): Likewise.
15365 (nds32_expand_prologue): Likewise.
15366 (nds32_expand_epilogue): Likewise.
15367 (nds32_dynamic_chain_address): New function.
15368 * config/nds32/nds32.h (machine_function): Add fields for dwarf
15369 exception handling.
15370 (DYNAMIC_CHAIN_ADDRESS): Define.
15371 (EH_RETURN_DATA_REGNO): Define.
15372 (EH_RETURN_STACKADJ_RTX): Define.
15373 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
15374 patterns for dwarf exception handling.
15376 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15378 * config/nds32/nds32.h: Clean up obsolete macros.
15380 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15382 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
15383 Add enum values for particular instructions.
15384 * config/nds32/nds32-intrinsic.c: Implementation of expanding
15385 particular intrinsic functions.
15386 * config/nds32/nds32-intrinsic.md: Likewise.
15387 * config/nds32/nds32_intrinsic.h: Likewise.
15388 * config/nds32/nds32.h (nds32_builtins): Likewise.
15389 * config/nds32/nds32.md (type): Add pbsad and pbsada.
15390 (btst, ave): New patterns for particular instructions.
15392 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15394 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
15395 Add enum values for atomic load/store and memory sync.
15396 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
15398 * config/nds32/nds32-intrinsic.md: Likewise.
15399 * config/nds32/nds32_intrinsic.h: Likewise.
15400 * config/nds32/nds32.h (nds32_builtins): Likewise.
15402 2018-04-07 Jakub Jelinek <jakub@redhat.com>
15404 PR tree-optimization/85257
15405 * fold-const.c (native_encode_vector): If not all elts could fit
15406 and off is -1, return 0 rather than offset.
15407 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
15408 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
15409 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
15410 adjust buffer in native_interpret_expr call.
15412 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15414 * config/nds32/constants.md (unspec_volatile_element): Add cache
15415 control enum values.
15416 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
15417 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
15418 * config/nds32/nds32.c (nds32_cctl_names): New.
15419 (nds32_print_operand): Handle cache control register names.
15420 * config/nds32/nds32.h (nds32_builtins): New enum values.
15421 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
15423 * config/nds32/nds32.md (type): Add mmu.
15424 * config/nds32/pipelines.md (simple_insn): Add mmu.
15426 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15428 * config/nds32/nds32.md (type): Remove call.
15429 * config/nds32/pipelines.md (simple_insn): Likewise.
15431 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15433 * config/nds32/constants.md (unspec_volatile_element): Add
15434 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
15435 UNSPEC_VOLATILE_FMFCFG.
15436 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
15437 description for fmfcfg and fmfcsr.
15438 (bdesc_1arg): Add fmtcsr.
15439 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
15440 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
15441 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
15442 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
15443 unspec_fmfcfg): New patterns.
15444 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
15445 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
15446 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
15447 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
15448 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
15449 __nds32__fmfcfg): Define.
15451 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15453 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
15454 intrinsic register names.
15455 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
15456 intrinsic register enum values and macros.
15458 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15460 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
15461 for load/store addressing form.
15462 (nds32_print_operand_address): Likewise.
15464 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
15467 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
15468 based on LABEL_REF. Remove useless assertion.
15469 (pic_address_needs_scratch): Fix formatting.
15470 (sparc_legitimize_pic_address): Minor tweaks.
15471 (sparc_delegitimize_address): Adjust assertion accordingly.
15472 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
15473 into symbolic_operand.
15474 (movsi_high_pic_label_ref): Likewise.
15475 (movsi_lo_sum_pic_label_ref): Likewise.
15476 (movdi_pic_label_ref): Likewise.
15477 (movdi_high_pic_label_ref): Likewise.
15478 (movdi_lo_sum_pic_label_ref): Likewise.
15480 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
15482 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
15483 custom LIB_SPEC setup.
15485 2018-04-06 Ruslan Bukin <br@bsdpad.com>
15486 Kito Cheng <kito.cheng@gmail.com>
15488 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
15489 * config/riscv/freebsd.h: New.
15491 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
15493 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
15494 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
15497 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
15498 Kito Cheng <kito.cheng@gmail.com>
15500 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
15501 nds32_output_call, nds32_symbol_binds_local_p): New functions.
15502 * config/nds32/nds32-protos.h (nds32_output_call,
15503 nds32_output_return): Declare.
15504 * config/nds32/nds32.md: Refine all the call and return patterns.
15506 2018-04-06 Jakub Jelinek <jakub@redhat.com>
15509 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
15510 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
15512 PR rtl-optimization/84872
15513 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
15514 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
15515 EDGE_CROSSING edge.
15517 2018-04-06 Tamar Christina <tamar.christina@arm.com>
15519 * expr.c (copy_blkmode_to_reg): Revert 254862.
15520 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
15522 2018-04-06 Richard Biener <rguenther@suse.de>
15524 PR middle-end/85244
15525 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
15526 after seeing a component reference with an adjacent field. Treat
15527 refs to arrays at struct end of external decls similar to
15528 refs to unconstrained commons.
15530 2018-04-06 Jakub Jelinek <jakub@redhat.com>
15533 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
15534 look through SAVE_EXPRs with non-side-effects argument. Adjust
15536 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
15539 2018-04-06 Richard Biener <rguenther@suse.de>
15541 PR middle-end/85180
15542 * alias.c (find_base_term): New wrapper around find_base_term
15543 unwinding CSELIB_VAL_PTR changes.
15544 (find_base_term): Do not restore CSELIB_VAL_PTR during the
15547 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15549 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
15551 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
15552 constant definitions.
15553 ("nop"): lr 0,0 -> nopr r0
15554 ("nop_lr0", "nop_lr1"): New insn definitions.
15556 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
15558 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
15559 NDS32_V3PUSH_AVAILABLE_P macro.
15561 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
15562 Chung-Ju Wu <jasonwucj@gmail.com>
15564 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
15565 (nds32*-*-*): Add float and fpu_config into supported_defaults.
15566 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
15567 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
15568 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
15569 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
15570 * config/nds32/constraints.md: New constraints and checking for hard
15571 float configuration.
15572 * config/nds32/iterators.md: New mode iterator and attribute for hard
15573 float configuration.
15574 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
15576 * config/nds32/nds32-fpu.md: New file.
15577 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
15578 deal with hard float code generation.
15579 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
15581 (abi_type, float_reg_number): New enum type.
15582 * config/nds32/nds32-predicates.c: New predicates for hard float.
15583 * config/nds32/nds32-protos.h: Declare functions for hard float.
15584 * config/nds32/nds32.c: Implementation for hard float configuration.
15585 * config/nds32/nds32.h: Definitions for hard float configuration.
15586 * config/nds32/nds32.md: Include hard float machine description and
15587 modify patterns for hard float configuration.
15588 * config/nds32/nds32.opt: New options for hard float configuration.
15589 * config/nds32/predicates.md: New predicates for hard float
15592 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
15594 * common/config/nds32/nds32-common.c
15595 (nds32_option_optimization_table): Enable -mreleax-hint by default.
15597 2018-04-05 Jakub Jelinek <jakub@redhat.com>
15599 PR middle-end/85195
15600 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
15601 CONSTRUCTOR_ELT (ctor, ...)->value.
15603 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
15606 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
15608 2018-04-05 Tom de Vries <tom@codesourcery.com>
15611 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
15614 2018-04-05 Shiva Chen <shiva0217@gmail.com>
15615 Kito Cheng <kito.cheng@gmail.com>
15617 * config/nds32/constraints.md (U33): Fine-tune checking condition.
15618 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
15619 * config/nds32/nds32.h (nds32_16bit_address_type): Add
15620 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
15622 2018-04-05 Shiva Chen <shiva0217@gmail.com>
15623 Kito Cheng <kito.cheng@gmail.com>
15625 * config/nds32/constraints.md (Ufe): New memory constraint.
15626 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
15627 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
15628 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
15630 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
15631 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
15633 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15635 * config/nds32/nds32.md: Use optimize_size in the condition for
15636 alu-shift instructions.
15638 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15640 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
15642 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15644 * config/nds32/nds32.md (negsi2): Refine pattern.
15646 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
15647 Chung-Ju Wu <jasonwucj@gmail.com>
15649 * config/nds32/iterators.md (shift_rotate): New code iterator.
15650 (shift): New code attribute.
15651 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
15652 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
15653 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
15654 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
15655 bit-wise operations.
15656 (andsi3, *andsi3): Ditto.
15657 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
15658 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
15659 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
15660 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
15661 nds32_ior_operand, nds32_xor_operand): New predicates.
15663 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15665 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
15666 (addsi3, subsi3): ... this.
15668 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15670 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
15672 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15674 * config/nds32/nds32.md: Adjust indention.
15676 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
15678 * config/nds32/nds32.md (feature): New attribute.
15680 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15682 * config/nds32/nds32.md (subtype): New attribute.
15684 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15687 * config/arm/arm-builtins.c (arm_expand_builtin): Change
15688 expansion to perform a bitwise AND of the argument followed by a
15689 boolean negation of the result.
15691 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
15693 PR rtl-optimization/84878
15694 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
15695 the basic block. Assert the use reference is not artificial and that
15696 it has an associated insn.
15698 2018-04-04 Michael Matz <matz@suse.de>
15700 * builtins.c (compute_objsize): Pass correct operand
15701 to array_at_struct_end_p.
15703 2018-04-04 Richard Biener <rguenther@suse.de>
15706 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
15707 from contexts for DINFO_LEVEL_TERSE and below.
15709 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
15711 * config/nds32/nds32-doubleword.md (move_<mode>): Require
15712 resiter_operand condition.
15713 * config/nds32/nds32.md (*move<mode>): Ditto.
15715 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
15716 Monk Chiang <sh.chiang04@gmail.com>
15718 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
15720 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15722 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
15724 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15725 Kito Cheng <kito.cheng@gmail.com>
15727 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
15728 nds32_cond_code_str, output_cond_branch,
15729 output_cond_branch_compare_zero, nds32_expand_cbranch,
15730 nds32_expand_cstore, nds32_expand_movcc,
15731 nds32_output_cbranchsi4_equality_zero,
15732 nds32_output_cbranchsi4_equality_reg,
15733 nds32_output_cbranchsi4_equality_reg_or_const_int,
15734 nds32_output_cbranchsi4_greater_less_zero: New functions.
15735 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
15736 nds32_expand_cstore, nds32_expand_movcc,
15737 nds32_output_cbranchsi4_equality_zero,
15738 nds32_output_cbranchsi4_equality_reg,
15739 nds32_output_cbranchsi4_equality_reg_or_const_int,
15740 nds32_output_cbranchsi4_greater_less_zero): Declare.
15741 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
15742 nds32_rimm11s_operand): New predicates.
15743 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
15744 * config/nds32/nds32.md: Rewrite all the branch and conditional move
15747 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
15749 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
15750 * config/nds32/nds32.md: Ditto.
15751 * config/nds32/pipelines.md: Ditto.
15753 2018-04-04 Richard Biener <rguenther@suse.de>
15755 PR tree-optimization/85168
15756 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
15757 propagating abnormals.
15759 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15761 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
15763 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15764 Kito Cheng <kito.cheng@gmail.com>
15766 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
15767 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
15768 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
15769 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
15770 * config/nds32/nds32.md (sibcall_internal): New.
15771 (sibcall_register): Remove.
15772 (sibcall_immediate): Remove.
15773 (sibcall_value_internal): New.
15774 (sibcall_value_register): Remove.
15775 (sibcall_value_immediate): Remove.
15776 * config/nds32/predicates.md (nds32_general_register_operand): New.
15777 (nds32_call_address_operand): New.
15779 2018-04-03 Jakub Jelinek <jakub@redhat.com>
15781 PR rtl-optimization/85167
15782 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
15783 bb_defs if *split_p, instead preinitialize it to NULL.
15785 PR tree-optimization/85156
15786 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
15787 evaluating the argument multiple times.
15789 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
15791 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
15793 (_mm_cvtpd_ps): Likewise.
15794 (_mm_cvttpd_epi32): Likewise.
15795 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
15796 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
15797 vector, pixel, and bool following altivec.h include.
15799 2018-04-03 Martin Sebor <msebor@redhat.com>
15801 * doc/extend.texi (Common Function Attributes): Clarify.
15802 (const attribute): Likewise.
15803 (pure attribute): Likewise.
15805 2018-04-03 Jakub Jelinek <jakub@redhat.com>
15808 * config/i386/i386.c (ix86_expand_vector_set): Use
15809 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
15811 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
15813 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
15814 instructions when changing rounding bits to preserve precision bits
15815 in the x87 control word.
15817 2018-04-03 Martin Liska <mliska@suse.cz>
15819 PR tree-optimization/82491
15820 * rtl.h (strip_offset_and_add): Replace += suboffset with
15821 poly_uint64 () + suboffset.
15823 2018-03-29 Martin Liska <mliska@suse.cz>
15824 Martin Jambor <mjambor@suse.cz>
15827 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
15828 param_type is not an integral or pointer type.
15830 2018-04-03 Richard Biener <rguenther@suse.de>
15832 * sese.h (recompute_all_dominators): Remove.
15834 2018-04-02 Martin Sebor <msebor@redhat.com>
15836 * doc/invoke.texi (-Wrestrict): Fix typos.
15838 2018-04-02 Jim Wilson <jimw@sifive.com>
15840 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
15841 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
15842 (<optab>di3, <optab>si3_extend): Likewise.
15843 (<optab>si3_mask, <optab>si3_mask_1): New.
15844 (<optab>di3_mask, <optab>di3_mask_1): New.
15845 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
15846 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
15847 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
15849 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
15851 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
15854 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
15856 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
15857 (nds32_canonicalize_comparison): New function.
15859 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
15860 Kito Cheng <kito.cheng@gmail.com>
15861 Kuan-Lin Chen <kuanlinchentw@gmail.com>
15863 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
15864 * config/nds32/constants.md (unspec_volatile_element): Add
15865 UNSPEC_VOLATILE_RELAX_GROUP.
15866 * config/nds32/nds32-relax-opt.c: New file.
15867 * config/nds32/nds32-predicates.c
15868 (nds32_symbol_load_store_p): New function.
15869 * config/nds32/nds32-protos.h
15870 (nds32_symbol_load_store_p): Declare function.
15871 (make_pass_nds32_relax_opt): Declare new rtl pass function.
15872 * config/nds32/nds32.c
15873 (nds32_register_pass): New function to register pass.
15874 (nds32_register_passes): New function to register passes.
15875 * config/nds32/nds32.md (relax_group): New pattern.
15876 * config/nds32/nds32.opt (mrelax-hint): New option.
15877 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
15879 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
15881 * config/nds32/t-nds32: Modify files dependency.
15883 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
15885 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
15886 (PROFILE_HOOK): Define its implementation.
15888 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
15890 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
15891 type and 32-bit size.
15893 2018-04-01 Jakub Jelinek <jakub@redhat.com>
15895 PR middle-end/85090
15896 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
15897 (V_128_256): New mode iterator.
15898 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
15899 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
15900 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
15902 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
15903 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
15905 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
15908 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
15909 NaN inputs correctly.
15911 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
15914 * config/rs6000/vsx.md (??r): New mode attribute.
15915 (*vsx_mov<mode>_64bit): Use it.
15916 (*vsx_mov<mode>_32bit): Likewise.
15918 2018-03-30 Martin Sebor <msebor@redhat.com>
15920 PR tree-optimization/84818
15921 * builtins.c (check_access): Use warning_n.
15923 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15926 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
15928 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
15931 2018-03-30 Julia Koval <julia.koval@intel.com>
15934 * x86-tune.def (movx, partial_reg_dependency): Enable for
15937 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
15939 PR inline-asm/84985
15940 * lra-constraints.c (process_alt_operands): Move setting
15941 this_alternative_matches below.
15943 2018-03-29 Martin Liska <mliska@suse.cz>
15946 * doc/invoke.texi: Document how LTO works with debug info.
15947 Describe auto-load support of binutils. Mention 'x86-64'
15948 as valid option value of -march option.
15950 2018-03-29 Jakub Jelinek <jakub@redhat.com>
15952 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
15955 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
15956 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
15957 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
15960 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
15963 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
15964 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
15965 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
15966 for RS6000_BTM_POWERPC64.
15967 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
15968 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
15969 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
15972 (DIVDEU): Likewise.
15974 2018-03-28 Carl Love <cel@us.ibm.com>
15977 2017-09-27 Carl Love <cel@us.ibm.com>
15979 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
15980 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
15981 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
15984 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15986 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
15987 instead of __vector bool.
15988 (_mm_max_pu8): Likewise.
15989 (_mm_min_pi16): Likewise.
15991 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
15994 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
15995 (DIVWEUO): Likewise.
15996 (DIVDEO): Likewise.
15997 (DIVDEUO): Likewise.
15998 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
15999 DIVWEUO and DIVDEUO.
16000 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
16001 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
16002 (div_extend): Likewise.
16003 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
16005 (__builtin_divweuo): Likewise.
16006 (__builtin_divdeo): Likewise.
16007 (__builtin_divdeuo): Likewise.
16009 2018-03-28 Jakub Jelinek <jakub@redhat.com>
16012 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
16013 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
16015 PR tree-optimization/82004
16016 * gimple-match-head.c (optimize_pow_to_exp): New function.
16017 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
16018 Don't fold to exp if optimize_pow_to_exp is false.
16020 2018-03-28 Martin Liska <mliska@suse.cz>
16023 * calls.c (initialize_argument_information): Fix trailing space.
16024 * common.opt: Fix typo and provide better explanation for
16025 -fsanitize-coverage option.
16026 * config/i386/i386.opt: Fix typo.
16028 2018-03-28 Jakub Jelinek <jakub@redhat.com>
16029 Martin Liska <mliska@suse.cz>
16032 * gimplify.c (asan_poison_variable): Don't do the check for
16033 gimplify_omp_ctxp here.
16034 (gimplify_decl_expr): Do it here.
16035 (gimplify_target_expr): Likewise.
16037 2018-03-28 Martin Liska <mliska@suse.cz>
16040 * config/i386/i386.c (ix86_function_arg_advance): Do not call
16041 chkp_type_bounds_count if MPX is not enabled.
16043 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
16045 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
16047 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
16050 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
16051 function to create the function decl for complex long double
16052 multiply and divide for -mabi=ieeelongdouble.
16053 (init_float128_ieee): Call it.
16055 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
16058 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
16059 -fcf-protection=branch -mibt.
16060 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
16062 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16065 * config/arm/arm.c (arm_valid_symbolic_address): Handle
16066 arm_word_relocations.
16068 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
16071 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
16072 extern array declarations.
16074 2018-03-27 Richard Biener <rguenther@suse.de>
16076 PR middle-end/84067
16077 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
16078 explicit single_use checks.
16080 2018-03-27 Richard Biener <rguenther@suse.de>
16082 PR tree-optimization/85082
16083 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
16086 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16088 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
16089 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
16090 Turn on fasynchronous-unwind-tables and funwind-tables.
16092 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
16095 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
16096 (*bmi_blsr_<mode>_ccz): Ditto.
16098 2018-03-26 Tom de Vries <tom@codesourcery.com>
16100 PR tree-optimization/85063
16101 * omp-general.c (offloading_function_p): New function. Factor out
16103 * omp-offload.c (pass_omp_target_link::gate): ... here.
16104 * omp-general.h (offloading_function_p): Declare.
16105 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
16106 with attribute omp declare target for offloading functions.
16108 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
16110 PR tree-optimization/84005
16111 * tree-data-ref.h (get_base_for_alignment): Declare.
16112 * tree-data-ref.c (get_base_for_alignment_1): New function.
16113 (get_base_for_alignment): Likewise.
16114 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
16115 get_base_for_alignment to find a suitable base object, instead
16116 of always using drb->base_address.
16118 2018-03-23 Jakub Jelinek <jakub@redhat.com>
16120 PR inline-asm/85022
16121 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
16122 known size by default.
16124 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
16126 PR inline-asm/85030
16127 * lra-constraints.c (process_alt_operands): Don't match BLKmode
16128 and non BLKmode operands.
16130 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16133 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
16134 Clean up attributes.
16136 2018-03-23 Richard Biener <rguenther@suse.de>
16139 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
16140 we are going to emit early debug for LTO.
16142 2018-03-23 Jakub Jelinek <jakub@redhat.com>
16144 PR inline-asm/85034
16145 * function.c (match_asm_constraints_1): Don't optimize if input
16146 doesn't satisfy general_operand predicate for output's mode.
16148 PR inline-asm/85022
16149 * alias.c (write_dependence_p): Don't require for x_canonicalized
16150 non-VOIDmode if x has VOIDmode.
16153 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
16154 just don't try to optimize it rather than assert it never happens.
16156 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
16158 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
16159 macro expansions for definition of ST_INTERNAL_<mode> and
16160 LD_INTERNAL_<mode> builtins.
16161 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
16163 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
16165 (altivec_expand_st_builtin): Likewise.
16166 (altivec_expand_builtin): Remove calls to deleted functions.
16167 (rs6000_address_for_altivec): Delete this function.
16168 * config/rs6000/vector.md: Remove expands for
16169 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
16171 2018-03-22 Sudakshina Das <sudi.das@arm.com>
16174 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
16175 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
16176 re-computing once computed.
16177 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
16178 (arm_init_machine_status): Initialize
16179 machine->static_chain_stack_bytes.
16181 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
16184 * doc/extend.texi: Add four new prototypes for vec_ld.
16185 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
16186 definitions for more logical presentation.
16187 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
16188 entries for V1TI variants of __builtin_altivec_ld builtin.
16189 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
16190 handling of V1TI variant of LVX icode pattern.
16191 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
16192 (rs6000_gimple_fold_builtin): Likewise.
16193 (altivec_init_builtins): Add code to define
16194 __builtin_altivec_lvx_v1ti function.
16196 2018-03-22 Jakub Jelinek <jakub@redhat.com>
16198 PR inline-asm/84941
16199 * function.c (match_asm_constraints_1): Don't do the optimization
16200 if input isn't a REG, SUBREG, MEM or constant.
16202 2018-03-22 Tom de Vries <tom@codesourcery.com>
16204 PR tree-optimization/84956
16205 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
16206 bb_has_abnormal_pred.
16208 2018-03-22 Jakub Jelinek <jakub@redhat.com>
16211 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
16212 DECL_INITIAL (decl) to decl at the end.
16213 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
16214 adjust the comment.
16216 2018-03-21 Joseph Myers <joseph@codesourcery.com>
16218 * doc/extend.texi (__builtin_tgmath): Document when complex
16219 integer types are treated as _Complex _Float64.
16221 2018-03-21 Tom de Vries <tom@codesourcery.com>
16223 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
16225 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16227 PR tree-optimization/84960
16228 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
16229 if it is ENTRY block, move them into single succ of ENTRY in that case.
16231 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
16233 PR tree-optimization/84811
16234 * poly-int.h (poly_span_traits): Remove the T3 parameter and
16235 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
16236 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
16237 (known_subrange_p): Update accordingly. Cast each value involved
16238 in the size comparison, rather than casting the result of the
16241 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16243 PR tree-optimization/84982
16244 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
16245 by flipping the least significant bit rather than all bits from
16246 bitpos to bitpos + bitsize - 1.
16248 2018-03-21 Nathan Sidwell <nathan@acm.org>
16250 * doc/extend.texi (Deprecated Features): Remove mention of
16251 long-deleted deprecations.
16253 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16256 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
16257 * configure: Regenerate.
16259 2018-03-21 Tom de Vries <tom@codesourcery.com>
16261 PR tree-optimization/83126
16262 * tree-parloops.c (num_phis): New function.
16263 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
16265 2018-03-21 Nathan Sidwell <nathan@acm.org>
16267 * doc/extend.texi (Deprecated Features): Update deprecated flags,
16268 mention anon-struct/union members and trailing attributes.
16270 2018-03-21 Bin Cheng <bin.cheng@arm.com>
16272 PR tree-optimization/84969
16273 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
16274 builtin memset partitions if they set different rhs values.
16276 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16278 PR rtl-optimization/84989
16279 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
16280 VEC_DUPLICATE with scalar result mode.
16282 2018-03-21 Martin Liska <mliska@suse.cz>
16285 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
16286 not intended return statement.
16288 2018-03-21 Martin Liska <mliska@suse.cz>
16291 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
16292 (chkp_find_bound_slots_1): Limit number of iterations.
16294 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
16297 * Minor grammar fixes for x86 options.
16299 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16302 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
16303 holding REG_CFA_RESTORE notes, instead turn them into a USE.
16305 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
16308 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
16309 (altivec_lvx_<mode>_1op): Likewise.
16310 (altivec_stvx_<mode>_2op): Likewise.
16311 (altivec_stvx_<mode>_1op): Likewise.
16312 (altivec_lvx_<VM2:mode>): New define_expand.
16313 (altivec_stvx_<VM2:mode>): Likewise.
16314 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
16315 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
16316 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
16317 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
16318 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
16319 (rs6000_gen_lvx): Likewise.
16320 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
16321 (altivec_expand_stv_builtin): Likewise.
16322 (altivec_expand_builtin): Likewise.
16323 * config/rs6000/vector.md: Likewise.
16325 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16328 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
16331 2018-03-20 Richard Biener <rguenther@suse.de>
16334 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
16335 sign-conversions as zero, fall back to standard scalar_stmt
16338 2018-03-20 Martin Liska <mliska@suse.cz>
16341 * predict.c (rebuild_frequencies): Handle case when we have
16342 PROFILE_ABSENT, but flag_guess_branch_prob is false.
16344 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16347 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
16348 flag_section_anchors.
16349 * varasm.c (use_blocks_for_decl_p): Remove hack for
16350 dw2_force_const_mem.
16353 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
16355 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
16356 be created, use lowpart_subreg of operands[0] rather than operands[0]
16358 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
16359 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
16360 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
16361 and n constraint instead of aarch64_shift_imm_di and Usd.
16362 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
16363 (*aarch64_<optab>_reg_minus<mode>3): ... this.
16365 2018-03-20 Sudakshina Das <sudi.das@arm.com>
16368 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
16369 to favor GPR over NEON registers.
16370 (<shift>di3_neon): Likewise.
16372 2018-03-20 Tom de Vries <tom@codesourcery.com>
16375 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
16376 (nvptx_process_pars): Emit bar.sync asap and alap.
16378 2018-03-20 Tom de Vries <tom@codesourcery.com>
16381 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
16382 seen_label if seen_label is already set.
16384 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16387 * config/i386/i386.c (fold_builtin_cpu): For features above 31
16388 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
16389 Use 1U instead of 1. Formatting fixes.
16392 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
16393 instead of TREE_TYPE (s1) for the return value.
16395 2018-03-19 Jakub Jelinek <jakub@redhat.com>
16397 PR tree-optimization/84946
16398 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
16399 bitsize + bitsize in poly_uint64 rather than poly_int64.
16402 * dwarf2asm.c: Include fold-const.c.
16403 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
16404 of decl rather than decl itself.
16406 PR rtl-optimization/84643
16407 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
16409 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
16412 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
16413 calling assemble_variable.
16415 2018-03-19 Sudakshina Das <sudi.das@arm.com>
16418 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
16419 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
16421 2018-03-19 Jim Wilson <jimw@sifive.com>
16424 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
16425 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
16426 (riscv_first_stack_step): Likewise.
16427 (riscv_option_override): Use STACK_BOUNDARY instead of
16428 MIN_STACK_BOUNDARY.
16429 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
16430 MIN_STACK_BOUNDARY.
16431 (BIGGEST_ALIGNMENT): Set to 128.
16432 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
16433 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
16436 2018-03-19 Richard Biener <rguenther@suse.de>
16438 PR tree-optimization/84933
16439 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
16440 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
16442 2018-03-19 Richard Biener <rguenther@suse.de>
16444 PR tree-optimization/84859
16445 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
16446 (cond_if_else_store_replacement): Perform sinking operation on
16447 single-store BBs regardless of MAX_STORES_TO_SINK setting.
16448 Generalize what a BB with a single eligible store is.
16450 2018-03-19 Richard Biener <rguenther@suse.de>
16452 PR tree-optimization/84929
16453 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
16454 chrec_is_positive against non-chrec arg.
16456 2018-03-19 Tamar Christina <tamar.christina@arm.com>
16459 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
16461 2018-03-18 Martin Liska <mliska@suse.cz>
16463 PR rtl-optimization/84635
16464 * regrename.c (build_def_use): Use matches_mode only when
16467 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
16469 PR tree-optimization/84913
16470 * tree-vect-loop.c (vectorizable_reduction): Don't try to
16471 vectorize chains of COND_EXPRs.
16473 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
16475 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
16477 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
16479 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
16481 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
16483 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
16485 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
16486 Kito Cheng <kito.cheng@gmail.com>
16488 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
16489 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
16490 (nds32_adjust_reg_alloc_order): New function.
16491 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
16493 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
16495 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
16496 nds32_print_operand, nds32_print_operand_address): Use
16497 HOST_WIDE_INT_PRINT_DEC instead.
16499 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
16501 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
16503 2018-03-17 Jakub Jelinek <jakub@redhat.com>
16506 * config/i386/i386.c (initial_ix86_tune_features,
16507 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
16508 unsigned long long.
16509 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
16510 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
16511 rather than 1u << ix86_tune. Formatting fix.
16512 (ix86_option_override_internal): Change ix86_arch_mask from
16513 unsigned int to unsigned HOST_WIDE_INT, initialize to
16514 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
16515 (ix86_function_specific_restore): Likewise.
16517 2018-03-16 Jakub Jelinek <jakub@redhat.com>
16520 * postreload.c (reload_combine_recognize_pattern): Perform
16521 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
16522 truncate_int_for_mode the result for the destination's mode.
16525 * hsa-gen.c (mem_type_for_type): Fix comment typo.
16526 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
16528 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
16531 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
16534 * lra-assigns.c (lra_split_hard_reg_for): Don't use
16535 regno_allocno_class_array and sorted_pseudos.
16536 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
16537 insns where regno is used.
16539 2018-03-16 Martin Liska <mliska@suse.cz>
16542 * multiple_target.c (create_dispatcher_calls): Redirect
16543 reference in the symbol table.
16545 2018-03-16 Martin Liska <mliska@suse.cz>
16548 * multiple_target.c (create_dispatcher_calls): Redirect also
16551 2018-03-16 Jakub Jelinek <jakub@redhat.com>
16555 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
16556 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
16557 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
16559 2018-03-16 Julia Koval <julia.koval@intel.com>
16561 * doc/invoke.texi (Skylake Server): Add CLWB.
16562 Cannonlake): Remove CLWB.
16564 2018-03-16 Jakub Jelinek <jakub@redhat.com>
16566 PR tree-optimization/84841
16567 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
16569 (FLOAT_ONE_CONST_TYPE): Define.
16570 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
16571 (sort_by_operand_rank): Put entries with higher constant_type last
16572 rather than first to match comments.
16574 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
16576 * config/nios2/nios2.md (movsi_internal): Fix thinko in
16579 2018-03-15 Jakub Jelinek <jakub@redhat.com>
16582 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
16583 check and use address of target always.
16585 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
16588 * config/i386/i386.c (indirect_thunk_needed): Update comments.
16589 (indirect_thunk_bnd_needed): Likewise.
16590 (indirect_thunks_used): Likewise.
16591 (indirect_thunks_bnd_used): Likewise.
16592 (indirect_return_needed): New.
16593 (indirect_return_bnd_needed): Likewise.
16594 (output_indirect_thunk_function): Add a bool argument for
16596 (output_indirect_thunk_function): Don't generate alias for
16597 function return thunk.
16598 (ix86_code_end): Call output_indirect_thunk_function to generate
16599 function return thunks.
16600 (ix86_output_function_return): Set indirect_return_bnd_needed
16601 and indirect_return_needed instead of indirect_thunk_bnd_needed
16602 and indirect_thunk_needed.
16604 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
16606 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
16607 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
16608 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
16610 2018-03-15 David Malcolm <dmalcolm@redhat.com>
16611 Paul Hua <paul.hua.gm@gmail.com>
16614 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
16616 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
16618 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
16619 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
16620 resp. SFmode cases.
16622 2018-03-15 Tamar Christina <tamar.christina@arm.com>
16625 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
16626 instead of GET_MODE_SIZE when comparing Units.
16628 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
16631 * varasm.c (hash_section): Return an unchangeble hash value
16632 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
16633 Return !aarch64_can_use_per_function_literal_pools_p ().
16635 2018-03-15 Jakub Jelinek <jakub@redhat.com>
16638 * optabs.c (emit_conditional_move): Pass address of cmode's copy
16639 rather than address of cmode as last argument to prepare_cmp_insn.
16641 2018-03-15 Julia Koval <julia.koval@intel.com>
16643 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
16644 F_AVX512VNNI, F_AVX512BITALG): New.
16646 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
16649 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
16650 insn for floating-point loads and stores.
16652 2018-03-14 Carl Love <cel@us.ibm.com>
16654 * config/rs6000/rs6000-c.c: Add macro definitions for
16655 ALTIVEC_BUILTIN_VEC_PERMXOR.
16656 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
16657 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
16658 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
16659 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
16661 * config/doc/extend.texi: Add prototypes for vec_permxor.
16663 2018-03-14 David Malcolm <dmalcolm@redhat.com>
16666 * diagnostic-show-locus.c (class layout_point): Convert m_line
16667 from int to linenum_type.
16668 (line_span::comparator): Use linenum "compare" function when
16669 comparing line numbers.
16670 (test_line_span): New function.
16671 (layout_range::contains_point): Convert param "row" from int to
16673 (layout_range::intersects_line_p): Likewise.
16674 (layout::will_show_line_p): Likewise.
16675 (layout::print_source_line): Likewise.
16676 (layout::should_print_annotation_line_p): Likewise.
16677 (layout::print_annotation_line): Likewise.
16678 (layout::print_leading_fixits): Likewise.
16679 (layout::annotation_line_showed_range_p): Likewise.
16680 (struct line_corrections): Likewise for field m_row.
16681 (line_corrections::line_corrections): Likewise for param "row".
16682 (layout::print_trailing_fixits): Likewise.
16683 (layout::get_state_at_point): Likewise.
16684 (layout::get_x_bound_for_row): Likewise.
16685 (layout::print_line): Likewise.
16686 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
16687 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
16688 * input.c (selftest::test_linenum_comparisons): New function.
16689 (selftest::input_c_tests): Call it.
16690 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
16691 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
16692 * selftest.h (ASSERT_GT): New macro.
16693 (ASSERT_GT_AT): New macro.
16694 (ASSERT_LT): New macro.
16695 (ASSERT_LT_AT): New macro.
16697 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
16699 PR rtl-optimization/84780
16700 * combine.c (distribute_links): Don't make a link based on pc_rtx.
16702 2018-03-14 Martin Liska <mliska@suse.cz>
16704 * tree.c (record_node_allocation_statistics): Use
16705 get_stats_node_kind.
16706 (get_stats_node_kind): New function extracted from
16707 record_node_allocation_statistics.
16708 (free_node): Use get_stats_node_kind.
16710 2018-03-14 Richard Biener <rguenther@suse.de>
16712 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
16713 that the value-set of ANTIC_IN doesn't grow.
16716 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
16718 (BB_VISITED_WITH_VISITED_SUCCS): New define.
16719 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
16721 2018-03-14 Julia Koval <julia.koval@intel.com>
16723 * config.gcc (icelake-client, icelake-server): New.
16725 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
16726 (initial_ix86_arch_features): Ditto.
16727 (PTA_SKYLAKE): Add SGX.
16728 (PTA_ICELAKE): Remove.
16729 (PTA_ICELAKE_CLIENT): New.
16730 (PTA_ICELAKE_SERVER): New.
16731 (ix86_option_override_internal): Split up icelake on icelake client and
16733 (get_builtin_code_for_version): Ditto.
16734 (fold_builtin_cpu): Ditto.
16735 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
16736 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
16737 * config/i386/i386.h (processor_type): Ditto.
16738 * doc/invoke.texi: Ditto.
16740 2018-03-14 Jakub Jelinek <jakub@redhat.com>
16743 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
16744 INTEGER_CST offset, add it together with bitpos / 8 and
16745 sign extend based on POINTER_SIZE.
16749 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
16752 * config/i386/constraints.md (Yc): New register constraint.
16753 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
16754 Use Yc constraint for alternative 2 of operand 0. Remove
16755 preferred_for_speed attribute.
16757 2018-03-14 Richard Biener <rguenther@suse.de>
16759 PR tree-optimization/84830
16760 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
16761 with the old one to avoid oscillations.
16763 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
16766 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
16768 (assign_by_spills): Return a flag of reload assignment failure.
16769 Do not process the reload assignment failures. Do not spill other
16770 reload pseudos if they has the same reg class. Update n if
16772 (lra_assign): Add a return arg. Set up from the result of
16773 assign_by_spills call.
16774 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
16775 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
16776 usage_insns if it is not NULL.
16777 (spill_hard_reg_in_range): New function.
16778 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
16779 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
16780 function prototypes.
16781 (lra_assign): Change prototype.
16782 * lra.c (lra): Add code to deal with fails by splitting hard reg
16785 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
16787 * config/riscv/riscv.opt (mrelax): New option.
16788 * config/riscv/riscv.c (riscv_file_start): Emit ".option
16789 "norelax" when riscv_mrelax is disabled.
16790 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
16792 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16795 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
16796 reassociation for int modes.
16798 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
16800 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
16801 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
16803 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
16804 * config/aarch64/aarch64-sve.md
16805 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
16806 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
16807 (*extend<mode><Vwide>2): Rename to...
16808 (aarch64_sve_extend<mode><Vwide>2): ...this.
16809 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
16810 renaming the old pattern to...
16811 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
16813 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
16814 define_expand, renaming the old pattern to...
16815 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
16816 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
16817 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
16818 account when deciding which SVE instruction the optab should use.
16819 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
16821 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
16823 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
16824 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
16825 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
16826 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
16827 (tlsdesc_small_<mode>): Turn a define_expand and use
16828 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
16829 (tlsdesc_small_advsimd_<mode>): ...this.
16830 (tlsdesc_small_sve_<mode>): New pattern.
16832 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
16834 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
16835 (UNSPEC_UMUL_HIGHPART): New constants.
16836 (MUL_HIGHPART): New int iteraor.
16837 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
16838 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
16840 (*<su>mul<mode>3_highpart): New define_insn.
16842 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
16845 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
16848 2018-03-13 Martin Liska <mliska@suse.cz>
16851 * (sem_item_optimizer::sem_item_optimizer): Initialize new
16853 (sem_item_optimizer::~sem_item_optimizer): Release it.
16854 (sem_item_optimizer::merge_classes): Register variable aliases.
16855 (sem_item_optimizer::fixup_pt_set): New function.
16856 (sem_item_optimizer::fixup_points_to_sets): Likewise.
16857 * ipa-icf.h: Declare new variables and functions.
16859 2018-03-13 Jakub Jelinek <jakub@redhat.com>
16861 PR middle-end/84834
16862 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
16863 integer_pow2p@2 and test integer_pow2p in condition.
16864 (A < 0 ? C : 0): Similarly for @1.
16866 PR middle-end/84831
16867 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
16868 characters starting at p contain '\0' character, don't look beyond
16872 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
16873 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
16876 * reg-stack.c (change_stack): Change update_end var from int to
16877 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
16878 also call set_block_for_insn on the newly added insns and rescan.
16881 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
16882 on the last operand.
16885 * tree.c (stabilize_reference_1): Return save_expr (e) for
16886 STATEMENT_LIST even if it doesn't have side-effects.
16888 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
16890 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
16892 2018-03-12 Renlin Li <renlin.li@arm.com>
16894 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
16895 aarch64_output_scalar_simd_mov_immediate.
16897 2018-03-12 Martin Sebor <msebor@redhat.com>
16899 PR tree-optimization/83456
16900 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
16901 for perfectly overlapping calls to memcpy.
16902 (gimple_fold_builtin_memory_chk): Same.
16903 (gimple_fold_builtin_strcpy): Handle no-warning.
16904 (gimple_fold_builtin_stxcpy_chk): Same.
16905 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
16907 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
16909 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
16910 parameter. Use it for SFmode.
16911 (rs6000_function_arg_advance_1): Adjust.
16912 (rs6000_function_arg): Adjust.
16913 (rs6000_gimplify_va_arg): Pass false for that new parameter.
16915 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
16917 PR rtl-optimization/84169
16918 PR rtl-optimization/84780
16919 * combine.c (can_combine_p): Check for a 2-insn combination whether
16920 the destination register is used between the two insns, too.
16922 2018-03-12 Richard Biener <rguenther@suse.de>
16924 PR tree-optimization/84803
16925 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
16926 for refs DR analysis didn't process.
16928 2018-03-12 Richard Biener <rguenther@suse.de>
16930 PR tree-optimization/84777
16931 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
16932 force-vectorize loops ignore whether we are optimizing for size.
16934 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
16936 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
16937 (TARGET_MD_ASM_ADJUST): Define.
16939 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
16940 Kito Cheng <kito.cheng@gmail.com>
16941 Chung-Ju Wu <jasonwucj@gmail.com>
16943 * config/nds32/nds32.c (nds32_compute_stack_frame,
16944 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
16945 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
16946 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
16947 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
16948 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
16949 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
16950 * config/nds32/nds32.md (prologue, epilogue): Use macro
16951 NDS32_V3PUSH_AVAILABLE_P to do checking.
16953 2018-03-11 Jakub Jelinek <jakub@redhat.com>
16956 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
16957 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
16958 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
16959 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
16960 addition of most attributes on !orig_type_die or the attribute not
16961 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
16963 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
16964 Chung-Ju Wu <jasonwucj@gmail.com>
16966 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
16967 __NDS32_VH__ macro.
16968 * config/nds32/nds32.opt (mvh): New option.
16970 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
16971 Chung-Ju Wu <jasonwucj@gmail.com>
16973 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
16975 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
16976 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
16979 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
16980 Chung-Ju Wu <jasonwucj@gmail.com>
16982 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
16984 * config/nds32/nds32-multiple.md (strlensi): New pattern.
16985 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
16987 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
16988 Kito Cheng <kito.cheng@gmail.com>
16989 Chung-Ju Wu <jasonwucj@gmail.com>
16991 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
16992 UNSPEC_FFMISM and UNSPEC_FLMISM.
16993 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
16994 for ffb, ffmism and flmism.
16995 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
16996 (unspec_ffmism): Ditto.
16997 (unspec_flmism): Ditto.
16998 (nds32_expand_builtin_impl): Check if string extension is available.
16999 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
17000 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
17002 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
17005 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
17008 * lra-assigns.c (assign_by_spills): Return a flag of reload
17009 assignment failure. Do not process the reload assignment
17010 failures. Do not spill other reload pseudos if they has the same
17012 (lra_assign): Add a return arg. Set up from the result of
17013 assign_by_spills call.
17014 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
17015 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
17016 usage_insns if it is not NULL.
17017 (spill_hard_reg_in_range): New function.
17018 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
17019 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
17020 function prototypes.
17021 (lra_assign): Change prototype.
17022 * lra.c (lra): Add code to deal with fails by splitting hard reg
17025 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
17028 * config/i386/i386.opt: Replace Enforcment with Enforcement.
17030 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
17033 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
17034 (dw_val_node): Add val_symbolic_view.
17035 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
17036 (symview_upper_bound): New.
17037 (new_line_info_table): Initialize symviews_since_reset.
17038 (dwarf2out_source_line): Count symviews_since_reset and set
17039 symview_upper_bound.
17040 (dw_val_equal_p): Handle symview.
17041 (add_AT_symview): New.
17042 (print_dw_val): Handle symview.
17043 (attr_checksum, attr_checksum_ordered): Likewise.
17044 (same_dw_val_p, size_of_die): Likewise.
17045 (value_format, output_die): Likewise.
17046 (add_high_low_attributes): Use add_AT_symview for entry_view.
17047 (dwarf2out_finish): Reset symview_upper_bound, clear
17050 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
17053 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
17054 Add strict argument and use it.
17055 (rs6000_split_multireg_move): Update for new strict argument.
17056 (mem_operand_gpr): Disallow all non-offsettable addresses.
17057 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
17059 2018-03-09 Jakub Jelinek <jakub@redhat.com>
17062 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
17063 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
17064 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
17067 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
17068 decl, use remap_type if we want to use the type.
17070 2018-03-09 Martin Sebor <msebor@redhat.com>
17072 PR tree-optimization/84526
17073 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
17075 (builtin_access::generic_overlap): Be prepared to handle non-array
17078 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
17080 PR rtl-optimization/84682
17081 * lra-constraints.c (process_address_1): Check is_address flag
17082 for address constraints.
17083 (process_alt_operands): Likewise.
17084 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
17085 preprocess_constraints.
17086 * recog.h (preprocess_constraints): Add oploc parameter.
17088 * recog.c (preprocess_constraints): Test address_operand for
17089 CT_ADDRESS constraints.
17091 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
17094 * lra-assigns.c (assign_by_spills): Return a flag of reload
17095 assignment failure. Do not process the reload assignment
17096 failures. Do not spill other reload pseudos if they has the same
17098 (lra_assign): Add a return arg. Set up from the result of
17099 assign_by_spills call.
17100 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
17101 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
17102 usage_insns if it is not NULL.
17103 (spill_hard_reg_in_range): New function.
17104 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
17105 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
17106 function prototypes.
17107 (lra_assign): Change prototype.
17108 * lra.c (lra): Add code to deal with fails by splitting hard reg
17111 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17114 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
17115 Accept complain bool parameter. Only emit errors if it is true.
17116 (arm_parse_cpu_option_name): Likewise.
17117 (arm_target_thumb_only): Adjust callers of the above.
17118 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
17119 prototype to take a default true bool parameter.
17120 (arm_parse_arch_option_name): Likewise.
17122 2018-03-09 David Malcolm <dmalcolm@redhat.com>
17123 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
17127 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
17128 * configure: Regenerate.
17129 * configure.ac ("linker --version-script option"): New.
17130 ("linker soname option"): New.
17132 2018-03-09 Richard Biener <rguenther@suse.de>
17134 PR tree-optimization/84775
17135 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
17136 immediate uses of predicate stmts and mark them modified.
17139 PR tree-optimization/84178
17140 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
17142 (version_loop_for_if_conversion): Delay update_ssa call.
17143 (tree_if_conversion): Delay update_ssa until after predicate
17146 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
17149 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
17150 when the function accesses prior frames.
17152 2018-03-08 Jakub Jelinek <jakub@redhat.com>
17155 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
17156 gen_llsym, otherwise call maybe_gen_llsym.
17158 PR inline-asm/84742
17159 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
17160 has ',' character inside of it.
17162 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17165 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
17166 as clobbering CC_REGNUM.
17168 2018-03-08 Richard Biener <rguenther@suse.de>
17170 PR middle-end/84552
17171 * tree-scalar-evolution.c: Include tree-into-ssa.h.
17172 (follow_copies_to_constant): Do not follow SSA names registered
17175 2018-03-08 Richard Biener <rguenther@suse.de>
17177 PR tree-optimization/84178
17178 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
17180 (version_loop_for_if_conversion): Delay update_ssa call.
17181 (tree_if_conversion): Delay update_ssa until after predicate
17184 2018-03-08 David Malcolm <dmalcolm@redhat.com>
17186 PR tree-optimization/84178
17187 * tree-if-conv.c (release_bb_predicate): Remove the
17188 the assertion that the stmts have NULL use_ops.
17189 Discard the statements, asserting that they haven't
17190 yet been added to a BB.
17192 2018-03-08 Richard Biener <rguenther@suse.de>
17194 PR tree-optimization/84746
17195 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
17196 (phi_translate): Pass in destination ANTIC_OUT set.
17197 (phi_translate_1): Likewise. For a simplified result lookup
17198 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
17199 (phi_translate_set): Adjust.
17200 (do_pre_regular_insertion): Likewise.
17201 (do_pre_partial_partial_insertion): Likewise.
17203 2018-03-08 Martin Liska <mliska@suse.cz>
17205 PR gcov-profile/84735
17206 * doc/gcov.texi: Document usage of profile files.
17207 * gcov-io.h: Document changes in the format.
17209 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
17213 * dwarf2out.c (struct dw_line_info_table): Update comments for
17215 (FORCE_RESET_NEXT_VIEW): New.
17216 (FORCE_RESETTING_VIEW_P): New.
17217 (RESETTING_VIEW_P): Check for -1 too.
17218 (ZERO_VIEW_P): Likewise.
17219 (new_line_info_table): Force-reset next view.
17220 (dwarf2out_begin_function): Likewise.
17221 (dwarf2out_source_line): Simplify zero_view_p initialization.
17222 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
17223 view directly. Omit view when omitting .loc at line 0.
17225 2018-03-08 Jakub Jelinek <jakub@redhat.com>
17227 PR tree-optimization/84740
17228 * tree-switch-conversion.c (process_switch): Call build_constructors
17229 only if info.phi_count is non-zero.
17231 PR tree-optimization/84739
17232 * tree-tailcall.c (find_tail_calls): Check call arguments against
17233 DECL_ARGUMENTS (current_function_decl) rather than
17234 DECL_ARGUMENTS (func) when checking for tail recursion.
17236 2018-03-07 Jakub Jelinek <jakub@redhat.com>
17238 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
17239 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
17240 Volker Reichelt's entry and add entries for people that perform
17241 GCC fuzzy testing and report numerous bugs.
17243 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
17246 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
17247 readonly data in sdata, if that is disabled.
17248 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
17249 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
17250 -mreadonly-in-sdata option.
17252 2018-03-07 Martin Sebor <msebor@redhat.com>
17254 PR tree-optimization/84468
17255 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
17256 basic block when looking for nul assignment.
17258 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
17261 * except.h (output_function_exception_table): Adjust prototype.
17262 * except.c (output_function_exception_table): Remove FNNAME parameter
17263 and add SECTION parameter. Ouput one part of the table at a time.
17264 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
17265 the first part of the exception table and emit unwind directives.
17266 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
17267 (i386_pe_seh_cold_init): Likewise.
17268 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
17269 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
17270 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
17271 (ix86_output_call_insn): Emit a nop in one more case for SEH.
17272 * config/i386/winnt.c: Include except.h.
17273 (struct seh_frame_state): Add reg_offset, after_prologue and
17274 in_cold_section fields.
17275 (i386_pe_seh_end_prologue): Set seh->after_prologue.
17276 (i386_pe_seh_cold_init): New function.
17277 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
17278 to seh->in_cold_section.
17279 (seh_emit_push): Record the offset of the push.
17280 (seh_emit_save): Record the offet of the save.
17281 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
17282 Test seh->after_prologue to disregard the epilogue.
17283 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
17284 (i386_pe_end_cold_function): New function.
17286 2018-03-07 Jakub Jelinek <jakub@redhat.com>
17289 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
17290 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
17293 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
17295 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
17296 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
17298 PR middle-end/84723
17299 * multiple_target.c: Include tree-inline.h and intl.h.
17300 (expand_target_clones): Diagnose and fail if node->definition and
17301 !tree_versionable_function_p (node->decl).
17303 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
17305 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
17307 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
17308 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17309 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
17311 2018-03-06 Jakub Jelinek <jakub@redhat.com>
17314 * combine.c (try_combine): Use reg_or_subregno instead of handling
17315 just paradoxical SUBREGs and REGs.
17317 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
17319 * config/arc/arc.c (arc_finalize_pic): Remove function.
17320 (arc_must_save_register): We use single base PIC register, remove
17321 checks to save/restore the PIC register.
17322 (arc_expand_prologue): Likewise.
17323 * config/arc/arc-protos.h (arc_set_default_type_attributes):
17325 (arc_verify_short): Likewise.
17326 (arc_attr_type): Likewise.
17327 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
17328 (walk_stores): Likewise.
17329 (arc_address_cost): Make it static.
17330 (arc_verify_short): Likewise.
17331 (branch_dest): Likewise.
17332 (arc_attr_type): Likewise.
17333 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
17334 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
17335 (arc_final_prescan_insn): Remove inserting the nops due to
17336 hardware hazards. It is done in reorg step.
17337 (insn_length_variant_t): Remove.
17338 (insn_length_parameters_t): Likewise.
17339 (arc_insn_length_parameters): Likewise.
17340 (arc_get_insn_variants): Likewise.
17341 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
17343 2018-03-06 Jakub Jelinek <jakub@redhat.com>
17345 PR inline-asm/84683
17346 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
17349 PR tree-optimization/84687
17350 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
17352 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
17354 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17356 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
17357 Rename to ppc_speculation_barrier.
17358 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
17359 __builtin_ppc_speculation_barrier.
17361 2018-03-05 Jakub Jelinek <jakub@redhat.com>
17364 * combine.c (combine_simplify_rtx): Don't try to simplify if
17365 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
17368 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
17370 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
17371 to 32 bytes when compiling for POWER9.
17373 2018-03-05 Jakub Jelinek <jakub@redhat.com>
17376 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
17377 regparm >= 3 with no arg reg available also for calls with
17378 flag_force_indirect_call. Pass decl to ix86_function_regparm.
17381 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
17383 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
17385 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
17388 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
17390 2018-03-05 Richard Biener <rguenther@suse.de>
17392 PR tree-optimization/84486
17393 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
17394 When inserting a __builtin_assume_aligned call set the LHS
17395 SSA name alignment info accordingly.
17397 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
17399 PR tree-optimization/84114
17400 * config/aarch64/aarch64.c (aarch64_reassociation_width)
17401 Avoid reassociation of FLOAT_MODE addition.
17403 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
17405 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
17406 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
17407 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
17408 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
17409 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
17410 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
17411 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
17413 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
17414 __builtin_ia32_wbinvd): New builtins.
17415 (SPECIAL_ARGS2): New.
17416 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
17417 (SPECIAL_ARGS2): New.
17418 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
17419 (ix86_valid_target_attribute_inner_p): Ditto.
17420 (ix86_init_mmx_sse_builtins): Add special_args2.
17421 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
17422 TARGET_WBNOINVD_P): New.
17423 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
17424 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
17425 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
17426 * config/i386/immintrin.h (_wbinvd): New intrinsic.
17427 * config/i386/pconfigintrin.h: New file.
17428 * config/i386/wbnoinvdintrin.h: Ditto.
17429 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
17431 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
17433 2018-03-05 Richard Biener <rguenther@suse.de>
17435 PR tree-optimization/84670
17436 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
17438 (BB_VISITED_WITH_VISITED_SUCCS): New define.
17439 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
17440 (compute_antic_aux): Only assert the number of values in ANTIC_IN
17441 doesn't grow if all successors (recursively) were visited at least
17444 2018-03-05 Richard Biener <rguenther@suse.de>
17446 PR tree-optimization/84650
17447 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
17448 if executed in the loop pipeline.
17450 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
17452 * doc/configfiles.texi (Configuration Files): Move info about
17453 conditionalizing $target-protos.h to...
17454 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
17455 differs from $target-protos.h.
17457 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
17458 Chung-Ju Wu <jasonwucj@gmail.com>
17460 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
17461 * config/nds32/nds32-multiple.md (setmemsi): Define.
17462 * config/nds32/nds32-memory-manipulation.c
17463 (nds32_gen_dup_4_byte_to_word_value): New.
17464 (emit_setmem_word_loop): New.
17465 (emit_setmem_byte_loop): New.
17466 (nds32_expand_setmem_loop): New.
17467 (nds32_expand_setmem_loop_v3m): New.
17468 (nds32_expand_setmem_unroll): New.
17469 (nds32_expand_setmem): New.
17471 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
17472 Chung-Ju Wu <jasonwucj@gmail.com>
17474 * config/nds32/nds32-memory-manipulation.c
17475 (nds32_emit_load_store): New.
17476 (nds32_emit_post_inc_load_store): New.
17477 (nds32_emit_mem_move): New.
17478 (nds32_emit_mem_move_block): New.
17479 (nds32_expand_movmemsi_loop_unknown_size): New.
17480 (nds32_expand_movmemsi_loop_known_size): New.
17481 (nds32_expand_movmemsi_loop): New.
17482 (nds32_expand_movmemsi_unroll): New.
17483 (nds32_expand_movmemqi): Rename ...
17484 (nds32_expand_movmemsi): ... to this.
17485 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
17486 (movmemsi): ... to this.
17487 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
17488 (nds32_expand_movmemsi): ... to this.
17490 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
17491 Monk Chiang <sh.chiang04@gmail.com>
17492 Chung-Ju Wu <jasonwucj@gmail.com>
17494 * config/nds32/nds32-protos.h
17495 (nds32_expand_load_multiple): New arguments.
17496 (nds32_expand_store_multiple): Ditto.
17497 (nds32_valid_multiple_load_store): Rename ...
17498 (nds32_valid_multiple_load_store_p): ... to this.
17499 * config/nds32/nds32-memory-manipulation.c
17500 (nds32_expand_load_multiple): Refine implementation.
17501 (nds32_expand_store_multiple): Ditto.
17502 * config/nds32/nds32-multiple.md
17503 (load_multiple): Update nds32_expand_load_multiple interface.
17504 (store_multiple): Update nds32_expand_store_multiple interface.
17505 * config/nds32/nds32-predicates.c
17506 (nds32_valid_multiple_load_store): Rename ...
17507 (nds32_valid_multiple_load_store_p): ... to this and refine
17509 * config/nds32/predicates.md
17510 (nds32_load_multiple_and_update_address_operation): New predicate.
17511 (nds32_store_multiple_and_update_address_operation): New predicate.
17513 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
17514 Chung-Ju Wu <jasonwucj@gmail.com>
17516 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
17517 (combo): New attribute.
17518 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
17520 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
17522 * config/nds32/nds32.opt: Change -mcmodel= default value.
17524 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
17525 Monk Chiang <sh.chiang04@gmail.com>
17526 Chung-Ju Wu <jasonwucj@gmail.com>
17528 * config/nds32/constants.md (unspec_element): New enum.
17529 * config/nds32/constraints.md (Umw): New constraint.
17530 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
17531 * config/nds32/nds32-intrinsic.md: Likewise.
17532 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
17533 (nds32_valid_smw_lwm_base_p): New.
17534 (nds32_output_smw_single_word): New.
17535 (nds32_output_lmw_single_word): New.
17536 (nds32_expand_unaligned_load): New.
17537 (nds32_expand_unaligned_store): New.
17538 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
17539 (nds32_output_smw_single_word): Declare.
17540 (nds32_output_lmw_single_word): Declare.
17541 (nds32_expand_unaligned_load): Declare.
17542 (nds32_expand_unaligned_store): Declare.
17543 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
17544 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
17545 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
17546 NDS32_BUILTIN_UASTORE_DW.
17547 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
17550 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
17551 Kito Cheng <kito.cheng@gmail.com>
17552 Chung-Ju Wu <jasonwucj@gmail.com>
17554 * config/nds32/nds32-intrinsic.c
17555 (nds32_expand_builtin_null_ftype_reg): Delete.
17556 (nds32_expand_builtin_reg_ftype_imm): Ditto.
17557 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
17558 (nds32_read_argument): New.
17559 (nds32_legitimize_target): Ditto.
17560 (nds32_legitimize_argument): Ditto.
17561 (nds32_check_constant_argument): Ditto.
17562 (nds32_expand_unop_builtin): Ditto.
17563 (nds32_expand_unopimm_builtin): Ditto.
17564 (nds32_expand_binop_builtin): Ditto.
17565 (nds32_builtin_decl_impl): Ditto.
17566 (builtin_description): Ditto.
17567 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
17568 (nds32_init_builtins_impl): Ditto.
17569 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
17570 (nds32_builtin_decl): New.
17571 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
17572 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
17574 2018-03-02 Jeff Law <law@redhat.com>
17576 * reorg.c (stop_search_p): Handle DEBUG_INSN.
17577 (redundant_insn, fill_simple_delay_slots): Likewise.
17578 (fill_slots_from_thread): Likewise.
17579 * resource.c (mark_referenced_resources): Likewise.
17580 (mark_set_resources, find_dead_or_set_registers): Likewise.
17582 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17584 * substring-locations.h (format_warning_va): Formatting fix for
17585 ATTRIBUTE_GCC_DIAG.
17586 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
17588 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
17589 * substring-locations.c: Include intl.h.
17590 (format_warning_va): Turned into small wrapper around
17591 format_warning_n_va, renamed to ...
17592 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
17593 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
17595 (format_warning_at_substring_n): New function.
17596 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
17597 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
17598 format_warning_at_substring with just a shorter name instead of
17599 const function pointer.
17600 (fmtwarn_n): New function.
17601 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
17602 appropriate, get rid of all the fmtstr temporaries, move conditionals
17603 with G_() wrapped string literals directly into fmtwarn arguments,
17604 cast dir.len to (int), formatting fixes.
17606 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
17608 * doc/invoke.texi: Remove "Cilk Plus" references.
17610 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17611 Richard Biener <rguenther@suse.de>
17614 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
17615 for error or warning attributes if CALL_FROM_THUNK_P is set.
17618 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17621 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
17622 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
17625 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
17626 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
17628 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
17629 instead of -1U in last predictors element's probability member.
17631 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
17634 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
17635 arguments if they are comparable.
17637 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
17639 PR tree-optimization/84634
17640 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
17641 masks and masked_loop_p with a single loop_masks, making sure it's
17642 null for bb vectorization.
17644 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
17646 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
17647 (vect_analyze_data_ref_access): Use loop->safe_len rather than
17648 loop->force_vectorize to check whether there is no alias.
17650 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17653 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
17655 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
17657 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
17658 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
17659 instead of a loop around prev_real_insn.
17660 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
17663 PR inline-asm/84625
17664 * config/i386/i386.c (ix86_print_operand): Use conditional
17665 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
17668 2018-03-02 Richard Biener <rguenther@suse.de>
17670 PR tree-optimization/84427
17671 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
17672 (bitmap_set_subtract_values): Rewrite to handle multiple
17675 (prune_clobbered_mems): Likewise.
17676 (phi_translate): Take edge instead of pred/phiblock.
17677 (phi_translate_1): Likewise.
17678 (phi_translate_set): Likewise. Insert all translated
17679 exprs for a value into the set, keeping possibly multiple
17680 expressions per value.
17681 (compute_antic_aux): Adjust for phi_translate changes.
17682 When intersecting union the expressions and prune those
17683 not in the final value set, keeping possibly multiple
17684 expressions per value. Do not use value-insertion
17685 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
17686 all expressions. Add verification that the value-sets
17687 only shrink during iteration.
17688 (compute_partial_antic_aux): Adjust for the phi_translate changes.
17689 (do_pre_regular_insertion): Likewise.
17690 (do_pre_partial_partial_insertion): Likewise.
17692 2018-03-02 Richard Biener <rguenther@suse.de>
17695 * config/darwin.c (saved_debug_info_level): New static global.
17696 (darwin_asm_lto_start): Disable debug info generation for LTO out.
17697 (darwin_asm_lto_end): Restore debug info generation settings.
17699 2018-03-01 Martin Liska <mliska@suse.cz>
17702 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
17703 volatile arguments.
17705 2018-03-01 Richard Biener <rguenther@suse.de>
17708 * dwarf2out.c (gen_variable_die): Properly handle late VLA
17709 type annotation with LTO when debug was disabled at compile-time.
17711 2018-03-01 Matthew Fortune <mfortune@gmail.com>
17713 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
17715 (mips_final_postscan_insn): Likewise.
17717 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
17719 PR rtl-optimization/84528
17720 * alias.c (init_alias_target): Add commentary.
17721 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
17722 a unique base value if the frame pointer is not eliminated
17723 to the stack pointer.
17725 2018-03-01 Tom de Vries <tom@codesourcery.com>
17727 PR rtl-optimization/83327
17728 * lra-int.h (hard_regs_spilled_into): Declare.
17729 * lra.c (hard_regs_spilled_into): Define.
17730 (init_reg_info): Init hard_regs_spilled_into.
17731 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
17732 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
17733 (process_bb_lives): Handle hard_regs_spilled_into.
17734 (lra_create_live_ranges_1): Before doing liveness propagation, clear
17735 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
17737 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
17739 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
17740 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
17741 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
17742 * config/rs6000/aix72.h: New file.
17744 2018-02-28 Jakub Jelinek <jakub@redhat.com>
17746 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
17747 instead of warning_at with conditional singular and plural messages
17751 * stor-layout.c (update_alignment_for_field): For
17752 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
17753 && !DECL_PACKED (field), do the alignment update, just use
17754 only desired_align instead of MAX (type_align, desired_align)
17756 (place_field): Don't do known_align < desired_align handling
17757 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
17758 is non-NULL, instead do it after rli->prev_field handling and
17759 only if not within a bitfield word. For DECL_PACKED (field)
17760 use type_align of BITS_PER_UNIT.
17762 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
17764 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
17765 superfluous parentheses and trailing spaces.
17767 2018-02-28 Richard Biener <rguenther@suse.de>
17769 PR tree-optimization/84584
17770 * graphite-scop-detection.c (scop_detection::add_scop): Discard
17771 SCoPs with fake exit edge.
17773 2018-02-28 Martin Liska <mliska@suse.cz>
17776 * timevar.c (timer::print): Fix format to properly print 100%
17779 2018-02-28 Richard Biener <rguenther@suse.de>
17781 PR middle-end/84607
17782 * genmatch.c (capture_info::walk_match): Do not mark
17783 captured expressions without operands as expr_p given
17784 they act more like predicates and should be subject to
17785 "lost tail" side-effect preserving.
17787 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
17789 PR rtl-optimization/81611
17790 * auto-inc-dec.c (attempt_change): Move dead note from
17791 mem_insn if it's the next use of regno
17792 (find_address): Take address use of reg holding
17793 non-incremented value. Add parm to limit search to the named
17795 (merge_in_block): Attempt to use a mem insn that is the next
17796 use of the original regno.
17798 2018-02-27 Martin Sebor <msebor@redhat.com>
17801 * doc/invoke.texi (-Wmissing-attributes): New option.
17802 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
17804 2018-02-27 Martin Sebor <msebor@redhat.com>
17806 PR translation/84207
17807 * diagnostic-core.h (warning_n, error_n, inform_n): Change
17808 n argument to unsigned HOST_WIDE_INT.
17809 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
17810 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
17811 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
17812 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
17814 2018-02-27 Richard Biener <rguenther@suse.de>
17816 PR tree-optimization/84512
17817 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
17818 Do not use the estimate returned from record_stmt_cost for
17819 the scalar iteration cost but sum properly using add_stmt_cost.
17821 2018-02-27 Richard Biener <rguenther@suse.de>
17823 PR tree-optimization/84466
17824 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
17825 Adjust last change to less strictly validate use operands.
17827 2018-02-27 Martin Liska <mliska@suse.cz>
17829 PR gcov-profile/84548
17830 * gcov.c (process_file): Allow partial overlap and consider it
17831 also as group functions.
17832 (output_lines): Properly calculate range of lines for a group.
17834 2018-02-27 Martin Liska <mliska@suse.cz>
17836 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
17837 'ggc' suffixes. Change first column width.
17838 (timer::print): Fix formatting of the column.
17840 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
17842 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
17843 preserve inline entry blocks for the sake of debug inline
17844 entry point markers alone.
17845 (remove_unused_locals): Suggest in comments a better place to
17846 force the preservation of inline entry blocks that are
17847 otherwise unused, but do not preserve them.
17849 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
17851 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
17853 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
17856 * config/i386/constraints.md (Bs): Replace
17857 ix86_indirect_branch_register with
17858 TARGET_INDIRECT_BRANCH_REGISTER.
17860 * config/i386/i386.md (indirect_jump): Likewise.
17861 (tablejump): Likewise.
17862 (*sibcall_memory): Likewise.
17863 (*sibcall_value_memory): Likewise.
17864 Peepholes of indirect call and jump via memory: Likewise.
17865 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
17866 (*sibcall_value_GOT_32): Likewise.
17867 * config/i386/predicates.md (indirect_branch_operand): Likewise.
17868 (GOT_memory_operand): Likewise.
17869 (call_insn_operand): Likewise.
17870 (sibcall_insn_operand): Likewise.
17871 (GOT32_symbol_operand): Likewise.
17872 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
17874 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
17876 PR rtl-optimization/83496
17877 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
17878 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
17879 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
17880 redundant insn, if any.
17881 (relax_delay_slots): Likewise.
17882 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
17884 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
17886 PR tree-optimization/83965
17887 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
17888 that grouped statements are part of a reduction chain. Return
17889 true if the statement is not marked as a reduction itself but
17890 is part of a group.
17891 (vect_recog_dot_prod_pattern): Don't check whether the statement
17892 is part of a group here.
17893 (vect_recog_sad_pattern): Likewise.
17894 (vect_recog_widen_sum_pattern): Likewise.
17896 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
17899 * final.c (rest_of_clean_state): Also look for calls inside sequences.
17901 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
17904 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
17906 (ix86_output_indirect_function_return): New prototype.
17907 (ix86_split_simple_return_pop_internal): Likewise.
17908 * config/i386/i386.c (indirect_return_via_cx): New.
17909 (indirect_return_via_cx_bnd): Likewise.
17910 (indirect_thunk_name): Handle return va CX_REG.
17911 (output_indirect_thunk_function): Create alias for
17912 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
17913 (ix86_output_indirect_jmp): Remove the bool argument.
17914 (ix86_output_indirect_function_return): New function.
17915 (ix86_split_simple_return_pop_internal): Likewise.
17916 * config/i386/i386.md (*indirect_jump): Don't pass false
17917 to ix86_output_indirect_jmp.
17918 (*tablejump_1): Likewise.
17919 (simple_return_pop_internal): Change it to define_insn_and_split.
17920 Call ix86_split_simple_return_pop_internal to split it for
17921 -mfunction-return=.
17922 (simple_return_indirect_internal): Call
17923 ix86_output_indirect_function_return instead of
17924 ix86_output_indirect_jmp.
17926 2018-02-26 Jakub Jelinek <jakub@redhat.com>
17929 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
17930 memset and value initialization afterwards.
17932 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
17934 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
17936 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17939 * common/config/aarch64/aarch64-common.c
17940 (aarch_option_optimization_table[]): Switch
17941 off fomit-frame-pointer
17943 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
17944 Chung-Ju Wu <jasonwucj@gmail.com>
17946 * config/nds32/nds32-multiple.md (load_multiple): Disallow
17948 (store_multiple): Ditto.
17950 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
17952 * config.gcc: Add --with-cpu support for nds32 target.
17953 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
17954 * config/nds32/nds32.opt: Add -mcpu= option.
17956 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
17958 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
17959 isel=yes): Warn for these deprecated options.
17961 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
17963 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
17964 ISA_2_5_MASKS_EMBEDDED.
17966 2018-02-23 Jakub Jelinek <jakub@redhat.com>
17968 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
17969 p->max as pointers rather than using iterative_hash_expr.
17971 2018-02-23 Carl Love <cel@us.ibm.com>
17973 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
17974 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
17976 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
17977 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
17978 P8V_BUILTIN_VEC_VUNSIGNED2.
17980 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
17983 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
17984 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
17986 * lra-constraints.c (curr_insn_transform): Set up
17987 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
17989 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
17990 * lra-eliminations.c (spill_pseudos): Ditto.
17991 (process_insn_for_elimination): Ditto.
17992 * lra-lives.c (reg_early_clobber_p): Use the new macros.
17993 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
17994 LRA_NON_CLOBBERED_ALT.
17996 2018-02-22 Martin Sebor <msebor@redhat.com>
17998 PR tree-optimization/84480
17999 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
18000 to maybe_diag_stxncpy_trunc. Call it.
18001 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
18002 from gimple_fold_builtin_strcpy. Print inlining stack.
18003 (handle_builtin_stxncpy): Print inlining stack.
18004 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
18006 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
18009 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
18010 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
18011 and -fcheck-pointer-bounds are used together.
18012 (indirect_thunk_prefix): New enum.
18013 (indirect_thunk_need_prefix): New function.
18014 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
18015 "_nt" instead of "_bnd" for NOTRACK prefix.
18016 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
18017 (output_indirect_thunk_function): Likewise.
18019 (ix86_code_end): Update output_indirect_thunk_function calls.
18020 (ix86_output_indirect_branch_via_reg): Replace
18021 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
18022 (ix86_output_indirect_branch_via_push): Likewise.
18023 (ix86_output_function_return): Likewise.
18024 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
18025 incompatible with -fcf-protection=branch and
18026 -fcheck-pointer-bounds.
18028 2018-02-22 Steve Ellcey <sellcey@cavium.com>
18031 * config/aarch64/aarch64.c (aarch64_print_address_internal):
18032 Change gcc_assert call to output_operand_lossage.
18034 2018-02-22 Steve Ellcey <sellcey@cavium.com>
18036 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
18038 2018-02-22 DJ Delorie <dj@redhat.com>
18039 Sebastian Perta <sebastian.perta@renesas.com>
18040 Oleg Endo <olegendo@gcc.gnu.org>
18042 * config/rx/rx.c (rx_rtx_costs): New function.
18043 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
18045 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
18047 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
18049 2018-02-22 Martin Liska <mliska@suse.cz>
18052 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
18053 Add "native" as a possible value.
18055 2018-02-22 Martin Liska <mliska@suse.cz>
18058 * config/i386/i386.c (ix86_option_override_internal):
18059 Add "native" as a possible value for -march and -mtune.
18061 2018-02-22 Jakub Jelinek <jakub@redhat.com>
18064 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
18065 to all type variants.
18067 PR tree-optimization/84503
18068 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
18069 width as info->bitpos + info->bitsize - start.
18070 (merged_store_group::merge_overlapping): Simplify width computation.
18071 (check_no_overlap): New function.
18072 (imm_store_chain_info::try_coalesce_bswap): Compute expected
18073 start + width and last_order of the group, fail if check_no_overlap
18075 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
18076 to group if check_no_overlap fails.
18078 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
18080 * config/rs6000/altivec.md: Delete contraint arguments to
18081 define_expand, define_split, and define_peephole2, and in
18082 define_insn_and_split if always unused.
18083 * config/rs6000/darwin.md: Ditto.
18084 * config/rs6000/dfp.md: Ditto.
18085 * config/rs6000/rs6000.md: Ditto.
18086 * config/rs6000/sync.md: Ditto.
18087 * config/rs6000/vector.md: Ditto.
18088 * config/rs6000/vsx.md: Ditto.
18090 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
18092 * config/rs6000/altivec.md: Write output control strings as braced
18093 blocks instead of double-quoted strings.
18094 * config/rs6000/darwin.md: Ditto.
18095 * config/rs6000/rs6000.md: Ditto.
18096 * config/rs6000/vector.md: Ditto.
18097 * config/rs6000/vsx.md: Ditto.
18099 2018-02-21 Jason Merrill <jason@redhat.com>
18101 PR c++/84314 - ICE with templates and fastcall attribute.
18102 * attribs.c (build_type_attribute_qual_variant): Remove assert.
18104 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
18106 * ipa-cp.c (determine_versionability): Fix comment typos.
18108 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
18111 * ipa-cp.c (determine_versionability): Do not version functions caling
18114 2018-02-21 Martin Liska <mliska@suse.cz>
18117 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
18118 Add "native" as a possible value.
18119 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
18120 the macro when native cpu detection is available.
18122 2018-02-21 Martin Liska <mliska@suse.cz>
18125 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
18126 Add "native" as a possible value.
18127 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
18128 when native cpu detection is available.
18130 2018-02-21 Jakub Jelinek <jakub@redhat.com>
18131 Martin Sebor <msebor@redhat.com>
18133 PR tree-optimization/84478
18134 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
18136 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
18137 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
18138 support which is conservatively correct, for 2 only stay conservative
18139 for maxlen. Formatting and comment capitalization fixes. Add STRICT
18140 argument to the 2 argument get_range_strlen, adjust 6 arg
18141 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
18143 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
18144 (gimple_fold_builtin_strlen): Pass true as last argument to
18147 2018-02-20 Martin Sebor <msebor@redhat.com>
18149 PR middle-end/84095
18150 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
18151 (builtin_memref::set_base_and_offset): Same. Handle inner references.
18152 (builtin_memref::builtin_memref): Factor out parts into
18153 set_base_and_offset and call it.
18155 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
18157 PR middle-end/84406
18158 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
18159 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
18160 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
18161 search at the associated MODE_INT.
18163 2018-02-20 Jeff Law <law@redhat.com>
18165 PR middle-end/82123
18166 PR tree-optimization/81592
18167 PR middle-end/79257
18168 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
18169 for range data rather than using global data.
18170 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
18171 range data rather than using global data.
18172 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
18173 pass it to children as needed.
18174 (struct directive::fmtresult): Similarly.
18175 (struct directive::set_width): Similarly.
18176 (struct directive::set_precision): Similarly.
18177 (format_integer, format_directive, parse_directive): Similarly.
18178 (format_none): Accept unnamed vr_values parameter.
18179 (format_percent, format_floating, format_character): Similarly.
18180 (format_string, format_plain): Similarly.
18181 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
18182 the EVRP range analyzer for range data rather than using global data.
18183 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
18184 gimple-ssa-evrp-analyze.h
18185 (class sprintf_dom_walker): Add after_dom_children member function.
18186 Add evrp_range_analyzer member.
18187 (sprintf_dom_walker::before_dom_children): Call into the EVRP
18188 range analyzer as needed.
18189 (sprintf_dom_walker::after_dom_children): New member function.
18190 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
18192 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
18193 (evrp_range_analyzer::pop_to_marker): Likewise.
18195 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
18197 PR tree-optimization/84419
18198 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
18199 with the required type if its current type is compatible but
18202 2018-02-20 Jakub Jelinek <jakub@redhat.com>
18204 PR middle-end/82004
18205 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
18206 after vectorization.
18208 2018-02-20 Martin Liska <mliska@suse.cz>
18211 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
18212 possible values if we don't have a hint.
18214 2018-02-20 Martin Liska <mliska@suse.cz>
18218 * final.c (shorten_branches): Build align_tab array with one
18220 * opts.c (finish_options): Add alignment option limit check.
18221 (MAX_CODE_ALIGN): Likewise.
18222 (MAX_CODE_ALIGN_VALUE): Likewise.
18223 * doc/invoke.texi: Document maximum allowed option value for
18224 all -falign-* options.
18226 2018-02-19 Jakub Jelinek <jakub@redhat.com>
18229 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
18230 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
18231 * var-tracking.c (emit_note_insn_var_location): Remove all references
18232 to NOTE_INSN_CALL_ARG_LOCATION.
18233 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
18234 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
18235 Use copy_rtx_if_shared.
18236 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
18237 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
18238 (dwarf2out_var_location): Remove handling of
18239 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
18241 * final.c (final_scan_insn): Remove all references to
18242 NOTE_INSN_CALL_ARG_LOCATION.
18243 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
18244 before dumping final insns.
18245 * except.c (emit_note_eh_region_end): Remove all references to
18246 NOTE_INSN_CALL_ARG_LOCATION.
18247 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
18248 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
18249 * config/arc/arc.c (hwloop_optimize): Likewise.
18250 * config/arm/arm.c (create_fix_barrier): Likewise.
18251 * config/s390/s390.c (s390_chunkify_start): Likewise.
18252 * config/sh/sh.c (find_barrier): Likewise.
18253 * config/i386/i386.c (rest_of_insert_endbranch,
18254 ix86_seh_fixup_eh_fallthru): Likewise.
18255 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
18256 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
18257 * config/frv/frv.c (frv_function_prologue): Likewise.
18258 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
18260 (note_outside_basic_block_p): Remove all references to
18261 NOTE_INSN_CALL_ARG_LOCATION.
18262 * gengtype.c (adjust_field_rtx_def): Likewise.
18263 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
18265 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
18266 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
18269 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
18272 PR tree-optimization/84452
18273 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
18274 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
18277 2018-02-19 Martin Liska <mliska@suse.cz>
18280 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
18282 2018-02-19 Martin Liska <mliska@suse.cz>
18283 Richard Sandiford <richard.sandiford@linaro.org>
18285 PR tree-optimization/82491
18286 * gimple-fold.c (get_base_constructor): Make earlier bail out
18289 2018-02-19 Carl Love <cel@us.ibm.com>
18291 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
18292 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
18294 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
18295 P8V_BUILTIN_VEC_NEG.
18297 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
18299 * config/rl78/rl78.md (movdf): New define expand.
18301 2018-02-19 Martin Liska <mliska@suse.cz>
18304 * doc/invoke.texi: Fix typo.
18305 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
18307 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
18309 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
18310 handle rs6000_single_float and rs6000_double_float specially for
18313 2018-02-16 Jeff Law <law@redhat.com>
18315 * config/rx/rx.c (add_pop_cfi_notes): New function.;
18316 (pop_regs): Use it.
18318 2018-02-16 Jakub Jelinek <jakub@redhat.com>
18321 * ipa-inline.c (inline_small_functions): Fix a typo.
18323 2018-02-16 Nathan Sidwell <nathan@acm.org>
18325 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
18327 2018-02-16 Carl Love <cel@us.ibm.com>
18329 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
18330 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
18331 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
18332 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
18333 expansion to P8V_BUILTIN_VEC_FLOAT2.
18335 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
18337 PR rtl-optimization/70023
18338 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
18339 src_regno into account.
18341 2018-02-16 Carl Love <cel@us.ibm.com>
18343 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
18344 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
18345 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
18346 * config/rs6000/rs6000.c: Remove case statements for
18347 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
18348 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
18349 and P9V_BUILTIN_VEC_VINSERT4B.
18350 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
18351 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
18352 * config/rs6000/vsx.md:
18353 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
18356 2018-02-16 Carl Love <cel@us.ibm.com>
18358 * config/rs6000/altivec.h: Add builtin names vec_extract4b
18360 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
18362 * config/rs6000/rs6000-c.c: Add the definitions for
18363 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
18364 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
18365 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
18366 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
18367 definition for insert4b and define insn *insert3b_internal.
18368 * doc/extend.texi: Add documentation for vec_extract4b.
18370 2018-02-16 Nathan Sidwell <nathan@acm.org>
18372 * doc/extend.texi (Backwards Compatibility): Mention friend
18373 injection. Note for-scope is deprecated.
18374 * doc/invoke.texi (-ffriend-injection): Deprecate.
18376 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
18378 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
18379 that moved to I2, also allow destinations that are a paradoxical
18380 subreg (instead of a normal reg).
18382 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
18385 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
18388 2018-02-16 Richard Biener <rguenther@suse.de>
18390 PR tree-optimization/84037
18391 PR tree-optimization/84016
18393 * config/i386/i386.c (ix86_builtin_vectorization_cost):
18394 Adjust vec_construct for the fact we need additional higher latency
18395 128bit inserts for AVX256 and AVX512 vector builds.
18396 (ix86_add_stmt_cost): Scale vector construction cost for
18399 2018-02-16 Richard Biener <rguenther@suse.de>
18401 PR tree-optimization/84417
18402 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
18403 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
18404 (non_rewritable_lvalue_p): Likewise, use poly-ints.
18406 2018-02-16 Martin Liska <mliska@suse.cz>
18409 * internal-fn.def (ASAN_CHECK): Set proper flags.
18410 (ASAN_MARK): Likewise.
18412 2018-02-16 Julia Koval <julia.koval@intel.com>
18414 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
18415 from PTA_CANNONLAKE.
18417 2018-02-16 Jakub Jelinek <jakub@redhat.com>
18420 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
18421 Use ++iter rather than iter++ for std::list iterators.
18422 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
18423 defer deleting them until all nodes in the forest are processed. Do
18424 free even leaf nodes. Change to_process into auto_vec.
18427 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
18428 * vec.h (vec_default_construct): Use memset instead of placement new
18429 if BROKEN_VALUE_INITIALIZATION is defined.
18430 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
18431 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
18434 PR rtl-optimization/83723
18435 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
18436 * lra.c (lra_substitute_pseudo): Likewise. If true, use
18437 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
18439 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
18441 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
18443 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
18445 PR rtl-optimization/81443
18446 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
18447 from inner REGs to paradoxical SUBREGs.
18449 2018-02-16 Richard Biener <rguenther@suse.de>
18451 PR tree-optimization/84399
18452 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
18453 For operands we can analyze at their definition make sure we can
18454 analyze them at each use as well.
18456 2018-02-16 Richard Biener <rguenther@suse.de>
18458 PR tree-optimization/84190
18459 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
18460 volatile accesses if the decl isn't volatile.
18462 2018-02-15 Jason Merrill <jason@redhat.com>
18464 PR c++/84314 - ICE with templates and fastcall attribute.
18465 * attribs.c (build_type_attribute_qual_variant): Don't clobber
18466 TYPE_CANONICAL on an existing type.
18468 2018-02-15 Jakub Jelinek <jakub@redhat.com>
18470 PR tree-optimization/84383
18471 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
18472 dstoff nor call operand_equal_p if dstbase is NULL.
18474 PR tree-optimization/84334
18475 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
18476 also a CONSTANT_CLASS_P, punt.
18478 2018-02-14 Jim Wilson <jimw@sifive.com>
18480 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
18481 first SMALL_OPERAND check. New local min_second_step. Move assert
18482 to where locals are set. Add TARGET_RVC support.
18483 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
18485 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
18487 * doc/invoke.texi: Correct -Wformat-overflow code sample.
18489 2018-02-14 Martin Sebor <msebor@redhat.com>
18491 PR tree-optimization/83698
18492 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
18493 arrays constrain the offset range to their bounds.
18494 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
18495 (builtin_access::overlap): Avoid setting the size of overlap if it's
18497 (maybe_diag_overlap): Also consider arrays when deciding what values
18498 of offsets to include in diagnostics.
18500 2018-02-14 Martin Sebor <msebor@redhat.com>
18503 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
18504 that correspond to the kind of a declaration.
18506 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
18509 * config/pa/pa.md: Load address of PIC label using the linkage table
18510 if the label is nonlocal.
18512 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
18514 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
18515 warning message if user requests -maltivec=be.
18516 * doc/invoke.texi: Document deprecation of -maltivec=be.
18518 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
18521 * config/rs6000/rs6000-c.c: Update definitions for
18522 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
18523 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
18525 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
18528 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
18529 add _get_ssp intrinsics. Remove argument from
18530 __builtin_ia32_rdssp[d|q].
18531 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
18532 * config/i386/i386-builtin.def: Remove argument from
18533 __builtin_ia32_rdssp[d|q].
18534 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
18535 ix86_expand_special_args_builtin for _rdssp[d|q].
18536 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
18537 Clear register before usage.
18538 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
18539 Add documentation for new _get_ssp and _inc_ssp intrinsics.
18541 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
18543 PR tree-optimization/84357
18544 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
18545 operand 1 of an ARRAY_REF too.
18547 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
18550 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
18551 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
18553 (set_of_reg): New struct.
18554 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
18555 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
18556 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
18558 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
18559 Split into bitclr, bitset, bitinvert patterns if appropriate.
18560 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
18561 use rx_fuse_in_memory_bitop.
18562 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
18563 to named insn, correct maximum insn length.
18565 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
18568 * machmode.def: Define a complex mode for PARTIAL_INT.
18569 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
18571 * doc/rtl.texi: Document CSPImode.
18572 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
18574 (msp430_hard_regno_nregs_with_padding): Likewise.
18576 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
18579 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
18581 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
18583 PR rtl-optimization/84169
18584 * combine.c (try_combine): New variable split_i2i3. Set it to true if
18585 we generated a parallel as new i3 and we split that to new i2 and i3
18586 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
18587 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
18588 those to i2, not i1. Partially rewrite this scan code.
18590 2018-02-13 Jakub Jelinek <jakub@redhat.com>
18593 * stor-layout.c (place_field): For variable length fields, adjust
18594 offset_align afterwards not just based on the field's alignment,
18595 but also on the size.
18597 PR middle-end/84309
18598 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
18599 of exps and logs in the use_exp2 case.
18601 2018-02-13 Jeff Law <law@redhat.com>
18603 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
18604 entry for "vector".
18606 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
18609 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
18613 * common.opt (gas-loc-support, gas-locview-support): New.
18614 (ginline-points, ginternal-reset-location-views): New.
18615 * doc/invoke.texi: Document them. Use @itemx where intended.
18616 (gvariable-location-views): Adjust.
18617 * target.def (reset_location_view): New.
18618 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
18619 (TARGET_RESET_LOCATION_VIEW): New.
18620 * doc/tm.texi: Rebuilt.
18621 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
18622 (dwarf2out_default_as_locview_support): New.
18623 (output_asm_line_debug_info): Use option variables.
18624 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
18625 (output_loc_list): Likewise.
18626 (add_high_low_attributes): Check option variables.
18627 Don't output entry view attribute in strict mode.
18628 (gen_inlined_subroutine_die): Check option variables.
18629 (dwarf2out_inline_entry): Likewise.
18630 (init_sections_and_labels): Likewise.
18631 (dwarf2out_early_finish): Likewise.
18632 (maybe_reset_location_view): New, from...
18633 (dwarf2out_var_location): ... here. Call it.
18634 * debug.h (dwarf2out_default_as_loc_support): Declare.
18635 (dwarf2out_default_as_locview_support): Declare.
18636 * hooks.c (hook_int_rtx_insn_0): New.
18637 * hooks.h (hook_int_rtx_insn_0): Declare.
18638 * toplev.c (process_options): Take -gas-loc-support and
18639 -gas-locview-support from dwarf2out. Enable
18640 -gvariable-location-views by default only with locview
18641 assembler support. Enable -ginternal-reset-location-views by
18642 default only if the target defines the corresponding hook.
18643 Enable -ginline-points by default if location views are
18644 enabled; force it disabled if statement frontiers are
18646 * tree-inline.c (expand_call_inline): Check option variables.
18647 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
18649 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
18651 PR tree-optimization/84321
18652 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
18653 handling. Also check whether the anti-range contains any values
18654 that satisfy the mask; switch to a VR_RANGE if not.
18656 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
18659 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
18661 2018-02-13 Martin Jambor <mjambor@suse.cz>
18664 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
18665 of call statements, also set location of a load to a temporary.
18667 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
18669 * config/rl78/rl78.c (add_vector_labels): New function.
18670 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
18671 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
18672 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
18673 which checks that no arguments are passed.
18674 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
18675 * doc/extend.texi: Documentation for the new attribute.
18677 2018-02-13 Andreas Schwab <schwab@suse.de>
18679 * config/riscv/linux.h (CPP_SPEC): Define.
18681 2018-02-13 Jakub Jelinek <jakub@redhat.com>
18684 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
18685 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
18686 OPTION_MASK_ISA_AES as first argument to def_builtin_const
18687 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
18688 instead of OPTION_MASK_ISA_PCLMUL as first argument to
18689 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
18690 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
18691 temporarily for AES and PCLMUL builtins.
18693 PR tree-optimization/84339
18694 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
18695 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
18698 PR middle-end/84309
18699 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
18700 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
18701 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
18703 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
18705 * omp-simd-clone.h: New file.
18706 * omp-simd-clone.c: Include omp-simd-clone.h.
18707 (expand_simd_clones): No longer static.
18708 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
18709 cgraph.h and omp-simd-clone.h.
18710 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
18711 (vect_recog_widen_shift_pattern): Formatting fix.
18712 (vect_pattern_recog_1): Don't check optab for calls.
18715 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
18716 operands[2] into a REG before using gen_lowpart on it.
18718 2018-02-12 Jeff Law <law@redhat.com>
18721 * config/sh/sh.c (find_barrier): Consider a sibling call
18724 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
18725 successfully back substituting a reg.
18727 2018-02-12 Richard Biener <rguenther@suse.de>
18729 PR tree-optimization/84037
18730 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
18731 parameter, move visited init to caller.
18732 (vect_slp_analyze_operations): Separate cost from validity
18733 check, initialize visited once for all instances.
18734 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
18736 * tree-vect-stmts.c (vect_model_simple_cost): Make early
18738 (vect_model_promotion_demotion_cost): Likewise.
18739 (vectorizable_bswap): Guard cost modeling with !slp_node
18740 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
18742 (vectorizable_call): Likewise.
18743 (vectorizable_conversion): Likewise.
18744 (vectorizable_assignment): Likewise.
18745 (vectorizable_shift): Likewise.
18746 (vectorizable_operation): Likewise.
18747 (vectorizable_store): Likewise.
18748 (vectorizable_load): Likewise.
18749 (vectorizable_condition): Likewise.
18750 (vectorizable_comparison): Likewise.
18752 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
18755 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
18756 (ASAN_MARK): Fix fnspec to account for return value, change pointer
18757 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
18759 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
18761 PR middle-end/83665
18762 * params.def (inline-min-speedup): Increase from 8 to 15.
18763 (max-inline-insns-auto): Decrease from 40 to 30.
18764 * ipa-split.c (consider_split): Add some buffer for function to
18765 be considered inlining candidate.
18766 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
18769 2018-02-12 Richard Biener <rguenther@suse.de>
18771 PR tree-optimization/84037
18772 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
18773 matched stmts if we cannot swap the non-matched ones.
18775 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
18777 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
18778 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
18779 _mm_maskz_scalef_round_ss): New intrinsics.
18780 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
18781 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
18782 __builtin_ia32_scalefss_round): Remove.
18783 (__builtin_ia32_scalefsd_mask_round,
18784 __builtin_ia32_scalefss_mask_round): New intrinsics.
18785 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
18786 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
18787 ((match_operand:VF_128 2 "<round_nimm_predicate>"
18788 "<round_constraint>")): Changed to ...
18789 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
18790 "<round_scalar_constraint>")): ... this.
18791 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
18792 %0, %1, %2<round_op3>}"): Changed to ...
18793 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
18794 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
18795 %2<round_scalar_mask_op3>}"): ... this.
18796 * config/i386/subst.md (round_scalar_nimm_predicate): New.
18798 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
18800 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
18801 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
18802 (_mm_maskz_sqrt_round_ss): New intrinsics.
18803 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
18804 (__builtin_ia32_sqrtsd_mask_round)
18805 (__builtin_ia32_sqrtss_mask_round): New builtins.
18806 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
18807 (__builtin_ia32_sqrtss_round): Remove.
18808 (__builtin_ia32_sqrtsd_mask_round)
18809 (__builtin_ia32_sqrtss_mask_round): New builtins.
18810 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
18811 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
18812 ((match_operand:VF_128 1 "vector_operand"
18813 "xBm,<round_constraint>")): Changed to ...
18814 ((match_operand:VF_128 1 "vector_operand"
18815 "xBm,<round_scalar_constraint>")): ... this.
18816 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
18817 %0, %2, %<iptr>1<round_op3>}): Changed to ...
18818 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
18819 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
18820 %<iptr>1<round_scalar_mask_op3>}): ... this.
18821 ((set_attr "prefix" "<round_prefix>")): Changed to ...
18822 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
18824 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
18827 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
18828 Cast vec_cmpeq result to correct type.
18829 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
18830 Cast vec_cmpgt result to correct type.
18832 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
18834 * final.c (final_scan_insn_1): Renamed from...
18835 (final_scan_insn): ... this. New wrapper, to recover
18836 seen from the outermost call in recursive ones.
18837 * config/sparc/sparc.c (output_return): Drop seen from call.
18838 (output_sibcall): Likewise.
18839 * config/visium/visium.c (output_branch): Likewise.
18841 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
18843 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
18846 2018-02-10 Alan Modra <amodra@gmail.com>
18849 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
18850 Specify LR as an input.
18852 2018-02-10 Jakub Jelinek <jakub@redhat.com>
18855 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
18856 remove_member_access_dummy_vars): New functions.
18857 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
18858 lower_omp_1, execute_lower_omp): Use them.
18860 PR rtl-optimization/84308
18861 * shrink-wrap.c (spread_components): Release todo vector.
18863 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
18865 PR rtl-optimization/57193
18866 * ira-color.c (struct allocno_color_data): Add member
18867 conflict_allocno_hard_prefs.
18868 (update_conflict_allocno_hard_prefs): New.
18869 (bucket_allocno_compare_func): Add a preference based on
18870 conflict_allocno_hard_prefs.
18871 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
18872 (color_allocnos): Remove a dead code. Initiate
18873 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
18875 2018-02-09 Jakub Jelinek <jakub@redhat.com>
18878 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
18879 constraint from =wa to wa. Avoid a subreg on the output operand,
18880 instead use a pseudo and subreg it in a move.
18881 (p9_xxbrd_<mode>): Changed to ...
18882 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
18883 (p9_xxbrd_v2df): New expander.
18884 (p9_xxbrw_<mode>): Changed to ...
18885 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
18886 (p9_xxbrw_v4sf): New expander.
18888 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
18890 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
18892 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
18895 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
18896 multiply in 32-bit mode.
18897 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
18898 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
18901 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
18903 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
18904 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
18905 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
18906 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
18908 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
18911 * dwarf2out.c (is_trivial_indirect_ref): New function.
18912 (dwarf2out_late_global_decl): Do not generate a location
18913 attribute for variables that have a non-trivial DECL_VALUE_EXPR
18914 and that are not defined in the current unit.
18916 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
18918 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
18919 instead of a libcall for UNORDERED.
18921 2018-02-09 Tamar Christina <tamar.christina@arm.com>
18924 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
18925 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
18927 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18930 * config/s390/s390.c (s390_set_current_function): Invoke
18931 s390_indirect_branch_settings also if fndecl didn't change.
18933 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
18935 * config/rs6000/rs6000.md (blockage): Set length to zero.
18937 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
18939 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
18941 2018-02-09 Jakub Jelinek <jakub@redhat.com>
18944 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
18945 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
18949 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
18950 PARALLEL incoming that failed vt_get_decl_and_offset check.
18952 PR middle-end/84237
18953 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
18954 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
18956 (get_variable_section): For decls in named .bss* sections pass true as
18957 second argument to bss_initializer_p.
18959 2018-02-09 Marek Polacek <polacek@redhat.com>
18960 Jakub Jelinek <jakub@redhat.com>
18963 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
18964 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
18965 Sync some changes from cxx_fold_indirect_ref.
18967 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
18969 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
18971 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
18972 (BLOCK_INLINE_ENTRY_LABEL): New.
18973 (dwarf2out_var_location): Disregard inline entry markers.
18974 (inline_entry_data): New struct.
18975 (inline_entry_data_hasher): New hashtable type.
18976 (inline_entry_data_hasher::hash): New.
18977 (inline_entry_data_hasher::equal): New.
18978 (inline_entry_data_table): New variable.
18979 (add_high_low_attributes): Add DW_AT_entry_pc and
18980 DW_AT_GNU_entry_view attributes if a pending entry is found
18981 in inline_entry_data_table. Add old entry_pc attribute only
18982 if debug nonbinding markers are disabled.
18983 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
18984 markers are enabled.
18985 (block_within_block_p, dwarf2out_inline_entry): New.
18986 (dwarf2out_finish): Check that no entries remained in
18987 inline_entry_data_table.
18988 * final.c (reemit_insn_block_notes): Handle inline entry notes.
18989 (final_scan_insn, notice_source_line): Likewise.
18990 (rest_of_clean_state): Skip inline entry markers.
18991 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
18993 * gimple.c (gimple_build_debug_inline_entry): New.
18994 * gimple.h (enum gimple_debug_subcode): Add
18995 GIMPLE_DEBUG_INLINE_ENTRY.
18996 (gimple_build_debug_inline_entry): Declare.
18997 (gimple_debug_inline_entry_p): New.
18998 (gimple_debug_nonbind_marker_p): Adjust.
18999 * insn-notes.def (INLINE_ENTRY): New.
19000 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
19001 inline entry marker notes.
19002 (print_insn): Likewise.
19003 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
19004 (INSN_DEBUG_MARKER_KIND): Likewise.
19005 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
19006 * tree-inline.c (expand_call_inline): Build and insert
19007 debug_inline_entry stmt.
19008 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
19009 inline entry blocks early, if nonbind markers are enabled.
19010 (dump_scope_block): Dump fragment info.
19011 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
19012 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
19013 (gimple_build_debug_inline_entry): New.
19014 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
19015 Enable/disable inline entry points too.
19016 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
19017 (DEBUG_INSN): Describe inline entry markers.
19019 * common.opt (gvariable-location-views): New.
19020 (gvariable-location-views=incompat5): New.
19021 * config.in: Rebuilt.
19022 * configure: Rebuilt.
19023 * configure.ac: Test assembler for view support.
19024 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
19025 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
19026 * dwarf2out.c (var_loc_view): New typedef.
19027 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
19028 (dwarf2out_locviews_in_attribute): New.
19029 (dwarf2out_locviews_in_loclist): New.
19030 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
19031 (enum dw_line_info_opcode): Add LI_adv_address.
19032 (struct dw_line_info_table): Add view.
19033 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
19034 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
19035 (zero_view_p): New variable.
19036 (ZERO_VIEW_P): New macro.
19037 (output_asm_line_debug_info): New.
19038 (struct var_loc_node): Add view.
19039 (add_AT_view_list, AT_loc_list): New.
19040 (add_var_loc_to_decl): Add view param. Test it against last.
19041 (new_loc_list): Add view params. Record them.
19042 (AT_loc_list_ptr): Handle loc and view lists.
19043 (view_list_to_loc_list_val_node): New.
19044 (print_dw_val): Handle dw_val_class_view_list.
19045 (size_of_die): Likewise.
19046 (value_format): Likewise.
19047 (loc_list_has_views): New.
19048 (gen_llsym): Set vl_symbol too.
19049 (maybe_gen_llsym, skip_loc_list_entry): New.
19050 (dwarf2out_maybe_output_loclist_view_pair): New.
19051 (output_loc_list): Output view list or entries too.
19052 (output_view_list_offset): New.
19053 (output_die): Handle dw_val_class_view_list.
19054 (output_dwarf_version): New.
19055 (output_compilation_unit_header): Use it.
19056 (output_skeleton_debug_sections): Likewise.
19057 (output_rnglists, output_line_info): Likewise.
19058 (output_pubnames, output_aranges): Update version comments.
19059 (output_one_line_info_table): Output view numbers in asm comments.
19060 (dw_loc_list): Determine current endview, pass it to new_loc_list.
19061 Call maybe_gen_llsym.
19062 (loc_list_from_tree_1): Adjust.
19063 (add_AT_location_description): Create view list attribute if
19064 needed, check it's absent otherwise.
19065 (convert_cfa_to_fb_loc_list): Adjust.
19066 (maybe_emit_file): Call output_asm_line_debug_info for test.
19067 (dwarf2out_var_location): Reset views as needed. Precompute
19068 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
19069 attribute. Set view.
19070 (new_line_info_table): Reset next view.
19071 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
19072 (dwarf2out_source_line): Likewise. Output view resets and labels to
19073 the assembler, or select appropriate line info opcodes.
19074 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
19075 (optimize_string_length): Catch it. Adjust.
19076 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
19077 dw_val_class_view_list, and remove it if no longer needed.
19078 (hash_loc_list): Hash view numbers.
19079 (loc_list_hasher::equal): Compare them.
19080 (optimize_location_lists): Check whether a view list symbol is
19081 needed, and whether the locview attribute is present, and
19082 whether they match. Remove the locview attribute if no longer
19084 (index_location_lists): Call skip_loc_list_entry for test.
19085 (dwarf2out_finish): Call output_asm_line_debug_info for test.
19086 Use output_dwarf_version.
19087 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
19088 (struct dw_val_node): Add val_view_list.
19089 * final.c (SEEN_NEXT_VIEW): New.
19090 (set_next_view_needed): New.
19091 (clear_next_view_needed): New.
19092 (maybe_output_next_view): New.
19093 (final_start_function): Rename to...
19094 (final_start_function_1): ... this. Take pointer to FIRST,
19095 add SEEN parameter. Emit param bindings in the initial view.
19096 (final_start_function): Reintroduce SEEN-less interface.
19097 (final): Rename to...
19098 (final_1): ... this. Take SEEN parameter. Output final pending
19099 next view at the end.
19100 (final): Reintroduce seen-less interface.
19101 (final_scan_insn): Output pending next view before switching
19102 sections or ending a block. Mark the next view as needed when
19103 outputting variable locations. Notify debug backend of section
19104 changes, and of location view changes.
19105 (rest_of_handle_final): Adjust.
19106 * toplev.c (process_options): Autodetect value for debug variable
19107 location views option. Warn on incompat5 without -gdwarf-5.
19108 * doc/invoke.texi (gvariable-location-views): New.
19109 (gvariable-location-views=incompat5): New.
19110 (gno-variable-location-views): New.
19112 2018-02-08 David Malcolm <dmalcolm@redhat.com>
19114 PR tree-optimization/84136
19115 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
19116 that the result of find_edge is non-NULL.
19118 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
19121 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
19122 storing integer register in SImode. Fix cost of 256 and 512
19123 byte aligned SSE register store.
19125 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
19127 * config/i386/i386.c (ix86_multiplication_cost): Fix
19128 multiplication cost for TARGET_AVX512DQ.
19130 2018-02-08 Marek Polacek <polacek@redhat.com>
19132 PR tree-optimization/84238
19133 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
19136 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19138 PR tree-optimization/84265
19139 * tree-vect-stmts.c (vectorizable_store): Don't treat
19140 VMAT_CONTIGUOUS accesses as grouped.
19141 (vectorizable_load): Likewise.
19143 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19145 PR tree-optimization/81635
19146 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
19147 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
19148 (test_round_for_mask): New functions.
19149 (wide_int_cc_tests): Call test_round_for_mask.
19150 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
19151 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
19152 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
19153 range returned by get_range_info.
19155 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
19158 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
19159 * symtab.c: Include builtins.h
19160 (symtab_node::output_to_lto_symbol_table_p): Move here
19161 from lto-streamer-out.c:output_symbol_p.
19162 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
19163 (output_symbol_p): Move all logic to symtab.c
19164 (produce_symtab): Update.
19166 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19168 * config/s390/s390-opts.h (enum indirect_branch): Define.
19169 * config/s390/s390-protos.h (s390_return_addr_from_memory)
19170 (s390_indirect_branch_via_thunk)
19171 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
19172 (enum s390_indirect_branch_type): Define.
19173 * config/s390/s390.c (struct s390_frame_layout, struct
19174 machine_function): Remove.
19175 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
19176 (indirect_branch_table_label_no, indirect_branch_table_name):
19178 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
19179 (enum s390_indirect_branch_option): Define.
19180 (s390_return_addr_from_memory): New function.
19181 (s390_handle_string_attribute): New function.
19182 (s390_attribute_table): Add new attribute handler.
19183 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
19184 (s390_indirect_branch_via_thunk): New function.
19185 (s390_indirect_branch_via_inline_thunk): New function.
19186 (s390_function_ok_for_sibcall): When jumping via thunk disallow
19187 sibling call optimization for non z10 compiles.
19188 (s390_emit_call): Force indirect branch target to be a single
19189 register. Add r1 clobber for non-z10 compiles.
19190 (s390_emit_epilogue): Emit return jump via return_use expander.
19191 (s390_reorg): Handle JUMP_INSNs as execute targets.
19192 (s390_option_override_internal): Perform validity checks for the
19193 new command line options.
19194 (s390_indirect_branch_attrvalue): New function.
19195 (s390_indirect_branch_settings): New function.
19196 (s390_set_current_function): Invoke s390_indirect_branch_settings.
19197 (s390_output_indirect_thunk_function): New function.
19198 (s390_code_end): Implement target hook.
19199 (s390_case_values_threshold): Implement target hook.
19200 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
19202 * config/s390/s390.h (struct s390_frame_layout)
19203 (struct machine_function): Move here from s390.c.
19204 (TARGET_INDIRECT_BRANCH_NOBP_RET)
19205 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
19206 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
19207 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
19208 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
19209 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
19210 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
19211 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
19212 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
19213 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
19214 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
19215 (mnemonic attribute): Add values which aren't recognized
19217 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
19218 pattern for branch conversion. Fix mnemonic attribute.
19219 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
19220 indirect branch via thunk if requested.
19221 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
19222 ("*indirect_jump"): Disable for branch conversion using out of
19224 ("indirect_jump_via_thunk<mode>_z10")
19225 ("indirect_jump_via_thunk<mode>")
19226 ("indirect_jump_via_inlinethunk<mode>_z10")
19227 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
19228 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
19229 ("casesi_jump_via_inlinethunk<mode>_z10")
19230 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
19231 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
19232 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
19233 ("*indirect2_jump"): Disable for branch conversion.
19234 ("casesi_jump"): Turn into expander and expand patterns for branch
19236 ("return_use"): New expander.
19237 ("*return"): Emit return via thunk and rename it to ...
19238 ("*return<mode>"): ... this one.
19239 * config/s390/s390.opt: Add new options and and enum for the
19242 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19244 * lra-constraints.c (match_reload): Unconditionally use
19245 gen_lowpart_SUBREG, rather than selecting between that
19246 and equivalent gen_rtx_SUBREG code.
19248 2018-02-08 Richard Biener <rguenther@suse.de>
19250 PR tree-optimization/84233
19251 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
19252 changed flag instead of boguously re-using phi_inserted.
19254 2018-02-08 Martin Jambor <mjambor@suse.cz>
19256 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
19257 static local variables.
19259 2018-02-08 Richard Biener <rguenther@suse.de>
19261 PR tree-optimization/84278
19262 * tree-vect-stmts.c (vectorizable_store): When looking for
19263 smaller vector types to perform grouped strided loads/stores
19264 make sure the mode is supported by the target.
19265 (vectorizable_load): Likewise.
19267 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
19269 * config/aarch64/aarch64.c (aarch64_components_for_bb):
19270 Increase LDP/STP opportunities by adding adjacent callee-saves.
19272 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
19274 PR rtl-optimization/84068
19275 PR rtl-optimization/83459
19276 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
19278 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
19280 PR tree-optimization/84224
19281 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
19282 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
19283 non-zero arguments.
19285 2018-02-07 Iain Sandoe <iain@codesourcery.com>
19288 * config/rs6000/altivec.md (*restore_world): Remove LR use.
19289 * config/rs6000/predicates.md (restore_world_operation): Adjust op
19290 count, remove one USE.
19292 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19294 * doc/install.texi (Configuration): Document the
19295 --with-long-double-format={ibm,ieee} PowerPC configuration
19299 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
19300 Convert from define_expand to be define_insn_and_split. Rework
19301 float/double/_Float128 conversions to QI/HI/SImode to work with
19302 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
19303 conversions to QI/HImode types did a store and then a load to
19304 truncate the value. For conversions to VSX registers, don't split
19305 the insn, instead emit the code directly. Use the code iterator
19306 any_fix to combine signed and unsigned conversions.
19307 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
19308 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
19309 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
19310 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
19311 (fix_<mode>di2_hw): Likewise.
19312 (fixuns_<mode>di2_hw): Likewise.
19313 (fix_<mode>si2_hw): Likewise.
19314 (fixuns_<mode>si2_hw): Likewise.
19315 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
19316 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
19317 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
19318 fix<uns>_trunc<SFDF:mode>si2_p8.
19319 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
19320 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19321 (fix<uns>_<mode>_mem): Likewise.
19322 (fctiw<u>z_<mode>_mem): Likewise.
19323 (fix<uns>_<mode>_mem): Likewise.
19324 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
19325 the register allocator from doing a direct move to the GPRs to do
19326 a store, and instead use the ISA 3.0 store byte/half-word from
19327 vector register instruction. For IEEE 128-bit floating point,
19328 also optimize stores of 32-bit ints.
19329 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
19331 2018-02-07 Alan Hayward <alan.hayward@arm.com>
19333 * genextract.c (push_pathstr_operand): New function to support
19335 (walk_rtx): Call push_pathstr_operand.
19336 (print_path): Support [a-zA-Z].
19338 2018-02-07 Richard Biener <rguenther@suse.de>
19340 PR tree-optimization/84037
19341 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
19342 (cse_and_gimplify_to_preheader): Declare.
19343 (vect_get_place_in_interleaving_chain): Likewise.
19344 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
19346 (_loop_vec_info::~_loop_vec_info): Delete it.
19347 (cse_and_gimplify_to_preheader): New function.
19348 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
19349 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
19350 (vectorizable_load): Likewise. For grouped stores always base
19351 the IV on the first element.
19352 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
19353 condition before gimplifying.
19355 2018-02-07 Jakub Jelinek <jakub@redhat.com>
19357 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
19358 *DIV_EXPR and *MOD_EXPR.
19360 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
19363 * config/i386/i386.c (ix86_option_override_internal): Mask out
19364 the CF_SET bit when checking -fcf-protection.
19366 2018-02-07 Tom de Vries <tom@codesourcery.com>
19369 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
19372 2018-02-07 Richard Biener <rguenther@suse.de>
19374 PR tree-optimization/84204
19375 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
19378 PR tree-optimization/84205
19379 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
19380 special-case isl_ast_op_zdiv_r.
19382 PR tree-optimization/84223
19383 * graphite-scop-detection.c (gather_bbs::before_dom_children):
19384 Only add conditions from within the region.
19385 (gather_bbs::after_dom_children): Adjust.
19387 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
19390 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
19391 * config/avr/avr.md: Only post-reload split REG-REG moves if
19392 either register is GENERAL_REG_P.
19394 2018-02-07 Jakub Jelinek <jakub@redhat.com>
19396 PR tree-optimization/84235
19397 * tree-ssa-scopedtables.c
19398 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
19399 if the subtraction is performed in floating point type where NaNs are
19400 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
19401 build 1. Formatting fix.
19403 2018-02-06 Jakub Jelinek <jakub@redhat.com>
19406 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
19407 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
19408 and skip it regardless of bb boundaries. Use CALL_P macro,
19409 don't test INSN_P (insn) together with CALL_P or JUMP_P check
19410 unnecessarily, formatting fix.
19412 2018-02-06 Michael Collison <michael.collison@arm.com>
19414 * config/arm/thumb2.md:
19415 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
19416 (*thumb_mov_notscc): Ditto.
19418 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
19421 * config/rs6000/rs6000.md (su code attribute): Use "u" for
19422 unsigned_fix, not "s".
19424 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19426 * configure.ac (gcc_fn_eh_frame_ro): New function.
19427 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
19428 correct .eh_frame permissions.
19429 * configure: Regenerate.
19431 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
19433 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
19434 irrelevant options.
19436 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19438 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19439 Display warning message for -mno-speculate-indirect-jumps.
19441 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
19443 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
19445 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
19447 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
19449 PR tree-optimization/84225
19450 * tree-eh.c (find_trapping_overflow): Only call
19451 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
19453 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
19456 * config/i386/i386.c: Reimplement the check of possible options
19457 -mibt/-mshstk conbination. Change error messages.
19458 * doc/invoke.texi: Fix a typo: remove extra '='.
19460 2018-02-06 Marek Polacek <polacek@redhat.com>
19462 PR tree-optimization/84228
19463 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
19465 2018-02-06 Tamar Christina <tamar.christina@arm.com>
19468 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
19469 emitted arch directives.
19470 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
19471 __ARM_FEATURE_COPROC before changing architectures.
19473 2018-02-06 Richard Biener <rguenther@suse.de>
19475 * config/i386/i386.c (print_reg): Fix typo.
19476 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
19478 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
19480 * configure: Regenerate.
19482 2018-02-05 Martin Sebor <msebor@redhat.com>
19484 PR tree-optimization/83369
19485 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
19488 2018-02-05 Martin Liska <mliska@suse.cz>
19490 * doc/invoke.texi: Cherry-pick upstream r323995.
19492 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
19494 * ira.c (ira_init_register_move_cost): Adjust comment.
19496 2018-02-05 Martin Liska <mliska@suse.cz>
19498 PR gcov-profile/84137
19499 * doc/gcov.texi: Fix typo in documentation.
19501 2018-02-05 Martin Liska <mliska@suse.cz>
19503 PR gcov-profile/83879
19504 * doc/gcov.texi: Document necessity of --dynamic-list-data when
19505 using dlopen functionality.
19507 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
19509 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
19510 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
19511 _mm_maskz_range_ss, _mm_mask_range_round_ss,
19512 _mm_maskz_range_round_ss): New intrinsics.
19513 (__builtin_ia32_rangesd128_round)
19514 (__builtin_ia32_rangess128_round): Remove.
19515 (__builtin_ia32_rangesd128_mask_round,
19516 __builtin_ia32_rangess128_mask_round): New builtins.
19517 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
19518 __builtin_ia32_rangess128_round): Remove.
19519 (__builtin_ia32_rangesd128_mask_round,
19520 __builtin_ia32_rangess128_mask_round): New builtins.
19521 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
19522 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
19523 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
19524 "<round_saeonly_constraint>")): Changed to ...
19525 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
19526 "<round_saeonly_scalar_constraint>")): ... this.
19527 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
19528 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
19529 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
19530 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
19531 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
19533 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
19535 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
19537 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
19538 Remove all values except native, 8540 and 8548.
19540 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
19542 * config/i386/i386.c (ix86_output_function_return): Pass
19543 INVALID_REGNUM, instead of -1, as invalid register number to
19544 indirect_thunk_name and output_indirect_thunk.
19546 2018-02-02 Julia Koval <julia.koval@intel.com>
19548 * config.gcc: Add -march=icelake.
19549 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
19550 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
19551 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
19552 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
19553 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
19554 (processor_target_table): Add icelake.
19555 (ix86_option_override_internal): Handle new PTAs.
19556 (get_builtin_code_for_version): Handle icelake.
19557 (M_INTEL_COREI7_ICELAKE): New.
19558 (fold_builtin_cpu): Handle icelake.
19559 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
19560 * doc/invoke.texi: Add -march=icelake.
19562 2018-02-02 Julia Koval <julia.koval@intel.com>
19564 * config/i386/i386.c (ix86_option_override_internal): Change flags type
19565 to wide_int_bitmask.
19566 * wide-int-bitmask.h: New.
19568 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
19571 * config/i386/i386.md: Replace Pmode with word_mode in
19572 builtin_setjmp_setup and builtin_longjmp to support x32.
19574 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
19578 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
19580 (rs6000_supported_cpu_names): New static variable.
19581 (linux_cpu_translation_table): Likewise.
19582 (elf_platform) <cpu>: Define new static variable and use it.
19583 Translate kernel AT_PLATFORM name to canonical name if needed.
19584 Error if platform name is unknown.
19586 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
19589 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
19591 2018-02-01 Jeff Law <law@redhat.com>
19594 * config/i386/i386.c (release_scratch_register_on_entry): Add new
19595 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
19596 the scratch if RELEASE_VIA_POP is false.
19597 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
19598 If we have to save a temporary register, decrement SIZE appropriately.
19599 Pass new arguments to release_scratch_register_on_entry.
19600 (ix86_adjust_stack_and_probe): Likewise.
19601 (ix86_emit_probe_stack_range): Pass new arguments to
19602 release_scratch_register_on_entry.
19604 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
19606 PR rtl-optimization/84157
19607 * combine.c (change_zero_ext): Use REG_P predicate in
19608 front of HARD_REGISTER_P predicate.
19610 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
19612 * config/avr/avr.c (avr_option_override): Move disabling of
19613 -fdelete-null-pointer-checks to...
19614 * common/config/avr/avr-common.c (avr_option_optimization_table):
19617 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19619 PR tree-optimization/81635
19620 * tree-data-ref.c (split_constant_offset_1): For types that
19621 wrap on overflow, try to use range info to prove that wrapping
19624 2018-02-01 Renlin Li <renlin.li@arm.com>
19627 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
19628 TAILCALL_ADDR_REGS.
19629 (aarch64_register_move_cost): Likewise.
19630 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
19631 TAILCALL_ADDR_REGS.
19632 (REG_CLASS_NAMES): Likewise.
19633 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
19634 TAILCALL_ADDR_REGS. Remove IP registers.
19635 * config/aarch64/aarch64.md (Ucs): Update register constraint.
19637 2018-02-01 Richard Biener <rguenther@suse.de>
19639 * domwalk.h (dom_walker::dom_walker): Add additional constructor
19640 for specifying RPO order and allow NULL for that.
19641 * domwalk.c (dom_walker::dom_walker): Likewise.
19642 (dom_walker::walk): Handle NULL RPO order.
19643 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
19645 (rewrite_update_dom_walker): Likewise.
19646 (mark_def_dom_walker): Likewise.
19648 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19650 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
19651 (aarch64_maybe_expand_sve_subreg_move): Declare.
19652 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
19653 * config/aarch64/predicates.md (aarch64_any_register_operand): New
19655 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
19656 that are semantically a reverse operation.
19657 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
19658 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
19659 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
19661 (aarch64_can_change_mode_class): For big-endian, forbid changes
19662 between two SVE modes if they have different element sizes.
19664 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19666 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
19667 the TImode handling for big-endian targets.
19669 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19671 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
19672 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
19674 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
19675 Remove BSWAP handing for big-endian targets and use the form of
19676 LD1RQ appropariate for the mode.
19678 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19680 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
19681 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
19682 duplicated element.
19684 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19687 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
19688 check for operands that need to go through aarch64_sve_reload_be.
19690 2018-02-01 Jakub Jelinek <jakub@redhat.com>
19692 PR tree-optimization/81661
19693 PR tree-optimization/84117
19694 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
19695 * tree-eh.c: Include gimplify.h.
19696 (find_trapping_overflow, replace_trapping_overflow,
19697 rewrite_to_non_trapping_overflow): New functions.
19698 * tree-vect-loop.c: Include tree-eh.h.
19699 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
19700 * tree-data-ref.c: Include tree-eh.h.
19701 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
19703 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
19705 PR rtl-optimization/84123
19706 * combine.c (change_zero_ext): Check if hard register satisfies
19707 can_change_dest_mode before calling gen_lowpart_SUBREG.
19709 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
19712 * ira.c (ira_init_register_move_cost): Remove assert.
19714 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
19716 PR rtl-optimization/84071
19717 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
19718 * doc/tm.texi: Regenerate.
19720 2018-01-31 Richard Biener <rguenther@suse.de>
19722 PR tree-optimization/84132
19723 * tree-data-ref.c (analyze_miv_subscript): Properly
19724 check whether evolution_function_is_affine_multivariate_p
19725 before calling gcd_of_steps_may_divide_p.
19727 2018-01-31 Julia Koval <julia.koval@intel.com>
19730 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
19731 * config/i386/i386.md (rdpid_rex64) New.
19732 (rdpid): Make 32bit only.
19734 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
19737 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
19738 an IDENTIFIER_NODE for FUNCTION_TYPE's.
19740 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
19743 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
19745 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
19747 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
19749 PR rtl-optimization/84071
19750 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
19751 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
19753 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
19755 * config/arc/arc.c (arc_handle_aux_attribute): New function.
19756 (arc_attribute_table): Add 'aux' attribute.
19757 (arc_in_small_data_p): Consider aux like variables.
19758 (arc_is_aux_reg_p): New function.
19759 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
19760 (arc_get_aux_arg): New function.
19761 (prepare_move_operands): Handle aux-register access.
19762 (arc_handle_aux_attribute): New function.
19763 * doc/extend.texi (ARC Variable attributes): Add subsection.
19765 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
19767 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
19768 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
19769 (arc_attribute_table): Add 'uncached' attribute.
19770 (arc_print_operand): Print '.di' flag for uncached memory
19772 (arc_in_small_data_p): Do not consider for small data the uncached
19774 (arc_is_uncached_mem_p): New function.
19775 * config/arc/predicates.md (compact_store_memory_operand): Check
19776 for uncached memory accesses.
19777 (nonvol_nonimm_operand): Likewise.
19778 * doc/extend.texi (ARC Type Attribute): New subsection.
19780 2018-01-31 Jakub Jelinek <jakub@redhat.com>
19783 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
19784 falign-loops=): Add Optimization flag.
19786 2018-01-30 Jeff Law <law@redhat.com>
19789 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
19790 INT_REGISTERS_SAVED. Check it prior to calling
19791 get_scratch_register_on_entry.
19792 (ix86_adjust_stack_and_probe): Similarly.
19793 (ix86_emit_probe_stack_range): Similarly.
19794 (ix86_expand_prologue): Corresponding changes.
19796 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19799 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
19800 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
19802 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
19805 * lra-constraints.c (curr_insn_transform): Process AND in the
19808 2018-01-30 Jakub Jelinek <jakub@redhat.com>
19810 PR rtl-optimization/83986
19811 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
19812 dependence against last_pending_memory_flush in addition to
19813 pending_jump_insns.
19815 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
19817 PR tree-optimization/81611
19818 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
19821 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19824 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
19827 2018-01-30 Richard Biener <rguenther@suse.de>
19828 Jakub Jelinek <jakub@redhat.com>
19830 PR tree-optimization/84111
19831 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
19832 inner loops added during recursion, as they don't have up-to-date
19835 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
19838 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
19839 (can_inline_edge_by_limits_p): ... here.
19840 (can_early_inline_edge_p, check_callers,
19841 update_caller_keys, update_callee_keys, recursive_inlining,
19842 add_new_edges_to_heap, speculation_useful_p,
19843 inline_small_functions,
19844 inline_small_functions, flatten_function,
19845 inline_to_all_callers_1): Update.
19847 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
19849 * profile-count.c (profile_count::combine_with_ipa_count): Handle
19852 2018-01-30 Richard Biener <rguenther@suse.de>
19854 PR tree-optimization/83008
19855 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
19856 invariant and constant vector uses in stmts when they need
19857 more than one stmt.
19859 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19862 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
19863 * configure: Regenerate.
19865 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
19867 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
19869 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
19870 Use gen_rtx_REG rather than gen_lowpart.
19872 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
19874 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
19875 rather than 0 when creating partial subregs.
19877 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
19879 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
19882 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19885 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
19886 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
19887 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
19888 flags. This restores the settings used before the 2017-07-24.
19889 Turning off pre increment/decrement/modify allows IVOPTS to
19890 optimize DF/SF loops where the index is an int.
19892 2018-01-29 Richard Biener <rguenther@suse.de>
19893 Kelvin Nilsen <kelvin@gcc.gnu.org>
19896 * tree-vect-stmts.c (vectorizable_call): Don't call
19897 targetm.vectorize_builtin_md_vectorized_function if callee is
19900 2018-01-22 Carl Love <cel@us.ibm.com>
19902 * doc/extend.tex: Fix typo in second arg in
19903 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
19905 2018-01-29 Richard Biener <rguenther@suse.de>
19907 PR tree-optimization/84086
19908 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
19909 (flush_ssaname_freelist): When SSA names were released reset
19910 the SCEV hash table.
19912 2018-01-29 Richard Biener <rguenther@suse.de>
19914 PR tree-optimization/84057
19915 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
19916 removed paths when removing edges.
19918 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
19920 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
19921 -mfunction-return=@var{choice}.
19923 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
19925 PR diagnostic/84034
19926 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
19927 Handle CR like TAB.
19928 (layout::print_source_line): Likewise.
19929 (test_get_line_width_without_trailing_whitespace): Add test cases.
19931 2018-01-27 Jakub Jelinek <jakub@redhat.com>
19933 PR middle-end/84040
19934 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
19937 2018-01-26 Jim Wilson <jimw@sifive.com>
19939 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
19941 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
19944 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19946 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
19947 and CMP + SUB-immediate -> SUBS.
19949 2018-01-26 Martin Sebor <msebor@redhat.com>
19951 PR tree-optimization/83896
19952 * tree-ssa-strlen.c (get_string_len): Rename...
19953 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
19954 Avoid assuming length is constant.
19955 (handle_char_store): Use HOST_WIDE_INT for string length.
19957 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
19960 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
19961 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
19963 2018-01-26 Richard Biener <rguenther@suse.de>
19965 PR rtl-optimization/84003
19966 * dse.c (record_store): Only record redundant stores when
19967 the earlier store aliases at least all accesses the later one does.
19969 2018-01-26 Jakub Jelinek <jakub@redhat.com>
19971 PR rtl-optimization/83985
19972 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
19973 REG_CFA_RESTORE insns.
19974 (delete_unmarked_insns): Don't ignore separate shrink wrapping
19975 REG_CFA_RESTORE insns here.
19978 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
19979 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
19981 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
19983 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
19984 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
19985 (arc_init): Likewise.
19986 (arc_override_options): Likewise.
19987 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
19989 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
19991 * config/arc/arc.h (TARGET_DBNZ): Define.
19992 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
19993 properly set the tune attribute.
19994 (dbnz): Use TARGET_DBNZ guard.
19995 * config/arc/arc.opt (mtune): Add core3 option.
19997 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
19999 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
20000 recognize new pic like addresses.
20001 (arc_delegitimize_address): Clean up.
20003 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20005 * config/arc/arc-arches.def: Option mrf16 valid for all
20007 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
20008 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
20009 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
20010 * config/arc/arc-tables.opt: Regenerate.
20011 * config/arc/arc.c (arc_conditional_register_usage): Handle
20012 reduced register file case.
20013 (arc_file_start): Set must have build attributes.
20014 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
20015 mrf16 option value.
20016 * config/arc/arc.opt (mrf16): Add new option.
20017 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
20018 * config/arc/genmultilib.awk: Handle new mrf16 option.
20019 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
20020 * config/arc/t-multilib: Regenerate.
20021 * doc/invoke.texi (ARC Options): Document mrf16 option.
20023 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20025 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
20026 * config/arc/arc.c (arc_handle_secure_attribute): New function.
20027 (arc_attribute_table): Add 'secure_call' attribute.
20028 (arc_print_operand): Print secure call operand.
20029 (arc_function_ok_for_sibcall): Don't optimize tail calls when
20031 (arc_is_secure_call_p): New function. * config/arc/arc.md
20032 (call_i): Add support for sjli instruction.
20033 (call_value_i): Likewise.
20034 * config/arc/constraints.md (Csc): New constraint.
20036 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20037 John Eric Martin <John.Martin@emmicro-us.com>
20039 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
20040 * config/arc/arc.c (_arc_jli_section): New struct.
20041 (arc_jli_section): New type.
20042 (rc_jli_sections): New static variable.
20043 (arc_handle_jli_attribute): New function.
20044 (arc_attribute_table): Add jli_always and jli_fixed attribute.
20045 (arc_file_end): New function.
20046 (TARGET_ASM_FILE_END): Define.
20047 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
20048 (arc_add_jli_section): New function.
20049 (jli_call_scan): Likewise.
20050 (arc_reorg): Call jli_call_scan.
20051 (arc_output_addsi): Remove 'S' from printing asm operand.
20052 (arc_is_jli_call_p): New function.
20053 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
20055 (movhi_insn): Likewise.
20056 (movsi_insn): Likewise.
20057 (movsi_set_cc_insn): Likewise.
20058 (loadqi_update): Likewise.
20059 (load_zeroextendqisi_update): Likewise.
20060 (load_signextendqisi_update): Likewise.
20061 (loadhi_update): Likewise.
20062 (load_zeroextendhisi_update): Likewise.
20063 (load_signextendhisi_update): Likewise.
20064 (loadsi_update): Likewise.
20065 (loadsf_update): Likewise.
20066 (movsicc_insn): Likewise.
20067 (bset_insn): Likewise.
20068 (bxor_insn): Likewise.
20069 (bclr_insn): Likewise.
20070 (bmsk_insn): Likewise.
20071 (bicsi3_insn): Likewise.
20072 (cmpsi_cc_c_insn): Likewise.
20073 (movsi_ne): Likewise.
20074 (movsi_cond_exec): Likewise.
20075 (clrsbsi2): Likewise.
20076 (norm_f): Likewise.
20086 (call_i): Remove 'S' asm letter, add jli instruction.
20087 (call_value_i): Likewise.
20088 * config/arc/arc.op (mjli-always): New option.
20089 * config/arc/constraints.md (Cji): New constraint.
20090 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
20092 (subsf3_fpx): Likewise.
20093 (mulsf3_fpx): Likewise.
20094 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
20096 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
20097 function attrbutes.
20098 * doc/invoke.texi (ARC): Document mjli-always option.
20100 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
20102 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
20103 avoid addition with 0 and use incw and decw where possible.
20105 2018-01-26 Richard Biener <rguenther@suse.de>
20107 PR tree-optimization/81082
20108 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
20109 association if it requires casting to unsigned.
20110 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
20111 from fold_plusminus_mult_expr to catch important cases late when
20112 range info is available.
20114 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20116 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
20117 * configure.ac (hidden_linkonce): New test.
20118 * configure: Regenerate.
20119 * config.in: Regenerate.
20121 2018-01-26 Julia Koval <julia.koval@intel.com>
20123 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
20124 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
20125 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
20126 _mm_mask_bitshuffle_epi64_mask): Fix type.
20127 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
20128 USI_FTYPE_V4DI_V4DI_USI): Remove.
20129 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
20130 __builtin_ia32_vpshufbitqmb256_mask,
20131 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
20132 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
20133 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
20135 2018-01-26 Alan Modra <amodra@gmail.com>
20138 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
20139 UNSPEC_VBPERMQ. Sort other unspecs.
20141 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
20143 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
20145 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
20147 PR middle-end/83055
20148 * predict.c (drop_profile): Do not push/pop cfun; update also
20150 (handle_missing_profiles): Fix logic looking for zero profiles.
20152 2018-01-25 Jakub Jelinek <jakub@redhat.com>
20154 PR middle-end/83977
20155 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
20156 on functions with #pragma omp declare simd or functions with simd
20158 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
20159 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
20160 Remove trailing \n from warning_at calls.
20162 2018-01-25 Tom de Vries <tom@codesourcery.com>
20165 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
20166 for neutered workers.
20168 2018-01-24 Joseph Myers <joseph@codesourcery.com>
20171 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
20172 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
20174 2018-01-24 Jeff Law <law@redhat.com>
20177 * i386.c (get_probe_interval): Move to earlier point.
20178 (ix86_compute_frame_layout): If -fstack-clash-protection and
20179 the frame is larger than the probe interval, then use pushes
20180 to save registers rather than reg->mem moves.
20181 (ix86_expand_prologue): Remove conditional for int_registers_saved
20184 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
20187 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
20188 min/max for never referenced object.
20190 2018-01-24 Jakub Jelinek <jakub@redhat.com>
20192 PR middle-end/83977
20193 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
20195 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
20196 attributes from DECL_ATTRIBUTES (decl) without affecting
20197 DECL_ATTRIBUTES (current_function_decl).
20198 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
20199 functions with non-NULL DECL_ABSTRACT_ORIGIN.
20201 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
20203 PR tree-optimization/83979
20204 * fold-const.c (fold_comparison): Use constant_boolean_node
20205 instead of boolean_{true,false}_node.
20207 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
20209 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
20212 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20214 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
20215 Simplify the clause that sets the length attribute.
20216 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
20217 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
20218 clause that sets the length attribute.
20219 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
20221 2018-01-24 Tom de Vries <tom@codesourcery.com>
20224 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
20225 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
20226 Add strict parameter.
20227 (prevent_branch_around_nothing): Insert dummy insn between branch to
20228 label and label with no ptx insn inbetween.
20229 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
20231 2018-01-24 Tom de Vries <tom@codesourcery.com>
20234 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
20235 for neutered threads in warp.
20236 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
20238 2018-01-24 Richard Biener <rguenther@suse.de>
20240 PR tree-optimization/83176
20241 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
20244 2018-01-24 Richard Biener <rguenther@suse.de>
20246 PR tree-optimization/82819
20247 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
20248 code generating pluses that are no-ops in the target precision.
20250 2018-01-24 Richard Biener <rguenther@suse.de>
20252 PR middle-end/84000
20253 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
20255 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
20257 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
20258 to merge probabilities.
20259 * predict.c (probably_never_executed): Also mark as cold functions
20260 with global 0 profile and guessed local profile.
20261 * profile-count.c (profile_probability::combine_with_count): New
20263 * profile-count.h (profile_probability::operator*,
20264 profile_probability::operator*=, profile_probability::operator/,
20265 profile_probability::operator/=): Reduce precision to adjusted
20266 and set value to guessed on contradictory divisions.
20267 (profile_probability::combine_with_freq): Remove.
20268 (profile_probability::combine_wiht_count): Declare.
20269 (profile_count::force_nonzero):: Set to adjusted.
20270 (profile_count::probability_in):: Set quality to adjusted.
20271 * tree-ssa-tail-merge.c (replace_block_by): Use
20272 combine_with_count.
20274 2018-01-23 Andrew Waterman <andrew@sifive.com>
20275 Jim Wilson <jimw@sifive.com>
20277 * config/riscv/riscv.c (riscv_stack_boundary): New.
20278 (riscv_option_override): Set riscv_stack_boundary. Handle
20279 riscv_preferred_stack_boundary_arg.
20280 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
20281 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
20282 (STACK_BOUNDARY): Set to riscv_stack_boundary.
20283 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
20284 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
20285 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
20287 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
20290 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
20291 of struct ix86_frame.
20292 (ix86_expand_epilogue): Likewise. Add a local variable for
20293 the reg_save_offset field in struct ix86_frame.
20295 2018-01-23 Bin Cheng <bin.cheng@arm.com>
20297 PR tree-optimization/82604
20298 * tree-loop-distribution.c (enum partition_kind): New enum item
20299 PKIND_PARTIAL_MEMSET.
20300 (partition_builtin_p): Support above new enum item.
20301 (generate_code_for_partition): Ditto.
20302 (compute_access_range): Differentiate cases that equality can be
20303 proven at all loops, the innermost loops or no loops.
20304 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
20305 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
20306 (finalize_partitions, distribute_loop): Don't fuse partition of
20307 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
20308 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
20309 parloop is enabled.
20311 2018-01-23 Martin Liska <mliska@suse.cz>
20313 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
20314 order to ignore the predictor.
20315 (PRED_POLYMORPHIC_CALL): Likewise.
20316 (PRED_RECURSIVE_CALL): Likewise.
20318 2018-01-23 Martin Liska <mliska@suse.cz>
20320 * tree-profile.c (tree_profiling): Print function header to
20321 aware reader which function we are working on.
20322 * value-prof.c (gimple_find_values_to_profile): Do not print
20323 not interesting value histograms.
20325 2018-01-23 Martin Liska <mliska@suse.cz>
20327 * profile-count.h (enum profile_quality): Add
20328 profile_uninitialized as the first value. Do not number values
20329 as they are zero based.
20330 (profile_count::verify): Update sanity check.
20331 (profile_probability::verify): Likewise.
20333 2018-01-23 Nathan Sidwell <nathan@acm.org>
20335 * doc/invoke.texi (ffor-scope): Deprecate.
20337 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20339 PR tree-optimization/83510
20340 * domwalk.c (set_all_edges_as_executable): New function.
20341 (dom_walker::dom_walker): Convert bool param
20342 "skip_unreachable_blocks" to enum reachability. Move setup of
20343 edge flags to set_all_edges_as_executable and only do it when
20344 reachability is REACHABLE_BLOCKS.
20345 * domwalk.h (enum dom_walker::reachability): New enum.
20346 (dom_walker::dom_walker): Convert bool param
20347 "skip_unreachable_blocks" to enum reachability.
20348 (set_all_edges_as_executable): New decl.
20349 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
20350 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
20352 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
20353 but converting true to REACHABLE_BLOCKS.
20354 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
20356 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
20357 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
20358 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
20360 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
20361 if check_all_array_refs will be called.
20363 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20365 * tree.c (selftest::test_location_wrappers): Add more test
20368 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20370 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
20371 (selftest::test_bit_in_range): Likewise.
20373 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
20376 * doc/sourcebuild.texi (vect_float): Say that the selector
20377 only describes the situation when -funsafe-math-optimizations is on.
20378 (vect_float_strict): Document.
20380 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
20382 PR tree-optimization/83965
20383 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
20384 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
20385 instead of checking only for a reduction.
20386 (vect_recog_widen_sum_pattern): Likewise.
20388 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
20390 * predict.c (probably_never_executed): Only use precise profile info.
20391 (compute_function_frequency): Skip after inlining hack since we now
20392 have quality checking.
20394 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
20396 * profile-count.h (profile_probability::very_unlikely,
20397 profile_probability::unlikely, profile_probability::even): Set
20398 precision to guessed.
20400 2018-01-23 Richard Biener <rguenther@suse.de>
20402 PR tree-optimization/83963
20403 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
20404 Properly terminate dominator walk when crossing the exit edge not
20405 when visiting its source block.
20407 2018-01-23 Jakub Jelinek <jakub@redhat.com>
20410 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
20411 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
20413 2018-01-22 Jakub Jelinek <jakub@redhat.com>
20415 PR tree-optimization/83957
20416 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
20417 semicolon after for body surrounded by braces.
20419 PR tree-optimization/83081
20420 * profile-count.h (profile_probability::split): New method.
20421 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
20422 Use profile_probability::split.
20423 (do_compare_rtx_and_jump): Fix adjustment of probabilities
20424 when splitting a single conditional jump into 2.
20426 2018-01-22 David Malcolm <dmalcolm@redhat.com>
20428 PR tree-optimization/69452
20429 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
20432 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20434 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
20435 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
20436 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
20438 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20440 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
20442 * config/rl78/rl78.md (movdi): New define_expand.
20443 * config/rl78/rl78.c (rl78_split_movdi): New function.
20445 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
20448 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
20450 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
20451 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
20452 128-bit to produce an UNSPEC move to get the double word with the
20453 signbit and then a shift directly to do signbit.
20454 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
20455 implementation with a new version that just does either a direct
20456 move or a regular move. Move memory interface to separate insns.
20457 Move insns so they are next to the expander.
20458 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
20459 with signbit move. Split big and little endian case.
20460 (signbit<mode>2_dm_mem_le): Likewise.
20461 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
20462 (signbit<mode>2_dm2): Likewise.
20464 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20466 * config/rl78/rl78.md (anddi3): New define_expand.
20468 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20470 * config/rl78/rl78.md (umindi3): New define_expand.
20472 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20474 * config/rl78/rl78.md (smindi3): New define_expand.
20476 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20478 * config/rl78/rl78.md (smaxdi3): New define_expand.
20480 2018-01-22 Carl Love <cel@us.ibm.com>
20482 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
20483 LVX_V1TI): Add macro expansion.
20484 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
20485 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
20486 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
20487 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
20488 Change check to determine if the instruction is a byte reversing
20489 entry. Fix typo in comment.
20490 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
20491 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
20492 Add def_builtin calls for new builtins.
20493 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
20494 Add define_insn expansion.
20496 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20498 * config/rl78/rl78.md (umaxdi3): New define_expand.
20500 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20502 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
20503 for non-QImode registers.
20505 2018-01-22 Richard Biener <rguenther@suse.de>
20507 PR tree-optimization/83963
20508 * graphite-scop-detection.c (scop_detection::get_sese): Delay
20509 including the loop exit block.
20510 (scop_detection::merge_sese): Likewise.
20511 (scop_detection::add_scop): Do it here instead.
20513 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20515 * doc/sourcebuild.texi (arm_softfloat): Document.
20517 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
20520 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
20521 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
20522 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
20524 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20525 David Edelsohn <dje.gcc@gmail.com>
20528 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
20529 Change "crset eq" to "crset 2".
20530 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
20531 (*call_indirect_aix<mode>_nospec): Likewise.
20532 (*call_value_indirect_aix<mode>_nospec): Likewise.
20533 (*call_indirect_elfv2<mode>_nospec): Likewise.
20534 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
20535 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
20536 change assembly output from . to $.
20537 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
20538 (indirect_jump<mode>_nospec): Change assembly output from . to $.
20539 (*tablejump<mode>_internal1_nospec): Likewise.
20541 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
20544 * config/sh/sh_optimize_sett_clrt.cc:
20545 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
20547 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
20549 PR tree-optimization/83940
20550 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
20551 offset_dt to vect_constant_def rather than vect_unknown_def_type.
20552 (vect_check_load_store_mask): Add a mask_dt_out parameter and
20553 use it to pass back the definition type.
20554 (vect_check_store_rhs): Likewise rhs_dt_out.
20555 (vect_build_gather_load_calls): Add a mask_dt argument and use
20556 it instead of a call to vect_is_simple_use.
20557 (vectorizable_store): Update calls to vect_check_load_store_mask
20558 and vect_check_store_rhs. Use the dt returned by the latter instead
20559 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
20560 instead of calls to vect_is_simple_use. Pass the scalar rather
20561 than the vector operand to vect_is_simple_use when handling
20562 second and subsequent copies of an rhs value.
20563 (vectorizable_load): Update calls to vect_check_load_store_mask
20564 and vect_build_gather_load_calls. Use the cached mask_dt and
20565 gs_info.offset_dt instead of calls to vect_is_simple_use.
20567 2018-01-20 Jakub Jelinek <jakub@redhat.com>
20569 PR middle-end/83945
20570 * tree-emutls.c: Include gimplify.h.
20571 (lower_emutls_2): New function.
20572 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
20573 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
20574 it before further processing.
20577 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
20578 UINTVAL (trueop1) instead of INTVAL (op1).
20580 2018-01-19 Jakub Jelinek <jakub@redhat.com>
20584 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
20585 INCOMING_FRAME_SP_OFFSET if not defined.
20586 (scan_trace): Add ENTRY argument. If true and
20587 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
20588 emit a note to adjust the CFA offset.
20589 (create_cfi_notes): Adjust scan_trace callers.
20590 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
20591 INCOMING_FRAME_SP_OFFSET in the CIE.
20592 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
20593 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
20595 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
20596 * doc/tm.texi: Regenerated.
20598 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20600 PR rtl-optimization/83147
20601 * lra-constraints.c (remove_inheritance_pseudos): Use
20602 lra_substitute_pseudo_within_insn.
20604 2018-01-19 Tom de Vries <tom@codesourcery.com>
20605 Cesar Philippidis <cesar@codesourcery.com>
20608 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
20610 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
20613 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
20614 spaces for function labels.
20616 2018-01-19 Martin Liska <mliska@suse.cz>
20618 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
20619 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
20620 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
20621 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
20622 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
20623 (PRED_CONST_RETURN): Change from 69 to 65.
20624 (PRED_NULL_RETURN): Change from 91 to 71.
20625 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
20626 (PRED_LOOP_GUARD): Change from 66 to 73.
20628 2018-01-19 Martin Liska <mliska@suse.cz>
20630 * predict.c (predict_insn_def): Add new assert.
20631 (struct branch_predictor): Change type to signed integer.
20632 (test_prediction_value_range): Amend test to cover
20633 PROB_UNINITIALIZED.
20634 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
20635 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
20636 (PRED_LOOP_ITERATIONS_MAX): Likewise.
20637 (PRED_LOOP_IV_COMPARE): Likewise.
20638 * predict.h (PROB_UNINITIALIZED): Define new constant.
20640 2018-01-19 Martin Liska <mliska@suse.cz>
20642 * predict.c (dump_prediction): Add new format for
20643 analyze_brprob.py script which is enabled with -details
20645 * profile-count.h (precise_p): New function.
20647 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
20649 PR tree-optimization/83922
20650 * tree-vect-loop.c (vect_verify_full_masking): Return false if
20651 there are no statements that need masking.
20652 (vect_active_double_reduction_p): New function.
20653 (vect_analyze_loop_operations): Use it when handling phis that
20654 are not in the loop header.
20656 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
20658 PR tree-optimization/83914
20659 * tree-vect-loop.c (vectorizable_induction): Don't convert
20660 init_expr or apply the peeling adjustment for inductions
20661 that are nested within the vectorized loop.
20663 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20665 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
20668 2018-01-18 Jakub Jelinek <jakub@redhat.com>
20672 * function.h (gimplify_parameters): Add gimple_seq * argument.
20673 * function.c: Include gimple.h and options.h.
20674 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
20675 for the added local temporaries if needed.
20676 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
20677 if there are any parameter cleanups, wrap whole body into a
20678 try/finally with the cleanups.
20680 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
20683 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20684 Use GET_MODE_CLASS for scalar floating point.
20686 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
20690 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
20691 Fix call of call_cgraph_insertion_hooks.
20693 2018-01-18 Martin Sebor <msebor@redhat.com>
20695 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
20697 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
20700 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
20703 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
20706 * common.opt: (-ffile-prefix-map): New option.
20707 * opts.c (common_handle_option): Defer it.
20708 * opts-global.c (handle_common_deferred_options): Handle it.
20709 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
20710 * file-prefix-map.h: New file.
20711 (remap_debug_filename, add_debug_prefix_map): ...here.
20712 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
20713 * final.c (debug_prefix_map, add_debug_prefix_map
20714 remap_debug_filename): Move to...
20715 * file-prefix-map.c: New file.
20716 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
20717 generalize, get rid of alloca(), use strrchr() instead of strchr().
20718 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
20719 Implement in terms of add_prefix_map().
20720 (remap_macro_filename, remap_debug_filename): Implement in term of
20722 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
20723 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
20724 * dbxout.c: Include file-prefix-map.h.
20725 * varasm.c: Likewise.
20726 * vmsdbgout.c: Likewise.
20727 * xcoffout.c: Likewise.
20728 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
20729 * doc/cppopts.texi (-fmacro-prefix-map): Document.
20730 * doc/invoke.texi (-ffile-prefix-map): Document.
20731 (-fdebug-prefix-map): Update description.
20733 2018-01-18 Martin Liska <mliska@suse.cz>
20735 * config/i386/i386.c (indirect_thunk_name): Document that also
20737 (output_indirect_thunk): Document why both instructions
20738 (pause and lfence) are generated.
20740 2018-01-18 Richard Biener <rguenther@suse.de>
20742 PR tree-optimization/83887
20743 * graphite-scop-detection.c
20744 (scop_detection::get_nearest_dom_with_single_entry): Remove.
20745 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
20746 (scop_detection::merge_sese): Re-implement with a flood-fill
20747 algorithm that properly finds a SESE region if it exists.
20749 2018-01-18 Jakub Jelinek <jakub@redhat.com>
20752 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
20753 pointer_diff optimizations use view_convert instead of convert.
20755 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20757 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
20758 Generate different code for -mno-speculate-indirect-jumps.
20759 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
20760 (*call_indirect_aix<mode>): Disable for
20761 -mno-speculate-indirect-jumps.
20762 (*call_indirect_aix<mode>_nospec): New define_insn.
20763 (*call_value_indirect_aix<mode>): Disable for
20764 -mno-speculate-indirect-jumps.
20765 (*call_value_indirect_aix<mode>_nospec): New define_insn.
20766 (*sibcall_nonlocal_sysv<mode>): Generate different code for
20767 -mno-speculate-indirect-jumps.
20768 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
20770 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
20772 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
20773 long double type, set the flags for noting the default long double
20774 type, even if we don't pass or return a long double type.
20776 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
20779 * ipa-inline.c (flatten_function): Do not overwrite final inlining
20782 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
20784 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
20785 support for merge[hl].
20786 (fold_mergehl_helper): New helper function.
20787 (tree-vector-builder.h): New #include for tree_vector_builder usage.
20788 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
20789 (altivec_vmrglw_direct): Add xxmrglw insn.
20791 2018-01-17 Andrew Waterman <andrew@sifive.com>
20793 * config/riscv/riscv.c (riscv_conditional_register_usage): If
20794 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
20796 2018-01-17 David Malcolm <dmalcolm@redhat.com>
20799 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
20800 call the lto_location_cache before reading the
20801 DECL_SOURCE_LOCATION of the types.
20803 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
20804 Richard Sandiford <richard.sandiford@linaro.org>
20806 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
20807 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
20808 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
20809 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
20810 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
20812 * config/aarch64/constraints.md (aarch64_movti_operand):
20814 * config/aarch64/predicates.md (Uti): Add new constraint.
20816 2018-01-17 Carl Love <cel@us.ibm.com>
20818 * config/rs6000/vsx.md (define_expand xl_len_r,
20819 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
20820 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
20822 (define_expand, define_insn): Move the shift left from the
20823 define_insn to the define_expand for lxvl and stxvl instructions.
20824 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
20825 and XL_LEN_R definitions to PURE.
20827 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
20829 * config/i386/i386.c (indirect_thunk_name): Declare regno
20830 as unsigned int. Compare regno with INVALID_REGNUM.
20831 (output_indirect_thunk): Ditto.
20832 (output_indirect_thunk_function): Ditto.
20833 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
20834 in the call to output_indirect_thunk_function.
20836 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
20838 PR middle-end/83884
20839 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
20840 rather than the size of inner_type to determine the stack slot size
20841 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
20843 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
20846 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
20849 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
20851 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
20852 endian Linux systems to optionally enable multilibs for selecting
20853 the long double type if the user configured an explicit type.
20854 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
20855 have no long double multilibs if not defined.
20856 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
20857 warn if the user used -mabi={ieee,ibm}longdouble and we built
20858 multilibs for long double.
20859 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
20860 appropriate multilib option.
20861 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
20863 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
20864 for building long double multilibs.
20865 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
20867 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
20869 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
20872 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
20874 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
20877 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
20880 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
20883 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
20885 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
20886 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
20888 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
20890 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
20891 different rtl trees depending on TARGET_64BIT.
20892 (rs6000_gen_lvx): Likewise.
20894 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
20896 * config/visium/visium.md (nop): Tweak comment.
20897 (hazard_nop): Likewise.
20899 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20901 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
20902 -mspeculate-indirect-jumps.
20903 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
20904 for -mno-speculate-indirect-jumps.
20905 (*call_indirect_elfv2<mode>_nospec): New define_insn.
20906 (*call_value_indirect_elfv2<mode>): Disable for
20907 -mno-speculate-indirect-jumps.
20908 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
20909 (indirect_jump): Emit different RTL for
20910 -mno-speculate-indirect-jumps.
20911 (*indirect_jump<mode>): Disable for
20912 -mno-speculate-indirect-jumps.
20913 (*indirect_jump<mode>_nospec): New define_insn.
20914 (tablejump): Emit different RTL for
20915 -mno-speculate-indirect-jumps.
20916 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
20917 (tablejumpsi_nospec): New define_expand.
20918 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
20919 (tablejumpdi_nospec): New define_expand.
20920 (*tablejump<mode>_internal1): Disable for
20921 -mno-speculate-indirect-jumps.
20922 (*tablejump<mode>_internal1_nospec): New define_insn.
20923 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
20926 2018-01-16 Artyom Skrobov tyomitch@gmail.com
20928 * caller-save.c (insert_save): Drop unnecessary parameter. All
20931 2018-01-16 Jakub Jelinek <jakub@redhat.com>
20932 Richard Biener <rguenth@suse.de>
20935 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
20936 return early, inline manually is_gimple_sizepos. Make sure if we
20937 call gimplify_expr we don't end up with a gimple constant.
20938 * tree.c (variably_modified_type_p): Don't return true for
20939 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
20940 * gimplify.h (is_gimple_sizepos): Remove.
20942 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
20944 PR tree-optimization/83857
20945 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
20946 vectorizable_live_operation for pure SLP statements.
20947 (vectorizable_live_operation): Handle PHIs.
20949 2018-01-16 Richard Biener <rguenther@suse.de>
20951 PR tree-optimization/83867
20952 * tree-vect-stmts.c (vect_transform_stmt): Precompute
20953 nested_in_vect_loop_p since the scalar stmt may get invalidated.
20955 2018-01-16 Jakub Jelinek <jakub@redhat.com>
20958 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
20959 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
20960 If off is not INTEGER_CST, issue a may not be aligned warning
20961 rather than isn't aligned. Use isn%'t rather than isn't.
20962 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
20964 <case MULT_EXPR>: Improve the case when bottom and one of the
20965 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
20966 operand, in that case check if the other operand is multiple of
20967 bottom divided by the INTEGER_CST operand.
20969 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
20972 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
20973 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
20974 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
20975 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
20976 * config/pa/pa.c (pa_function_arg_advance): Likewise.
20977 (pa_function_arg, pa_arg_partial_bytes): Likewise.
20978 (pa_function_arg_size): New function.
20980 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
20982 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
20983 in a separate statement.
20985 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
20987 PR tree-optimization/83847
20988 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
20989 group gathers and scatters.
20991 2018-01-16 Jakub Jelinek <jakub@redhat.com>
20993 PR rtl-optimization/86620
20994 * params.def (max-sched-ready-insns): Bump minimum value to 1.
20996 PR rtl-optimization/83213
20997 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
20998 to last if both are JUMP_INSNs.
21000 PR tree-optimization/83843
21001 * gimple-ssa-store-merging.c
21002 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
21003 store_immediate_info for bswap/nop orig_stores.
21005 2018-01-15 Andrew Waterman <andrew@sifive.com>
21007 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
21009 <UDIV>: Increase cost if !TARGET_DIV.
21011 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
21013 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
21014 (define_attr "cr_logical_3op"): New.
21015 (cceq_ior_compare): Adjust.
21016 (cceq_ior_compare_complement): Adjust.
21017 (*cceq_rev_compare): Adjust.
21018 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
21019 (is_cracked_insn): Adjust.
21020 (insn_must_be_first_in_group): Adjust.
21021 * config/rs6000/40x.md: Adjust.
21022 * config/rs6000/440.md: Adjust.
21023 * config/rs6000/476.md: Adjust.
21024 * config/rs6000/601.md: Adjust.
21025 * config/rs6000/603.md: Adjust.
21026 * config/rs6000/6xx.md: Adjust.
21027 * config/rs6000/7450.md: Adjust.
21028 * config/rs6000/7xx.md: Adjust.
21029 * config/rs6000/8540.md: Adjust.
21030 * config/rs6000/cell.md: Adjust.
21031 * config/rs6000/e300c2c3.md: Adjust.
21032 * config/rs6000/e500mc.md: Adjust.
21033 * config/rs6000/e500mc64.md: Adjust.
21034 * config/rs6000/e5500.md: Adjust.
21035 * config/rs6000/e6500.md: Adjust.
21036 * config/rs6000/mpc.md: Adjust.
21037 * config/rs6000/power4.md: Adjust.
21038 * config/rs6000/power5.md: Adjust.
21039 * config/rs6000/power6.md: Adjust.
21040 * config/rs6000/power7.md: Adjust.
21041 * config/rs6000/power8.md: Adjust.
21042 * config/rs6000/power9.md: Adjust.
21043 * config/rs6000/rs64.md: Adjust.
21044 * config/rs6000/titan.md: Adjust.
21046 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21048 * config/i386/predicates.md (indirect_branch_operand): Rewrite
21049 ix86_indirect_branch_register logic.
21051 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21053 * config/i386/constraints.md (Bs): Update
21054 ix86_indirect_branch_register check. Don't check
21055 ix86_indirect_branch_register with GOT_memory_operand.
21057 * config/i386/predicates.md (GOT_memory_operand): Don't check
21058 ix86_indirect_branch_register here.
21059 (GOT32_symbol_operand): Likewise.
21061 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21063 * config/i386/predicates.md (constant_call_address_operand):
21064 Rewrite ix86_indirect_branch_register logic.
21065 (sibcall_insn_operand): Likewise.
21067 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21069 * config/i386/constraints.md (Bs): Replace
21070 ix86_indirect_branch_thunk_register with
21071 ix86_indirect_branch_register.
21073 * config/i386/i386.md (indirect_jump): Likewise.
21074 (tablejump): Likewise.
21075 (*sibcall_memory): Likewise.
21076 (*sibcall_value_memory): Likewise.
21077 Peepholes of indirect call and jump via memory: Likewise.
21078 * config/i386/i386.opt: Likewise.
21079 * config/i386/predicates.md (indirect_branch_operand): Likewise.
21080 (GOT_memory_operand): Likewise.
21081 (call_insn_operand): Likewise.
21082 (sibcall_insn_operand): Likewise.
21083 (GOT32_symbol_operand): Likewise.
21085 2018-01-15 Jakub Jelinek <jakub@redhat.com>
21087 PR middle-end/83837
21088 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
21089 type rather than type addr's type points to.
21090 (expand_omp_atomic_mutex): Likewise.
21091 (expand_omp_atomic): Likewise.
21093 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21096 * config/i386/i386.c (output_indirect_thunk_function): Use
21097 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
21098 for __x86_return_thunk.
21100 2018-01-15 Richard Biener <rguenther@suse.de>
21102 PR middle-end/83850
21103 * expmed.c (extract_bit_field_1): Fix typo.
21105 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21108 * config/arm/iterators.md (VF): New mode iterator.
21109 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
21110 Remove integer-related logic from pattern.
21111 (neon_vabd<mode>_3): Likewise.
21113 2018-01-15 Jakub Jelinek <jakub@redhat.com>
21115 PR middle-end/82694
21116 * common.opt (fstrict-overflow): No longer an alias.
21117 (fwrapv-pointer): New option.
21118 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
21119 also for pointer types based on flag_wrapv_pointer.
21120 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
21121 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
21122 opts->x_flag_wrapv got set.
21123 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
21124 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
21125 POINTER_TYPE_OVERFLOW_UNDEFINED.
21126 * match.pd: Likewise in address comparison pattern.
21127 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
21129 2018-01-15 Richard Biener <rguenther@suse.de>
21132 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
21133 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
21134 Reset type names to their identifier if their TYPE_DECL doesn't
21135 have linkage (and thus is used for ODR and devirt).
21136 (save_debug_info_for_decl): Remove.
21137 (save_debug_info_for_type): Likewise.
21138 (add_tree_to_fld_list): Adjust.
21139 * tree-pretty-print.c (dump_generic_node): Make dumping of
21140 type names more robust.
21142 2018-01-15 Richard Biener <rguenther@suse.de>
21144 * BASE-VER: Bump to 8.0.1.
21146 2018-01-14 Martin Sebor <msebor@redhat.com>
21149 * builtins.c (check_access): Avoid warning when the no-warning bit
21152 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
21154 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
21155 * ira-color (allocno_hard_regs_compare): Likewise.
21157 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
21160 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
21161 Use .pushsection/.popsection.
21163 2018-01-14 Martin Sebor <msebor@redhat.com>
21166 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
21168 2018-01-14 Jakub Jelinek <jakub@redhat.com>
21170 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
21171 entry from extra_headers.
21172 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
21173 extra_headers, make the list bitwise identical to the i?86-*-* one.
21175 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21177 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
21178 -mcmodel=large with -mindirect-branch=thunk,
21179 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
21180 -mfunction-return=thunk-extern.
21181 * doc/invoke.texi: Document -mcmodel=large is incompatible with
21182 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
21183 -mfunction-return=thunk and -mfunction-return=thunk-extern.
21185 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21187 * config/i386/i386.c (print_reg): Print the name of the full
21188 integer register without '%'.
21189 (ix86_print_operand): Handle 'V'.
21190 * doc/extend.texi: Document 'V' modifier.
21192 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21194 * config/i386/constraints.md (Bs): Disallow memory operand for
21195 -mindirect-branch-register.
21197 * config/i386/predicates.md (indirect_branch_operand): Likewise.
21198 (GOT_memory_operand): Likewise.
21199 (call_insn_operand): Likewise.
21200 (sibcall_insn_operand): Likewise.
21201 (GOT32_symbol_operand): Likewise.
21202 * config/i386/i386.md (indirect_jump): Call convert_memory_address
21203 for -mindirect-branch-register.
21204 (tablejump): Likewise.
21205 (*sibcall_memory): Likewise.
21206 (*sibcall_value_memory): Likewise.
21207 Disallow peepholes of indirect call and jump via memory for
21208 -mindirect-branch-register.
21209 (*call_pop): Replace m with Bw.
21210 (*call_value_pop): Likewise.
21211 (*sibcall_pop_memory): Replace m with Bs.
21212 * config/i386/i386.opt (mindirect-branch-register): New option.
21213 * doc/invoke.texi: Document -mindirect-branch-register option.
21215 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21217 * config/i386/i386-protos.h (ix86_output_function_return): New.
21218 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
21219 set function_return_type.
21220 (indirect_thunk_name): Add ret_p to indicate thunk for function
21222 (output_indirect_thunk_function): Pass false to
21223 indirect_thunk_name.
21224 (ix86_output_indirect_branch_via_reg): Likewise.
21225 (ix86_output_indirect_branch_via_push): Likewise.
21226 (output_indirect_thunk_function): Create alias for function
21227 return thunk if regno < 0.
21228 (ix86_output_function_return): New function.
21229 (ix86_handle_fndecl_attribute): Handle function_return.
21230 (ix86_attribute_table): Add function_return.
21231 * config/i386/i386.h (machine_function): Add
21232 function_return_type.
21233 * config/i386/i386.md (simple_return_internal): Use
21234 ix86_output_function_return.
21235 (simple_return_internal_long): Likewise.
21236 * config/i386/i386.opt (mfunction-return=): New option.
21237 (indirect_branch): Mention -mfunction-return=.
21238 * doc/extend.texi: Document function_return function attribute.
21239 * doc/invoke.texi: Document -mfunction-return= option.
21241 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21243 * config/i386/i386-opts.h (indirect_branch): New.
21244 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
21245 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
21246 with local indirect jump when converting indirect call and jump.
21247 (ix86_set_indirect_branch_type): New.
21248 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
21249 (indirectlabelno): New.
21250 (indirect_thunk_needed): Likewise.
21251 (indirect_thunk_bnd_needed): Likewise.
21252 (indirect_thunks_used): Likewise.
21253 (indirect_thunks_bnd_used): Likewise.
21254 (INDIRECT_LABEL): Likewise.
21255 (indirect_thunk_name): Likewise.
21256 (output_indirect_thunk): Likewise.
21257 (output_indirect_thunk_function): Likewise.
21258 (ix86_output_indirect_branch_via_reg): Likewise.
21259 (ix86_output_indirect_branch_via_push): Likewise.
21260 (ix86_output_indirect_branch): Likewise.
21261 (ix86_output_indirect_jmp): Likewise.
21262 (ix86_code_end): Call output_indirect_thunk_function if needed.
21263 (ix86_output_call_insn): Call ix86_output_indirect_branch if
21265 (ix86_handle_fndecl_attribute): Handle indirect_branch.
21266 (ix86_attribute_table): Add indirect_branch.
21267 * config/i386/i386.h (machine_function): Add indirect_branch_type
21268 and has_local_indirect_jump.
21269 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
21271 (tablejump): Likewise.
21272 (*indirect_jump): Use ix86_output_indirect_jmp.
21273 (*tablejump_1): Likewise.
21274 (simple_return_indirect_internal): Likewise.
21275 * config/i386/i386.opt (mindirect-branch=): New option.
21276 (indirect_branch): New.
21279 (thunk-inline): Likewise.
21280 (thunk-extern): Likewise.
21281 * doc/extend.texi: Document indirect_branch function attribute.
21282 * doc/invoke.texi: Document -mindirect-branch= option.
21284 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
21287 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
21289 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
21291 * ipa-inline.c (want_inline_small_function_p): Return false if
21292 inlining has already failed with CIF_FINAL_ERROR.
21293 (update_caller_keys): Call want_inline_small_function_p before
21295 (update_callee_keys): Likewise.
21297 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21299 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
21301 (rs6000_quadword_masked_address_p): Likewise.
21302 (quad_aligned_load_p): Likewise.
21303 (quad_aligned_store_p): Likewise.
21304 (const_load_sequence_p): Add comment to describe the outer-most loop.
21305 (mimic_memory_attributes_and_flags): New function.
21306 (rs6000_gen_stvx): Likewise.
21307 (replace_swapped_aligned_store): Likewise.
21308 (rs6000_gen_lvx): Likewise.
21309 (replace_swapped_aligned_load): Likewise.
21310 (replace_swapped_load_constant): Capitalize argument name in
21311 comment describing this function.
21312 (rs6000_analyze_swaps): Add a third pass to search for vector loads
21313 and stores that access quad-word aligned addresses and replace
21314 with stvx or lvx instructions when appropriate.
21315 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
21316 New function prototype.
21317 (rs6000_quadword_masked_address_p): Likewise.
21318 (rs6000_gen_lvx): Likewise.
21319 (rs6000_gen_stvx): Likewise.
21320 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
21321 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
21322 when memory address is aligned.
21323 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
21324 this split to select lvx instruction when memory address is aligned.
21325 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
21326 instruction when memory address is aligned.
21327 (*vsx_le_perm_load_v16qi): Likewise.
21328 (four unnamed splitters): Modify to select the stvx instruction
21329 when memory is aligned.
21331 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
21333 * predict.c (determine_unlikely_bbs): Handle correctly BBs
21334 which appears in the queue multiple times.
21336 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21337 Alan Hayward <alan.hayward@arm.com>
21338 David Sherwood <david.sherwood@arm.com>
21340 * tree-vectorizer.h (vec_lower_bound): New structure.
21341 (_loop_vec_info): Add check_nonzero and lower_bounds.
21342 (LOOP_VINFO_CHECK_NONZERO): New macro.
21343 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
21344 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
21345 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
21346 fields. Make seg_len the distance travelled, not including the
21348 (dr_direction_indicator): Declare.
21349 (dr_zero_step_indicator): Likewise.
21350 (dr_known_forward_stride_p): Likewise.
21351 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
21353 (runtime_alias_check_p): Allow runtime alias checks with
21355 (operator ==): Compare access_size and align.
21356 (prune_runtime_alias_test_list): Rework for new distinction between
21357 the access_size and seg_len.
21358 (create_intersect_range_checks_index): Likewise. Cope with polynomial
21360 (get_segment_min_max): New function.
21361 (create_intersect_range_checks): Use it.
21362 (dr_step_indicator): New function.
21363 (dr_direction_indicator): Likewise.
21364 (dr_zero_step_indicator): Likewise.
21365 (dr_known_forward_stride_p): Likewise.
21366 * tree-loop-distribution.c (data_ref_segment_size): Return
21367 DR_STEP * (niters - 1).
21368 (compute_alias_check_pairs): Update call to the dr_with_seg_len
21370 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
21371 (vect_preserves_scalar_order_p): New function, split out from...
21372 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
21373 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
21374 (vect_vfa_access_size): New function.
21375 (vect_vfa_align): Likewise.
21376 (vect_compile_time_alias): Take access_size_a and access_b arguments.
21377 (dump_lower_bound): New function.
21378 (vect_check_lower_bound): Likewise.
21379 (vect_small_gap_p): Likewise.
21380 (vectorizable_with_step_bound_p): Likewise.
21381 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
21382 depencies if the vectorization factor is 1. Convert the checks
21383 for nonzero steps into checks on the bounds of DR_STEP. Try using
21384 a bunds check for variable steps if the minimum required step is
21385 relatively small. Update calls to the dr_with_seg_len
21386 constructor and to vect_compile_time_alias.
21387 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
21389 (vect_loop_versioning): Call it.
21390 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
21392 (vect_estimate_min_profitable_iters): Account for any bounds checks.
21394 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21395 Alan Hayward <alan.hayward@arm.com>
21396 David Sherwood <david.sherwood@arm.com>
21398 * doc/sourcebuild.texi (vect_scatter_store): Document.
21399 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
21401 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
21403 * genopinit.c (main): Add supports_vec_scatter_store and
21404 supports_vec_scatter_store_cached to target_optabs.
21405 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
21406 IFN_MASK_SCATTER_STORE.
21407 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
21409 * internal-fn.h (internal_store_fn_p): Declare.
21410 (internal_fn_stored_value_index): Likewise.
21411 * internal-fn.c (scatter_store_direct): New macro.
21412 (expand_scatter_store_optab_fn): New function.
21413 (direct_scatter_store_optab_supported_p): New macro.
21414 (internal_store_fn_p): New function.
21415 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
21416 IFN_MASK_SCATTER_STORE.
21417 (internal_fn_mask_index): Likewise.
21418 (internal_fn_stored_value_index): New function.
21419 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
21420 for scatter stores.
21421 * optabs-query.h (supports_vec_scatter_store_p): Declare.
21422 * optabs-query.c (supports_vec_scatter_store_p): New function.
21423 * tree-vectorizer.h (vect_get_store_rhs): Declare.
21424 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
21425 true for scatter stores.
21426 (vect_gather_scatter_fn_p): Handle scatter stores too.
21427 (vect_check_gather_scatter): Consider using scatter stores if
21428 supports_vec_scatter_store_p.
21429 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
21430 scatter stores too.
21431 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
21432 internal_fn_stored_value_index.
21433 (check_load_store_masking): Handle scatter stores too.
21434 (vect_get_store_rhs): Make public.
21435 (vectorizable_call): Use internal_store_fn_p.
21436 (vectorizable_store): Handle scatter store internal functions.
21437 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
21438 when deciding whether the end of the group has been reached.
21439 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
21440 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
21441 (mask_scatter_store<mode>): New insns.
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_gather_scatter_fn_p): Declare.
21448 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
21449 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
21451 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
21452 Use vect_truncate_gather_scatter_offset if we can't treat the
21453 operation as a normal gather load or scatter store.
21454 (get_group_load_store_type): Take the gather_scatter_info
21455 as argument. Try using a gather load or scatter store for
21456 single-element groups.
21457 (get_load_store_type): Update calls to get_group_load_store_type
21458 and vect_use_strided_gather_scatters_p.
21460 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21461 Alan Hayward <alan.hayward@arm.com>
21462 David Sherwood <david.sherwood@arm.com>
21464 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
21465 optional tree argument.
21466 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
21468 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
21469 but continue to use the current value as a fallback.
21470 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
21471 to compare the updates.
21472 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
21473 (get_load_store_type): Use it when handling a strided access.
21474 (vect_get_strided_load_store_ops): New function.
21475 (vect_get_data_ptr_increment): Likewise.
21476 (vectorizable_load): Handle strided gather loads. Always pass
21477 a step to vect_create_data_ref_ptr and bump_vector_ptr.
21479 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21480 Alan Hayward <alan.hayward@arm.com>
21481 David Sherwood <david.sherwood@arm.com>
21483 * doc/md.texi (gather_load@var{m}): Document.
21484 (mask_gather_load@var{m}): Likewise.
21485 * genopinit.c (main): Add supports_vec_gather_load and
21486 supports_vec_gather_load_cached to target_optabs.
21487 * optabs-tree.c (init_tree_optimization_optabs): Use
21488 ggc_cleared_alloc to allocate target_optabs.
21489 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
21490 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
21492 * internal-fn.h (internal_load_fn_p): Declare.
21493 (internal_gather_scatter_fn_p): Likewise.
21494 (internal_fn_mask_index): Likewise.
21495 (internal_gather_scatter_fn_supported_p): Likewise.
21496 * internal-fn.c (gather_load_direct): New macro.
21497 (expand_gather_load_optab_fn): New function.
21498 (direct_gather_load_optab_supported_p): New macro.
21499 (direct_internal_fn_optab): New function.
21500 (internal_load_fn_p): Likewise.
21501 (internal_gather_scatter_fn_p): Likewise.
21502 (internal_fn_mask_index): Likewise.
21503 (internal_gather_scatter_fn_supported_p): Likewise.
21504 * optabs-query.c (supports_at_least_one_mode_p): New function.
21505 (supports_vec_gather_load_p): Likewise.
21506 * optabs-query.h (supports_vec_gather_load_p): Declare.
21507 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
21508 and memory_type field.
21509 (NUM_PATTERNS): Bump to 15.
21510 * tree-vect-data-refs.c: Include internal-fn.h.
21511 (vect_gather_scatter_fn_p): New function.
21512 (vect_describe_gather_scatter_call): Likewise.
21513 (vect_check_gather_scatter): Try using internal functions for
21514 gather loads. Recognize existing calls to a gather load function.
21515 (vect_analyze_data_refs): Consider using gather loads if
21516 supports_vec_gather_load_p.
21517 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
21518 (vect_get_gather_scatter_offset_type): Likewise.
21519 (vect_convert_mask_for_vectype): Likewise.
21520 (vect_add_conversion_to_patterm): Likewise.
21521 (vect_try_gather_scatter_pattern): Likewise.
21522 (vect_recog_gather_scatter_pattern): New pattern recognizer.
21523 (vect_vect_recog_func_ptrs): Add it.
21524 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
21525 internal_fn_mask_index and internal_gather_scatter_fn_p.
21526 (check_load_store_masking): Take the gather_scatter_info as an
21527 argument and handle gather loads.
21528 (vect_get_gather_scatter_ops): New function.
21529 (vectorizable_call): Check internal_load_fn_p.
21530 (vectorizable_load): Likewise. Handle gather load internal
21532 (vectorizable_store): Update call to check_load_store_masking.
21533 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
21534 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
21535 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
21536 (aarch64_gather_scale_operand_d): New predicates.
21537 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
21538 (mask_gather_load<mode>): New insns.
21540 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21541 Alan Hayward <alan.hayward@arm.com>
21542 David Sherwood <david.sherwood@arm.com>
21544 * optabs.def (fold_left_plus_optab): New optab.
21545 * doc/md.texi (fold_left_plus_@var{m}): Document.
21546 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
21547 * internal-fn.c (fold_left_direct): Define.
21548 (expand_fold_left_optab_fn): Likewise.
21549 (direct_fold_left_optab_supported_p): Likewise.
21550 * fold-const-call.c (fold_const_fold_left): New function.
21551 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
21552 * tree-parloops.c (valid_reduction_p): New function.
21553 (gather_scalar_reductions): Use it.
21554 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
21555 (vect_finish_replace_stmt): Declare.
21556 * tree-vect-loop.c (fold_left_reduction_fn): New function.
21557 (needs_fold_left_reduction_p): New function, split out from...
21558 (vect_is_simple_reduction): ...here. Accept reductions that
21559 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
21560 (vect_force_simple_reduction): Also store the reduction type in
21561 the assignment's STMT_VINFO_REDUC_TYPE.
21562 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
21563 (merge_with_identity): New function.
21564 (vect_expand_fold_left): Likewise.
21565 (vectorize_fold_left_reduction): Likewise.
21566 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
21567 scalar phi in place for it. Check for target support and reject
21568 cases that would reassociate the operation. Defer the transform
21569 phase to vectorize_fold_left_reduction.
21570 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
21571 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
21572 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
21574 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21576 * tree-if-conv.c (predicate_mem_writes): Remove redundant
21577 call to ifc_temp_var.
21579 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21580 Alan Hayward <alan.hayward@arm.com>
21581 David Sherwood <david.sherwood@arm.com>
21583 * target.def (legitimize_address_displacement): Take the original
21584 offset as a poly_int.
21585 * targhooks.h (default_legitimize_address_displacement): Update
21587 * targhooks.c (default_legitimize_address_displacement): Likewise.
21588 * doc/tm.texi: Regenerate.
21589 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
21590 as an argument, moving assert of ad->disp == ad->disp_term to...
21591 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
21592 Try calling targetm.legitimize_address_displacement before expanding
21593 the address rather than afterwards, and adjust for the new interface.
21594 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
21595 Match the new hook interface. Handle SVE addresses.
21596 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
21597 new hook interface.
21599 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21601 * Makefile.in (OBJS): Add early-remat.o.
21602 * target.def (select_early_remat_modes): New hook.
21603 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
21604 * doc/tm.texi: Regenerate.
21605 * targhooks.h (default_select_early_remat_modes): Declare.
21606 * targhooks.c (default_select_early_remat_modes): New function.
21607 * timevar.def (TV_EARLY_REMAT): New timevar.
21608 * passes.def (pass_early_remat): New pass.
21609 * tree-pass.h (make_pass_early_remat): Declare.
21610 * early-remat.c: New file.
21611 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
21613 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
21615 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21616 Alan Hayward <alan.hayward@arm.com>
21617 David Sherwood <david.sherwood@arm.com>
21619 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
21620 vfm1 with a bound_epilog parameter.
21621 (vect_do_peeling): Update calls accordingly, and move the prologue
21622 call earlier in the function. Treat the base bound_epilog as 0 for
21623 fully-masked loops and retain vf - 1 for other loops. Add 1 to
21624 this base when peeling for gaps.
21625 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
21626 with fully-masked loops.
21627 (vect_estimate_min_profitable_iters): Handle the single peeled
21628 iteration in that case.
21630 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21631 Alan Hayward <alan.hayward@arm.com>
21632 David Sherwood <david.sherwood@arm.com>
21634 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
21635 single-element interleaving even if the size is not a power of 2.
21636 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
21637 accesses for single-element interleaving if the group size is
21640 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21641 Alan Hayward <alan.hayward@arm.com>
21642 David Sherwood <david.sherwood@arm.com>
21644 * doc/md.texi (fold_extract_last_@var{m}): Document.
21645 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
21646 * optabs.def (fold_extract_last_optab): New optab.
21647 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
21648 * internal-fn.c (fold_extract_direct): New macro.
21649 (expand_fold_extract_optab_fn): Likewise.
21650 (direct_fold_extract_optab_supported_p): Likewise.
21651 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
21652 * tree-vect-loop.c (vect_model_reduction_cost): Handle
21653 EXTRACT_LAST_REDUCTION.
21654 (get_initial_def_for_reduction): Do not create an initial vector
21655 for EXTRACT_LAST_REDUCTION reductions.
21656 (vectorizable_reduction): Leave the scalar phi in place for
21657 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
21658 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
21659 epilogue code for EXTRACT_LAST_REDUCTION and defer the
21660 transform phase to vectorizable_condition.
21661 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
21663 (vect_finish_stmt_generation): ...here.
21664 (vect_finish_replace_stmt): New function.
21665 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
21666 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
21668 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
21670 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21671 Alan Hayward <alan.hayward@arm.com>
21672 David Sherwood <david.sherwood@arm.com>
21674 * doc/md.texi (extract_last_@var{m}): Document.
21675 * optabs.def (extract_last_optab): New optab.
21676 * internal-fn.def (EXTRACT_LAST): New internal function.
21677 * internal-fn.c (cond_unary_direct): New macro.
21678 (expand_cond_unary_optab_fn): Likewise.
21679 (direct_cond_unary_optab_supported_p): Likewise.
21680 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
21681 loops using EXTRACT_LAST.
21682 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
21683 (extract_last_<mode>): ...this optab.
21684 (vec_extract<mode><Vel>): Update accordingly.
21686 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21687 Alan Hayward <alan.hayward@arm.com>
21688 David Sherwood <david.sherwood@arm.com>
21690 * target.def (empty_mask_is_expensive): New hook.
21691 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
21692 * doc/tm.texi: Regenerate.
21693 * targhooks.h (default_empty_mask_is_expensive): Declare.
21694 * targhooks.c (default_empty_mask_is_expensive): New function.
21695 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
21696 if the target says that empty masks are expensive.
21697 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
21699 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
21701 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21702 Alan Hayward <alan.hayward@arm.com>
21703 David Sherwood <david.sherwood@arm.com>
21705 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
21706 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
21707 (vect_use_loop_mask_for_alignment_p): New function.
21708 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
21709 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
21710 niters_skip argument. Make sure that the first niters_skip elements
21711 of the first iteration are inactive.
21712 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
21713 Update call to vect_set_loop_masks_directly.
21714 (get_misalign_in_elems): New function, split out from...
21715 (vect_gen_prolog_loop_niters): ...here.
21716 (vect_update_init_of_dr): Take a code argument that specifies whether
21717 the adjustment should be added or subtracted.
21718 (vect_update_init_of_drs): Likewise.
21719 (vect_prepare_for_masked_peels): New function.
21720 (vect_do_peeling): Skip prologue peeling if we're using a mask
21721 instead. Update call to vect_update_inits_of_drs.
21722 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
21724 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
21725 alignment. Do not include the number of peeled iterations in
21726 the minimum threshold in that case.
21727 (vectorizable_induction): Adjust the start value down by
21728 LOOP_VINFO_MASK_SKIP_NITERS iterations.
21729 (vect_transform_loop): Call vect_prepare_for_masked_peels.
21730 Take the number of skipped iterations into account when calculating
21732 * tree-vect-stmts.c (vect_gen_while_not): New function.
21734 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21735 Alan Hayward <alan.hayward@arm.com>
21736 David Sherwood <david.sherwood@arm.com>
21738 * doc/sourcebuild.texi (vect_fully_masked): Document.
21739 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
21740 default value to 0.
21741 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
21743 (vect_analyze_loop_2): ...here. Don't check the vectorization
21744 factor against the number of loop iterations if the loop is
21747 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21748 Alan Hayward <alan.hayward@arm.com>
21749 David Sherwood <david.sherwood@arm.com>
21751 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
21752 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
21753 (dump_groups): Update accordingly.
21754 (iv_use::mem_type): New member variable.
21755 (address_p): New function.
21756 (record_use): Add a mem_type argument and initialize the new
21758 (record_group_use): Add a mem_type argument. Use address_p.
21759 Remove obsolete null checks of base_object. Update call to record_use.
21760 (find_interesting_uses_op): Update call to record_group_use.
21761 (find_interesting_uses_cond): Likewise.
21762 (find_interesting_uses_address): Likewise.
21763 (get_mem_type_for_internal_fn): New function.
21764 (find_address_like_use): Likewise.
21765 (find_interesting_uses_stmt): Try find_address_like_use before
21766 calling find_interesting_uses_op.
21767 (addr_offset_valid_p): Use the iv mem_type field as the type
21768 of the addressed memory.
21769 (add_autoinc_candidates): Likewise.
21770 (get_address_cost): Likewise.
21771 (split_small_address_groups_p): Use address_p.
21772 (split_address_groups): Likewise.
21773 (add_iv_candidate_for_use): Likewise.
21774 (autoinc_possible_for_pair): Likewise.
21775 (rewrite_groups): Likewise.
21776 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
21777 (determine_group_iv_cost): Update after split of USE_ADDRESS.
21778 (get_alias_ptr_type_for_ptr_address): New function.
21779 (rewrite_use_address): Rewrite address uses in calls that were
21780 identified by find_address_like_use.
21782 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21783 Alan Hayward <alan.hayward@arm.com>
21784 David Sherwood <david.sherwood@arm.com>
21786 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
21788 * gimple-expr.h (is_gimple_addressable: Likewise.
21789 * gimple-expr.c (is_gimple_address): Likewise.
21790 * internal-fn.c (expand_call_mem_ref): New function.
21791 (expand_mask_load_optab_fn): Use it.
21792 (expand_mask_store_optab_fn): Likewise.
21794 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21795 Alan Hayward <alan.hayward@arm.com>
21796 David Sherwood <david.sherwood@arm.com>
21798 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
21799 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
21800 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
21801 (cond_umax@var{mode}): Document.
21802 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
21803 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
21804 (cond_umin_optab, cond_umax_optab): New optabs.
21805 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
21806 (COND_IOR, COND_XOR): New internal functions.
21807 * internal-fn.h (get_conditional_internal_fn): Declare.
21808 * internal-fn.c (cond_binary_direct): New macro.
21809 (expand_cond_binary_optab_fn): Likewise.
21810 (direct_cond_binary_optab_supported_p): Likewise.
21811 (get_conditional_internal_fn): New function.
21812 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
21813 Cope with reduction statements that are vectorized as calls rather
21815 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
21816 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
21817 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
21818 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
21819 (UNSPEC_COND_EOR): New unspecs.
21820 (optab): Add mappings for them.
21821 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
21822 (sve_int_op, sve_fp_op): New int attributes.
21824 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21825 Alan Hayward <alan.hayward@arm.com>
21826 David Sherwood <david.sherwood@arm.com>
21828 * optabs.def (while_ult_optab): New optab.
21829 * doc/md.texi (while_ult@var{m}@var{n}): Document.
21830 * internal-fn.def (WHILE_ULT): New internal function.
21831 * internal-fn.h (direct_internal_fn_supported_p): New override
21832 that takes two types as argument.
21833 * internal-fn.c (while_direct): New macro.
21834 (expand_while_optab_fn): New function.
21835 (convert_optab_supported_p): Likewise.
21836 (direct_while_optab_supported_p): New macro.
21837 * wide-int.h (wi::udiv_ceil): New function.
21838 * tree-vectorizer.h (rgroup_masks): New structure.
21839 (vec_loop_masks): New typedef.
21840 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
21841 and fully_masked_p.
21842 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
21843 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
21844 (vect_max_vf): New function.
21845 (slpeel_make_loop_iterate_ntimes): Delete.
21846 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
21847 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
21848 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
21849 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
21850 internal-fn.h, stor-layout.h and optabs-query.h.
21851 (vect_set_loop_mask): New function.
21852 (add_preheader_seq): Likewise.
21853 (add_header_seq): Likewise.
21854 (interleave_supported_p): Likewise.
21855 (vect_maybe_permute_loop_masks): Likewise.
21856 (vect_set_loop_masks_directly): Likewise.
21857 (vect_set_loop_condition_masked): Likewise.
21858 (vect_set_loop_condition_unmasked): New function, split out from
21859 slpeel_make_loop_iterate_ntimes.
21860 (slpeel_make_loop_iterate_ntimes): Rename to..
21861 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
21862 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
21863 (vect_do_peeling): Update call accordingly.
21864 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
21866 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
21867 mask_compare_type, can_fully_mask_p and fully_masked_p.
21868 (release_vec_loop_masks): New function.
21869 (_loop_vec_info): Use it to free the loop masks.
21870 (can_produce_all_loop_masks_p): New function.
21871 (vect_get_max_nscalars_per_iter): Likewise.
21872 (vect_verify_full_masking): Likewise.
21873 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
21874 retries, and free the mask rgroups before retrying. Check loop-wide
21875 reasons for disallowing fully-masked loops. Make the final decision
21876 about whether use a fully-masked loop or not.
21877 (vect_estimate_min_profitable_iters): Do not assume that peeling
21878 for the number of iterations will be needed for fully-masked loops.
21879 (vectorizable_reduction): Disable fully-masked loops.
21880 (vectorizable_live_operation): Likewise.
21881 (vect_halve_mask_nunits): New function.
21882 (vect_double_mask_nunits): Likewise.
21883 (vect_record_loop_mask): Likewise.
21884 (vect_get_loop_mask): Likewise.
21885 (vect_transform_loop): Handle the case in which the final loop
21886 iteration might handle a partial vector. Call vect_set_loop_condition
21887 instead of slpeel_make_loop_iterate_ntimes.
21888 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
21889 (check_load_store_masking): New function.
21890 (prepare_load_store_mask): Likewise.
21891 (vectorizable_store): Handle fully-masked loops.
21892 (vectorizable_load): Likewise.
21893 (supportable_widening_operation): Use vect_halve_mask_nunits for
21895 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
21896 (vect_gen_while): New function.
21897 * config/aarch64/aarch64.md (umax<mode>3): New expander.
21898 (aarch64_uqdec<mode>): New insn.
21900 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21901 Alan Hayward <alan.hayward@arm.com>
21902 David Sherwood <david.sherwood@arm.com>
21904 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
21905 (reduc_xor_scal_optab): New optabs.
21906 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
21907 (reduc_xor_scal_@var{m}): Document.
21908 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
21909 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
21910 internal functions.
21911 * fold-const-call.c (fold_const_call): Handle them.
21912 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
21913 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
21914 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
21915 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
21916 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
21917 (UNSPEC_XORV): New unspecs.
21918 (optab): Add entries for them.
21919 (BITWISEV): New int iterator.
21920 (bit_reduc_op): New int attributes.
21922 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21923 Alan Hayward <alan.hayward@arm.com>
21924 David Sherwood <david.sherwood@arm.com>
21926 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
21927 * internal-fn.def (VEC_SHL_INSERT): New internal function.
21928 * optabs.def (vec_shl_insert_optab): New optab.
21929 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
21930 (duplicate_and_interleave): Likewise.
21931 * tree-vect-loop.c: Include internal-fn.h.
21932 (neutral_op_for_slp_reduction): New function, split out from
21933 get_initial_defs_for_reduction.
21934 (get_initial_def_for_reduction): Handle option 2 for variable-length
21935 vectors by loading the neutral value into a vector and then shifting
21936 the initial value into element 0.
21937 (get_initial_defs_for_reduction): Replace the code argument with
21938 the neutral value calculated by neutral_op_for_slp_reduction.
21939 Use gimple_build_vector for constant-length vectors.
21940 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
21941 but the first group_size elements have a neutral value.
21942 Use duplicate_and_interleave otherwise.
21943 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
21944 Update call to get_initial_defs_for_reduction. Handle SLP
21945 reductions for variable-length vectors by creating one vector
21946 result for each scalar result, with the elements associated
21947 with other scalar results stubbed out with the neutral value.
21948 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
21949 Require IFN_VEC_SHL_INSERT for double reductions on
21950 variable-length vectors, or SLP reductions that have
21951 a neutral value. Require can_duplicate_and_interleave_p
21952 support for variable-length unchained SLP reductions if there
21953 is no neutral value, such as for MIN/MAX reductions. Also require
21954 the number of vector elements to be a multiple of the number of
21955 SLP statements when doing variable-length unchained SLP reductions.
21956 Update call to vect_create_epilog_for_reduction.
21957 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
21958 and remove initial values.
21959 (duplicate_and_interleave): Make public.
21960 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
21961 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
21963 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21964 Alan Hayward <alan.hayward@arm.com>
21965 David Sherwood <david.sherwood@arm.com>
21967 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
21968 (can_duplicate_and_interleave_p): New function.
21969 (vect_get_and_check_slp_defs): Take the vector of statements
21970 rather than just the current one. Remove excess parentheses.
21971 Restriction rejectinon of vect_constant_def and vect_external_def
21972 for variable-length vectors to boolean types, or types for which
21973 can_duplicate_and_interleave_p is false.
21974 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
21975 (duplicate_and_interleave): New function.
21976 (vect_get_constant_vectors): Use gimple_build_vector for
21977 constant-length vectors and suitable variable-length constant
21978 vectors. Use duplicate_and_interleave for other variable-length
21979 vectors. Don't defer the update when inserting new statements.
21981 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21982 Alan Hayward <alan.hayward@arm.com>
21983 David Sherwood <david.sherwood@arm.com>
21985 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
21986 min_profitable_iters doesn't go negative.
21988 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21989 Alan Hayward <alan.hayward@arm.com>
21990 David Sherwood <david.sherwood@arm.com>
21992 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
21993 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
21994 * optabs.def (vec_mask_load_lanes_optab): New optab.
21995 (vec_mask_store_lanes_optab): Likewise.
21996 * internal-fn.def (MASK_LOAD_LANES): New internal function.
21997 (MASK_STORE_LANES): Likewise.
21998 * internal-fn.c (mask_load_lanes_direct): New macro.
21999 (mask_store_lanes_direct): Likewise.
22000 (expand_mask_load_optab_fn): Handle masked operations.
22001 (expand_mask_load_lanes_optab_fn): New macro.
22002 (expand_mask_store_optab_fn): Handle masked operations.
22003 (expand_mask_store_lanes_optab_fn): New macro.
22004 (direct_mask_load_lanes_optab_supported_p): Likewise.
22005 (direct_mask_store_lanes_optab_supported_p): Likewise.
22006 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
22008 (vect_load_lanes_supported): Likewise.
22009 * tree-vect-data-refs.c (strip_conversion): New function.
22010 (can_group_stmts_p): Likewise.
22011 (vect_analyze_data_ref_accesses): Use it instead of checking
22012 for a pair of assignments.
22013 (vect_store_lanes_supported): Take a masked_p parameter.
22014 (vect_load_lanes_supported): Likewise.
22015 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
22016 vect_store_lanes_supported and vect_load_lanes_supported.
22017 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
22018 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
22019 parameter. Don't allow gaps for masked accesses.
22020 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
22021 and vect_load_lanes_supported.
22022 (get_load_store_type): Take a masked_p parameter and update
22023 call to get_group_load_store_type.
22024 (vectorizable_store): Update call to get_load_store_type.
22025 Handle IFN_MASK_STORE_LANES.
22026 (vectorizable_load): Update call to get_load_store_type.
22027 Handle IFN_MASK_LOAD_LANES.
22029 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22030 Alan Hayward <alan.hayward@arm.com>
22031 David Sherwood <david.sherwood@arm.com>
22033 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
22035 * config/aarch64/aarch64-protos.h
22036 (aarch64_sve_struct_memory_operand_p): Declare.
22037 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
22038 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
22039 (VPRED, vpred): Handle SVE structure modes.
22040 * config/aarch64/constraints.md (Utx): New constraint.
22041 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
22042 (aarch64_sve_struct_nonimmediate_operand): New predicates.
22043 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
22044 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
22045 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
22046 structure modes. Split into pieces after RA.
22047 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
22048 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
22050 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
22051 SVE structure modes.
22052 (aarch64_classify_address): Likewise.
22053 (sizetochar): Move earlier in file.
22054 (aarch64_print_operand): Handle SVE register lists.
22055 (aarch64_array_mode): New function.
22056 (aarch64_sve_struct_memory_operand_p): Likewise.
22057 (TARGET_ARRAY_MODE): Redefine.
22059 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22060 Alan Hayward <alan.hayward@arm.com>
22061 David Sherwood <david.sherwood@arm.com>
22063 * target.def (array_mode): New target hook.
22064 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
22065 * doc/tm.texi: Regenerate.
22066 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
22067 * hooks.c (hook_optmode_mode_uhwi_none): New function.
22068 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
22069 targetm.array_mode.
22070 * stor-layout.c (mode_for_array): Likewise. Support polynomial
22073 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22074 Alan Hayward <alan.hayward@arm.com>
22075 David Sherwood <david.sherwood@arm.com>
22077 * fold-const.c (fold_binary_loc): Check the argument types
22078 rather than the result type when testing for a vector operation.
22080 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22082 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
22083 * doc/tm.texi: Regenerate.
22085 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22086 Alan Hayward <alan.hayward@arm.com>
22087 David Sherwood <david.sherwood@arm.com>
22089 * doc/invoke.texi (-msve-vector-bits=): Document new option.
22090 (sve): Document new AArch64 extension.
22091 * doc/md.texi (w): Extend the description of the AArch64
22092 constraint to include SVE vectors.
22093 (Upl, Upa): Document new AArch64 predicate constraints.
22094 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
22096 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
22097 (msve-vector-bits=): New option.
22098 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
22099 SVE when these are disabled.
22100 (sve): New extension.
22101 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
22102 modes. Adjust their number of units based on aarch64_sve_vg.
22103 (MAX_BITSIZE_MODE_ANY_MODE): Define.
22104 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
22105 aarch64_addr_query_type.
22106 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
22107 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
22108 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
22109 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
22110 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
22111 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
22112 (aarch64_simd_imm_zero_p): Delete.
22113 (aarch64_check_zero_based_sve_index_immediate): Declare.
22114 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
22115 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
22116 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
22117 (aarch64_sve_float_mul_immediate_p): Likewise.
22118 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
22119 rather than an rtx.
22120 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
22121 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
22122 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
22123 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
22124 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
22125 (aarch64_regmode_natural_size): Likewise.
22126 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
22127 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
22129 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
22130 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
22131 for VG and the SVE predicate registers.
22132 (V_ALIASES): Add a "z"-prefixed alias.
22133 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
22134 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
22135 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
22136 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
22137 (REG_CLASS_NAMES): Add entries for them.
22138 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
22139 and the predicate registers.
22140 (aarch64_sve_vg): Declare.
22141 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
22142 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
22143 (REGMODE_NATURAL_SIZE): Define.
22144 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
22146 * config/aarch64/aarch64.c: Include cfgrtl.h.
22147 (simd_immediate_info): Add a constructor for series vectors,
22148 and an associated step field.
22149 (aarch64_sve_vg): New variable.
22150 (aarch64_dbx_register_number): Handle VG and the predicate registers.
22151 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
22152 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
22153 (VEC_ANY_DATA, VEC_STRUCT): New constants.
22154 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
22155 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
22156 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
22157 (aarch64_get_mask_mode): New functions.
22158 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
22159 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
22160 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
22161 predicate modes and predicate registers. Explicitly restrict
22162 GPRs to modes of 16 bytes or smaller. Only allow FP registers
22163 to store a vector mode if it is recognized by
22164 aarch64_classify_vector_mode.
22165 (aarch64_regmode_natural_size): New function.
22166 (aarch64_hard_regno_caller_save_mode): Return the original mode
22168 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
22169 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
22170 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
22171 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
22173 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
22174 does not overlap dest if the function is frame-related. Handle
22176 (aarch64_split_add_offset): New function.
22177 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
22178 them aarch64_add_offset.
22179 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
22180 and update call to aarch64_sub_sp.
22181 (aarch64_add_cfa_expression): New function.
22182 (aarch64_expand_prologue): Pass extra temporary registers to the
22183 functions above. Handle the case in which we need to emit new
22184 DW_CFA_expressions for registers that were originally saved
22185 relative to the stack pointer, but now have to be expressed
22186 relative to the frame pointer.
22187 (aarch64_output_mi_thunk): Pass extra temporary registers to the
22189 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
22190 IP0 and IP1 values for SVE frames.
22191 (aarch64_expand_vec_series): New function.
22192 (aarch64_expand_sve_widened_duplicate): Likewise.
22193 (aarch64_expand_sve_const_vector): Likewise.
22194 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
22195 Handle SVE constants. Use emit_move_insn to move a force_const_mem
22196 into the register, rather than emitting a SET directly.
22197 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
22198 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
22199 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
22200 (offset_9bit_signed_scaled_p): New functions.
22201 (aarch64_replicate_bitmask_imm): New function.
22202 (aarch64_bitmask_imm): Use it.
22203 (aarch64_cannot_force_const_mem): Reject expressions involving
22204 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
22205 (aarch64_classify_index): Handle SVE indices, by requiring
22206 a plain register index with a scale that matches the element size.
22207 (aarch64_classify_address): Handle SVE addresses. Assert that
22208 the mode of the address is VOIDmode or an integer mode.
22209 Update call to aarch64_classify_symbol.
22210 (aarch64_classify_symbolic_expression): Update call to
22211 aarch64_classify_symbol.
22212 (aarch64_const_vec_all_in_range_p): New function.
22213 (aarch64_print_vector_float_operand): Likewise.
22214 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
22215 "vN" for FP registers with SVE modes. Handle (const ...) vectors
22216 and the FP immediates 1.0 and 0.5.
22217 (aarch64_print_address_internal): Handle SVE addresses.
22218 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
22219 (aarch64_regno_regclass): Handle predicate registers.
22220 (aarch64_secondary_reload): Handle big-endian reloads of SVE
22222 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
22223 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
22224 (aarch64_convert_sve_vector_bits): New function.
22225 (aarch64_override_options): Use it to handle -msve-vector-bits=.
22226 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
22227 rather than an rtx.
22228 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
22229 Handle SVE vector and predicate modes. Accept VL-based constants
22230 that need only one temporary register, and VL offsets that require
22231 no temporary registers.
22232 (aarch64_conditional_register_usage): Mark the predicate registers
22233 as fixed if SVE isn't available.
22234 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
22235 Return true for SVE vector and predicate modes.
22236 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
22237 rather than an unsigned int. Handle SVE modes.
22238 (aarch64_preferred_simd_mode): Update call accordingly. Handle
22240 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
22242 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
22243 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
22244 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
22245 (aarch64_sve_float_mul_immediate_p): New functions.
22246 (aarch64_sve_valid_immediate): New function.
22247 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
22248 Explicitly reject structure modes. Check for INDEX constants.
22249 Handle PTRUE and PFALSE constants.
22250 (aarch64_check_zero_based_sve_index_immediate): New function.
22251 (aarch64_simd_imm_zero_p): Delete.
22252 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
22253 vector modes. Accept constants in the range of CNT[BHWD].
22254 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
22255 ask for an Advanced SIMD mode.
22256 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
22257 (aarch64_simd_vector_alignment): Handle SVE predicates.
22258 (aarch64_vectorize_preferred_vector_alignment): New function.
22259 (aarch64_simd_vector_alignment_reachable): Use it instead of
22261 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
22262 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
22264 (MAX_VECT_LEN): Delete.
22265 (expand_vec_perm_d): Add a vec_flags field.
22266 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
22267 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
22268 (aarch64_evpc_ext): Don't apply a big-endian lane correction
22270 (aarch64_evpc_rev): Rename to...
22271 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
22272 (aarch64_evpc_rev_global): New function.
22273 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
22274 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
22276 (aarch64_evpc_sve_tbl): New function.
22277 (aarch64_expand_vec_perm_const_1): Update after rename of
22278 aarch64_evpc_rev. Handle SVE permutes too, trying
22279 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
22280 than aarch64_evpc_tbl.
22281 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
22282 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
22283 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
22284 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
22285 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
22286 (aarch64_expand_sve_vcond): New functions.
22287 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
22288 of aarch64_vector_mode_p.
22289 (aarch64_dwarf_poly_indeterminate_value): New function.
22290 (aarch64_compute_pressure_classes): Likewise.
22291 (aarch64_can_change_mode_class): Likewise.
22292 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
22293 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
22294 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
22295 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
22296 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
22297 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
22298 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
22299 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
22301 (Dn, Dl, Dr): Accept const as well as const_vector.
22302 (Dz): Likewise. Compare against CONST0_RTX.
22303 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
22304 of "vector" where appropriate.
22305 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
22306 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
22307 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
22308 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
22309 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
22310 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
22311 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
22312 (v_int_equiv): Extend to SVE modes.
22313 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
22315 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
22316 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
22317 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
22318 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
22319 (SVE_COND_FP_CMP): New int iterators.
22320 (perm_hilo): Handle the new unpack unspecs.
22321 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
22323 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
22324 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
22325 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
22326 (aarch64_equality_operator, aarch64_constant_vector_operand)
22327 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
22328 (aarch64_sve_nonimmediate_operand): Likewise.
22329 (aarch64_sve_general_operand): Likewise.
22330 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
22331 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
22332 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
22333 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
22334 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
22335 (aarch64_sve_float_arith_immediate): Likewise.
22336 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
22337 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
22338 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
22339 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
22340 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
22341 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
22342 (aarch64_sve_float_arith_operand): Likewise.
22343 (aarch64_sve_float_arith_with_sub_operand): Likewise.
22344 (aarch64_sve_float_mul_operand): Likewise.
22345 (aarch64_sve_vec_perm_operand): Likewise.
22346 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
22347 (aarch64_mov_operand): Accept const_poly_int and const_vector.
22348 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
22349 as well as const_vector.
22350 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
22351 in file. Use CONST0_RTX and CONSTM1_RTX.
22352 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
22353 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
22354 Use aarch64_simd_imm_zero.
22355 * config/aarch64/aarch64-sve.md: New file.
22356 * config/aarch64/aarch64.md: Include it.
22357 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
22358 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
22359 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
22360 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
22361 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
22362 (sve): New attribute.
22363 (enabled): Disable instructions with the sve attribute unless
22365 (movqi, movhi): Pass CONST_POLY_INT operaneds through
22366 aarch64_expand_mov_immediate.
22367 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
22368 CNT[BHSD] immediates.
22369 (movti): Split CONST_POLY_INT moves into two halves.
22370 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
22371 Split additions that need a temporary here if the destination
22372 is the stack pointer.
22373 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
22374 (*add<mode>3_poly_1): New instruction.
22375 (set_clobber_cc): New expander.
22377 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22379 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
22380 parameter and use it instead of GET_MODE_SIZE (innermode). Use
22381 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
22382 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
22383 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
22384 Change innermode from fixed_mode_size to machine_mode.
22385 (simplify_subreg): Update call accordingly. Handle a constant-sized
22386 subreg of a variable-length CONST_VECTOR.
22388 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22389 Alan Hayward <alan.hayward@arm.com>
22390 David Sherwood <david.sherwood@arm.com>
22392 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
22393 (add_offset_to_base): New function, split out from...
22394 (create_mem_ref): ...here. When handling a scale other than 1,
22395 check first whether the address is valid without the offset.
22396 Add it into the base if so, leaving the index and scale as-is.
22398 2018-01-12 Jakub Jelinek <jakub@redhat.com>
22401 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
22402 fold_for_warn before checking if arg2 is INTEGER_CST.
22404 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
22406 * config/rs6000/predicates.md (load_multiple_operation): Delete.
22407 (store_multiple_operation): Delete.
22408 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
22409 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
22410 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
22411 guarded by TARGET_STRING.
22412 (rs6000_output_load_multiple): Delete.
22413 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22414 OPTION_MASK_STRING / TARGET_STRING handling.
22415 (print_operand) <'N', 'O'>: Add comment that these are unused now.
22416 (const rs6000_opt_masks) <"string">: Change mask to 0.
22417 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
22418 (MASK_STRING): Delete.
22419 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
22421 (load_multiple): Delete.
22428 (store_multiple): Delete.
22435 (movmemsi_8reg): Delete.
22436 (corresponding unnamed define_insn): Delete.
22437 (movmemsi_6reg): Delete.
22438 (corresponding unnamed define_insn): Delete.
22439 (movmemsi_4reg): Delete.
22440 (corresponding unnamed define_insn): Delete.
22441 (movmemsi_2reg): Delete.
22442 (corresponding unnamed define_insn): Delete.
22443 (movmemsi_1reg): Delete.
22444 (corresponding unnamed define_insn): Delete.
22445 * config/rs6000/rs6000.opt (mno-string): New.
22446 (mstring): Replace by deprecation warning stub.
22447 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
22449 2018-01-12 Jakub Jelinek <jakub@redhat.com>
22451 * regrename.c (regrename_do_replace): If replacing the same
22452 reg multiple times, try to reuse last created gen_raw_REG.
22455 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
22456 main to workaround a bug in GDB.
22458 2018-01-12 Tom de Vries <tom@codesourcery.com>
22461 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
22463 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
22465 PR rtl-optimization/80481
22466 * ira-color.c (get_cap_member): New function.
22467 (allocnos_conflict_by_live_ranges_p): Use it.
22468 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
22469 (setup_slot_coalesced_allocno_live_ranges): Ditto.
22471 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
22474 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
22475 (*saddl_se_1): Ditto.
22476 (*ssubsi_1): Ditto.
22477 (*ssubl_se_1): Ditto.
22479 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
22481 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
22482 rather than wi::to_widest for DR_INITs.
22483 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
22484 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
22485 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
22487 (vect_analyze_group_access_1): Note that here.
22489 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
22491 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
22492 polynomial type sizes.
22494 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
22496 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
22497 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
22498 (gimple_add_tmp_var): Likewise.
22500 2018-01-12 Martin Liska <mliska@suse.cz>
22502 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
22503 (gimple_alloc_sizes): Likewise.
22504 (dump_gimple_statistics): Use PRIu64 in printf format.
22505 * gimple.h: Change uint64_t to int.
22507 2018-01-12 Martin Liska <mliska@suse.cz>
22509 * tree-core.h: Use uint64_t instead of int.
22510 * tree.c (tree_node_counts): Likewise.
22511 (tree_node_sizes): Likewise.
22512 (dump_tree_statistics): Use PRIu64 in printf format.
22514 2018-01-12 Martin Liska <mliska@suse.cz>
22516 * Makefile.in: As qsort_chk is implemented in vec.c, add
22517 vec.o to linkage of gencfn-macros.
22518 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
22519 passing the info to record_node_allocation_statistics.
22520 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
22522 * ggc-common.c (struct ggc_usage): Add operator== and use
22523 it in operator< and compare function.
22524 * mem-stats.h (struct mem_usage): Likewise.
22525 * vec.c (struct vec_usage): Remove operator< and compare
22526 function. Can be simply inherited.
22528 2018-01-12 Martin Jambor <mjambor@suse.cz>
22531 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
22532 * tree-ssa-math-opts.c: Include domwalk.h.
22533 (convert_mult_to_fma_1): New function.
22534 (fma_transformation_info): New type.
22535 (fma_deferring_state): Likewise.
22536 (cancel_fma_deferring): New function.
22537 (result_of_phi): Likewise.
22538 (last_fma_candidate_feeds_initial_phi): Likewise.
22539 (convert_mult_to_fma): Added deferring logic, split actual
22540 transformation to convert_mult_to_fma_1.
22541 (math_opts_dom_walker): New type.
22542 (math_opts_dom_walker::after_dom_children): New method, body moved
22543 here from pass_optimize_widening_mul::execute, added deferring logic
22545 (pass_optimize_widening_mul::execute): Moved most of code to
22546 math_opts_dom_walker::after_dom_children.
22547 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
22548 * config/i386/i386.c (ix86_option_override_internal): Added
22549 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
22551 2018-01-12 Richard Biener <rguenther@suse.de>
22554 * dwarf2out.c (gen_variable_die): Do not reset old_die for
22555 inline instance vars.
22557 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
22560 * config/rx/rx.c (rx_is_restricted_memory_address):
22561 Handle SUBREG case.
22563 2018-01-12 Richard Biener <rguenther@suse.de>
22565 PR tree-optimization/80846
22566 * target.def (split_reduction): New target hook.
22567 * targhooks.c (default_split_reduction): New function.
22568 * targhooks.h (default_split_reduction): Declare.
22569 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
22570 target requests first reduce vectors by combining low and high
22572 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
22573 (get_vectype_for_scalar_type_and_size): Export.
22574 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
22575 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
22576 * doc/tm.texi: Regenerate.
22577 * config/i386/i386.c (ix86_split_reduction): Implement
22578 TARGET_VECTORIZE_SPLIT_REDUCTION.
22580 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
22583 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
22584 in PIC mode except for TARGET_VXWORKS_RTP.
22585 * config/sparc/sparc.c: Include cfgrtl.h.
22586 (TARGET_INIT_PIC_REG): Define.
22587 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
22588 (sparc_pic_register_p): New predicate.
22589 (sparc_legitimate_address_p): Use it.
22590 (sparc_legitimize_pic_address): Likewise.
22591 (sparc_delegitimize_address): Likewise.
22592 (sparc_mode_dependent_address_p): Likewise.
22593 (gen_load_pcrel_sym): Remove 4th parameter.
22594 (load_got_register): Adjust call to above. Remove obsolete stuff.
22595 (sparc_expand_prologue): Do not call load_got_register here.
22596 (sparc_flat_expand_prologue): Likewise.
22597 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
22598 (sparc_use_pseudo_pic_reg): New function.
22599 (sparc_init_pic_reg): Likewise.
22600 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
22601 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
22603 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
22605 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22606 Add item for branch_cost.
22608 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
22610 PR rtl-optimization/83565
22611 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
22612 not extend the result to a larger mode for rotate operations.
22613 (num_sign_bit_copies1): Likewise.
22615 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22618 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
22620 Use values-Xc.o for -pedantic.
22621 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
22623 2018-01-12 Martin Liska <mliska@suse.cz>
22626 * ipa-devirt.c (final_warning_record::grow_type_warnings):
22628 (possible_polymorphic_call_targets): Use it.
22629 (ipa_devirt): Likewise.
22631 2018-01-12 Martin Liska <mliska@suse.cz>
22633 * profile-count.h (enum profile_quality): Use 0 as invalid
22634 enum value of profile_quality.
22636 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
22638 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
22639 -mext-string options.
22641 2018-01-12 Richard Biener <rguenther@suse.de>
22643 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
22644 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
22645 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
22647 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
22649 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
22651 * configure.ac (--with-long-double-format): Add support for the
22652 configuration option to change the default long double format on
22654 * config.gcc (powerpc*-linux*-*): Likewise.
22655 * configure: Regenerate.
22656 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
22657 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
22658 used without modification.
22660 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22662 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
22663 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
22664 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
22665 MISC_BUILTIN_SPEC_BARRIER.
22666 (rs6000_init_builtins): Likewise.
22667 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
22669 (speculation_barrier): New define_insn.
22670 * doc/extend.texi: Document __builtin_speculation_barrier.
22672 2018-01-11 Jakub Jelinek <jakub@redhat.com>
22675 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
22676 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
22677 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
22679 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
22680 integral modes instead of "ss" and "sd".
22681 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
22682 vectors with 32-bit and 64-bit elements.
22683 (vecdupssescalarmodesuffix): New mode attribute.
22684 (vec_dup<mode>): Use it.
22686 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
22689 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
22690 frame if argument is passed on stack.
22692 2018-01-11 Jakub Jelinek <jakub@redhat.com>
22695 * ree.c (combine_reaching_defs): Optimize also
22696 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
22697 reg2=any_extend(exp); reg1=reg2;, formatting fix.
22699 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
22701 PR middle-end/83189
22702 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
22704 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
22706 PR middle-end/83718
22707 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
22708 after they are computed.
22710 2018-01-11 Bin Cheng <bin.cheng@arm.com>
22712 PR tree-optimization/83695
22713 * gimple-loop-linterchange.cc
22714 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
22715 reset cached scev information after interchange.
22716 (pass_linterchange::execute): Remove call to scev_reset_htab.
22718 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22720 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
22721 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
22722 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
22723 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
22724 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
22725 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
22726 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
22727 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
22728 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
22729 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
22730 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
22731 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
22732 (V_lane_reg): Likewise.
22733 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
22735 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
22736 (vfmal_lane_low<mode>_intrinsic,
22737 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
22738 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
22739 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
22740 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
22741 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
22742 vfmsl_lane_high<mode>_intrinsic): New define_insns.
22744 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22746 * config/arm/arm-cpus.in (fp16fml): New feature.
22747 (ALL_SIMD): Add fp16fml.
22748 (armv8.2-a): Add fp16fml as an option.
22749 (armv8.3-a): Likewise.
22750 (armv8.4-a): Add fp16fml as part of fp16.
22751 * config/arm/arm.h (TARGET_FP16FML): Define.
22752 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
22754 * config/arm/arm-modes.def (V2HF): Define.
22755 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
22756 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
22757 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
22758 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
22759 vfmsl_low, vfmsl_high): New set of builtins.
22760 * config/arm/iterators.md (PLUSMINUS): New code iterator.
22761 (vfml_op): New code attribute.
22762 (VFMLHALVES): New int iterator.
22763 (VFML, VFMLSEL): New mode attributes.
22764 (V_reg): Define mapping for V2HF.
22765 (V_hi, V_lo): New mode attributes.
22766 (VF_constraint): Likewise.
22767 (vfml_half, vfml_half_selector): New int attributes.
22768 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
22770 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
22771 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
22773 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
22774 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
22775 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
22776 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
22778 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
22779 Document new effective target and option set.
22781 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22783 * config/arm/arm-cpus.in (armv8_4): New feature.
22784 (ARMv8_4a): New fgroup.
22785 (armv8.4-a): New arch.
22786 * config/arm/arm-tables.opt: Regenerate.
22787 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
22788 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
22789 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
22790 Add matching rules for -march=armv8.4-a and extensions.
22791 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
22793 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
22796 * config/rx/rx.md (BW): New mode attribute.
22797 (sync_lock_test_and_setsi): Add mode suffix to insn output.
22799 2018-01-11 Richard Biener <rguenther@suse.de>
22801 PR tree-optimization/83435
22802 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
22803 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
22804 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
22806 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22807 Alan Hayward <alan.hayward@arm.com>
22808 David Sherwood <david.sherwood@arm.com>
22810 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
22812 (aarch64_classify_address): Initialize it. Track polynomial offsets.
22813 (aarch64_print_address_internal): Use it to check for a zero offset.
22815 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22816 Alan Hayward <alan.hayward@arm.com>
22817 David Sherwood <david.sherwood@arm.com>
22819 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
22820 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
22821 Return a poly_int64 rather than a HOST_WIDE_INT.
22822 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
22823 rather than a HOST_WIDE_INT.
22824 * config/aarch64/aarch64.h (aarch64_frame): Protect with
22825 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
22826 hard_fp_offset, frame_size, initial_adjust, callee_offset and
22827 final_offset from HOST_WIDE_INT to poly_int64.
22828 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
22829 to_constant when getting the number of units in an Advanced SIMD
22831 (aarch64_builtin_vectorized_function): Check for a constant number
22833 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
22835 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
22836 attribute instead of GET_MODE_NUNITS.
22837 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
22838 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
22839 GET_MODE_SIZE for fixed-size registers.
22840 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
22841 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
22842 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
22843 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
22844 (aarch64_print_operand, aarch64_print_address_internal)
22845 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
22846 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
22847 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
22848 Handle polynomial GET_MODE_SIZE.
22849 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
22850 wider than SImode without modification.
22851 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
22852 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
22853 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
22854 passing and returning SVE modes.
22855 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
22856 rather than GEN_INT.
22857 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
22858 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
22859 (aarch64_allocate_and_probe_stack_space): Likewise.
22860 (aarch64_layout_frame): Cope with polynomial offsets.
22861 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
22862 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
22863 polynomial offsets.
22864 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
22865 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
22866 poly_int64 rather than a HOST_WIDE_INT.
22867 (aarch64_get_separate_components, aarch64_process_components)
22868 (aarch64_expand_prologue, aarch64_expand_epilogue)
22869 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
22870 (aarch64_anchor_offset): New function, split out from...
22871 (aarch64_legitimize_address): ...here.
22872 (aarch64_builtin_vectorization_cost): Handle polynomial
22873 TYPE_VECTOR_SUBPARTS.
22874 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
22876 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
22877 number of elements from the PARALLEL rather than the mode.
22878 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
22879 rather than GET_MODE_BITSIZE.
22880 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
22881 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
22882 (aarch64_expand_vec_perm_const_1): Handle polynomial
22883 d->perm.length () and d->perm elements.
22884 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
22885 Apply to_constant to d->perm elements.
22886 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
22887 polynomial CONST_VECTOR_NUNITS.
22888 (aarch64_move_pointer): Take amount as a poly_int64 rather
22890 (aarch64_progress_pointer): Avoid temporary variable.
22891 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
22892 the mode attribute instead of GET_MODE.
22894 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22895 Alan Hayward <alan.hayward@arm.com>
22896 David Sherwood <david.sherwood@arm.com>
22898 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
22899 x exists before using it.
22900 (aarch64_add_constant_internal): Rename to...
22901 (aarch64_add_offset_1): ...this. Replace regnum with separate
22902 src and dest rtxes. Handle the case in which they're different,
22903 including when the offset is zero. Replace scratchreg with an rtx.
22904 Use 2 additions if there is no spare register into which we can
22905 move a 16-bit constant.
22906 (aarch64_add_constant): Delete.
22907 (aarch64_add_offset): Replace reg with separate src and dest
22908 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
22909 Use aarch64_add_offset_1.
22910 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
22911 an rtx rather than an int. Take the delta as a poly_int64
22912 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
22913 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
22914 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
22915 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
22916 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
22917 and aarch64_add_sp.
22918 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
22919 aarch64_add_constant.
22921 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22923 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22924 Use scalar_float_mode.
22926 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22928 * config/aarch64/aarch64-simd.md
22929 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
22930 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
22931 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
22932 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
22933 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
22934 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
22935 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
22936 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
22937 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
22938 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
22940 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22943 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
22944 targ_options->x_arm_arch_string is non NULL.
22946 2018-01-11 Tamar Christina <tamar.christina@arm.com>
22948 * config/aarch64/aarch64.h
22949 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
22951 2018-01-11 Sudakshina Das <sudi.das@arm.com>
22954 * expmed.c (emit_store_flag_force): Swap if const op0
22955 and change VOIDmode to mode of op0.
22957 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22959 PR rtl-optimization/83761
22960 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
22961 than bytes to mode_for_size.
22963 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
22965 PR middle-end/83189
22966 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
22967 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
22970 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
22972 PR middle-end/83575
22973 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
22974 when in layout mode.
22975 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
22976 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
22979 2018-01-10 Michael Collison <michael.collison@arm.com>
22981 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
22982 * config/aarch64/aarch64-option-extension.def: Add
22983 AARCH64_OPT_EXTENSION of 'fp16fml'.
22984 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
22985 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
22986 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
22987 * config/aarch64/constraints.md (Ui7): New constraint.
22988 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
22989 (VFMLA_SEL_W): Ditto.
22992 (VFMLA16_LOW): New int iterator.
22993 (VFMLA16_HIGH): Ditto.
22994 (UNSPEC_FMLAL): New unspec.
22995 (UNSPEC_FMLSL): Ditto.
22996 (UNSPEC_FMLAL2): Ditto.
22997 (UNSPEC_FMLSL2): Ditto.
22998 (f16mac): New code attribute.
22999 * config/aarch64/aarch64-simd-builtins.def
23000 (aarch64_fmlal_lowv2sf): Ditto.
23001 (aarch64_fmlsl_lowv2sf): Ditto.
23002 (aarch64_fmlalq_lowv4sf): Ditto.
23003 (aarch64_fmlslq_lowv4sf): Ditto.
23004 (aarch64_fmlal_highv2sf): Ditto.
23005 (aarch64_fmlsl_highv2sf): Ditto.
23006 (aarch64_fmlalq_highv4sf): Ditto.
23007 (aarch64_fmlslq_highv4sf): Ditto.
23008 (aarch64_fmlal_lane_lowv2sf): Ditto.
23009 (aarch64_fmlsl_lane_lowv2sf): Ditto.
23010 (aarch64_fmlal_laneq_lowv2sf): Ditto.
23011 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
23012 (aarch64_fmlalq_lane_lowv4sf): Ditto.
23013 (aarch64_fmlsl_lane_lowv4sf): Ditto.
23014 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
23015 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
23016 (aarch64_fmlal_lane_highv2sf): Ditto.
23017 (aarch64_fmlsl_lane_highv2sf): Ditto.
23018 (aarch64_fmlal_laneq_highv2sf): Ditto.
23019 (aarch64_fmlsl_laneq_highv2sf): Ditto.
23020 (aarch64_fmlalq_lane_highv4sf): Ditto.
23021 (aarch64_fmlsl_lane_highv4sf): Ditto.
23022 (aarch64_fmlalq_laneq_highv4sf): Ditto.
23023 (aarch64_fmlsl_laneq_highv4sf): Ditto.
23024 * config/aarch64/aarch64-simd.md:
23025 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
23026 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
23027 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
23028 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
23029 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
23030 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
23031 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
23032 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
23033 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
23034 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
23035 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
23036 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
23037 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
23038 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
23039 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
23040 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
23041 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
23042 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
23043 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
23044 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
23045 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
23046 (vfmlsl_low_u32): Ditto.
23047 (vfmlalq_low_u32): Ditto.
23048 (vfmlslq_low_u32): Ditto.
23049 (vfmlal_high_u32): Ditto.
23050 (vfmlsl_high_u32): Ditto.
23051 (vfmlalq_high_u32): Ditto.
23052 (vfmlslq_high_u32): Ditto.
23053 (vfmlal_lane_low_u32): Ditto.
23054 (vfmlsl_lane_low_u32): Ditto.
23055 (vfmlal_laneq_low_u32): Ditto.
23056 (vfmlsl_laneq_low_u32): Ditto.
23057 (vfmlalq_lane_low_u32): Ditto.
23058 (vfmlslq_lane_low_u32): Ditto.
23059 (vfmlalq_laneq_low_u32): Ditto.
23060 (vfmlslq_laneq_low_u32): Ditto.
23061 (vfmlal_lane_high_u32): Ditto.
23062 (vfmlsl_lane_high_u32): Ditto.
23063 (vfmlal_laneq_high_u32): Ditto.
23064 (vfmlsl_laneq_high_u32): Ditto.
23065 (vfmlalq_lane_high_u32): Ditto.
23066 (vfmlslq_lane_high_u32): Ditto.
23067 (vfmlalq_laneq_high_u32): Ditto.
23068 (vfmlslq_laneq_high_u32): Ditto.
23069 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
23070 (AARCH64_FL_FOR_ARCH8_4): New.
23071 (AARCH64_ISA_F16FML): New ISA flag.
23072 (TARGET_F16FML): New feature flag for fp16fml.
23073 (doc/invoke.texi): Document new fp16fml option.
23075 2018-01-10 Michael Collison <michael.collison@arm.com>
23077 * config/aarch64/aarch64-builtins.c:
23078 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
23079 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23080 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
23081 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
23082 (AARCH64_ISA_SHA3): New ISA flag.
23083 (TARGET_SHA3): New feature flag for sha3.
23084 * config/aarch64/iterators.md (sha512_op): New int attribute.
23085 (CRYPTO_SHA512): New int iterator.
23086 (UNSPEC_SHA512H): New unspec.
23087 (UNSPEC_SHA512H2): Ditto.
23088 (UNSPEC_SHA512SU0): Ditto.
23089 (UNSPEC_SHA512SU1): Ditto.
23090 * config/aarch64/aarch64-simd-builtins.def
23091 (aarch64_crypto_sha512hqv2di): New builtin.
23092 (aarch64_crypto_sha512h2qv2di): Ditto.
23093 (aarch64_crypto_sha512su0qv2di): Ditto.
23094 (aarch64_crypto_sha512su1qv2di): Ditto.
23095 (aarch64_eor3qv8hi): Ditto.
23096 (aarch64_rax1qv2di): Ditto.
23097 (aarch64_xarqv2di): Ditto.
23098 (aarch64_bcaxqv8hi): Ditto.
23099 * config/aarch64/aarch64-simd.md:
23100 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
23101 (aarch64_crypto_sha512su0qv2di): Ditto.
23102 (aarch64_crypto_sha512su1qv2di): Ditto.
23103 (aarch64_eor3qv8hi): Ditto.
23104 (aarch64_rax1qv2di): Ditto.
23105 (aarch64_xarqv2di): Ditto.
23106 (aarch64_bcaxqv8hi): Ditto.
23107 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
23108 (vsha512h2q_u64): Ditto.
23109 (vsha512su0q_u64): Ditto.
23110 (vsha512su1q_u64): Ditto.
23111 (veor3q_u16): Ditto.
23112 (vrax1q_u64): Ditto.
23113 (vxarq_u64): Ditto.
23114 (vbcaxq_u16): Ditto.
23115 * config/arm/types.md (crypto_sha512): New type attribute.
23116 (crypto_sha3): Ditto.
23117 (doc/invoke.texi): Document new sha3 option.
23119 2018-01-10 Michael Collison <michael.collison@arm.com>
23121 * config/aarch64/aarch64-builtins.c:
23122 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
23123 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23124 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
23125 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
23126 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
23127 (AARCH64_ISA_SM4): New ISA flag.
23128 (TARGET_SM4): New feature flag for sm4.
23129 * config/aarch64/aarch64-simd-builtins.def
23130 (aarch64_sm3ss1qv4si): Ditto.
23131 (aarch64_sm3tt1aq4si): Ditto.
23132 (aarch64_sm3tt1bq4si): Ditto.
23133 (aarch64_sm3tt2aq4si): Ditto.
23134 (aarch64_sm3tt2bq4si): Ditto.
23135 (aarch64_sm3partw1qv4si): Ditto.
23136 (aarch64_sm3partw2qv4si): Ditto.
23137 (aarch64_sm4eqv4si): Ditto.
23138 (aarch64_sm4ekeyqv4si): Ditto.
23139 * config/aarch64/aarch64-simd.md:
23140 (aarch64_sm3ss1qv4si): Ditto.
23141 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
23142 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
23143 (aarch64_sm4eqv4si): Ditto.
23144 (aarch64_sm4ekeyqv4si): Ditto.
23145 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
23146 (sm3part_op): Ditto.
23147 (CRYPTO_SM3TT): Ditto.
23148 (CRYPTO_SM3PART): Ditto.
23149 (UNSPEC_SM3SS1): New unspec.
23150 (UNSPEC_SM3TT1A): Ditto.
23151 (UNSPEC_SM3TT1B): Ditto.
23152 (UNSPEC_SM3TT2A): Ditto.
23153 (UNSPEC_SM3TT2B): Ditto.
23154 (UNSPEC_SM3PARTW1): Ditto.
23155 (UNSPEC_SM3PARTW2): Ditto.
23156 (UNSPEC_SM4E): Ditto.
23157 (UNSPEC_SM4EKEY): Ditto.
23158 * config/aarch64/constraints.md (Ui2): New constraint.
23159 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
23160 * config/arm/types.md (crypto_sm3): New type attribute.
23161 (crypto_sm4): Ditto.
23162 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
23163 (vsm3tt1aq_u32): Ditto.
23164 (vsm3tt1bq_u32): Ditto.
23165 (vsm3tt2aq_u32): Ditto.
23166 (vsm3tt2bq_u32): Ditto.
23167 (vsm3partw1q_u32): Ditto.
23168 (vsm3partw2q_u32): Ditto.
23169 (vsm4eq_u32): Ditto.
23170 (vsm4ekeyq_u32): Ditto.
23171 (doc/invoke.texi): Document new sm4 option.
23173 2018-01-10 Michael Collison <michael.collison@arm.com>
23175 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
23176 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
23177 (AARCH64_FL_FOR_ARCH8_4): New.
23178 (AARCH64_FL_V8_4): New flag.
23179 (doc/invoke.texi): Document new armv8.4-a option.
23181 2018-01-10 Michael Collison <michael.collison@arm.com>
23183 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23184 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
23185 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
23186 * config/aarch64/aarch64-option-extension.def: Add
23187 AARCH64_OPT_EXTENSION of 'sha2'.
23188 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
23189 (crypto): Disable sha2 and aes if crypto disabled.
23190 (crypto): Enable aes and sha2 if enabled.
23191 (simd): Disable sha2 and aes if simd disabled.
23192 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
23194 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
23195 (TARGET_SHA2): New feature flag for sha2.
23196 (TARGET_AES): New feature flag for aes.
23197 * config/aarch64/aarch64-simd.md:
23198 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
23199 conditional on TARGET_AES.
23200 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
23201 (aarch64_crypto_sha1hsi): Make pattern conditional
23203 (aarch64_crypto_sha1hv4si): Ditto.
23204 (aarch64_be_crypto_sha1hv4si): Ditto.
23205 (aarch64_crypto_sha1su1v4si): Ditto.
23206 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
23207 (aarch64_crypto_sha1su0v4si): Ditto.
23208 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
23209 (aarch64_crypto_sha256su0v4si): Ditto.
23210 (aarch64_crypto_sha256su1v4si): Ditto.
23211 (doc/invoke.texi): Document new aes and sha2 options.
23213 2018-01-10 Martin Sebor <msebor@redhat.com>
23215 PR tree-optimization/83781
23216 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
23219 2018-01-11 Martin Sebor <msebor@gmail.com>
23220 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23222 PR tree-optimization/83501
23223 PR tree-optimization/81703
23225 * tree-ssa-strlen.c (get_string_cst): Rename...
23226 (get_string_len): ...to this. Handle global constants.
23227 (handle_char_store): Adjust.
23229 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
23230 Jim Wilson <jimw@sifive.com>
23232 * config/riscv/riscv-protos.h (riscv_output_return): New.
23233 * config/riscv/riscv.c (struct machine_function): New naked_p field.
23234 (riscv_attribute_table, riscv_output_return),
23235 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
23236 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
23237 (riscv_compute_frame_info): Only compute frame->mask if not a naked
23239 (riscv_expand_prologue): Add early return for naked function.
23240 (riscv_expand_epilogue): Likewise.
23241 (riscv_function_ok_for_sibcall): Return false for naked function.
23242 (riscv_set_current_function): New.
23243 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
23244 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
23245 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
23246 * doc/extend.texi (RISC-V Function Attributes): New.
23248 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
23250 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
23251 check for 128-bit long double before checking TCmode.
23252 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
23253 128-bit long doubles before checking TFmode or TCmode.
23254 (FLOAT128_IBM_P): Likewise.
23256 2018-01-10 Martin Sebor <msebor@redhat.com>
23258 PR tree-optimization/83671
23259 * builtins.c (c_strlen): Unconditionally return zero for the empty
23261 Use -Warray-bounds for warnings.
23262 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
23263 for non-constant array indices with COMPONENT_REF, arrays of
23264 arrays, and pointers to arrays.
23265 (gimple_fold_builtin_strlen): Determine and set length range for
23266 non-constant character arrays.
23268 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
23270 PR middle-end/81897
23271 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
23274 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
23276 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
23278 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
23281 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
23282 VECTOR_MEM_ALTIVEC_OR_VSX_P.
23283 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
23284 indexed_or_indirect_operand predicate.
23285 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
23286 (*vsx_le_perm_load_v8hi): Likewise.
23287 (*vsx_le_perm_load_v16qi): Likewise.
23288 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
23289 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
23290 (*vsx_le_perm_store_v8hi): Likewise.
23291 (*vsx_le_perm_store_v16qi): Likewise.
23292 (eight unnamed splitters): Likewise.
23294 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
23296 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
23297 * config/rs6000/emmintrin.h: Likewise.
23298 * config/rs6000/mmintrin.h: Likewise.
23299 * config/rs6000/xmmintrin.h: Likewise.
23301 2018-01-10 David Malcolm <dmalcolm@redhat.com>
23304 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
23306 * tree.c (tree_nop_conversion): Return true for location wrapper
23308 (maybe_wrap_with_location): New function.
23309 (selftest::check_strip_nops): New function.
23310 (selftest::test_location_wrappers): New function.
23311 (selftest::tree_c_tests): Call it.
23312 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
23313 (maybe_wrap_with_location): New decl.
23314 (EXPR_LOCATION_WRAPPER_P): New macro.
23315 (location_wrapper_p): New inline function.
23316 (tree_strip_any_location_wrapper): New inline function.
23318 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
23321 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
23322 stack_realign_offset for the largest alignment of stack slot
23324 (ix86_find_max_used_stack_alignment): New function.
23325 (ix86_finalize_stack_frame_flags): Use it. Set
23326 max_used_stack_alignment if we don't realign stack.
23327 * config/i386/i386.h (machine_function): Add
23328 max_used_stack_alignment.
23330 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
23332 * config/arm/arm.opt (-mbranch-cost): New option.
23333 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
23336 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
23339 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
23340 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
23342 2018-01-10 Richard Biener <rguenther@suse.de>
23345 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
23346 early out so it also covers the case where we have a non-NULL
23349 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
23351 PR tree-optimization/83753
23352 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
23353 for non-strided grouped accesses if the number of elements is 1.
23355 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
23358 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
23359 * i386.h (TARGET_USE_GATHER): Define.
23360 * x86-tune.def (X86_TUNE_USE_GATHER): New.
23362 2018-01-10 Martin Liska <mliska@suse.cz>
23365 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
23366 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
23368 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
23369 CLEANUP_NO_PARTITIONING is not set.
23371 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
23373 * doc/rtl.texi: Remove documentation of (const ...) wrappers
23374 for vectors, as a partial revert of r254296.
23375 * rtl.h (const_vec_p): Delete.
23376 (const_vec_duplicate_p): Don't test for vector CONSTs.
23377 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
23378 * expmed.c (make_tree): Likewise.
23381 * common.md (E, F): Use CONSTANT_P instead of checking for
23383 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
23384 checking for CONST_VECTOR.
23386 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
23388 PR middle-end/83575
23389 * predict.c (force_edge_cold): Handle in more sane way edges
23390 with no prediction.
23392 2018-01-09 Carl Love <cel@us.ibm.com>
23394 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
23396 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
23397 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
23398 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
23399 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
23400 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
23401 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
23402 * config/rs6000/rs6000-protos.h: Add extern defition for
23403 rs6000_generate_float2_double_code.
23404 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
23406 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
23407 (float2_v2df): Add define_expand.
23409 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
23412 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
23413 op_mode in the force_to_mode call.
23415 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
23417 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
23418 instead of checking each element individually.
23419 (aarch64_evpc_uzp): Likewise.
23420 (aarch64_evpc_zip): Likewise.
23421 (aarch64_evpc_ext): Likewise.
23422 (aarch64_evpc_rev): Likewise.
23423 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
23424 instead of checking each element individually. Return true without
23426 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
23427 whether all selected elements come from the same input, instead of
23428 checking each element individually. Remove calls to gen_rtx_REG,
23429 start_sequence and end_sequence and instead assert that no rtl is
23432 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
23434 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
23435 order of HIGH and CONST checks.
23437 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
23439 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
23440 if the destination isn't an SSA_NAME.
23442 2018-01-09 Richard Biener <rguenther@suse.de>
23444 PR tree-optimization/83668
23445 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
23447 (canonicalize_loop_form): ... here, renamed from ...
23448 (canonicalize_loop_closed_ssa_form): ... this and amended to
23449 swap successor edges for loop exit blocks to make us use
23450 the RPO order we need for initial schedule generation.
23452 2018-01-09 Joseph Myers <joseph@codesourcery.com>
23454 PR tree-optimization/64811
23455 * match.pd: When optimizing comparisons with Inf, avoid
23456 introducing or losing exceptions from comparisons with NaN.
23458 2018-01-09 Martin Liska <mliska@suse.cz>
23461 * asan.c (shadow_mem_size): Add gcc_assert.
23463 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
23465 Don't save registers in main().
23468 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
23469 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
23470 * config/avr/avr.c (avr_set_current_function): Don't error if
23471 naked, OS_task or OS_main are specified at the same time.
23472 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
23474 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
23476 * common/config/avr/avr-common.c (avr_option_optimization_table):
23477 Switch on -mmain-is-OS_task for optimizing compilations.
23479 2018-01-09 Richard Biener <rguenther@suse.de>
23481 PR tree-optimization/83572
23482 * graphite.c: Include cfganal.h.
23483 (graphite_transform_loops): Connect infinite loops to exit
23484 and remove fake edges at the end.
23486 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
23488 * ipa-inline.c (edge_badness): Revert accidental checkin.
23490 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
23493 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
23494 symbols; not inline clones.
23496 2018-01-09 Jakub Jelinek <jakub@redhat.com>
23499 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
23500 hard registers. Formatting fixes.
23502 PR preprocessor/83722
23503 * gcc.c (try_generate_repro): Pass
23504 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
23505 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
23508 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
23509 Kito Cheng <kito.cheng@gmail.com>
23511 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
23512 (riscv_leaf_function_p): Delete.
23513 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
23515 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23517 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
23519 (do_ifelse): New function.
23520 (do_isel): New function.
23521 (do_sub3): New function.
23522 (do_add3): New function.
23523 (do_load_mask_compare): New function.
23524 (do_overlap_load_compare): New function.
23525 (expand_compare_loop): New function.
23526 (expand_block_compare): Call expand_compare_loop() when appropriate.
23527 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
23528 option description.
23529 (-mblock-compare-inline-loop-limit): New option.
23531 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23534 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
23535 Reverse order of second and third operands in first alternative.
23536 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
23537 of first and second elements in UNSPEC_VPERMR vector.
23538 (altivec_expand_vec_perm_le): Likewise.
23540 2018-01-08 Jeff Law <law@redhat.com>
23542 PR rtl-optimizatin/81308
23543 * tree-switch-conversion.c (cfg_altered): New file scoped static.
23544 (process_switch): If group_case_labels makes a change, then set
23546 (pass_convert_switch::execute): If a switch is converted, then
23547 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
23549 PR rtl-optimization/81308
23550 * recog.c (split_all_insns): Conditionally cleanup the CFG after
23553 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
23555 PR target/83663 - Revert r255946
23556 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
23557 generation for cases where splatting a value is not useful.
23558 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
23559 across a vec_duplicate and a paradoxical subreg forming a vector
23560 mode to a vec_concat.
23562 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23564 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
23565 -march=armv8.3-a variants.
23566 * config/arm/t-multilib: Likewise.
23567 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
23569 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23571 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
23573 (cceq_ior_compare_complement): Give it a name so I can use it, and
23574 change boolean_or_operator predicate to boolean_operator so it can
23575 be used to generate a crand.
23576 (eqne): New code iterator.
23577 (bd/bd_neg): New code_attrs.
23578 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
23579 a single define_insn.
23580 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
23581 decrement (bdnzt/bdnzf/bdzt/bdzf).
23582 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
23583 with the new names of the branch decrement patterns, and added the
23584 names of the branch decrement conditional patterns.
23586 2018-01-08 Richard Biener <rguenther@suse.de>
23588 PR tree-optimization/83563
23589 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
23592 2018-01-08 Richard Biener <rguenther@suse.de>
23594 PR middle-end/83713
23595 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
23597 2018-01-08 Richard Biener <rguenther@suse.de>
23599 PR tree-optimization/83685
23600 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
23601 references to abnormals.
23603 2018-01-08 Richard Biener <rguenther@suse.de>
23606 * dwarf2out.c (output_indirect_strings): Handle empty
23607 skeleton_debug_str_hash.
23608 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
23610 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
23612 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
23613 (emit_store_direct): Likewise.
23614 (arc_trampoline_adjust_address): Likewise.
23615 (arc_asm_trampoline_template): New function.
23616 (arc_initialize_trampoline): Use asm_trampoline_template.
23617 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
23618 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
23619 * config/arc/arc.md (flush_icache): Delete pattern.
23621 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
23623 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
23624 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
23627 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
23630 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
23631 by not USED_FOR_TARGET.
23632 (make_pass_resolve_sw_modes): Likewise.
23634 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
23636 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
23639 2018-01-08 Richard Biener <rguenther@suse.de>
23641 PR middle-end/83580
23642 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
23644 2018-01-08 Richard Biener <rguenther@suse.de>
23646 PR middle-end/83517
23647 * match.pd ((t * 2) / 2) -> t): Add missing :c.
23649 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
23651 PR middle-end/81897
23652 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
23653 basic blocks with a small number of successors.
23654 (convert_control_dep_chain_into_preds): Improve handling of
23656 (dump_predicates): Split apart into...
23657 (dump_pred_chain): ...here...
23658 (dump_pred_info): ...and here.
23659 (can_one_predicate_be_invalidated_p): Add debugging printfs.
23660 (can_chain_union_be_invalidated_p): Improve check for invalidation
23662 (uninit_uses_cannot_happen): Avoid unnecessary if
23663 convert_control_dep_chain_into_preds yielded nothing.
23665 2018-01-06 Martin Sebor <msebor@redhat.com>
23667 PR tree-optimization/83640
23668 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
23669 subtracting negative offset from size.
23670 (builtin_access::overlap): Adjust offset bounds of the access to fall
23671 within the size of the object if possible.
23673 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
23675 PR rtl-optimization/83699
23676 * expmed.c (extract_bit_field_1): Restrict the vector usage of
23677 extract_bit_field_as_subreg to cases in which the extracted
23678 value is also a vector.
23680 * lra-constraints.c (process_alt_operands): Test for the equivalence
23681 substitutions when detecting a possible reload cycle.
23683 2018-01-06 Jakub Jelinek <jakub@redhat.com>
23686 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
23687 by default if flag_selective_schedling{,2}. Formatting fixes.
23689 PR rtl-optimization/83682
23690 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
23691 if it has non-VECTOR_MODE element mode.
23692 (vec_duplicate_p): Likewise.
23694 PR middle-end/83694
23695 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
23696 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
23698 2018-01-05 Jakub Jelinek <jakub@redhat.com>
23701 * config/i386/i386-builtin.def
23702 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
23703 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
23704 Require also OPTION_MASK_ISA_AVX512F in addition to
23705 OPTION_MASK_ISA_GFNI.
23706 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
23707 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
23708 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
23709 to OPTION_MASK_ISA_GFNI.
23710 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
23711 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
23712 OPTION_MASK_ISA_AVX512BW.
23713 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
23714 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
23715 addition to OPTION_MASK_ISA_GFNI.
23716 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
23717 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
23718 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
23719 to OPTION_MASK_ISA_GFNI.
23720 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
23721 a requirement for all ISAs rather than any of them with a few
23723 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
23725 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
23726 bitmasks to be enabled with 3 exceptions, instead of requiring any
23727 enabled ISA with lots of exceptions.
23728 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
23729 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
23730 Change avx512bw in isa attribute to avx512f.
23731 * config/i386/sgxintrin.h: Add license boilerplate.
23732 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
23733 to __AVX512F__ and __AVX512VL to __AVX512VL__.
23734 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
23735 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
23737 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
23738 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
23739 temporarily sse2 rather than sse if not enabled already.
23742 * config/i386/sse.md (VI248_VLBW): Rename to ...
23743 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
23744 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
23745 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
23746 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
23747 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
23748 mode iterator instead of VI248_VLBW.
23750 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
23752 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
23753 (record_modified): Skip clobbers; add debug output.
23754 (param_change_prob): Use sreal frequencies.
23756 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
23758 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
23759 punt for user-aligned variables.
23761 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
23763 * tree-chrec.c (chrec_contains_symbols): Return true for
23766 2018-01-05 Sudakshina Das <sudi.das@arm.com>
23769 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
23770 of (x|y) == x for BICS pattern.
23772 2018-01-05 Jakub Jelinek <jakub@redhat.com>
23774 PR tree-optimization/83605
23775 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
23776 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
23779 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
23781 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
23782 * config/epiphany/rtems.h: New file.
23784 2018-01-04 Jakub Jelinek <jakub@redhat.com>
23785 Uros Bizjak <ubizjak@gmail.com>
23788 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
23789 QIreg_operand instead of register_operand predicate.
23790 * config/i386/i386.c (ix86_rop_should_change_byte_p,
23791 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
23792 comments instead of -fmitigate[-_]rop.
23794 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23797 * cgraphunit.c (symbol_table::compile): Switch to text_section
23798 before calling assembly_start debug hook.
23799 * run-rtl-passes.c (run_rtl_passes): Likewise.
23802 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23804 * tree-vrp.c (extract_range_from_binary_expr_1): Check
23805 range_int_cst_p rather than !symbolic_range_p before calling
23806 extract_range_from_multiplicative_op_1.
23808 2018-01-04 Jeff Law <law@redhat.com>
23810 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
23811 redundant test in assertion.
23813 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23815 * doc/rtl.texi: Document machine_mode wrapper classes.
23817 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23819 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
23820 using tree_to_uhwi.
23822 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23824 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
23825 the VEC_PERM_EXPR fold to fail.
23827 2018-01-04 Jakub Jelinek <jakub@redhat.com>
23830 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
23831 to switched_sections.
23833 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23836 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
23837 test for d.testing.
23839 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
23842 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
23843 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
23845 2018-01-04 Jakub Jelinek <jakub@redhat.com>
23848 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
23849 is BLKmode and bitpos not zero or mode change is needed.
23851 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23854 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
23857 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
23860 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
23861 instead of MULT rtx. Update all corresponding splitters.
23862 (*saddl_se): Ditto.
23863 (*ssub<modesuffix>): Ditto.
23864 (*ssubl_se): Ditto.
23865 (*cmp_sadd_di): Update split patterns.
23866 (*cmp_sadd_si): Ditto.
23867 (*cmp_sadd_sidi): Ditto.
23868 (*cmp_ssub_di): Ditto.
23869 (*cmp_ssub_si): Ditto.
23870 (*cmp_ssub_sidi): Ditto.
23871 * config/alpha/predicates.md (const23_operand): New predicate.
23872 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
23873 Look for ASHIFT, not MULT inner operand.
23874 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
23876 2018-01-04 Martin Liska <mliska@suse.cz>
23878 PR gcov-profile/83669
23879 * gcov.c (output_intermediate_file): Add version to intermediate
23881 * doc/gcov.texi: Document new field 'version' in intermediate
23882 file format. Fix location of '-k' option of gcov command.
23884 2018-01-04 Martin Liska <mliska@suse.cz>
23887 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
23889 2018-01-04 Jakub Jelinek <jakub@redhat.com>
23891 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
23893 2018-01-03 Martin Sebor <msebor@redhat.com>
23895 PR tree-optimization/83655
23896 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
23897 checking calls with invalid arguments.
23899 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23901 * tree-vect-stmts.c (vect_get_store_rhs): New function.
23902 (vectorizable_mask_load_store): Delete.
23903 (vectorizable_call): Return false for masked loads and stores.
23904 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
23905 instead of gimple_assign_rhs1.
23906 (vectorizable_load): Handle IFN_MASK_LOAD.
23907 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
23909 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23911 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
23913 (vectorizable_mask_load_store): ...here.
23914 (vectorizable_load): ...and here.
23916 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23918 * tree-vect-stmts.c (vect_build_all_ones_mask)
23919 (vect_build_zero_merge_argument): New functions, split out from...
23920 (vectorizable_load): ...here.
23922 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23924 * tree-vect-stmts.c (vect_check_store_rhs): New function,
23926 (vectorizable_mask_load_store): ...here.
23927 (vectorizable_store): ...and here.
23929 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23931 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
23933 (vectorizable_mask_load_store): ...here.
23935 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23937 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
23938 (vect_model_store_cost): Take a vec_load_store_type instead of a
23940 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
23941 (vect_model_store_cost): Take a vec_load_store_type instead of a
23943 (vectorizable_mask_load_store): Update accordingly.
23944 (vectorizable_store): Likewise.
23945 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
23947 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23949 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
23950 IFN_MASK_LOAD calls here rather than...
23951 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
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 * expmed.c (extract_bit_field_1): For vector extracts,
23958 fall back to extract_bit_field_as_subreg if vec_extract
23961 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23962 Alan Hayward <alan.hayward@arm.com>
23963 David Sherwood <david.sherwood@arm.com>
23965 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
23966 they are variable or constant sized.
23967 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
23968 slots for constant-sized data.
23970 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23971 Alan Hayward <alan.hayward@arm.com>
23972 David Sherwood <david.sherwood@arm.com>
23974 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
23975 handling COND_EXPRs with boolean comparisons, try to find a better
23976 basis for the mask type than the boolean itself.
23978 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23980 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
23981 is calculated and how it can be overridden.
23982 * genmodes.c (max_bitsize_mode_any_mode): New variable.
23983 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
23985 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
23988 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23989 Alan Hayward <alan.hayward@arm.com>
23990 David Sherwood <david.sherwood@arm.com>
23992 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
23993 Remove the mode argument.
23994 (aarch64_simd_valid_immediate): Remove the mode and inverse
23996 * config/aarch64/iterators.md (bitsize): New iterator.
23997 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
23998 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
23999 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
24000 aarch64_simd_valid_immediate.
24001 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
24002 (aarch64_reg_or_bic_imm): Likewise.
24003 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
24004 with an insn_type enum and msl with a modifier_type enum.
24005 Replace element_width with a scalar_mode. Change the shift
24006 to unsigned int. Add constructors for scalar_float_mode and
24007 scalar_int_mode elements.
24008 (aarch64_vect_float_const_representable_p): Delete.
24009 (aarch64_can_const_movi_rtx_p)
24010 (aarch64_simd_scalar_immediate_valid_for_move)
24011 (aarch64_simd_make_constant): Update call to
24012 aarch64_simd_valid_immediate.
24013 (aarch64_advsimd_valid_immediate_hs): New function.
24014 (aarch64_advsimd_valid_immediate): Likewise.
24015 (aarch64_simd_valid_immediate): Remove mode and inverse
24016 arguments. Rewrite to use the above. Use const_vec_duplicate_p
24017 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
24018 and aarch64_float_const_representable_p on the result.
24019 (aarch64_output_simd_mov_immediate): Remove mode argument.
24020 Update call to aarch64_simd_valid_immediate and use of
24021 simd_immediate_info.
24022 (aarch64_output_scalar_simd_mov_immediate): Update call
24025 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24026 Alan Hayward <alan.hayward@arm.com>
24027 David Sherwood <david.sherwood@arm.com>
24029 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
24030 (mode_nunits): Likewise CONST_MODE_NUNITS.
24031 * machmode.def (ADJUST_NUNITS): Document.
24032 * genmodes.c (mode_data::need_nunits_adj): New field.
24033 (blank_mode): Update accordingly.
24034 (adj_nunits): New variable.
24035 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
24037 (emit_mode_size_inline): Set need_bytesize_adj for all modes
24038 listed in adj_nunits.
24039 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
24040 listed in adj_nunits. Don't emit case statements for such modes.
24041 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
24042 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
24043 nothing if adj_nunits is nonnull.
24044 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
24045 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
24046 (emit_mode_fbit): Update use of print_maybe_const_decl.
24047 (emit_move_size): Likewise. Treat the array as non-const
24049 (emit_mode_adjustments): Handle adj_nunits.
24051 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24053 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
24054 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
24055 (VECTOR_MODES): Use it.
24056 (make_vector_modes): Take the prefix as an argument.
24058 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24059 Alan Hayward <alan.hayward@arm.com>
24060 David Sherwood <david.sherwood@arm.com>
24062 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
24063 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
24064 for MODE_VECTOR_BOOL.
24065 * machmode.def (VECTOR_BOOL_MODE): Document.
24066 * genmodes.c (VECTOR_BOOL_MODE): New macro.
24067 (make_vector_bool_mode): New function.
24068 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
24070 * lto-streamer-in.c (lto_input_mode_table): Likewise.
24071 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
24073 * stor-layout.c (int_mode_for_mode): Likewise.
24074 * tree.c (build_vector_type_for_mode): Likewise.
24075 * varasm.c (output_constant_pool_2): Likewise.
24076 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
24077 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
24078 for MODE_VECTOR_BOOL.
24079 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
24080 of mode class checks.
24081 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
24082 instead of a list of mode class checks.
24083 (expand_vector_scalar_condition): Likewise.
24084 (type_for_widest_vector_mode): Handle BImode as an inner mode.
24086 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24087 Alan Hayward <alan.hayward@arm.com>
24088 David Sherwood <david.sherwood@arm.com>
24090 * machmode.h (mode_size): Change from unsigned short to
24092 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
24093 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
24094 or if measurement_type is not polynomial.
24095 (fixed_size_mode::includes_p): Check for constant-sized modes.
24096 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
24097 return a poly_uint16 rather than an unsigned short.
24098 (emit_mode_size): Change the type of mode_size from unsigned short
24099 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
24100 (emit_mode_adjustments): Cope with polynomial vector sizes.
24101 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24103 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24105 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
24106 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
24107 * caller-save.c (setup_save_areas): Likewise.
24108 (replace_reg_with_saved_mem): Likewise.
24109 * calls.c (emit_library_call_value_1): Likewise.
24110 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
24111 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
24112 (gen_lowpart_for_combine): Likewise.
24113 * convert.c (convert_to_integer_1): Likewise.
24114 * cse.c (equiv_constant, cse_insn): Likewise.
24115 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
24116 (cselib_subst_to_values): Likewise.
24117 * dce.c (word_dce_process_block): Likewise.
24118 * df-problems.c (df_word_lr_mark_ref): Likewise.
24119 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
24120 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
24121 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
24122 (rtl_for_decl_location): Likewise.
24123 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
24124 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
24125 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
24126 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
24127 (expand_expr_real_1): Likewise.
24128 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
24129 (pad_below): Likewise.
24130 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
24131 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
24132 * ira.c (get_subreg_tracking_sizes): Likewise.
24133 * ira-build.c (ira_create_allocno_objects): Likewise.
24134 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
24135 (ira_sort_regnos_for_alter_reg): Likewise.
24136 * ira-costs.c (record_operand_costs): Likewise.
24137 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
24138 (resolve_simple_move): Likewise.
24139 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
24140 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
24141 (lra_constraints): Likewise.
24142 (CONST_POOL_OK_P): Reject variable-sized modes.
24143 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
24144 (add_pseudo_to_slot, lra_spill): Likewise.
24145 * omp-low.c (omp_clause_aligned_alignment): Likewise.
24146 * optabs-query.c (get_best_extraction_insn): Likewise.
24147 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
24148 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
24149 (expand_mult_highpart, valid_multiword_target_p): Likewise.
24150 * recog.c (offsettable_address_addr_space_p): Likewise.
24151 * regcprop.c (maybe_mode_change): Likewise.
24152 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
24153 * regrename.c (build_def_use): Likewise.
24154 * regstat.c (dump_reg_info): Likewise.
24155 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
24156 (find_reloads, find_reloads_subreg_address): Likewise.
24157 * reload1.c (eliminate_regs_1): Likewise.
24158 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
24159 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
24160 (simplify_binary_operation_1, simplify_subreg): Likewise.
24161 * targhooks.c (default_function_arg_padding): Likewise.
24162 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
24163 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
24164 (verify_gimple_assign_ternary): Likewise.
24165 * tree-inline.c (estimate_move_cost): Likewise.
24166 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
24167 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
24168 (get_address_cost_ainc): Likewise.
24169 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
24170 (vect_supportable_dr_alignment): Likewise.
24171 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24172 (vectorizable_reduction): Likewise.
24173 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
24174 (vectorizable_operation, vectorizable_load): Likewise.
24175 * tree.c (build_same_sized_truth_vector_type): Likewise.
24176 * valtrack.c (cleanup_auto_inc_dec): Likewise.
24177 * var-tracking.c (emit_note_insn_var_location): Likewise.
24178 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
24179 (ADDR_VEC_ALIGN): Likewise.
24181 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24182 Alan Hayward <alan.hayward@arm.com>
24183 David Sherwood <david.sherwood@arm.com>
24185 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
24187 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
24188 or if measurement_type is polynomial.
24189 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
24190 * combine.c (make_extraction): Likewise.
24191 * dse.c (find_shift_sequence): Likewise.
24192 * dwarf2out.c (mem_loc_descriptor): Likewise.
24193 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
24194 (extract_bit_field, extract_low_bits): Likewise.
24195 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
24196 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
24197 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
24198 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
24199 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
24200 * reload.c (find_reloads): Likewise.
24201 * reload1.c (alter_reg): Likewise.
24202 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
24203 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
24204 * tree-if-conv.c (predicate_mem_writes): Likewise.
24205 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
24206 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
24207 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
24208 * valtrack.c (dead_debug_insert_temp): Likewise.
24209 * varasm.c (mergeable_constant_section): Likewise.
24210 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
24212 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24213 Alan Hayward <alan.hayward@arm.com>
24214 David Sherwood <david.sherwood@arm.com>
24216 * expr.c (expand_assignment): Cope with polynomial mode sizes
24217 when assigning to a CONCAT.
24219 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24220 Alan Hayward <alan.hayward@arm.com>
24221 David Sherwood <david.sherwood@arm.com>
24223 * machmode.h (mode_precision): Change from unsigned short to
24225 (mode_to_precision): Return a poly_uint16 rather than an unsigned
24227 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
24228 or if measurement_type is not polynomial.
24229 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
24230 in which the mode is already known to be a scalar_int_mode.
24231 * genmodes.c (emit_mode_precision): Change the type of mode_precision
24232 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
24234 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24235 for GET_MODE_PRECISION.
24236 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24237 for GET_MODE_PRECISION.
24238 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
24240 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
24241 (expand_field_assignment, make_extraction): Likewise.
24242 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
24243 (get_last_value): Likewise.
24244 * convert.c (convert_to_integer_1): Likewise.
24245 * cse.c (cse_insn): Likewise.
24246 * expr.c (expand_expr_real_1): Likewise.
24247 * lra-constraints.c (simplify_operand_subreg): Likewise.
24248 * optabs-query.c (can_atomic_load_p): Likewise.
24249 * optabs.c (expand_atomic_load): Likewise.
24250 (expand_atomic_store): Likewise.
24251 * ree.c (combine_reaching_defs): Likewise.
24252 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
24253 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
24254 * tree.h (type_has_mode_precision_p): Likewise.
24255 * ubsan.c (instrument_si_overflow): Likewise.
24257 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24258 Alan Hayward <alan.hayward@arm.com>
24259 David Sherwood <david.sherwood@arm.com>
24261 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
24262 polynomial numbers of units.
24263 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
24264 (valid_vector_subparts_p): New function.
24265 (build_vector_type): Remove temporary shim and take the number
24266 of units as a poly_uint64 rather than an int.
24267 (build_opaque_vector_type): Take the number of units as a
24268 poly_uint64 rather than an int.
24269 * tree.c (build_vector_from_ctor): Handle polynomial
24270 TYPE_VECTOR_SUBPARTS.
24271 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
24272 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
24273 (build_vector_from_val): If the number of units is variable,
24274 use build_vec_duplicate_cst for constant operands and
24275 VEC_DUPLICATE_EXPR otherwise.
24276 (make_vector_type): Remove temporary is_constant ().
24277 (build_vector_type, build_opaque_vector_type): Take the number of
24278 units as a poly_uint64 rather than an int.
24279 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
24281 * cfgexpand.c (expand_debug_expr): Likewise.
24282 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
24283 (store_constructor, expand_expr_real_1): Likewise.
24284 (const_scalar_mask_from_tree): Likewise.
24285 * fold-const-call.c (fold_const_reduction): Likewise.
24286 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
24287 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
24288 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
24289 (fold_relational_const): Likewise.
24290 (native_interpret_vector): Likewise. Change the size from an
24291 int to an unsigned int.
24292 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
24293 TYPE_VECTOR_SUBPARTS.
24294 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
24295 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
24296 duplicating a non-constant operand into a variable-length vector.
24297 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
24298 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
24299 * ipa-icf.c (sem_variable::equals): Likewise.
24300 * match.pd: Likewise.
24301 * omp-simd-clone.c (simd_clone_subparts): Likewise.
24302 * print-tree.c (print_node): Likewise.
24303 * stor-layout.c (layout_type): Likewise.
24304 * targhooks.c (default_builtin_vectorization_cost): Likewise.
24305 * tree-cfg.c (verify_gimple_comparison): Likewise.
24306 (verify_gimple_assign_binary): Likewise.
24307 (verify_gimple_assign_ternary): Likewise.
24308 (verify_gimple_assign_single): Likewise.
24309 * tree-pretty-print.c (dump_generic_node): Likewise.
24310 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
24311 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
24312 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
24313 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
24314 (vect_shift_permute_load_chain): Likewise.
24315 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
24316 (expand_vector_condition, optimize_vector_constructor): Likewise.
24317 (lower_vec_perm, get_compute_type): Likewise.
24318 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24319 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
24320 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
24321 (vect_recog_mask_conversion_pattern): Likewise.
24322 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
24323 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
24324 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
24325 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
24326 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
24327 (vectorizable_shift, vectorizable_operation, vectorizable_store)
24328 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
24329 (supportable_widening_operation): Likewise.
24330 (supportable_narrowing_operation): Likewise.
24331 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
24333 * varasm.c (output_constant): Likewise.
24335 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24336 Alan Hayward <alan.hayward@arm.com>
24337 David Sherwood <david.sherwood@arm.com>
24339 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
24340 so that both the length == 3 and length != 3 cases set up their
24341 own permute vectors. Add comments explaining why we know the
24342 number of elements is constant.
24343 (vect_permute_load_chain): Likewise.
24345 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24346 Alan Hayward <alan.hayward@arm.com>
24347 David Sherwood <david.sherwood@arm.com>
24349 * machmode.h (mode_nunits): Change from unsigned char to
24351 (ONLY_FIXED_SIZE_MODES): New macro.
24352 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
24353 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
24354 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
24356 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
24357 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
24358 or if measurement_type is not polynomial.
24359 * genmodes.c (ZERO_COEFFS): New macro.
24360 (emit_mode_nunits_inline): Make mode_nunits_inline return a
24362 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
24363 Use ZERO_COEFFS when emitting initializers.
24364 * data-streamer.h (bp_pack_poly_value): New function.
24365 (bp_unpack_poly_value): Likewise.
24366 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24367 for GET_MODE_NUNITS.
24368 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24369 for GET_MODE_NUNITS.
24370 * tree.c (make_vector_type): Remove temporary shim and make
24371 the real function take the number of units as a poly_uint64
24372 rather than an int.
24373 (build_vector_type_for_mode): Handle polynomial nunits.
24374 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
24375 * emit-rtl.c (const_vec_series_p_1): Likewise.
24376 (gen_rtx_CONST_VECTOR): Likewise.
24377 * fold-const.c (test_vec_duplicate_folding): Likewise.
24378 * genrecog.c (validate_pattern): Likewise.
24379 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
24380 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
24381 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
24382 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
24383 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
24384 * rtlanal.c (subreg_get_info): Likewise.
24385 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
24386 (vect_grouped_load_supported): Likewise.
24387 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
24388 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
24389 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
24390 (simplify_const_unary_operation, simplify_binary_operation_1)
24391 (simplify_const_binary_operation, simplify_ternary_operation)
24392 (test_vector_ops_duplicate, test_vector_ops): Likewise.
24393 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
24394 instead of CONST_VECTOR_NUNITS.
24395 * varasm.c (output_constant_pool_2): Likewise.
24396 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
24397 explicit-encoded elements in the XVEC for variable-length vectors.
24399 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24401 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
24403 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24404 Alan Hayward <alan.hayward@arm.com>
24405 David Sherwood <david.sherwood@arm.com>
24407 * coretypes.h (fixed_size_mode): Declare.
24408 (fixed_size_mode_pod): New typedef.
24409 * builtins.h (target_builtins::x_apply_args_mode)
24410 (target_builtins::x_apply_result_mode): Change type to
24411 fixed_size_mode_pod.
24412 * builtins.c (apply_args_size, apply_result_size, result_vector)
24413 (expand_builtin_apply_args_1, expand_builtin_apply)
24414 (expand_builtin_return): Update accordingly.
24416 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24418 * cse.c (hash_rtx_cb): Hash only the encoded elements.
24419 * cselib.c (cselib_hash_rtx): Likewise.
24420 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
24421 CONST_VECTOR encoding.
24423 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24424 Jeff Law <law@redhat.com>
24427 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
24428 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
24429 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
24430 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
24433 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
24434 explicitly probe *sp in a noreturn function if there were any callee
24435 register saves or frame pointer is needed.
24437 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24440 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
24441 BLKmode for ternary, binary or unary expressions.
24444 * var-tracking.c (delete_vta_debug_insn): New inline function.
24445 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
24446 insns from get_insns () to NULL instead of each bb separately.
24447 Use delete_vta_debug_insn. No longer static.
24448 (vt_debug_insns_local, variable_tracking_main_1): Adjust
24449 delete_vta_debug_insns callers.
24450 * rtl.h (delete_vta_debug_insns): Declare.
24451 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
24452 instead of variable_tracking_main.
24454 2018-01-03 Martin Sebor <msebor@redhat.com>
24456 PR tree-optimization/83603
24457 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
24458 arguments past the endof the argument list in functions declared
24459 without a prototype.
24460 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
24461 Avoid checking when arguments are null.
24463 2018-01-03 Martin Sebor <msebor@redhat.com>
24466 * doc/extend.texi (attribute const): Fix a typo.
24467 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
24468 issuing -Wsuggest-attribute for void functions.
24470 2018-01-03 Martin Sebor <msebor@redhat.com>
24472 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
24473 offset_int::from instead of wide_int::to_shwi.
24474 (maybe_diag_overlap): Remove assertion.
24475 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
24476 * gimple-ssa-sprintf.c (format_directive): Same.
24477 (parse_directive): Same.
24478 (sprintf_dom_walker::compute_format_length): Same.
24479 (try_substitute_return_value): Same.
24481 2018-01-03 Jeff Law <law@redhat.com>
24483 PR middle-end/83654
24484 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
24485 non-constant residual for zero at runtime and avoid probing in
24486 that case. Reorganize code for trailing problem to mirror handling
24489 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24491 PR tree-optimization/83501
24492 * tree-ssa-strlen.c (get_string_cst): New.
24493 (handle_char_store): Call get_string_cst.
24495 2018-01-03 Martin Liska <mliska@suse.cz>
24497 PR tree-optimization/83593
24498 * tree-ssa-strlen.c: Include tree-cfg.h.
24499 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
24500 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
24501 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
24503 (strlen_dom_walker::before_dom_children): Call
24504 gimple_purge_dead_eh_edges. Dump tranformation with details
24506 (strlen_dom_walker::before_dom_children): Update call by adding
24507 new argument cleanup_eh.
24508 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
24510 2018-01-03 Martin Liska <mliska@suse.cz>
24513 * cif-code.def (VARIADIC_THUNK): New enum value.
24514 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
24517 2018-01-03 Jan Beulich <jbeulich@suse.com>
24519 * sse.md (mov<mode>_internal): Tighten condition for when to use
24520 vmovdqu<ssescalarsize> for TI and OI modes.
24522 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24524 Update copyright years.
24526 2018-01-03 Martin Liska <mliska@suse.cz>
24529 * ipa-visibility.c (function_and_variable_visibility): Skip
24530 functions with noipa attribure.
24532 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24534 * gcc.c (process_command): Update copyright notice dates.
24535 * gcov-dump.c (print_version): Ditto.
24536 * gcov.c (print_version): Ditto.
24537 * gcov-tool.c (print_version): Ditto.
24538 * gengtype.c (create_file): Ditto.
24539 * doc/cpp.texi: Bump @copying's copyright year.
24540 * doc/cppinternals.texi: Ditto.
24541 * doc/gcc.texi: Ditto.
24542 * doc/gccint.texi: Ditto.
24543 * doc/gcov.texi: Ditto.
24544 * doc/install.texi: Ditto.
24545 * doc/invoke.texi: Ditto.
24547 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24549 * vector-builder.h (vector_builder::m_full_nelts): Change from
24550 unsigned int to poly_uint64.
24551 (vector_builder::full_nelts): Update prototype accordingly.
24552 (vector_builder::new_vector): Likewise.
24553 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
24554 (vector_builder::operator ==): Likewise.
24555 (vector_builder::finalize): Likewise.
24556 * int-vector-builder.h (int_vector_builder::int_vector_builder):
24557 Take the number of elements as a poly_uint64 rather than an
24559 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
24560 from unsigned int to poly_uint64.
24561 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
24562 (vec_perm_indices::new_vector): Likewise.
24563 (vec_perm_indices::length): Likewise.
24564 (vec_perm_indices::nelts_per_input): Likewise.
24565 (vec_perm_indices::input_nelts): Likewise.
24566 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
24567 number of elements per input as a poly_uint64 rather than an
24568 unsigned int. Use the original encoding for variable-length
24569 vectors, rather than clamping each individual element.
24570 For the second and subsequent elements in each pattern,
24571 clamp the step and base before clamping their sum.
24572 (vec_perm_indices::series_p): Handle polynomial element counts.
24573 (vec_perm_indices::all_in_range_p): Likewise.
24574 (vec_perm_indices_to_tree): Likewise.
24575 (vec_perm_indices_to_rtx): Likewise.
24576 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
24577 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
24578 (tree_vector_builder::new_binary_operation): Handle polynomial
24579 element counts. Return false if we need to know the number
24580 of elements at compile time.
24581 * fold-const.c (fold_vec_perm): Punt if the number of elements
24582 isn't known at compile time.
24584 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24586 * vec-perm-indices.h (vec_perm_builder): Change element type
24587 from HOST_WIDE_INT to poly_int64.
24588 (vec_perm_indices::element_type): Update accordingly.
24589 (vec_perm_indices::clamp): Handle polynomial element_types.
24590 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
24591 (vec_perm_indices::all_in_range_p): Likewise.
24592 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
24594 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
24595 polynomial vec_perm_indices element types.
24596 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
24597 * fold-const.c (fold_vec_perm): Likewise.
24598 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
24599 * tree-vect-generic.c (lower_vec_perm): Likewise.
24600 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
24601 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
24602 element type to HOST_WIDE_INT.
24604 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24605 Alan Hayward <alan.hayward@arm.com>
24606 David Sherwood <david.sherwood@arm.com>
24608 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
24609 rather than an int. Use plus_constant.
24610 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
24611 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
24613 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24614 Alan Hayward <alan.hayward@arm.com>
24615 David Sherwood <david.sherwood@arm.com>
24617 * calls.c (emit_call_1, expand_call): Change struct_value_size from
24618 a HOST_WIDE_INT to a poly_int64.
24620 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24621 Alan Hayward <alan.hayward@arm.com>
24622 David Sherwood <david.sherwood@arm.com>
24624 * calls.c (load_register_parameters): Cope with polynomial
24625 mode sizes. Require a constant size for BLKmode parameters
24626 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
24627 forces a parameter to be padded at the lsb end in order to
24628 fill a complete number of words, require the parameter size
24629 to be ordered wrt UNITS_PER_WORD.
24631 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24632 Alan Hayward <alan.hayward@arm.com>
24633 David Sherwood <david.sherwood@arm.com>
24635 * reload1.c (spill_stack_slot_width): Change element type
24636 from unsigned int to poly_uint64_pod.
24637 (alter_reg): Treat mode sizes as polynomial.
24639 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24640 Alan Hayward <alan.hayward@arm.com>
24641 David Sherwood <david.sherwood@arm.com>
24643 * reload.c (complex_word_subreg_p): New function.
24644 (reload_inner_reg_of_subreg, push_reload): Use it.
24646 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24647 Alan Hayward <alan.hayward@arm.com>
24648 David Sherwood <david.sherwood@arm.com>
24650 * lra-constraints.c (process_alt_operands): Reject matched
24651 operands whose sizes aren't ordered.
24652 (match_reload): Refer to this check here.
24654 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24655 Alan Hayward <alan.hayward@arm.com>
24656 David Sherwood <david.sherwood@arm.com>
24658 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
24659 that the mode size is in the set {1, 2, 4, 8, 16}.
24661 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24662 Alan Hayward <alan.hayward@arm.com>
24663 David Sherwood <david.sherwood@arm.com>
24665 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
24666 Use plus_constant instead of gen_rtx_PLUS.
24668 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24669 Alan Hayward <alan.hayward@arm.com>
24670 David Sherwood <david.sherwood@arm.com>
24672 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
24673 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
24674 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
24675 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
24676 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
24677 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
24678 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
24679 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
24680 * config/i386/i386.c (ix86_push_rounding): ...this new function.
24681 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
24683 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
24684 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
24685 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
24686 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
24687 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
24688 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
24689 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
24690 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
24691 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
24692 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
24694 * expr.c (emit_move_resolve_push): Treat the input and result
24695 of PUSH_ROUNDING as a poly_int64.
24696 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
24697 (emit_push_insn): Likewise.
24698 * lra-eliminations.c (mark_not_eliminable): Likewise.
24699 * recog.c (push_operand): Likewise.
24700 * reload1.c (elimination_effects): Likewise.
24701 * rtlanal.c (nonzero_bits1): Likewise.
24702 * calls.c (store_one_arg): Likewise. Require the padding to be
24703 known at compile time.
24705 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24706 Alan Hayward <alan.hayward@arm.com>
24707 David Sherwood <david.sherwood@arm.com>
24709 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
24710 Use plus_constant instead of gen_rtx_PLUS.
24712 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24713 Alan Hayward <alan.hayward@arm.com>
24714 David Sherwood <david.sherwood@arm.com>
24716 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
24717 rather than an int.
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 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
24724 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
24725 via stack temporaries. Treat the mode size as polynomial too.
24727 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24728 Alan Hayward <alan.hayward@arm.com>
24729 David Sherwood <david.sherwood@arm.com>
24731 * expr.c (expand_expr_real_2): When handling conversions involving
24732 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
24733 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
24734 as a poly_uint64 too.
24736 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24737 Alan Hayward <alan.hayward@arm.com>
24738 David Sherwood <david.sherwood@arm.com>
24740 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
24742 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24743 Alan Hayward <alan.hayward@arm.com>
24744 David Sherwood <david.sherwood@arm.com>
24746 * combine.c (can_change_dest_mode): Handle polynomial
24747 REGMODE_NATURAL_SIZE.
24748 * expmed.c (store_bit_field_1): Likewise.
24749 * expr.c (store_constructor): Likewise.
24750 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
24751 and polynomial REGMODE_NATURAL_SIZE.
24752 (gen_lowpart_common): Likewise.
24753 * reginfo.c (record_subregs_of_mode): Likewise.
24754 * rtlanal.c (read_modify_subreg_p): Likewise.
24756 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24757 Alan Hayward <alan.hayward@arm.com>
24758 David Sherwood <david.sherwood@arm.com>
24760 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
24761 numbers of elements.
24763 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24764 Alan Hayward <alan.hayward@arm.com>
24765 David Sherwood <david.sherwood@arm.com>
24767 * match.pd: Cope with polynomial numbers of vector elements.
24769 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24770 Alan Hayward <alan.hayward@arm.com>
24771 David Sherwood <david.sherwood@arm.com>
24773 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
24774 in a POINTER_PLUS_EXPR.
24776 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24777 Alan Hayward <alan.hayward@arm.com>
24778 David Sherwood <david.sherwood@arm.com>
24780 * omp-simd-clone.c (simd_clone_subparts): New function.
24781 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
24782 (ipa_simd_modify_function_body): Likewise.
24784 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24785 Alan Hayward <alan.hayward@arm.com>
24786 David Sherwood <david.sherwood@arm.com>
24788 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
24789 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
24790 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
24791 (expand_vector_condition, vector_element): Likewise.
24792 (subparts_gt): New function.
24793 (get_compute_type): Use subparts_gt.
24794 (count_type_subparts): Delete.
24795 (expand_vector_operations_1): Use subparts_gt instead of
24796 count_type_subparts.
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-data-refs.c (vect_no_alias_p): Replace with...
24803 (vect_compile_time_alias): ...this new function. Do the calculation
24804 on poly_ints rather than trees.
24805 (vect_prune_runtime_alias_test_list): Update call accordingly.
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-slp.c (vect_build_slp_tree_1): Handle polynomial
24813 (vect_schedule_slp_instance): Likewise.
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-slp.c (vect_get_and_check_slp_defs): Reject
24820 constant and extern definitions for variable-length vectors.
24821 (vect_get_constant_vectors): Note that the number of units
24822 is known to be constant.
24824 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24825 Alan Hayward <alan.hayward@arm.com>
24826 David Sherwood <david.sherwood@arm.com>
24828 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
24829 of units as polynomial. Choose between WIDE and NARROW based
24832 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24833 Alan Hayward <alan.hayward@arm.com>
24834 David Sherwood <david.sherwood@arm.com>
24836 * tree-vect-stmts.c (simd_clone_subparts): New function.
24837 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
24839 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24840 Alan Hayward <alan.hayward@arm.com>
24841 David Sherwood <david.sherwood@arm.com>
24843 * tree-vect-stmts.c (vectorizable_call): Treat the number of
24844 vectors as polynomial. Use build_index_vector for
24845 IFN_GOMP_SIMD_LANE.
24847 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24848 Alan Hayward <alan.hayward@arm.com>
24849 David Sherwood <david.sherwood@arm.com>
24851 * tree-vect-stmts.c (get_load_store_type): Treat the number of
24852 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
24853 for variable-length vectors.
24854 (vectorizable_mask_load_store): Treat the number of units as
24855 polynomial, asserting that it is constant if the condition has
24856 already been enforced.
24857 (vectorizable_store, vectorizable_load): Likewise.
24859 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24860 Alan Hayward <alan.hayward@arm.com>
24861 David Sherwood <david.sherwood@arm.com>
24863 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
24864 of units as polynomial. Punt if we can't tell at compile time
24865 which vector contains the final result.
24867 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24868 Alan Hayward <alan.hayward@arm.com>
24869 David Sherwood <david.sherwood@arm.com>
24871 * tree-vect-loop.c (vectorizable_induction): Treat the number
24872 of units as polynomial. Punt on SLP inductions. Use an integer
24873 VEC_SERIES_EXPR for variable-length integer reductions. Use a
24874 cast of such a series for variable-length floating-point
24877 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24878 Alan Hayward <alan.hayward@arm.com>
24879 David Sherwood <david.sherwood@arm.com>
24881 * tree.h (build_index_vector): Declare.
24882 * tree.c (build_index_vector): New function.
24883 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
24884 of units as polynomial, forcibly converting it to a constant if
24885 vectorizable_reduction has already enforced the condition.
24886 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
24887 to create a {1,2,3,...} vector.
24888 (vectorizable_reduction): Treat the number of units as polynomial.
24889 Choose vectype_in based on the largest scalar element size rather
24890 than the smallest number of units. Enforce the restrictions
24893 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24894 Alan Hayward <alan.hayward@arm.com>
24895 David Sherwood <david.sherwood@arm.com>
24897 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
24898 number of units as polynomial.
24900 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24901 Alan Hayward <alan.hayward@arm.com>
24902 David Sherwood <david.sherwood@arm.com>
24904 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
24905 * target.def (autovectorize_vector_sizes): Return the vector sizes
24906 by pointer, using vector_sizes rather than a bitmask.
24907 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
24908 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
24909 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
24911 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
24912 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
24913 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
24914 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
24915 * omp-general.c (omp_max_vf): Likewise.
24916 * omp-low.c (omp_clause_aligned_alignment): Likewise.
24917 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
24918 * tree-vect-loop.c (vect_analyze_loop): Likewise.
24919 * tree-vect-slp.c (vect_slp_bb): Likewise.
24920 * doc/tm.texi: Regenerate.
24921 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
24923 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
24924 the vector size as a poly_uint64 rather than an unsigned int.
24925 (current_vector_size): Change from an unsigned int to a poly_uint64.
24926 (get_vectype_for_scalar_type): Update accordingly.
24927 * tree.h (build_truth_vector_type): Take the size and number of
24928 units as a poly_uint64 rather than an unsigned int.
24929 (build_vector_type): Add a temporary overload that takes
24930 the number of units as a poly_uint64 rather than an unsigned int.
24931 * tree.c (make_vector_type): Likewise.
24932 (build_truth_vector_type): Take the number of units as a poly_uint64
24933 rather than an unsigned int.
24935 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24936 Alan Hayward <alan.hayward@arm.com>
24937 David Sherwood <david.sherwood@arm.com>
24939 * target.def (get_mask_mode): Take the number of units and length
24940 as poly_uint64s rather than unsigned ints.
24941 * targhooks.h (default_get_mask_mode): Update accordingly.
24942 * targhooks.c (default_get_mask_mode): Likewise.
24943 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
24944 * doc/tm.texi: Regenerate.
24946 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24947 Alan Hayward <alan.hayward@arm.com>
24948 David Sherwood <david.sherwood@arm.com>
24950 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
24951 * omp-general.c (omp_max_vf): Likewise.
24952 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
24953 (expand_omp_simd): Handle polynomial safelen.
24954 * omp-low.c (omplow_simd_context): Add a default constructor.
24955 (omplow_simd_context::max_vf): Change from int to poly_uint64.
24956 (lower_rec_simd_input_clauses): Update accordingly.
24957 (lower_rec_input_clauses): Likewise.
24959 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24960 Alan Hayward <alan.hayward@arm.com>
24961 David Sherwood <david.sherwood@arm.com>
24963 * tree-vectorizer.h (vect_nunits_for_cost): New function.
24964 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
24965 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
24966 (vect_analyze_slp_cost): Likewise.
24967 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
24968 (vect_model_load_cost): Likewise.
24970 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24971 Alan Hayward <alan.hayward@arm.com>
24972 David Sherwood <david.sherwood@arm.com>
24974 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
24975 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
24976 from an unsigned int * to a poly_uint64_pod *.
24977 (calculate_unrolling_factor): New function.
24978 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
24980 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24981 Alan Hayward <alan.hayward@arm.com>
24982 David Sherwood <david.sherwood@arm.com>
24984 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
24985 from an unsigned int to a poly_uint64.
24986 (_loop_vec_info::slp_unrolling_factor): Likewise.
24987 (_loop_vec_info::vectorization_factor): Change from an int
24989 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
24990 (vect_get_num_vectors): New function.
24991 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
24992 (vect_get_num_copies): Use vect_get_num_vectors.
24993 (vect_analyze_data_ref_dependences): Change max_vf from an int *
24994 to an unsigned int *.
24995 (vect_analyze_data_refs): Change min_vf from an int * to a
24997 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
24998 than an unsigned HOST_WIDE_INT.
24999 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
25000 (vect_analyze_data_ref_dependence): Change max_vf from an int *
25001 to an unsigned int *.
25002 (vect_analyze_data_ref_dependences): Likewise.
25003 (vect_compute_data_ref_alignment): Handle polynomial vf.
25004 (vect_enhance_data_refs_alignment): Likewise.
25005 (vect_prune_runtime_alias_test_list): Likewise.
25006 (vect_shift_permute_load_chain): Likewise.
25007 (vect_supportable_dr_alignment): Likewise.
25008 (dependence_distance_ge_vf): Take the vectorization factor as a
25009 poly_uint64 rather than an unsigned HOST_WIDE_INT.
25010 (vect_analyze_data_refs): Change min_vf from an int * to a
25012 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
25013 vfm1 as a poly_uint64 rather than an int. Make the same change
25014 for the returned bound_scalar.
25015 (vect_gen_vector_loop_niters): Handle polynomial vf.
25016 (vect_do_peeling): Likewise. Update call to
25017 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
25018 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
25020 * tree-vect-loop.c (vect_determine_vectorization_factor)
25021 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
25022 (vect_get_known_peeling_cost): Likewise.
25023 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
25024 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
25025 (vect_transform_loop): Likewise. Use the lowest possible VF when
25026 updating the upper bounds of the loop.
25027 (vect_min_worthwhile_factor): Make static. Return an unsigned int
25028 rather than an int.
25029 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
25030 polynomial unroll factors.
25031 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
25032 (vect_make_slp_decision): Likewise.
25033 (vect_supported_load_permutation_p): Likewise, and polynomial
25035 (vect_analyze_slp_cost): Handle polynomial vf.
25036 (vect_slp_analyze_node_operations): Likewise.
25037 (vect_slp_analyze_bb_1): Likewise.
25038 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
25039 than an unsigned HOST_WIDE_INT.
25040 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
25041 (vectorizable_load): Handle polynomial vf.
25042 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
25044 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
25046 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25047 Alan Hayward <alan.hayward@arm.com>
25048 David Sherwood <david.sherwood@arm.com>
25050 * match.pd: Handle bit operations involving three constants
25051 and try to fold one pair.
25053 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25055 * tree-vect-loop-manip.c: Include gimple-fold.h.
25056 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
25057 niters_maybe_zero parameters. Handle other cases besides a step of 1.
25058 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
25059 Add a path that uses a step of VF instead of 1, but disable it
25061 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
25062 and niters_no_overflow parameters. Update calls to
25063 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
25064 Create a new SSA name if the latter choses to use a ste other
25065 than zero, and return it via niters_vector_mult_vf_var.
25066 * tree-vect-loop.c (vect_transform_loop): Update calls to
25067 vect_do_peeling, vect_gen_vector_loop_niters and
25068 slpeel_make_loop_iterate_ntimes.
25069 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
25070 (vect_gen_vector_loop_niters): Update declarations after above changes.
25072 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
25074 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
25075 128-bit round to integer instructions.
25076 (ceil<mode>2): Likewise.
25077 (btrunc<mode>2): Likewise.
25078 (round<mode>2): Likewise.
25080 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25082 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
25083 unaligned VSX load/store on P8/P9.
25084 (expand_block_clear): Allow the use of unaligned VSX
25085 load/store on P8/P9.
25087 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25089 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
25091 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
25092 swap associated with both a load and a store.
25094 2018-01-02 Andrew Waterman <andrew@sifive.com>
25096 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
25097 * config/riscv/riscv.md (clear_cache): Use it.
25099 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
25101 * web.c: Remove out-of-date comment.
25103 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25105 * expr.c (fixup_args_size_notes): Check that any existing
25106 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
25107 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
25108 (emit_single_push_insn): ...here.
25110 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25112 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
25113 (const_vector_encoded_nelts): New function.
25114 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
25115 (const_vector_int_elt, const_vector_elt): Declare.
25116 * emit-rtl.c (const_vector_int_elt_1): New function.
25117 (const_vector_elt): Likewise.
25118 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
25119 of CONST_VECTOR_ELT.
25121 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25123 * expr.c: Include rtx-vector-builder.h.
25124 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
25125 directly on the tree encoding.
25126 (const_vector_from_tree): Likewise.
25127 * optabs.c: Include rtx-vector-builder.h.
25128 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
25129 sequence of "u" values.
25130 * vec-perm-indices.c: Include rtx-vector-builder.h.
25131 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
25132 directly on the vec_perm_indices encoding.
25134 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25136 * doc/rtl.texi (const_vector): Describe new encoding scheme.
25137 * Makefile.in (OBJS): Add rtx-vector-builder.o.
25138 * rtx-vector-builder.h: New file.
25139 * rtx-vector-builder.c: Likewise.
25140 * rtl.h (rtx_def::u2): Add a const_vector field.
25141 (CONST_VECTOR_NPATTERNS): New macro.
25142 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
25143 (CONST_VECTOR_DUPLICATE_P): Likewise.
25144 (CONST_VECTOR_STEPPED_P): Likewise.
25145 (CONST_VECTOR_ENCODED_ELT): Likewise.
25146 (const_vec_duplicate_p): Check for a duplicated vector encoding.
25147 (unwrap_const_vec_duplicate): Likewise.
25148 (const_vec_series_p): Check for a non-duplicated vector encoding.
25149 Say that the function only returns true for integer vectors.
25150 * emit-rtl.c: Include rtx-vector-builder.h.
25151 (gen_const_vec_duplicate_1): Delete.
25152 (gen_const_vector): Call gen_const_vec_duplicate instead of
25153 gen_const_vec_duplicate_1.
25154 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
25155 (gen_const_vec_duplicate): Use rtx_vector_builder.
25156 (gen_const_vec_series): Likewise.
25157 (gen_rtx_CONST_VECTOR): Likewise.
25158 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
25159 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
25160 Build a new vector rather than modifying a CONST_VECTOR in-place.
25161 (handle_special_swappables): Update call accordingly.
25162 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
25163 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
25164 Build a new vector rather than modifying a CONST_VECTOR in-place.
25165 (handle_special_swappables): Update call accordingly.
25167 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25169 * simplify-rtx.c (simplify_const_binary_operation): Use
25170 CONST_VECTOR_ELT instead of XVECEXP.
25172 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25174 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
25175 the selector elements to be different from the data elements
25176 if the selector is a VECTOR_CST.
25177 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
25178 ssizetype for the selector.
25180 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25182 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
25183 before testing each element individually.
25184 * tree-vect-generic.c (lower_vec_perm): Likewise.
25186 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25188 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
25189 * selftest-run-tests.c (selftest::run_tests): Call it.
25190 * vector-builder.h (vector_builder::operator ==): New function.
25191 (vector_builder::operator !=): Likewise.
25192 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
25193 (vec_perm_indices::all_from_input_p): New function.
25194 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
25195 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
25196 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
25197 instead of reading the VECTOR_CST directly. Detect whether both
25198 vector inputs are the same before constructing the vec_perm_indices,
25199 and update the number of inputs argument accordingly. Use the
25200 utility functions added above. Only construct sel2 if we need to.
25202 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25204 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
25205 the broadcast of the low byte.
25206 (expand_mult_highpart): Use an explicit encoding for the permutes.
25207 * optabs-query.c (can_mult_highpart_p): Likewise.
25208 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25209 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25210 (vectorizable_bswap): Likewise.
25211 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
25212 explicit encoding for the power-of-2 permutes.
25213 (vect_permute_store_chain): Likewise.
25214 (vect_grouped_load_supported): Likewise.
25215 (vect_permute_load_chain): Likewise.
25217 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25219 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
25220 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
25221 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
25222 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
25223 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
25224 (vect_gen_perm_mask_any): Likewise.
25226 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25228 * int-vector-builder.h: New file.
25229 * vec-perm-indices.h: Include int-vector-builder.h.
25230 (vec_perm_indices): Redefine as an int_vector_builder.
25231 (auto_vec_perm_indices): Delete.
25232 (vec_perm_builder): Redefine as a stand-alone class.
25233 (vec_perm_indices::vec_perm_indices): New function.
25234 (vec_perm_indices::clamp): Likewise.
25235 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
25236 (vec_perm_indices::new_vector): New function.
25237 (vec_perm_indices::new_expanded_vector): Update for new
25238 vec_perm_indices class.
25239 (vec_perm_indices::rotate_inputs): New function.
25240 (vec_perm_indices::all_in_range_p): Operate directly on the
25241 encoded form, without computing elided elements.
25242 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
25243 encoding. Update for new vec_perm_indices class.
25244 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
25245 the given vec_perm_builder.
25246 (expand_vec_perm_var): Update vec_perm_builder constructor.
25247 (expand_mult_highpart): Use vec_perm_builder instead of
25248 auto_vec_perm_indices.
25249 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
25250 vec_perm_indices instead of auto_vec_perm_indices. Use a single
25251 or double series encoding as appropriate.
25252 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
25253 vec_perm_indices instead of auto_vec_perm_indices.
25254 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
25255 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
25256 (vect_permute_store_chain): Likewise.
25257 (vect_grouped_load_supported): Likewise.
25258 (vect_permute_load_chain): Likewise.
25259 (vect_shift_permute_load_chain): Likewise.
25260 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
25261 (vect_transform_slp_perm_load): Likewise.
25262 (vect_schedule_slp_instance): Likewise.
25263 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25264 (vectorizable_mask_load_store): Likewise.
25265 (vectorizable_bswap): Likewise.
25266 (vectorizable_store): Likewise.
25267 (vectorizable_load): Likewise.
25268 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
25269 vec_perm_indices instead of auto_vec_perm_indices. Use
25270 tree_to_vec_perm_builder to read the vector from a tree.
25271 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
25272 vec_perm_builder instead of a vec_perm_indices.
25273 (have_whole_vector_shift): Use vec_perm_builder and
25274 vec_perm_indices instead of auto_vec_perm_indices. Leave the
25275 truncation to calc_vec_perm_mask_for_shift.
25276 (vect_create_epilog_for_reduction): Likewise.
25277 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
25278 from auto_vec_perm_indices to vec_perm_indices.
25279 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
25280 instead of changing individual elements.
25281 (aarch64_vectorize_vec_perm_const): Use new_vector to install
25282 the vector in d.perm.
25283 * config/arm/arm.c (expand_vec_perm_d::perm): Change
25284 from auto_vec_perm_indices to vec_perm_indices.
25285 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
25286 instead of changing individual elements.
25287 (arm_vectorize_vec_perm_const): Use new_vector to install
25288 the vector in d.perm.
25289 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
25290 Update vec_perm_builder constructor.
25291 (rs6000_expand_interleave): Likewise.
25292 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
25293 (rs6000_expand_interleave): Likewise.
25295 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25297 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
25298 to qimode could truncate the indices.
25299 * optabs.c (expand_vec_perm_var): Likewise.
25301 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25303 * Makefile.in (OBJS): Add vec-perm-indices.o.
25304 * vec-perm-indices.h: New file.
25305 * vec-perm-indices.c: Likewise.
25306 * target.h (vec_perm_indices): Replace with a forward class
25308 (auto_vec_perm_indices): Move to vec-perm-indices.h.
25309 * optabs.h: Include vec-perm-indices.h.
25310 (expand_vec_perm): Delete.
25311 (selector_fits_mode_p, expand_vec_perm_var): Declare.
25312 (expand_vec_perm_const): Declare.
25313 * target.def (vec_perm_const_ok): Replace with...
25314 (vec_perm_const): ...this new hook.
25315 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
25316 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
25317 * doc/tm.texi: Regenerate.
25318 * optabs.def (vec_perm_const): Delete.
25319 * doc/md.texi (vec_perm_const): Likewise.
25320 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
25321 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
25322 expand_vec_perm for constant permutation vectors. Assert that
25323 the mode of variable permutation vectors is the integer equivalent
25324 of the mode that is being permuted.
25325 * optabs-query.h (selector_fits_mode_p): Declare.
25326 * optabs-query.c: Include vec-perm-indices.h.
25327 (selector_fits_mode_p): New function.
25328 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
25329 is defined, instead of checking whether the vec_perm_const_optab
25330 exists. Use targetm.vectorize.vec_perm_const instead of
25331 targetm.vectorize.vec_perm_const_ok. Check whether the indices
25332 fit in the vector mode before using a variable permute.
25333 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
25334 vec_perm_indices instead of an rtx.
25335 (expand_vec_perm): Replace with...
25336 (expand_vec_perm_const): ...this new function. Take the selector
25337 as a vec_perm_indices rather than an rtx. Also take the mode of
25338 the selector. Update call to shift_amt_for_vec_perm_mask.
25339 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
25340 Use vec_perm_indices::new_expanded_vector to expand the original
25341 selector into bytes. Check whether the indices fit in the vector
25342 mode before using a variable permute.
25343 (expand_vec_perm_var): Make global.
25344 (expand_mult_highpart): Use expand_vec_perm_const.
25345 * fold-const.c: Includes vec-perm-indices.h.
25346 * tree-ssa-forwprop.c: Likewise.
25347 * tree-vect-data-refs.c: Likewise.
25348 * tree-vect-generic.c: Likewise.
25349 * tree-vect-loop.c: Likewise.
25350 * tree-vect-slp.c: Likewise.
25351 * tree-vect-stmts.c: Likewise.
25352 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
25354 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
25355 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
25356 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
25357 (aarch64_vectorize_vec_perm_const): ...this new function.
25358 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25359 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25360 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
25361 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
25362 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25363 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25364 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
25366 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
25367 check for NEON modes.
25368 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
25369 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
25370 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
25371 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
25373 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
25374 the old VEC_PERM_CONST conditions.
25375 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
25376 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
25377 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
25378 (ia64_vectorize_vec_perm_const_ok): Merge into...
25379 (ia64_vectorize_vec_perm_const): ...this new function.
25380 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
25381 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
25382 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
25383 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
25384 * config/mips/mips.c (mips_expand_vec_perm_const)
25385 (mips_vectorize_vec_perm_const_ok): Merge into...
25386 (mips_vectorize_vec_perm_const): ...this new function.
25387 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
25388 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
25389 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
25390 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
25391 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
25392 (rs6000_expand_vec_perm_const): Delete.
25393 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
25395 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25396 (altivec_expand_vec_perm_const_le): Take each operand individually.
25397 Operate on constant selectors rather than rtxes.
25398 (altivec_expand_vec_perm_const): Likewise. Update call to
25399 altivec_expand_vec_perm_const_le.
25400 (rs6000_expand_vec_perm_const): Delete.
25401 (rs6000_vectorize_vec_perm_const_ok): Delete.
25402 (rs6000_vectorize_vec_perm_const): New function.
25403 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
25404 an element count and rtx array.
25405 (rs6000_expand_extract_even): Update call accordingly.
25406 (rs6000_expand_interleave): Likewise.
25407 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
25408 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
25409 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
25410 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
25411 (rs6000_expand_vec_perm_const): Delete.
25412 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25413 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25414 (altivec_expand_vec_perm_const_le): Take each operand individually.
25415 Operate on constant selectors rather than rtxes.
25416 (altivec_expand_vec_perm_const): Likewise. Update call to
25417 altivec_expand_vec_perm_const_le.
25418 (rs6000_expand_vec_perm_const): Delete.
25419 (rs6000_vectorize_vec_perm_const_ok): Delete.
25420 (rs6000_vectorize_vec_perm_const): New function. Remove stray
25421 reference to the SPE evmerge intructions.
25422 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
25423 an element count and rtx array.
25424 (rs6000_expand_extract_even): Update call accordingly.
25425 (rs6000_expand_interleave): Likewise.
25426 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
25427 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
25429 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25431 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25433 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
25434 vector mode and that that mode matches the mode of the data
25436 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
25437 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
25438 directly using expand_vec_perm_1 when forcing selectors into
25440 (expand_vec_perm_var): New function, split out from expand_vec_perm.
25442 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25444 * optabs-query.h (can_vec_perm_p): Delete.
25445 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
25446 * optabs-query.c (can_vec_perm_p): Split into...
25447 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
25448 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
25449 particular selector is valid.
25450 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
25451 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
25452 (vect_grouped_load_supported): Likewise.
25453 (vect_shift_permute_load_chain): Likewise.
25454 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
25455 (vect_transform_slp_perm_load): Likewise.
25456 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25457 (vectorizable_bswap): Likewise.
25458 (vect_gen_perm_mask_checked): Likewise.
25459 * fold-const.c (fold_ternary_loc): Likewise. Don't take
25460 implementations of variable permutation vectors into account
25461 when deciding which selector to use.
25462 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
25463 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
25464 with a false third argument.
25465 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
25466 to test whether the constant selector is valid and can_vec_perm_var_p
25467 to test whether a variable selector is valid.
25469 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25471 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
25472 * optabs-query.c (can_vec_perm_p): Likewise.
25473 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
25474 instead of vec_perm_indices.
25475 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
25476 (vect_gen_perm_mask_checked): Likewise,
25477 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
25478 (vect_gen_perm_mask_checked): Likewise,
25480 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25482 * optabs-query.h (qimode_for_vec_perm): Declare.
25483 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
25484 (qimode_for_vec_perm): ...this new function.
25485 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
25487 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25489 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
25490 does not have a conditional at the top.
25492 2018-01-02 Richard Biener <rguenther@suse.de>
25494 * ipa-inline.c (big_speedup_p): Fix expression.
25496 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
25499 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
25502 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
25506 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
25507 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
25508 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
25509 cond_taken_branch_cost 3->4.
25511 2018-01-01 Jakub Jelinek <jakub@redhat.com>
25513 PR tree-optimization/83581
25514 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
25515 TODO_cleanup_cfg if any changes have been made.
25517 PR middle-end/83608
25518 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
25519 convert_modes if target mode has the right side, but different mode
25522 PR middle-end/83609
25523 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
25524 last argument when extracting from CONCAT. If either from_real or
25525 from_imag is NULL, use expansion through memory. If result is not
25526 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
25527 the parts directly to inner mode, if even that fails, use expansion
25530 PR middle-end/83623
25531 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
25532 check for bswap in mode rather than HImode and use that in expand_unop
25535 Copyright (C) 2018 Free Software Foundation, Inc.
25537 Copying and distribution of this file, with or without modification,
25538 are permitted in any medium without royalty provided the copyright
25539 notice and this notice are preserved.