1 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
3 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
4 (get_constant_size): Don't make STRING_CSTs larger than they are.
5 (check_string_literal): New check function for STRING_CSTs.
6 (output_constant): Use it.
8 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
11 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
13 2018-09-13 Richard Biener <rguenther@suse.de>
15 PR tree-optimization/87263
16 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
17 (struct unwind_state): Add max_rpo field.
18 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
19 Compute max_rpo, the max RPO number a block can be backwards reached
20 from. Re-write non-iterating mode to a RPO ordered worklist approach,
21 separating it from the iterating mode.
23 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
25 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
26 (rfs_decision): New scheduling decision.
28 2018-09-13 Richard Biener <rguenther@suse.de>
31 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
32 (vn_nary_op_insert_pieces_predicated): Do not write useless
33 valid_dominated_by_p entry outside of the allocated storage.
35 2018-09-13 Omar Sandoval <osandov@osandov.com>
36 Tom de Vries <tdevries@suse.de>
39 * dwarf2out.c (is_c): New function.
40 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
42 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
45 * config/aarch64/aarch64.md (*aarch64_bfxil):
47 * config/aarch64/constraints.md (Ulc): Define.
48 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
50 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
53 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
55 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
56 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
57 aarch64_layout_frame call.
58 (aarch64_expand_epilogue): Likewise.
59 (aarch64_initial_elimination_offset): Likewise.
60 (aarch64_get_separate_components): Likewise.
61 (aarch64_use_return_insn_p): Likewise.
62 (aarch64_layout_frame): Remove unneeded check.
64 2018-09-13 Jakub Jelinek <jakub@redhat.com>
66 * configure.ac: Only append
67 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
68 gcc_config_arguments if it was never reconfigured or last reconfigure
69 was with different arguments.
70 * configure: Regenerated.
72 2018-09-13 Jakub Jelinek <jakub@redhat.com>
73 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
76 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
77 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
79 2018-09-13 Jakub Jelinek <jakub@redhat.com>
81 PR tree-optimization/87287
82 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
83 X % C == 0 to X % (unsigned) C == 0 optimization to ...
84 * match.pd (X % C == 0): ... here. New optimization.
86 2018-09-12 Jakub Jelinek <jakub@redhat.com>
89 * expr.h (maybe_optimize_mod_cmp): Declare.
90 * expr.c (mod_inv): New function.
91 (maybe_optimize_mod_cmp): New function.
92 (do_store_flag): Use it.
93 * cfgexpand.c (expand_gimple_cond): Likewise.
95 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
96 Julian Brown <julian@codesourcery.com>
99 * gimplify.c (gimplify_scan_omp_clauses): Set
100 target_firstprivatize_array_bases in OpenACC parallel and kernels
101 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
102 OpenACC data regions.
104 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
106 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
107 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
109 2018-09-12 Richard Biener <rguenther@suse.de>
111 PR tree-optimization/87280
112 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
113 edge but unreachable target.
114 (do_rpo_vn): For conservatively handling a PHI only mark
115 the backedge executable but not the block reachable.
117 2018-09-12 Richard Biener <rguenther@suse.de>
119 PR tree-optimization/87266
120 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
123 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
125 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
127 ("trunc<BFP:mode><DFP_ALL:mode>2")
128 ("trunc<DFP_ALL:mode><BFP:mode>2")
129 ("extend<BFP:mode><DFP_ALL:mode>2")
130 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
131 according to the target operand type.
133 2018-09-12 Jakub Jelinek <jakub@redhat.com>
134 Andreas Krebbel <krebbel@linux.ibm.com>
136 PR tree-optimization/86844
137 * gimple-ssa-store-merging.c
138 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
139 there are any overlapping stores in between them, make sure they are
140 also coalesced or we give up completely.
142 2018-09-12 Jakub Jelinek <jakub@redhat.com>
145 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
146 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
148 2018-09-12 Tom de Vries <tdevries@suse.de>
150 * common.opt (gdescribe-dies): Add option.
151 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
152 attribute for artifical and nameless decls.
153 (dwarf2out_register_external_die): Add description attribute to
154 external reference die.
155 (add_desc_attribute): New functions.
156 (gen_subprogram_die): Add description attribute to
157 DW_TAG_call_site_parameter.
158 * tree-pretty-print.c (print_generic_expr_to_str): New function.
159 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
160 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
162 (@item -gdescribe-dies): Add.
164 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
166 * tree-vrp.c (vrp_shift_undefined_p): Remove.
167 (extract_range_from_binary_expr_1: Call
168 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
169 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
172 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
174 * gimple-ssa-warn-alloca.c
175 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
176 field for ALLOCA_BOUND_*_LARGE.
178 2018-09-11 Nathan Sidwell <nathan@acm.org>
180 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
182 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
184 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
185 for clobbers. Remove obsolete comment.
187 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
189 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
190 mpxchk, mpxld and mpxst types.
191 (define_attr length_immediate): Remove all processing of mpx types.
192 (define_attr prefix_0f): Ditto.
193 (define_attr memory): Ditto.
195 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
197 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
198 (log<mode>2): Change operand 1 predicate to general_operand.
199 Extend operand 1 to XFmode and generate logxf3 insn.
200 (log10<mode>2): Change operand 1 predicate to general_operand.
201 Extend operand 1 to XFmode and generate log10xf3 insn.
202 (log2<mode>2): Change operand 1 predicate to general_operand.
203 Extend operand 1 to XFmode and generate log2xf3 insn.
204 (fyl2xp1_extend<mode>xf3_i387): Remove.
205 (log1p<mode>2): Change operand 1 predicate to general_operand.
206 Extend operand 1 to XFmode and generate log1pxf3 insn.
207 (fxtract_extend<mode>xf3_i387): Remove.
208 (logb<mode>2): Change operand 1 predicate to general_operand.
209 Extend operand 1 to XFmode and generate logbxf3 insn.
210 (ilogb<mode>2): Change operand 1 predicate to general_operand.
211 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
212 (significand<mode>2): Change operand 1 predicate to general_operand.
213 Extend operand 1 to XFmode and generate significandxf3 insn.
215 2018-09-11 Nathan Sidwell <nathan@acm.org>
217 * gcc.c (perror_with_name, pfatal_with_name): Delete.
218 (load_specs): Use fatal_error.
219 (DELETE_IF_ORDINARY, process_command): Use error.
220 (execute, run_attempt): Use fatal_error.
222 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
224 * diagnostic-core.h (sorry_at): New prototype.
225 * diagnostic.c (sorry_at): New function.
227 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
229 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
230 by zero as VR_UNDEFINED.
232 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
234 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
235 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
236 (<sincos>mode2): New expander.
237 (sincos_extend<mode>xf3_i387): Remove insn pattern.
238 (sincos -> sin, cos splitters): Remove splitter patterns.
239 (sincos<mode>3): Change operand 2 predicate to general_operand.
240 Extend operand 2 to XFmode and generate sincosxf3 insn.
241 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
242 Change operand 3 predicate to const1_operand.
243 (fptan_extend<mode>xf4_i387): Remove insn pattern.
244 (tanxf2): Update operands in the call to fptanxf4_i387.
245 (tan<mode>2): Change operand 1 predicate to general_operand.
246 Extend operand 1 to XFmode and generate tanxf3 insn.
247 (atan2xf3): Rename from *fpatanxf3_i387.
248 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
249 (atan2xf3): Remove expander.
250 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
251 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
252 (atan<mode>2): Change operand 1 predicate to general_operand.
253 Extend operand 1 to XFmode and generate atanxf3 insn.
255 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
257 * config/i386/i386.md (x87/SSE constant load splitter): Use
258 memory_operand instead of nonimmediate_operand for input operand
261 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
263 * config/i386/i386.md (float partial SSE register stall splitter): Move
264 splitter near its instruction pattern.
265 (float_extend partial SSE register stall splitter): Ditto.
266 (float_truncate partial SSE register stall splitter): Ditto.
268 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
271 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
272 to speculation_safe_value_not_needed.
275 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
276 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
277 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
278 leaf_function_p, instead use has_hard_reg_initial_val.
280 2018-09-09 Nathan Sidwell <nathan@acm.org>
282 * gcc.h (pfatal_with_name): Don't declare here.
283 * gcc.c (pfatal_with_name): Make static.
285 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
287 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
290 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
292 PR rtl-optimization/85458
293 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
294 priority hook to reduce the priority of EXPR.
296 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
298 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
299 DImode for x87 on 32bit targets. Conditionally disable x87 modes
300 with X87_ENABLE_FLOAT. Remove preparation code.
301 (*float<SWI48:mode><MODEF:mode>2): Rename from
302 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
303 math using "enabled" attribute.
304 (*floatdi<MODEF:mode>2_i387): Rename from
305 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
306 enable for 32bit targets only.
307 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
309 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
310 as operand 1 predicate. Rewrite as define_insn_and_split.
311 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
313 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
315 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
316 to fallthru to FLOAT case.
318 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
321 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
322 around folding of vec_sl to handle out of range shift values.
324 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
326 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
327 Update callers to gen_fix_trunc<mode>_i387_fisttp
328 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
329 nonimmediate_operand.
330 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
331 and corresponding splitters.
332 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
333 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
334 (fix_truncdi_i387_with_temp): Remove insn pattern
335 and corresponding splitters.
336 (fix_trunc<mode>_i387): Change operand 0 predicate to
337 nonimmediate_operand.
338 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
339 and corresponding splitters.
340 (*fistdi2_1): Remove.
342 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
343 (lrintxfdi2): Remove expander. Reimplement as define_insn.
344 (*fist<mode>2_1): Remove.
345 (fist<mode>2): Ditto.
346 (fist<mode>2_with_temp): Remove insn pattern and corresponding
348 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
349 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
350 (fistdi2_<rounding>): Change operand 0 predicate to
351 nonimmediate_operand.
352 (fistdi2_<rounding>_with_temp): Remove insn pattern
353 and corresponding splitters.
354 (fist<mode>2_<rounding>): Change operand 0 predicate to
355 nonimmediate_operand.
356 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
357 and corresponding splitters.
359 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
361 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
363 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
366 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
368 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
369 early gimple folding of vec_splat().
370 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
371 * gimple-fold.h: Add an extern define for tree_vec_extract().
373 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
375 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
376 wrappers around TREE_TYPE comparisons.
378 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
381 * config/s390/predicates.md: Add nonsym_memory_operand.
382 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
383 contains a SYMBOL_REF, load it into an intermediate pseudo.
384 (s390_emit_compare_and_swap): Legitimize operand.
385 * config/s390/s390.md: Use the new nonsym_memory_operand
386 with UNSPECV_CAS patterns.
388 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
391 * config/s390/s390-passes.def: New file.
392 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
394 (make_pass_s390_early_mach): Add declaration.
395 * config/s390/s390.c (make_pass_s390_early_mach):
396 (s390_option_override): Remove dynamic registration.
397 * config/s390/t-s390: Add s390-passes.def.
399 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
401 * config/s390/s390.c (s390_decompose_constant_pool_ref):
402 Remove UNSPEC_LTREL_BASE check.
403 (annotate_constant_pool_refs): Likewise.
404 (find_constant_pool_ref): Likewise.
405 (find_ltrel_base): Removed.
406 (replace_ltrel_base): Removed.
407 (s390_mainpool_finish): Remove replace_ltrel_base call.
408 (s390_chunkify_start): Remove pending LTREL_BASE logic.
409 (s390_chunkify_finish): Remove replace_ltrel_base call.
410 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
412 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
415 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
416 to speculation_safe_value_not_needed.
418 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
419 Bernd Schmidt <bernds_cb1@t-online.de>
421 * config/nvptx/nvptx-opts.h: New file.
422 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
423 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
425 (TARGET_SM35): New macro.
426 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
428 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
431 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
433 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
435 * config/i386/i386.md (truncdfsf2): Remove expander.
436 (truncdfsf2_with_temp): Ditto.
437 (truncxf<mode>2): Ditto.
438 (*truncdfsf_fast_mixed): Remove insn pattern.
439 (*truncdfsf_fast_i387): Ditto.
440 (*truncdfsf_mixed): Ditto.
441 (*truncdfsf_i387): Ditto.
442 (*truncdfsf2_i387_1): Ditto.
443 (*truncxfsf2_mixed): Ditto.
444 (*truncxfdf2_mixed): Ditto.
445 (*truncxf<mode>2_i387_noop): Ditto. Update callers
446 to call gen_truncxf<mode>2 instead.
447 (*truncxf<mode>2_i387): Remove.
448 (reg->reg splitters): Remove splitter pattern.
449 (reg->mem splitters): Ditto.
451 (truncdfsf2): New insn pattern.
452 (truncxf<mode>2): Ditto.
454 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
456 * tree-ssa-math-opts.c (is_mult_by): New function.
457 (is_square_of): Use the above.
458 (optimize_recip_sqrt): New function.
459 (pass_cse_reciprocals::execute): Use the above.
461 2018-09-05 Richard Biener <rguenther@suse.de>
464 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
465 to zero-init the emplaced vec.
467 2018-09-05 Martin Liska <mliska@suse.cz>
469 PR tree-optimization/87205
470 * tree-switch-conversion.c (pass_lower_switch::execute):
471 Group cases for switch statements.
473 2018-09-05 Richard Biener <rguenther@suse.de>
475 PR tree-optimization/87217
476 * tree-ssa-sccvn.c (vuse_valueize): New.
477 (vn_reference_lookup_pieces): Use it.
478 (vn_reference_lookup): Likewise.
480 2018-09-05 Nathan Sidwell <nathan@acm.org>
483 * stor-layout.c (place_field): Scan forwards to check last
484 bitfield when ms_bitfield_placement is in effect.
486 2018-09-05 Richard Biener <rguenther@suse.de>
489 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
492 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
493 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
495 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
496 * config.gcc (extra_objs): Build it.
497 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
499 * config/aarch64/aarch64-passes.def
500 (pass_tag_collision_avoidance): New pass.
501 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
502 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
503 (aarch64_classify_address): Remove static qualifier.
504 (aarch64_address_info, aarch64_address_type): Move to...
505 * config/aarch64/aarch64-protos.h: ... here.
506 (make_pass_tag_collision_avoidance): New function.
507 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
510 2018-09-05 Martin Liska <mliska@suse.cz>
512 * doc/gcov.texi: Update documentation of humar
514 * gcov.c (format_count): Print one decimal place, it provides
515 more fine number of situations like '1G' vs. '1.4G'.
517 2018-09-05 Martin Liska <mliska@suse.cz>
520 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
521 * optc-gen.awk: Allow 'Var' for Deprecated options in order
522 to generate a MASK value.
524 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
527 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
528 if frame pointer isn't used.
529 (compute_frame_pointer_to_fb_displacement): Likewise.
531 2018-09-04 Jakub Jelinek <jakub@redhat.com>
534 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
535 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
536 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
537 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
538 and OPTION_MASK_ISA_XSAVEC_UNSET.
540 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
542 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
543 XOR operations in NAND case.
545 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
547 * wide-int-range.cc (wide_int_range_convert): New.
548 * wide-int-range.h (wide_int_range_convert): New.
549 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
550 code into wide_int_range_convert.
551 (extract_range_into_wide_ints): Do not munge anti range constants
552 into the entire domain. Just return the range back.
554 2018-09-04 Martin Liska <mliska@suse.cz>
556 * genmatch.c (output_line_directive): Add new argument
558 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
560 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
562 * doc/invoke.texi (Option Summary): Add whitespace.
564 * doc/invoke.texi (Option Summary): Add -Waligned-new.
566 2018-09-04 Richard Biener <rguenther@suse.de>
568 PR tree-optimization/87211
569 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
570 backedge value we're supposed to treat as VARYING also number
571 the PHI to VARYING in case it got a different value-number already.
573 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
575 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
576 (extract_range_from_binary_expr_1): Do not call
577 vrp_can_optimize_bit_op.
578 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
580 (wide_int_range_get_mask_and_bounds): New.
581 (wide_int_range_optimize_bit_op): New.
582 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
583 (wide_int_range_bit_and): Same.
584 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
585 (wide_int_range_optimize_bit_op): New.
586 (wide_int_range_get_mask_and_bounds): New.
588 2018-09-04 Richard Biener <rguenther@suse.de>
590 PR tree-optimization/87176
591 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
592 variable. When value-numbering a virtual PHI node make sure
593 to not value-number to the backedge value.
595 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
597 * doc/extend.texi (Long Long, Hex Floats): Document support for
598 long long and hex floats in more recent versions of ISO C++.
600 2018-09-03 Richard Biener <rguenther@suse.de>
602 PR tree-optimization/87177
603 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
606 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
608 * bb-reorder.c (edge_order): Convert to C-qsort-style
609 tri-state comparator.
610 (reorder_basic_blocks_simple): Change std::stable_sort to gcc_stablesort.
612 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
614 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
615 tri-state comparator.
616 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
618 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
620 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
621 (mergesort): ... here as maximum count for using netsort.
622 (gcc_qsort): Set nlim to 3 if stable sort is requested.
623 (gcc_stablesort): New.
624 * system.h (gcc_stablesort): Declare.
626 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
628 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
629 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
630 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
632 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
634 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
635 lxsdx and stxsdx alternatives.
636 (*mov<mode>_hardfloat64): Ditto.
637 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
639 2018-09-03 Richard Biener <rguenther@suse.de>
641 PR tree-optimization/87200
642 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
645 2018-09-03 Martin Liska <mliska@suse.cz>
647 PR tree-optimization/87201
648 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
649 Fix parenthesis in an expression.
651 2018-09-03 Richard Biener <rguenther@suse.de>
653 PR tree-optimization/87197
654 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
655 visited. CSE the VN_INFO hashtable lookup.
657 PR tree-optimization/87169
658 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
659 iterating make sure there's no extra backedges from irreducible
660 regions feeding the header. Mark the destination block
663 2018-09-03 Martin Liska <mliska@suse.cz>
666 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
667 * common/common-targhooks.c (default_get_valid_option_values):
669 * common/common-targhooks.h (default_get_valid_option_values):
671 * common/config/i386/i386-common.c: Move processor_target_table
673 (ix86_get_valid_option_values): New function.
674 (TARGET_GET_VALID_OPTION_VALUES): New macro.
675 * config/i386/i386.c (struct ptt): Move to i386-common.c.
676 (PTA_*): Move all defined masks into i386-common.c.
677 (ix86_function_specific_restore): Use new processor_cost_table.
678 * config/i386/i386.h (struct ptt): Moved from i386.c.
679 (struct pta): Likewise.
680 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
681 * doc/tm.texi.in: Likewise.
682 * opt-suggestions.c (option_proposer::suggest_option):
683 Pass prefix to build_option_suggestions.
684 (option_proposer::get_completions): Likewise.
685 (option_proposer::build_option_suggestions): Use the new target
687 * opts.c (struct option_help_tuple): New struct.
688 (print_filtered_help): Use the new target hook.
690 2018-09-03 Martin Liska <mliska@suse.cz>
693 * predict.c (set_even_probabilities): Add likely_edges
694 argument and handle cases where we have precisely one
696 (combine_predictions_for_bb): Catch also likely_edges.
697 (tree_predict_by_opcode): Handle gswitch statements.
698 * tree-cfg.h (find_case_label_for_value): New declaration.
699 (find_taken_edge_switch_expr): Likewise.
700 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
701 Find pivot in decision tree based on probabily, not by number of
704 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
706 * doc/standards.texi (Standards): Update Objective-C reference.
708 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
710 * doc/install.texi (Prerequisites): Update link for MPC.
712 2018-09-01 Michael Matz <matz@suse.de>
714 PR tree-optimization/87074
715 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
716 PHIs for outer-loop uses.
718 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
720 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
721 * doc/invoke.texi (C Dialect Options): Ditto.
723 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
725 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
727 2018-08-31 Richard Biener <rguenther@suse.de>
729 PR tree-optimization/87168
730 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
731 (rpo_elim::eliminate_avail): When OP was not visited it must
734 2018-08-31 David Malcolm <dmalcolm@redhat.com>
736 * tree-vrp.c (copy_value_range): Convert param "from" from
737 "value_range *" to "const value_range *".
738 (range_is_null): Likewise for param "vr".
739 (range_int_cst_p): Likewise.
740 (range_int_cst_singleton_p): Likewise.
741 (symbolic_range_p): Likewise.
742 (value_ranges_intersect_p): Likewise for both params.
743 (value_range_nonnegative_p): Likewise for param "vr".
744 (value_range_constant_singleton): Likewise.
745 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
746 (extract_range_into_wide_ints): Likewise for param "vr".
747 (extract_range_from_multiplicative_op): Likewise for params "vr0"
749 (vrp_can_optimize_bit_op): Likewise.
750 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
752 (extract_range_from_unary_expr): Likewise.
753 (debug_value_range): Likewise for param "vr".
754 (value_range::dump): Add "const" qualifier.
755 (vrp_prop::check_array_ref): Convert local "vr" from
756 "value_range *" to "const value_range *".
757 (vrp_prop::check_mem_ref): Likewise.
758 (vrp_prop::visit_stmt): Likewise for local "old_vr".
759 (vrp_intersect_ranges_1): Likewise for param "vr_1".
760 (vrp_intersect_ranges): Likewise.
761 (simplify_stmt_for_jump_threading): Likewise for local "vr".
762 (vrp_prop::vrp_finalize): Likewise.
763 * tree-vrp.h (value_range::dump): Add "const" qualifier.
764 (vrp_intersect_ranges): Add "const" qualifier to params as above.
765 (extract_range_from_unary_expr): Likewise.
766 (value_range_constant_singleton): Likewise.
767 (symbolic_range_p): Likewise.
768 (copy_value_range): Likewise.
769 (extract_range_from_binary_expr_1): Likewise.
770 (range_int_cst_p): Likewise.
771 (vrp_set_zero_nonzero_bits): Likewise.
772 (range_int_cst_singleton_p): Likewise.
774 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
776 * config/aarch64/arm_neon.h (vabsd_s64): New.
777 (vnegd_s64): Likewise.
779 2018-08-31 Martin Jambor <mjambor@suse.cz>
781 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
783 2018-08-31 Martin Liska <mliska@suse.cz>
785 * ipa-icf.c (sem_item::add_type): Use
786 sem_item::m_type_hash_cache.
787 * ipa-icf.h: Move the cache from sem_item_optimizer
790 2018-08-31 Nathan Sidwell <nathan@acm.org>
792 * doc/extend.texi (Backwards Compatibility): Remove implicit
793 extern C leeway of () being (...).
795 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
797 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
799 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
803 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
805 2018-08-31 Jakub Jelinek <jakub@redhat.com>
808 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
809 gen_int_mode. Formatting fixes.
811 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
813 * target.def (custom_function_descriptors): Improve documentation.
814 * doc/tm.texi.in (Trampolines): Expand discussion of function
815 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
816 beginning of the section.
817 * doc/tm.texi: Regenerated.
819 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
821 * cfg.h (class auto_edge_flag): Spell out the template-id of the
822 base class in the initializer list. This is a workaround for
823 building with older GCC.
824 (class auto_bb_flag): Likewise.
826 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
828 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
829 (altivec_vcmpequ<VI_char>_p): Remove star.
830 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
832 (expand_strncmp_vec_sequence): New function.
833 (emit_final_str_compare_vec): New function.
834 (expand_strn_compare): Add support for vector strncmp.
835 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
836 length specification to bytes.
837 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
838 (vcmpnezb_p): New pattern.
839 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
840 for option -mstring-compare-inline-limit.
842 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
844 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
845 (PTA_SKYLAKE): Add PTA_AES.
846 (PTA_GOLDMONT): Likewise.
848 2018-08-29 Jan Hubicka <jh@suse.cz>
851 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
852 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
854 2018-08-29 Jan Hubicka <jh@suse.cz>
856 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
858 (hash_tree): Do not visit TYPE_STUB_DECL.
859 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
860 stream TYPE_STUB_DECL.
861 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
862 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
863 after free_lang_data.
864 (type_in_anonymous_namespace_p): Likewise.
866 2018-08-29 Jan Hubicka <jh@suse.cz>
868 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
869 comment that it has to be even number.
870 (class sreal): Change m_sig type to int32_t.
871 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
872 int64_t for temporary calculations.
873 (sreal_verify_basics): Drop one bit from minimum and maximum.
875 2018-08-30 Richard Biener <rguenther@suse.de>
877 PR tree-optimization/87147
878 * tree-ssa-sccvn.c (SSA_VISITED): New function.
879 (visit_phi): When the degenerate result is from the backedge and
880 we didn't visit its definition yet drop to VARYING.
881 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
883 2018-08-29 Jan Hubicka <jh@suse.cz>
885 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
887 (hash_tree): Likewise.
889 2018-08-29 Jan Hubicka <jh@suse.cz>
891 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
892 and TYPE_NEXT_REF_TO.
894 2018-08-29 Jan Hubicka <jh@suse.cz>
896 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
897 comment that it has to be even number.
898 (class sreal): Change m_sig type to int32_t.
899 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
900 int64_t for temporary calculations.
901 (sreal_verify_basics): Drop one bit from minimum and maximum.
903 2018-08-30 Tamar Christina <tamar.christina@arm.com>
905 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
907 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
910 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
911 if to_add is negative.
913 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
916 * builtins.c (c_strlen): Improve range checks.
918 2018-08-29 Martin Sebor <msebor@redhat.com>
919 Jeff Law <law@redhat.com>
921 PR tree-optimization/86714
922 PR tree-optimization/86711
923 * builtins.c (c_strlen): Add arguments to call to string_constant.
924 * expr.c (string_constant): Add argument. Detect missing nul
925 terminator and outermost declaration it's missing in.
926 * expr.h (string_constant): Add argument.
927 * fold-const.c (read_from_constant_string): Add arguments to call to
929 (c_getstr): Likewise.
930 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
932 * tree-ssa-strlen.c (get_stridx): Likewise.
934 2018-08-29 Jan Hubicka <jh@suse.cz>
936 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
937 Do not stream DECL_VINDEX.
938 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
939 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
940 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
942 2018-08-29 Richard Biener <rguenther@suse.de>
944 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
945 virtual operands that are not default defs to honor region
947 (rpo_vn_valueize): Remove ineffective code here.
949 2018-08-29 Richard Biener <rguenther@suse.de>
951 PR tree-optimization/87132
952 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
953 when skipping defs reachable over backedges.
955 2018-08-29 Richard Biener <rguenther@suse.de>
957 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
958 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
959 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
960 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
961 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
962 (vn_reference_lookup_3): Remove use of const_parms.
963 (free_rpo_vn): Do not free const_parms.
964 (do_rpo_vn): Do not call init_const_parms.
965 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
966 SSA_NAME_POINTS_TO_READONLY_MEMORY.
967 (call_may_clobber_ref_p_1): Likewise.
969 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
972 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
974 (-ftree-scev-cprop): Document.
976 2018-08-29 Jan Hubicka <jh@suse.cz>
978 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
980 (sreal constructor): Update.
981 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
982 sreal:operator/): Update.
984 2018-08-29 Martin Liska <mliska@suse.cz>
986 * tree-switch-conversion.c (switch_conversion::expand):
987 Strenghten assumption about gswitch statements.
989 2018-08-29 Richard Biener <rguenther@suse.de>
991 PR tree-optimization/87117
992 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
993 re-value-number released SSA VDEFs.
995 2018-08-29 Richard Biener <rguenther@suse.de>
997 PR tree-optimization/87126
998 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
1000 2018-08-28 Jim Wilson <jimw@sifive.com>
1002 * config/riscv/pic.md: Rewrite.
1003 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
1005 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
1006 (SOFTF, default_load, softload, softstore): New.
1008 2018-08-28 Jeff Law <law@redhat.com>
1010 * fold-const.c (fold_binary_loc): Remove recently added assert.
1012 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
1014 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
1015 to OP parmeter of generated function.
1017 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
1019 PR tree-optimization/87009
1020 * match.pd: Add boolean optimizations.
1022 2018-08-28 Martin Sebor <msebor@redhat.com>
1025 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
1026 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
1027 (pass_walloca::gate): Use it.
1028 (alloca_call_type): Same.
1029 (pass_walloca::execute): Same.
1030 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
1032 2018-08-28 David Malcolm <dmalcolm@redhat.com>
1034 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
1035 GCC_VERSION for usage of "__gcc_dump_printf__" format from
1038 2018-08-28 Richard Biener <rguenther@suse.de>
1040 PR tree-optimization/87124
1041 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
1042 constants before looking up avail.
1044 2018-08-28 Jakub Jelinek <jakub@redhat.com>
1047 * calls.c (maybe_warn_nonstring_arg): Punt early if
1048 warn_stringop_overflow is zero. Don't call get_range_strlen
1049 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
1050 Swap comparison operands to have constants on rhs. Only use
1051 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
1052 increment lenrng[0].
1054 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
1056 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
1057 use of tree_to_shwi. Remove duplicated test for the size being
1058 a whole number of bytes.
1060 2018-08-28 Richard Biener <rguenther@suse.de>
1062 PR tree-optimization/87117
1063 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
1064 Handle removed stmt without LHS (GIMPLE_NOP).
1066 2018-08-28 Richard Biener <rguenther@suse.de>
1068 PR tree-optimization/87117
1069 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
1070 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
1072 2018-08-28 Richard Biener <rguenther@suse.de>
1074 PR tree-optimization/87117
1075 * tree-ssa-pre.c (compute_avail): Do not make expressions
1076 with predicated values available.
1077 (get_expr_value_id): Assert we do not run into predicated value
1080 2018-08-28 Richard Biener <rguenther@suse.de>
1082 PR tree-optimization/87117
1083 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
1084 get virtual operands.
1085 (get_expr_operands): Handle STRING_CST like other decls.
1087 2018-08-28 Martin Liska <mliska@suse.cz>
1089 * tree.h: Update documentation of fndecl_built_in_p
1093 2018-08-27 Jeff Law <law@redhat.com>
1094 PR tree-optimization/87110
1095 * tree-ssa-dse.c (compute_trims): Handle non-constant
1098 2018-08-27 Martin Sebor <msebor@redhat.com>
1100 PR tree-optimization/86914
1101 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
1103 2018-08-27 Martin Sebor <msebor@redhat.com>
1105 PR tree-optimization/87112
1106 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
1107 the type of the bound argument.
1109 2018-08-27 Jeff Law <law@redhat.com>
1111 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
1112 type does not have a TYPE_SIZE_UNIT.
1114 2018-08-27 Steve Ellcey <sellcey@cavium.com>
1116 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
1117 with include of backend.h.
1119 2018-08-27 Richard Biener <rguenther@suse.de>
1121 PR tree-optimization/86927
1122 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
1123 use const cond reduction code.
1125 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
1127 PR tree-optimization/85758
1128 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
1130 2018-08-27 David Malcolm <dmalcolm@redhat.com>
1133 * diagnostic-show-locus.c (class layout_range): Update for
1134 conversion of show_caret_p to a tri-state.
1135 (layout_range::layout_range): Likewise.
1136 (make_range): Likewise.
1137 (layout::maybe_add_location_range): Likewise.
1138 (layout::should_print_annotation_line_p): Don't show annotation
1139 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
1140 (layout::get_state_at_point): Update for conversion of
1141 show_caret_p to a tri-state. Bail out early for
1142 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
1143 underlining or source colorization.
1144 (gcc_rich_location::add_location_if_nearby): Update for conversion
1145 of show_caret_p to a tri-state.
1146 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
1147 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
1149 (selftest::test_one_liner_labels): Likewise.
1150 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
1151 conversion of show_caret_p to a tri-state.
1152 * pretty-print.c (text_info::set_location): Likewise.
1153 * pretty-print.h (text_info::set_location): Likewise.
1154 * substring-locations.c (format_warning_n_va): Likewise.
1155 * tree-diagnostic.c (default_tree_printer): Likewise.
1156 * tree-pretty-print.c (newline_and_indent): Likewise.
1158 2018-08-27 David Malcolm <dmalcolm@redhat.com>
1161 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
1162 line above for line-insertion fix-it hints.
1163 (selftest::test_fixit_insert_containing_newline): Update the
1164 expected results, and add a test with line-numbering enabled.
1166 2018-08-27 Martin Liska <mliska@suse.cz>
1169 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
1170 params with DECL_HAS_VALUE_EXPR_P.
1172 2018-08-27 Martin Liska <mliska@suse.cz>
1174 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
1175 selected expansion strategy.
1177 2018-08-27 Martin Liska <mliska@suse.cz>
1179 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
1180 * builtins.c (is_builtin_fn): Likewise.
1181 * attribs.c (diag_attr_exclusions): Use new function
1182 fndecl_built_in_p and remove check for FUNCTION_DECL if
1184 (builtin_mathfn_code): Likewise.
1185 (fold_builtin_expect): Likewise.
1186 (fold_call_expr): Likewise.
1187 (fold_builtin_call_array): Likewise.
1188 (fold_call_stmt): Likewise.
1189 (set_builtin_user_assembler_name): Likewise.
1190 (is_simple_builtin): Likewise.
1191 * calls.c (gimple_alloca_call_p): Likewise.
1192 (maybe_warn_nonstring_arg): Likewise.
1193 * cfgexpand.c (expand_call_stmt): Likewise.
1194 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
1195 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
1196 (cgraph_node::verify_node): Likewise.
1197 * cgraphclones.c (build_function_decl_skip_args): Likewise.
1198 (cgraph_node::create_clone): Likewise.
1199 * config/arm/arm.c (arm_insert_attributes): Likewise.
1200 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
1201 * dse.c (scan_insn): Likewise.
1202 * expr.c (expand_expr_real_1): Likewise.
1203 * fold-const.c (operand_equal_p): Likewise.
1204 (fold_binary_loc): Likewise.
1205 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
1206 * gimple-low.c (lower_stmt): Likewise.
1207 * gimple-pretty-print.c (dump_gimple_call): Likewise.
1208 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Likewise.
1209 * gimple.c (gimple_build_call_from_tree): Likewise.
1210 (gimple_call_builtin_p): Likewise.
1211 (gimple_call_combined_fn): Likewise.
1212 * gimplify.c (gimplify_call_expr): Likewise.
1213 (gimple_boolify): Likewise.
1214 (gimplify_modify_expr): Likewise.
1215 (gimplify_addr_expr): Likewise.
1216 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
1217 * ipa-cp.c (determine_versionability): Likewise.
1218 * ipa-fnsummary.c (compute_fn_summary): Likewise.
1219 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
1220 * ipa-split.c (visit_bb): Likewise.
1221 (split_function): Likewise.
1222 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
1223 * lto-cgraph.c (input_node): Likewise.
1224 * lto-streamer-out.c (write_symbol): Likewise.
1225 * omp-low.c (setjmp_or_longjmp_p): Likewise.
1226 (lower_omp_1): Likewise.
1227 * predict.c (strip_predict_hints): Likewise.
1228 * print-tree.c (print_node): Likewise.
1229 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
1230 * trans-mem.c (is_tm_irrevocable): Likewise.
1231 (is_tm_load): Likewise.
1232 (is_tm_simple_load): Likewise.
1233 (is_tm_store): Likewise.
1234 (is_tm_simple_store): Likewise.
1235 (is_tm_abort): Likewise.
1236 (tm_region_init_1): Likewise.
1237 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
1238 * tree-cfg.c (verify_gimple_call): Likewise.
1239 (move_stmt_r): Likewise.
1240 (stmt_can_terminate_bb_p): Likewise.
1241 * tree-eh.c (lower_eh_constructs_2): Likewise.
1242 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
1243 * tree-inline.c (remap_gimple_stmt): Likewise.
1244 (copy_bb): Likewise.
1245 (estimate_num_insns): Likewise.
1246 (fold_marked_statements): Likewise.
1247 * tree-sra.c (scan_function): Likewise.
1248 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
1249 (optimize_stack_restore): Likewise.
1250 (pass_fold_builtins::execute): Likewise.
1251 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
1252 (mark_all_reaching_defs_necessary_1): Likewise.
1253 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
1254 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
1255 (pass_forwprop::execute): Likewise.
1256 * tree-ssa-loop-im.c (stmt_cost): Likewise.
1257 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1258 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
1259 * tree-ssa-strlen.c (get_string_length): Likewise.
1260 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
1261 (find_func_aliases_for_call): Likewise.
1262 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
1263 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
1264 * tree-tailcall.c (find_tail_calls): Likewise.
1265 * tree.c (need_assembler_name_p): Likewise.
1266 (free_lang_data_in_decl): Likewise.
1267 (get_call_combined_fn): Likewise.
1268 * ubsan.c (is_ubsan_builtin_p): Likewise.
1269 * varasm.c (incorporeal_function_p): Likewise.
1270 * tree.h (DECL_BUILT_IN): Remove and replace with
1272 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
1273 (fndecl_built_in_p): New.
1275 2018-08-27 Martin Liska <mliska@suse.cz>
1277 PR tree-optimization/86847
1278 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
1279 Dump also subtree probability.
1280 (switch_decision_tree::do_jump_if_equal): New function.
1281 (switch_decision_tree::emit_case_nodes): Handle special
1282 situations in balanced tree that can be emitted much simpler.
1283 Fix calculation of probabilities that happen in tree expansion.
1284 * tree-switch-conversion.h (struct cluster): Add
1286 (struct simple_cluster): Likewise.
1287 (struct case_tree_node): Add new function has_child.
1288 (do_jump_if_equal): New.
1290 2018-08-27 Martin Liska <mliska@suse.cz>
1292 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
1293 Add new argument to bit_test_cluster constructor.
1294 (bit_test_cluster::emit): Set bits really number of values
1296 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
1297 probability argument.
1298 * tree-switch-conversion.h (struct bit_test_cluster):
1299 Add m_handles_entire_switch.
1301 2018-08-27 Martin Liska <mliska@suse.cz>
1303 PR tree-optimization/86702
1304 * tree-switch-conversion.c (jump_table_cluster::emit):
1305 Make probabilities even for values in jump table
1306 according to number of cases handled.
1307 (switch_decision_tree::compute_cases_per_edge): Pass
1308 argument to reset_out_edges_aux function.
1309 (switch_decision_tree::analyze_switch_statement): Likewise.
1310 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
1313 2018-08-27 Martin Liska <mliska@suse.cz>
1315 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
1316 cfun argument explicitly.
1317 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
1318 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
1319 function gimple_switch_default_bb.
1320 (convert_switch_statements):
1322 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
1323 * stmt.c (label_to_block_fn): Use label_to_block and pass
1324 cfun argument explicitly and use gimple_switch_label_bb.
1325 (expand_case): Likewise.
1326 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
1327 cfun argument explicitly. Likewise.
1328 (make_edges_bb): Likewise.
1329 (make_cond_expr_edges): Likewise.
1330 (get_cases_for_edge): Likewise.
1331 (make_gimple_switch_edges): Likewise.
1332 (label_to_block_fn): Likewise.
1333 (label_to_block): Likewise.
1334 (make_goto_expr_edges): Likewise.
1335 (make_gimple_asm_edges): Likewise.
1336 (main_block_label): Likewise.
1337 (group_case_labels_stmt): Likewise.
1338 (find_taken_edge_computed_goto): Likewise.
1339 (find_taken_edge_switch_expr): Likewise.
1340 (gimple_verify_flow_info): Likewise.
1341 (gimple_redirect_edge_and_branch): Likewise.
1342 (gimple_switch_label_bb): New function.
1343 (gimple_switch_default_bb): Likewise.
1344 (gimple_switch_edge): Likewise.
1345 (gimple_switch_default_edge): Likewise.
1346 * tree-cfg.h (label_to_block_fn): Remove and replace ...
1347 (label_to_block): ... with this.
1348 (gimple_switch_label_bb): New.
1349 (gimple_switch_default_bb): Likewise.
1350 (gimple_switch_edge): Likewise.
1351 (gimple_switch_default_edge): Likewise.
1352 * tree-cfgcleanup.c (convert_single_case_switch): Use
1353 new gimple functions and pass new argument to label_to_block.
1354 (cleanup_control_flow_bb):
1355 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
1356 cfun argument explicitly.
1357 (make_eh_edges): Likewise.
1358 (redirect_eh_dispatch_edge): Likewise.
1359 (lower_resx): Likewise.
1360 (lower_eh_dispatch): Likewise.
1361 (maybe_remove_unreachable_handlers): Likewise.
1362 (unsplit_eh): Likewise.
1363 (cleanup_empty_eh): Likewise.
1364 (verify_eh_edges): Likewise.
1365 (verify_eh_dispatch_edge): Likewise.
1366 * tree-ssa-dom.c (record_edge_info): Likewise.
1367 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
1368 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
1369 (thread_through_normal_block): Likewise.
1370 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
1371 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
1372 * tree-switch-conversion.c (switch_conversion::collect): Use new
1374 (switch_conversion::check_final_bb): Likewise.
1375 (switch_conversion::gather_default_values): Pass new argument
1377 (switch_conversion::build_constructors): Likewise.
1378 (switch_decision_tree::compute_cases_per_edge): Use new
1379 gimple_switch_edge function.
1380 (switch_decision_tree::analyze_switch_statement): Pass new argument
1382 (switch_decision_tree::try_switch_expansion): Use
1383 gimple_switch_default_edge.
1384 * tree-vrp.c (find_switch_asserts): Pass new argument
1386 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
1387 (vr_values::simplify_switch_using_ranges): Likewise.
1389 2018-08-27 Richard Biener <rguenther@suse.de>
1391 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
1392 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
1394 * tree-ssa-sccvn.h (struct vn_pval): New structure.
1395 (struct vn_nary_op_s): Add unwind_to member. Add
1396 predicated_values flag and put result into a union together
1397 with a linked list of vn_pval.
1398 (struct vn_ssa_aux): Add name member to make maintaining
1399 a map of SSA name to vn_ssa_aux possible. Remove no longer
1400 needed info, dfsnum, low, visited, on_sccstack, use_processed
1401 and range_info_anti_range_p members.
1402 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
1403 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
1405 (vn_valueize): New global.
1406 (vn_context_bb): Likewise.
1407 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
1408 VN_INFO_PTR_INFO): Remove.
1409 * tree-ssa-sccvn.c: ... (rewrite)
1410 (pass_fre::execute): For -O2+ initialize loops and run
1411 RPO VN in optimistic mode (iterating). For -O1 and -Og
1412 run RPO VN in non-optimistic mode.
1413 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
1414 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
1415 * doc/invoke.texi (sccvn-max-scc-size): Remove.
1416 (rpo-vn-max-loop-depth): Document.
1417 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
1418 when valuezing the VUSE signals we walked out of the region.
1419 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
1420 (phi_translate): Set VN context block to use for availability
1422 (compute_avail): Likewise.
1423 (pre_valueize): New function.
1424 (pass_pre::execute): Adjust to the RPO VN API.
1426 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
1427 (propagate_constants_for_unrolling): Remove.
1428 (tree_unroll_loops_completely): Perform value-numbering
1429 on the unrolled bodies loop parent.
1431 2018-08-27 Richard Biener <rguenther@suse.de>
1433 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
1434 for partial antic compute.
1436 2018-08-27 Jakub Jelinek <jakub@redhat.com>
1438 PR rtl-optimization/87065
1439 * combine.c (simplify_if_then_else): Formatting fix.
1440 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
1442 (known_cond): Don't return const_true_rtx for vector modes. Use
1443 CONST0_RTX instead of const0_rtx. Formatting fixes.
1445 2018-08-27 Martin Liska <mliska@suse.cz>
1447 PR gcov-profile/87069
1448 * gcov.c (process_file): Record files already processed
1449 and warn about a file being processed multiple times.
1451 2018-08-27 Martin Liska <mliska@suse.cz>
1454 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1455 Set default values for x_aarch64_*_string strings.
1456 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
1457 prefix. For -mabi do not print '=ABI' in help and use
1458 <option_value> format for -msve-vector-bits and -moverride
1461 2018-08-26 Jeff Law <law@redhat.com>
1463 * config/mips/frame-header-opt.c: Include "backend.h" rather than
1466 2018-08-26 Marek Polacek <polacek@redhat.com>
1468 PR c++/87029, Implement -Wredundant-move.
1469 * doc/invoke.texi: Document -Wredundant-move.
1471 2018-08-25 Martin Sebor <msebor@redhat.com>
1473 PR tree-optimization/87059
1474 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
1475 to the same type as the other.
1476 * fold-const.c (fold_binary_loc): Assert expectation.
1478 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
1480 * config/darwin.c (machopic_legitimize_pic_address): Clean up
1481 extraneous parentheses, dead code section and formatting.
1483 2018-08-24 David Malcolm <dmalcolm@redhat.com>
1486 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
1487 wide enough for jumps in the line-numbering to be visible.
1488 (layout::print_gap_in_line_numbering): New member function.
1489 (layout::calculate_line_spans): When using line numbering, merge
1490 line spans that are only 1 line apart.
1491 (diagnostic_show_locus): When printing line numbers, show gaps in
1492 line numbering directly, rather than printing headers.
1493 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
1494 line-numbering with multiple line spans.
1495 (selftest::test_fixit_insert_containing_newline_2): Add test of
1496 line-numbering, in which the spans are close enough to be merged.
1498 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
1500 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
1501 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
1502 * tree-vrp.c (range_is_nonnull): Remove.
1503 (range_includes_zero_p): Accept value_range instead of min/max.
1504 (extract_range_from_binary_expr_1): Do not early bail on
1506 Use range_includes_zero_p instead of range_is_nonnull.
1507 (extract_range_from_unary_expr): Use range_includes_zero_p instead
1508 of range_is_nonnull.
1509 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
1510 special case VR_ANTI_RANGE.
1511 (vrp_finalize): Same.
1512 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
1514 (range_is_nonnull): Remove.
1515 * vr-values.c (vrp_stmt_computes_nonzero): Use
1516 range_includes_zero_p instead of range_is_nonnull.
1517 (extract_range_basic): Pass value_range to range_includes_zero_p
1518 instead of range_is_nonnull.
1520 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
1522 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
1523 * emit-rtl.h (rtl_data): Remove return_bnd.
1524 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
1525 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
1526 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
1527 (POINTER_BOUNDS_MODE): Remove definition.
1528 (make_pointer_bounds_mode): Remove.
1529 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
1530 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
1531 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
1532 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
1533 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
1534 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
1535 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
1537 * config/i386/i386-modes.def (BND32, BND64): Remove.
1538 * config/i386/i386.c (dbx_register_map): Remove bound registers.
1539 (dbx64_register_map): Ditto.
1540 (svr4_dbx_register_map): Ditto.
1541 (indirect_thunk_bnd_needed): Remove.
1542 (indirect_thunks_bnd_used): Ditto.
1543 (indirect_return_bnd_needed): Ditto.
1544 (indirect_return_via_cx_bnd): Ditto.
1545 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
1546 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
1547 (output_indirect_thunk): Ditto. Remove need_prefix argument.
1548 (output_indirect_thunk_function): Remove handling of
1549 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
1550 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
1551 (ix86_save_reg): Remove handling of crtl->return_bnd.
1552 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
1553 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
1554 and UNSPEC_BNDLX_ADDR.
1555 (ix86_output_indirect_branch_via_reg): Remove handling of
1556 indirect_thunk_prefix_bnd.
1557 (ix86_output_indirect_branch_via_push): Ditto.
1558 (ix86_output_function_return): Ditto.
1559 (ix86_output_indirect_function_return): Ditto.
1560 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
1561 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
1562 (CALL_USED_REGISTERS): Ditto.
1563 (REG_ALLOC_ORDER): Update for removal of bound registers.
1564 (HI_REGISTER_NAMES): Ditto.
1565 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
1566 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
1567 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
1568 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
1569 (FIRST_PSEUDO_REG): Update.
1570 (BND): Remove mode iterator.
1571 * config/i386/predicates.md (bnd_mem_operator): Remove.
1573 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
1575 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
1578 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
1580 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
1581 the case in which the permute needs only a single element and
1582 repeats for every vector of the result. Extend that case to
1583 handle variable-length vectors.
1584 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
1586 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
1589 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
1590 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
1592 2018-08-24 Richard Biener <rguenther@suse.de>
1594 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
1595 bb_flags_allocated members.
1596 (auto_flag): New RAII class for allocating flags.
1597 (auto_edge_flag): New RAII class for allocating edge flags.
1598 (auto_bb_flag): New RAII class for allocating bb flags.
1599 * cfgloop.c (verify_loop_structure): Allocate temporary edge
1601 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
1602 in favor of temporarily allocated BB flag.
1603 * hsa-brig.c: Re-order includes.
1604 * hsa-dump.c: Likewise.
1605 * hsa-regalloc.c: Likewise.
1606 * print-rtl.c: Likewise.
1607 * profile-count.c: Likewise.
1609 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
1612 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
1615 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
1618 * wide-int-range.cc (wide_int_range_div): Do not ignore result
1619 from wide_int_range_multiplicative_op.
1621 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1623 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
1626 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
1628 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
1631 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
1633 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
1634 full GENERIC expression used for replacement.
1636 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
1638 * tree-vrp.c (abs_extent_range): Remove.
1639 (extract_range_into_wide_ints): Pass wide ints by reference.
1640 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
1641 Pass wide ints by reference in all calls to
1642 extract_range_into_wide_ints.
1643 * wide-int-range.cc (wide_int_range_div): New.
1644 * wide-int-range.h (wide_int_range_div): New.
1645 (wide_int_range_includes_zero_p): New.
1646 (wide_int_range_zero_p): New.
1648 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
1650 * config/aarch64/aarch64.md (arches): New enum.
1651 (arch): New enum attr.
1652 (arch_enabled): New attr.
1653 (enabled): Now uses arch_enabled only.
1654 (simd, sve, fp16): Removed attribute.
1655 (fp): Attr now defined in terms of 'arch'.
1656 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
1657 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
1658 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
1659 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
1660 attributes into 'arch'.
1661 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
1662 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
1663 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
1664 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
1665 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
1666 'simd' attribute into 'arch'.
1667 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
1668 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
1669 Convert use of 'fp' attribute to 'arch'.
1670 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
1671 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
1673 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
1674 (different modes) Merge 'fp' and 'simd' into 'arch'.
1675 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
1678 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
1680 PR rtl-optimization/87026
1681 * expmed.c (canonicalize_comparison): If we can no longer create
1682 pseudoregisters, don't.
1684 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
1687 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
1689 * config/arm/arm.c (speculation_barrier_libfunc): New static
1691 (arm_init_libfuncs): Initialize it.
1692 (arm_emit_speculation_barrier): New function.
1693 * config/arm/arm.md (speculation_barrier): Call
1694 arm_emit_speculation_barrier for architectures that do not have
1696 (speculation_barrier_insn): Only match on Armv7 or later.
1698 2018-08-23 Richard Biener <rguenther@suse.de>
1701 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
1704 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
1706 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
1707 of single-vector TBLs.
1708 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
1711 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
1714 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
1715 aarch64_evpc_tbl guard.
1717 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
1719 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
1722 2018-08-22 Martin Sebor <msebor@redhat.com>
1725 * tree-pretty-print.c (pretty_print_string): Add argument.
1726 (dump_generic_node): Call to pretty_print_string with string size.
1728 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
1730 PR rtl-optimization/86771
1731 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
1732 of two SETs into those two SETs, one to be placed at i2, if that SETs
1733 destination is modified between i2 and i3.
1735 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1737 PR tree-optimization/86725
1738 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
1740 (vect_analyze_scalar_cycles_1): Check it.
1742 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1744 PR tree-optimization/86725
1745 * tree-vect-loop.c (vect_is_simple_reduction): When treating
1746 an outer loop phi as a double reduction, make sure that the
1747 single user of the phi result is an inner loop phi.
1749 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1751 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
1752 grouped stores with gaps to a strided group.
1754 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1756 * tree-vect-stmts.c (get_group_load_store_type)
1757 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
1758 first statement in a group.
1760 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1762 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
1763 the sequence used in gcc/gcc.c.
1765 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1768 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
1771 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1773 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
1774 Darwin10-specific unwinder-shim.
1775 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
1776 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
1777 New to cater for Darwin10 Rosetta.
1779 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1781 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
1782 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
1784 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1789 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
1790 (dwarf2out_switch_text_section): Generate a local label for the
1791 second function sub-section and apply it as the second FDE start
1793 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
1794 second sub-section start.
1796 2018-08-22 Richard Biener <rguenther@suse.de>
1798 PR tree-optimization/86988
1799 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
1801 2018-08-22 Richard Biener <rguenther@suse.de>
1803 PR tree-optimization/86945
1804 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
1806 2018-08-22 Alexandre Oliva <oliva@adacore.com>
1808 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
1809 a comment about how uses of r2 for .sdata2 come about.
1811 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
1813 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
1815 2018-08-21 Marek Polacek <polacek@redhat.com>
1817 PR c++/86981, Implement -Wpessimizing-move.
1818 * doc/invoke.texi: Document -Wpessimizing-move.
1820 2018-08-21 Jan Hubicka <jh@suse.cz>
1822 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
1823 * tree.h (is_redundant_typedef): Remove.
1824 * dwarf2out.c (is_redundant_typedef): Turn into static function.
1826 2018-08-21 Jan Hubicka <jh@suse.cz>
1828 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
1831 2018-08-21 Tamar Christina <tamar.christina@arm.com>
1833 * expmed.c (extract_low_bits): Reject invalid subregs early.
1835 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
1838 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
1841 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
1843 * config/vxworks.h: Guard vxworks_asm_out_constructor and
1844 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
1845 * config/vxworks.c: Likewise.
1847 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
1849 * config/vxworks.c: Set targetm.have_ctors_dtors
1850 if HAVE_INITFINI_ARRAY_SUPPORT.
1851 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
1852 if HAVE_INITFINI_ARRAY_SUPPORT.
1854 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
1856 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
1857 default search path for VxWorks < 7.
1859 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
1861 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
1862 (get_format_string): Refer to c_getstr.
1864 2018-08-21 Tom de Vries <tdevries@suse.de>
1866 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
1867 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
1868 (debuginfo_early_stop): Declare.
1869 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
1870 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
1871 (debuginfo_early_stop): New function.
1872 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
1873 and debuginfo_early_stop.
1874 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
1875 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
1876 (general_init): Call debuginfo_early_init.
1877 (finalize): Call debuginfo_fini.
1878 (do_compile): Call debuginfo_init.
1879 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
1881 (@item -fdump-debug, @item -fdump-earlydebug): Add.
1883 2018-08-21 Tom de Vries <tdevries@suse.de>
1885 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
1886 flag_dump_noaddr and flag_dump_unnumbered.
1888 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
1890 * wide-int-range.cc (wide_int_range_abs): New.
1891 (wide_int_range_order_set): Rename from wide_int_range_min_max.
1892 * wide-int-range.h (wide_int_range_abs): New.
1893 (wide_int_range_min_max): New.
1894 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
1895 case to call wide_int_range_abs.
1896 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
1897 (extract_range_from_abs_expr): Delete.
1899 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
1902 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
1903 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
1906 2018-08-20 Nathan Sidwell <nathan@acm.org>
1907 Jeff Law <law@redhat.com>
1909 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
1910 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
1912 2018-08-20 David Malcolm <dmalcolm@redhat.com>
1915 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
1916 (decl_attributes): Likewise.
1917 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
1919 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
1920 * diagnostic-core.h (class auto_diagnostic_group): New class.
1921 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
1922 (diagnostic_report_diagnostic): Handle the first diagnostics within
1924 (emit_diagnostic): Add auto_diagnostic_group instance.
1926 (inform_n): Likewise.
1927 (warning): Likewise.
1928 (warning_at): Likewise.
1929 (warning_n): Likewise.
1930 (pedwarn): Likewise.
1931 (permerror): Likewise.
1933 (error_n): Likewise.
1934 (error_at): Likewise.
1936 (fatal_error): Likewise.
1937 (internal_error): Likewise.
1938 (internal_error_no_backtrace): Likewise.
1939 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
1940 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
1941 * diagnostic.h (struct diagnostic_context): Add fields
1942 "diagnostic_group_nesting_depth",
1943 "diagnostic_group_emission_count", "begin_group_cb",
1945 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
1946 Add auto_diagnostic_group instance(s).
1947 (find_explicit_erroneous_behavior): Likewise.
1948 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
1949 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
1950 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
1951 (gimplify_va_arg_expr): Likewise.
1952 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
1953 (HSA_SORRY_AT): Likewise.
1954 * ipa-devirt.c (compare_virtual_tables): Likewise.
1955 (warn_odr): Likewise.
1956 * multiple_target.c (expand_target_clones): Likewise.
1957 * opts-common.c (cmdline_handle_error): Likewise.
1958 * reginfo.c (globalize_reg): Likewise.
1959 * substring-locations.c (format_warning_n_va): Likewise.
1960 * tree-inline.c (expand_call_inline): Likewise.
1961 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
1962 * tree-ssa-loop-niter.c
1963 (do_warn_aggressive_loop_optimizations): Likewise.
1964 * tree-ssa-uninit.c (warn_uninit): Likewise.
1965 * tree.c (warn_deprecated_use): Likewise.
1967 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
1970 * config/i386/i386.md (eh_return): Always update EH return
1971 address in word_mode.
1973 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
1975 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
1976 TARGET_SPLIT_COMPLEX_ARG is defined.
1978 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
1980 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
1982 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
1985 * expr.c (expand_assignment): Assert that bitpos is positive.
1986 (store_field): Likewise
1987 (expand_expr_real_1): Make sure that bitpos is positive.
1988 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
1991 2018-08-20 Nathan Sidwell <nathan@acm.org>
1993 * Makefile.in (CPP_ID_DATA_H): Delete.
1994 (CPP_INTERNAL_H): Don't add it.
1995 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
1996 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
1998 2018-08-20 Richard Biener <rguenther@suse.de>
2000 PR tree-optimization/78655
2001 * tree-vrp.c (extract_range_from_binary_expr_1): Make
2002 pointer + offset nonnull if either operand is nonnull work.
2004 2018-08-20 Tom de Vries <tdevries@suse.de>
2006 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
2007 unless the referenced die describes the added property using
2008 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
2009 Otherwise, add a DW_AT_location to the referenced die.
2011 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
2014 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
2015 register_operand when calling ix86_set_reg_reg_cost.
2016 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
2017 Set *total to 0 for operands that satisfy x86_64_immediate_operand
2018 predicate and to 1 otherwise.
2020 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
2022 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
2023 emit a diagnostic that it is not supported and reset the option.
2024 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
2025 supported and consume the option. (ASM_FINAL_SPEC): New.
2027 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
2029 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
2032 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
2034 C-SKY port: Documentation
2036 * doc/extend.texi (C-SKY Function Attributes): New section.
2037 * doc/invoke.texi (Option Summary): Add C-SKY options.
2038 (C-SKY Options): New section.
2039 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
2041 2018-08-17 Jojo <jijie_rong@c-sky.com>
2042 Huibin Wang <huibin_wang@c-sky.com>
2043 Sandra Loosemore <sandra@codesourcery.com>
2044 Chung-Lin Tang <cltang@codesourcery.com>
2046 C-SKY port: Backend implementation
2048 * config/csky/*: New.
2049 * common/config/csky/*: New.
2051 2018-08-17 Jojo <jijie_rong@c-sky.com>
2052 Huibin Wang <huibin_wang@c-sky.com>
2053 Sandra Loosemore <sandra@codesourcery.com>
2054 Chung-Lin Tang <cltang@codesourcery.com>
2055 Andrew Jenner <andrew@codesourcery.com>
2057 C-SKY port: Configury
2059 * config.gcc (csky-*-*): New.
2060 * configure.ac: Add csky to targets for dwarf2 debug_line support.
2061 * configure: Regenerated.
2063 2018-08-17 David Malcolm <dmalcolm@redhat.com>
2065 * dump-context.h: Include "dumpfile.h".
2066 (dump_context::dump_printf_va): Convert final param from va_list
2067 to va_list *. Convert from ATTRIBUTE_PRINTF to
2068 ATTRIBUTE_GCC_DUMP_PRINTF.
2069 (dump_context::dump_printf_loc_va): Likewise.
2070 * dumpfile.c: Include "stringpool.h".
2071 (make_item_for_dump_printf_va): Delete.
2072 (make_item_for_dump_printf): Delete.
2073 (class dump_pretty_printer): New class.
2074 (dump_pretty_printer::dump_pretty_printer): New ctor.
2075 (dump_pretty_printer::emit_items): New member function.
2076 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
2078 (dump_pretty_printer::emit_item): New member function.
2079 (dump_pretty_printer::stash_item): New member function.
2080 (dump_pretty_printer::format_decoder_cb): New member function.
2081 (dump_pretty_printer::decode_format): New member function.
2082 (dump_context::dump_printf_va): Reimplement in terms of
2083 dump_pretty_printer.
2084 (dump_context::dump_printf_loc_va): Convert final param from va_list
2086 (dump_context::begin_scope): Reimplement call to
2087 make_item_for_dump_printf.
2088 (dump_printf): Update for change to dump_printf_va.
2089 (dump_printf_loc): Likewise.
2090 (selftest::test_capture_of_dump_calls): Convert "stmt" from
2091 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
2092 with %T, %E, and %G.
2093 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
2094 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
2095 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
2096 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
2097 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
2098 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
2099 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
2100 within a dump_printf_loc call to "%wu".
2101 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
2102 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
2103 missing space after "=".
2104 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
2105 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
2106 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
2107 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
2108 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
2109 duplicate "vectorized" from message.
2111 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
2113 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
2114 polyNxK_t element's TYPE_STRING_FLAG.
2116 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
2118 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
2119 (they were unnamed before). Fix comments.
2121 2018-08-17 Nathan Sidwell <nathan@acm.org>
2123 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
2125 2018-08-17 Richard Biener <rguenther@suse.de>
2127 PR tree-optimization/86841
2128 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
2130 2018-08-17 Martin Liska <mliska@suse.cz>
2132 * common.opt: Remove Warn, Init and Report for options with
2133 Ignore/Deprecated flag. Warning is done automatically for
2135 * config/i386/i386.opt: Likewise.
2136 * config/ia64/ia64.opt: Likewise.
2137 * config/rs6000/rs6000.opt: Likewise.
2138 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
2139 Remove usage of flag_check_pointer_bounds.
2140 * lto-wrapper.c (merge_and_complain): Do not handle
2141 OPT_fcheck_pointer_bounds.
2142 (append_compiler_options): Likewise.
2143 * opt-functions.awk: Do not handle Deprecated.
2144 * optc-gen.awk: Check that Var, Report and Init are not
2145 used for an option with Ignore/Deprecated flag.
2146 * opts-common.c (decode_cmdline_option): Do not report
2148 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
2150 * opts.h (struct cl_option): Remove cl_deprecated flag.
2151 (CL_ERR_DEPRECATED): Remove error enum value.
2153 2018-08-17 Richard Biener <rguenther@suse.de>
2156 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
2157 across a va-arg-pack using call adjust its return value accordingly.
2159 2018-08-16 Martin Sebor <msebor@redhat.com>
2161 PR tree-optimization/86853
2162 * gimple-ssa-sprintf.c (struct format_result): Rename member.
2163 (struct fmtresult): Add member and initialize it in ctors.
2164 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
2165 (format_string): Handle %S the same as %ls. Set MAYFAIL.
2166 (format_directive): Set POSUNDER4K when MAYFAIL is set.
2167 (parse_directive): Handle %C same as %c.
2168 (sprintf_dom_walker::compute_format_length): Adjust.
2169 (is_call_safe): Adjust.
2171 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
2173 * builtins.c (c_strlen): Add new parameter eltsize. Use it
2174 for determining how to count the elements.
2175 * builtins.h (c_strlen): Adjust prototype.
2176 * expr.c (string_constant): Add new parameter mem_size.
2177 Set *mem_size appropriately.
2178 * expr.h (string_constant): Adjust protoype.
2179 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
2180 * gimple-fold.h (get_range_strlen): Adjust prototype.
2181 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
2182 (format_string): Call get_string_length with eltsize.
2184 2018-08-16 David Malcolm <dmalcolm@redhat.com>
2186 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
2187 to emit the span, rather than setting it as the prefix.
2189 2018-08-16 David Malcolm <dmalcolm@redhat.com>
2191 * diagnostic-show-locus.c (layout::start_annotation_line): Add
2192 "margin_char" parameter, defaulting to space. Use it in place
2193 of pp_space for the initial part of the margin.
2194 (layout::print_leading_fixits): Use '+' when filling the margin
2195 of line-insertion fix-it hints.
2197 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2199 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
2202 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2204 * config/rs6000/altivec.md: Don't set length attribute to the default
2206 * config/rs6000/darwin.md: Ditto.
2207 * config/rs6000/dfp.md: Ditto.
2208 * config/rs6000/htm.md: Ditto.
2209 * config/rs6000/rs6000.md: Ditto.
2210 * config/rs6000/sync.md: Ditto.
2211 * config/rs6000/vsx.md: Ditto.
2213 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2215 * config/rs6000/altivec.md: Don't set length attribute to the default
2216 value, for branch instructions.
2217 * config/rs6000/darwin.md: Ditto.
2218 * config/rs6000/rs6000.md: Ditto.
2220 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2222 * config/rs6000/rs6000.md (length): Always define as const_int 4.
2223 (unnamed conditional branch define_insn): Set length to 4 or 8
2224 depending on offset.
2225 (<bd>_<mode>): Similar, for alternative 0.
2226 (<bd>tf_<mode>): Ditto.
2228 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2230 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
2232 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
2234 * doc/rtl.texi: Replace old RTX class names with new names.
2237 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
2239 * expmed.h (canonicalize_comparison): New declaration.
2240 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
2241 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
2242 * optabs.c (prepare_cmp_insn): Likewise.
2243 * rtl.h (unsigned_condition_p): New function which checks if a
2244 comparison operator is unsigned.
2246 2018-08-16 Nathan Sidwell <nathan@acm.org>
2248 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
2249 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
2251 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2254 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
2255 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
2256 (mov<mov>): ..this and enable unconditionally.
2258 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2260 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
2262 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
2264 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
2265 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
2266 "Common" with "Target".
2268 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
2270 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
2271 * doc/invoke.texi (mmitigate-rop): Remove.
2272 * config/i386/i386.c: Do not include "regrename.h".
2273 (ix86_rop_should_change_byte_p, reg_encoded_number)
2274 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
2276 (ix86_reorg): Remove call to ix86_mitigate_rop.
2277 * config/i386/i386.md (attr "modrm_class"): Remove.
2278 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
2279 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
2280 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
2282 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
2284 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
2285 allow folding of mergeh() and mergel() for the float and double types.
2286 (fold_mergehl_helper): Rework to handle building a permute tree
2289 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
2291 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
2294 2018-08-15 David Malcolm <dmalcolm@redhat.com>
2296 * common.opt (fdiagnostics-show-labels): New option.
2297 * diagnostic-show-locus.c (class layout_range): Add field
2299 (class layout): Add field "m_show_labels_p".
2300 (layout_range::layout_range): Add param "label" and use it to
2302 (make_range): Pass in NULL for new "label" param of layout_range's
2304 (layout::layout): Initialize m_show_labels_p.
2305 (layout::maybe_add_location_range): Pass in loc_range->m_label
2306 when constructing layout_range instances.
2307 (struct line_label): New struct.
2308 (layout::print_any_labels): New member function.
2309 (layout::print_line): Call it if label-printing is enabled.
2310 (selftest::test_one_liner_labels): New test.
2311 (selftest::test_diagnostic_show_locus_one_liner): Call it.
2312 * diagnostic.c (diagnostic_initialize): Initialize
2313 context->show_labels_p.
2314 * diagnostic.h (struct diagnostic_context): Add field
2316 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
2317 -fno-diagnostics-show-labels.
2318 * dwarf2out.c (gen_producer_string): Add
2319 OPT_fdiagnostics_show_labels to the ignored options.
2320 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
2322 (gcc_rich_location::maybe_add_expr): Likewise.
2323 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
2324 label" param, defaulting to NULL.
2325 (gcc_rich_location::add_expr): Add "label" param.
2326 (gcc_rich_location::maybe_add_expr): Likewise.
2327 (class text_range_label): New class.
2328 (class range_label_for_type_mismatch): New class.
2329 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
2330 of format_warning_va.
2331 (fmtwarn_n): Likewise for new params of format_warning_n_va.
2332 * lto-wrapper.c (merge_and_complain): Add
2333 OPT_fdiagnostics_show_labels to the "pick one setting" options.
2334 (append_compiler_options): Likewise to the dropped options.
2335 (append_diag_options): Likewise to the passed-on options.
2336 * opts.c (common_handle_option): Handle the new option.
2337 * selftest-diagnostic.c
2338 (test_diagnostic_context::test_diagnostic_context): Enable
2340 * substring-locations.c: Include "gcc-rich-location.h".
2341 (format_warning_n_va): Add "fmt_label" and "param_label" params
2342 and use them as appropriate.
2343 (format_warning_va): Add "fmt_label" and "param_label" params,
2344 passing them on to format_warning_n_va.
2345 (format_warning_at_substring): Likewise.
2346 (format_warning_at_substring_n): Likewise.
2347 * substring-locations.h (format_warning_va): Add "fmt_label" and
2348 "param_label" params.
2349 (format_warning_n_va): Likewise.
2350 (format_warning_at_substring): Likewise.
2351 (format_warning_at_substring_n): Likewise.
2352 * toplev.c (general_init): Initialize global_dc->show_labels_p.
2354 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
2357 * builtins.c (expand_builtin_memcmp): Do not expand the call
2358 when overflow is detected.
2360 2018-08-15 Martin Sebor <msebor@redhat.com>
2362 PR tree-optimization/71625
2363 * config/aarch64/aarch64-builtins.c
2364 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
2366 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
2368 * config/s390/s390.c (s390_reorg): Remove loop.
2370 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
2373 (darwin_function_switched_text_sections): Delete.
2374 * gcc/config/darwin.h
2375 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
2377 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
2380 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
2381 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
2382 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
2384 2018-08-15 Martin Liska <mliska@suse.cz>
2386 PR tree-optimization/86925
2387 * predict.c (expr_expected_value_1): When taking
2388 later predictor, assign also probability.
2389 Use fold_build2_initializer_loc in order to fold
2390 the expression in -frounding-math.
2392 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
2394 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
2396 (expand_vec_perm_1): Try the new method.
2398 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
2401 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
2402 Check whether lra_live_max_point is 0 before dividing.
2404 2018-08-14 Martin Sebor <msebor@redhat.com>
2406 PR tree-optimization/86650
2407 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
2408 (vrp_prop::check_mem_ref): Same.
2410 2018-08-13 Liu Hao <lh_mouse@126.com>
2412 * pretty-print.c (eat_esc_sequence): Swap the foreground and
2413 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
2414 and clear it thereafter, as it only works for DBCS.
2416 2018-08-13 Liu Hao <lh_mouse@126.com>
2418 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
2419 handle returned by _get_osf_handle().
2421 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2423 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
2424 for folding vec_perm.
2426 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2428 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
2429 Add support for gimple-folding of vec_pack() and vec_unpack()
2432 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2434 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
2435 vec_xst variants to the list.
2436 (rs6000_gimple_fold_builtin): Add support for folding unaligned
2437 vector loads and stores.
2439 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
2441 * config.gcc (rs6000-ibm-aix4.x): Delete.
2442 (rs6000-ibm-aix5.1): Delete.
2443 (rs6000-ibm-aix5.2): Delete.
2444 (rs6000-ibm-aix5.3): Delete.
2445 * config/rs6000/aix43.h: Delete.
2446 * config/rs6000/aix51.h: Delete.
2447 * config/rs6000/aix52.h: Delete.
2448 * config/rs6000/t-aix43: Delete.
2450 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
2452 * config/s390/s390.c (s390_decompose_constant_pool_ref):
2454 (s390_decompose_address): Factor out constant pool ref
2457 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2459 * config/nds32/nds32-predicates.c
2460 (nds32_can_use_bclr_p): Change return type as bool.
2461 (nds32_can_use_bset_p): Ditto.
2462 (nds32_can_use_btgl_p): Ditto.
2463 (nds32_can_use_bitci_p): Ditto.
2464 * config/nds32/nds32-protos.h
2465 (nds32_can_use_bclr_p): Change declaration.
2466 (nds32_can_use_bset_p): Ditto.
2467 (nds32_can_use_btgl_p): Ditto.
2468 (nds32_can_use_bitci_p): Ditto.
2470 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2472 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
2473 Support -msched-prolog-epilog option.
2474 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
2476 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2478 * common/config/nds32/nds32-common.c
2479 (nds32_option_optimization_table): Enalbe -malways-align.
2481 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2483 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
2485 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
2486 OPT_misr_secure_ case.
2487 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
2488 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
2489 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
2491 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
2492 (nds32_isr_info): New field security_level.
2493 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
2494 * config/nds32/nds32.md (return_internal): Consider critical attribute.
2495 * config/nds32/nds32.opt (misr-secure): New option.
2496 * config/nds32/nds32_init.inc: New file.
2497 * config/nds32/nds32_isr.h: New file.
2499 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
2501 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
2502 Update comment for atomic instructions.
2503 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
2504 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
2506 (atomic_loaddi): Revise fence expansion to only emit fence prior to
2507 load for __ATOMIC_SEQ_CST model.
2508 (atomic_loaddi_1): Remove float register target.
2509 (atomic_storedi): Handle CONST_INT values.
2510 (atomic_storedi_1): Remove float register source. Add special case
2512 (memory_barrier): New expander and insn.
2514 2018-08-11 Jakub Jelinek <jakub@redhat.com>
2516 PR tree-optimization/86835
2517 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
2518 new_stmt after def_gsi, make sure to insert new_square_stmt after
2519 that stmt, not 2 stmts before it.
2521 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
2524 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
2527 2018-08-10 Martin Liska <mliska@suse.cz>
2530 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
2532 * builtins.c (expand_builtin_expect_with_probability):
2534 (expand_builtin_expect_with_probability): New function.
2535 (build_builtin_expect_predicate): Add new argumnet probability
2536 for BUILT_IN_EXPECT_WITH_PROBABILITY.
2537 (fold_builtin_expect):
2540 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
2541 * builtins.h (fold_builtin_expect): Set new argument.
2542 * doc/extend.texi: Document __builtin_expect_with_probability.
2543 * doc/invoke.texi: Likewise.
2544 * gimple-fold.c (gimple_fold_call): Pass new argument.
2545 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
2546 also BUILT_IN_EXPECT_WITH_PROBABILITY.
2547 * predict.c (get_predictor_value): New function.
2548 (expr_expected_value): Add new argument probability. Assume
2549 that predictor and probability are always non-null.
2550 (expr_expected_value_1): Likewise. For __builtin_expect and
2551 __builtin_expect_with_probability set probability. Handle
2552 combination in binary expressions.
2553 (tree_predict_by_opcode): Simplify code by simply calling
2554 get_predictor_value.
2555 (pass_strip_predict_hints::execute): Add handling of
2556 BUILT_IN_EXPECT_WITH_PROBABILITY.
2557 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
2559 * tree.h (DECL_BUILT_IN_P): New function.
2561 2018-08-10 Martin Liska <mliska@suse.cz>
2563 PR tree-optimization/85799
2564 * passes.def: Add argument for pass_strip_predict_hints.
2565 * predict.c (class pass_strip_predict_hints): Add new argument
2567 (strip_predictor_early): New function.
2568 (pass_strip_predict_hints::execute): Call the function to
2570 (strip_predict_hints): New function.
2571 * predict.def: Fix comment.
2573 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
2575 * Makefile.in: Clarify which tm.texi to copy over to assert the
2576 right to grant a GFDL license for all.
2578 2018-08-09 Jeff Law <law@redhat.com>
2580 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
2583 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
2585 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
2588 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
2590 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
2591 reductions for variable-length vectors.
2593 2018-08-09 David Malcolm <dmalcolm@redhat.com>
2596 * common.opt (fdiagnostics-show-line-numbers): New option.
2597 * diagnostic-show-locus.c (class layout): Add fields
2598 "m_show_line_numbers_p" and "m_linenum_width";
2599 (num_digits): New function.
2600 (test_num_digits): New function.
2601 (layout::layout): Initialize new fields. Update m_x_offset
2602 logic to handle any left margin.
2603 (layout::print_source_line): Print line number when requested.
2604 (layout::start_annotation_line): New member function.
2605 (layout::print_annotation_line): Call it.
2606 (layout::print_leading_fixits): Likewise.
2607 (layout::print_trailing_fixits): Likewise. Update calls to
2608 move_to_column for new parameter.
2609 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
2610 it to potentially call start_annotation_line.
2611 (layout::show_ruler): Call start_annotation_line.
2612 (selftest::test_line_numbers_multiline_range): New selftest.
2613 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
2614 and selftest::test_line_numbers_multiline_range.
2615 * diagnostic.c (diagnostic_initialize): Initialize
2616 show_line_numbers_p.
2617 * diagnostic.h (struct diagnostic_context): Add field
2618 "show_line_numbers_p".
2619 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
2620 -fno-diagnostics-show-line-numbers.
2621 * dwarf2out.c (gen_producer_string): Add
2622 OPT_fdiagnostics_show_line_numbers to the ignored options.
2623 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
2624 one setting" options.
2625 (append_compiler_options): Likewise to the dropped options.
2626 (append_diag_options): Likewise to the passed-on options.
2627 * opts.c (common_handle_option): Handle the new option.
2628 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
2630 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
2632 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
2633 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
2634 third argument of type "const signed char" to descriptions of
2635 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
2636 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
2637 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
2638 __builtin_bcdsub_ov functions.
2640 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
2642 PR tree-optimization/86858
2643 * tree-vect-loop.c (vect_is_simple_reduction): Restore
2644 flow_bb_inside_loop_p calls.
2646 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
2648 PR tree-optimization/86871
2649 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
2650 instead of gimple_assign_lhs.
2652 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
2655 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
2656 register constraint to operand 0.
2657 (add<mode>3_carryinC): Likewise.
2658 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
2660 2018-08-09 Martin Liska <mliska@suse.cz>
2663 * common.opt: Remove extra line.
2665 2018-08-09 Martin Liska <mliska@suse.cz>
2667 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
2668 at the end of a line, make first letter capital and end up
2669 a sentence with a dot.
2670 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
2671 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
2672 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
2673 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
2674 (PARAM_MAX_ISL_OPERATIONS): Likewise.
2675 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
2676 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
2677 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
2678 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
2679 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
2680 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
2681 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
2682 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
2683 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
2684 (PARAM_TREE_REASSOC_WIDTH): Likewise.
2685 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
2686 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
2687 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
2689 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
2692 * config/s390/s390.c (s390_option_override_internal): Reduce the
2693 stack-clash-protection-probe-interval param if it would be too big
2696 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
2699 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
2700 * config/m68k/m68k.c (handle_move_double): Don't call
2701 m68k_final_prescan_insn.
2702 (m68k_adjust_decorated_operand): Renamed from
2703 m68k_final_prescan_insn, remove first and third operand and
2705 (print_operand): Call it.
2706 (print_operand_address): Call it.
2708 2018-08-08 Nathan Sidwell <nathan@acm.org>
2710 * diagnostic.c (diagnostic_report_current_module): Use
2711 linemap_included_from & linemap_included_from_linemap.
2713 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
2715 * config/aarch64/aarch64-cores.def: Add phecda core.
2716 * config/aarch64/aarch64-tune.md: Regenerate.
2717 * doc/invoke.texi: Add phecda core.
2719 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
2722 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
2724 * config/s390/s390.md ("movti"): Add more alternatives for
2725 constant to GPR copies.
2727 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
2729 * config/s390/s390.c: Fix whitespace damage throughout the file.
2730 * config/s390/s390.h: Likewise.
2731 * config/s390/tpf.h: Likewise.
2733 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
2735 * config/s390/s390.c (s390_loadrelative_operand_p):
2736 Remove TARGET_CPU_ZARCH usages.
2737 (s390_rtx_costs): Likewise.
2738 (s390_legitimate_constant_p): Likewise.
2739 (s390_cannot_force_const_mem): Likewise.
2740 (legitimate_reload_constant_p): Likewise.
2741 (s390_preferred_reload_class): Likewise.
2742 (legitimize_pic_address): Likewise.
2743 (legitimize_tls_address): Likewise.
2744 (s390_split_branches): Removed.
2745 (s390_add_execute): Removed.
2746 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
2747 (s390_mainpool_start): Likewise.
2748 (s390_mainpool_finish): Likewise.
2749 (s390_mainpool_cancel): Removed.
2750 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
2751 (s390_chunkify_cancel): Likewise.
2752 (s390_return_addr_rtx): Likewise.
2753 (s390_register_info): Remove split_branches_pending_p uages.
2754 (s390_optimize_register_info): Likewise.
2755 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
2756 split_branches_pending_p usages.
2757 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
2758 (s390_load_got): Likewise.
2759 (s390_expand_split_stack_prologue): Likewise.
2760 (output_asm_nops): Likewise.
2761 (s390_function_profiler): Likewise.
2762 (s390_emit_call): Likewise.
2763 (s390_conditional_register_usage): Likewise.
2764 (s390_optimize_prologue): Likewise.
2765 (s390_reorg): Remove TARGET_CPU_ZARCH and
2766 split_branches_pending_p usages.
2767 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
2769 (s390_output_indirect_thunk_function): Likewise.
2770 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
2771 (TARGET_CPU_ZARCH_P): Removed.
2772 (struct machine_function): Remove split_branches_pending_p.
2773 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
2775 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
2777 * common/config/s390/s390-common.c (processor_flags_table):
2779 * config.gcc: Remove with_arch/with_tune support.
2780 * config/s390/2064.md: Remove cpu attribute comparisons.
2781 * config/s390/driver-native.c (s390_host_detect_local_cpu):
2783 * config/s390/linux.h (ASM_SPEC):
2784 Remove -march support.
2785 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
2786 Use a table to get an arch level.
2787 * config/s390/s390-opts.h (enum processor_type):
2789 * config/s390/s390.c
2790 (processor_table): Remove entries, add arch_level values.
2791 (s390_issue_rate): Remove cases.
2792 (s390_option_override): Adjust
2793 s390_option_override_internal() call.
2794 (s390_option_override_internal): Remove deprecation warning.
2795 (s390_valid_target_attribute_tree): Adjust
2796 s390_option_override_internal() call.
2797 * config/s390/s390.h (struct s390_processor):
2798 Share with s390-c.c, add arch_level field.
2799 * config/s390/s390.md:
2800 Remove occurrences in cpu attribute.
2801 * config/s390/s390.opt: Remove -march/-mtune support.
2802 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
2803 * doc/invoke.texi: Remove deprecation warning.
2805 2018-08-08 Luis Machado <luis.machado@linaro.org>
2807 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
2809 (qdf24xx_tunings): Set vector cost structure to
2810 qdf24xx_vector_cost.
2812 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
2813 <register_sextend>: Set to 3.
2815 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
2818 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
2819 * config/aarch64/aarch64-simd.md
2820 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
2821 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
2822 (aarch64_frecpx<mode>): ...this new pattern.
2823 * config/aarch64/aarch64-simd-builtins.def: Remove comment
2824 about aarch64_frecp<FRECP:frecp_suffix><mode>.
2826 2018-08-07 Martin Liska <mliska@suse.cz>
2829 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
2830 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
2831 * predict.def (PRED_MALLOC_NONNULL): New predictor.
2832 * doc/extend.texi: Document that malloc attribute adds
2835 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
2838 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2839 Define to speculation_safe_value_not_needed.
2841 2018-08-06 Jeff Law <law@redhat.com>
2843 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
2844 the vr_values instance to cprop_into_stmt.
2845 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
2846 (cprop_operand): Also query EVRP to determine if OP is a constant.
2848 2018-08-06 Nathan Sidwell <nathan@acm.org>
2850 * diagnostic.c (diagnostic_report_current_module): Reroll
2851 included-at loop. Translate text.
2853 2018-08-06 David Malcolm <dmalcolm@redhat.com>
2855 * function-tests.c (selftest::test_expansion_to_rtl): Call
2856 free_after_compilation.
2858 2018-08-06 Alan Hayward <alan.hayward@arm.com>
2860 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
2862 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
2864 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
2865 loops with memory block operations from getting unrolled.
2867 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
2870 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2871 Define to speculation_safe_value_not_needed.
2873 2018-08-06 Jeff Law <law@redhat.com>
2875 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
2878 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
2881 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
2882 with all enabled __intN types.
2884 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
2886 2018-08-06 Alan Hayward <alan.hayward@arm.com>
2888 * alias.c (record_set): Check for clobber high.
2889 * cfgexpand.c (expand_gimple_stmt): Likewise.
2890 * combine-stack-adj.c (single_set_for_csa): Likewise.
2891 * combine.c (find_single_use_1): Likewise.
2892 (set_nonzero_bits_and_sign_copies): Likewise.
2893 (get_combine_src_dest): Likewise.
2894 (is_parallel_of_n_reg_sets): Likewise.
2895 (try_combine): Likewise.
2896 (record_dead_and_set_regs_1): Likewise.
2897 (reg_dead_at_p_1): Likewise.
2898 (reg_dead_at_p): Likewise.
2899 * dce.c (deletable_insn_p): Likewise.
2900 (mark_nonreg_stores_1): Likewise.
2901 (mark_nonreg_stores_2): Likewise.
2902 * df-scan.c (df_find_hard_reg_defs): Likewise.
2903 (df_uses_record): Likewise.
2904 (df_get_call_refs): Likewise.
2905 * dwarf2out.c (mem_loc_descriptor): Likewise.
2906 * haifa-sched.c (haifa_classify_rtx): Likewise.
2907 * ira-build.c (create_insn_allocnos): Likewise.
2908 * ira-costs.c (scan_one_insn): Likewise.
2909 * ira.c (equiv_init_movable_p): Likewise.
2910 (rtx_moveable_p): Likewise.
2911 (interesting_dest_for_shprep): Likewise.
2912 * jump.c (mark_jump_label_1): Likewise.
2913 * postreload-gcse.c (record_opr_changes): Likewise.
2914 * postreload.c (reload_cse_simplify): Likewise.
2915 (struct reg_use): Add source expr.
2916 (reload_combine): Check for clobber high.
2917 (reload_combine_note_use): Likewise.
2918 (reload_cse_move2add): Likewise.
2919 (move2add_note_store): Likewise.
2920 * print-rtl.c (print_pattern): Likewise.
2921 * recog.c (decode_asm_operands): Likewise.
2922 (store_data_bypass_p): Likewise.
2923 (if_test_bypass_p): Likewise.
2924 * regcprop.c (kill_clobbered_value): Likewise.
2925 (kill_set_value): Likewise.
2926 * reginfo.c (reg_scan_mark_refs): Likewise.
2927 * reload1.c (maybe_fix_stack_asms): Likewise.
2928 (eliminate_regs_1): Likewise.
2929 (elimination_effects): Likewise.
2930 (mark_not_eliminable): Likewise.
2931 (scan_paradoxical_subregs): Likewise.
2932 (forget_old_reloads_1): Likewise.
2933 * reorg.c (find_end_label): Likewise.
2934 (try_merge_delay_insns): Likewise.
2935 (redundant_insn): Likewise.
2936 (own_thread_p): Likewise.
2937 (fill_simple_delay_slots): Likewise.
2938 (fill_slots_from_thread): Likewise.
2939 (dbr_schedule): Likewise.
2940 * resource.c (update_live_status): Likewise.
2941 (mark_referenced_resources): Likewise.
2942 (mark_set_resources): Likewise.
2943 * rtl.c (copy_rtx): Likewise.
2944 * rtlanal.c (reg_referenced_p): Likewise.
2945 (single_set_2): Likewise.
2946 (noop_move_p): Likewise.
2947 (note_stores): Likewise.
2948 * sched-deps.c (sched_analyze_reg): Likewise.
2949 (sched_analyze_insn): Likewise.
2951 2018-08-06 Alan Hayward <alan.hayward@arm.com>
2953 * cse.c (invalidate_reg): New function extracted from...
2954 (invalidate): ...here.
2955 (canonicalize_insn): Check for clobber high.
2956 (invalidate_from_clobbers): invalidate clobber highs.
2957 (invalidate_from_sets_and_clobbers): Likewise.
2958 (count_reg_usage): Check for clobber high.
2959 (insn_live_p): Likewise.
2960 * cselib.c (cselib_expand_value_rtx_1):Likewise.
2961 (cselib_invalidate_regno): Check for clobber in setter.
2962 (cselib_invalidate_rtx): Pass through setter.
2963 (cselib_invalidate_rtx_note_stores):
2964 (cselib_process_insn): Check for clobber high.
2965 * cselib.h (cselib_invalidate_rtx): Add operand.
2967 2018-08-06 Alan Hayward <alan.hayward@arm.com>
2969 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
2970 (mark_not_eliminable): Likewise.
2971 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
2972 * lra-lives.c (process_bb_lives): Check for clobber high.
2973 * lra.c (new_insn_reg): Remember clobber highs.
2974 (collect_non_operand_hard_regs): Check for clobber high.
2975 (lra_set_insn_recog_data): Likewise.
2976 (add_regs_to_insn_regno_info): Likewise.
2977 (lra_update_insn_regno_info): Likewise.
2979 2018-08-06 Alan Hayward <alan.hayward@arm.com>
2981 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
2982 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
2984 2018-08-06 Alan Hayward <alan.hayward@arm.com>
2986 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
2987 (copy_insn_1): Likewise.
2988 (gen_hard_reg_clobber_high): New gen function.
2989 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
2990 * genemit.c (gen_exp): Likewise.
2991 (gen_emit_seq): Pass through info.
2992 (gen_insn): Check for CLOBBER_HIGH.
2993 (gen_expand): Pass through info.
2994 (gen_split): Likewise.
2995 (output_add_clobbers): Likewise.
2996 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
2997 (remove_clobbers): Likewise.
2998 * rtl.h (gen_hard_reg_clobber_high): New declaration.
3000 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3002 * doc/rtl.texi (clobber_high): Add.
3003 (parallel): Add in clobber high
3004 * rtl.c (rtl_check_failed_code3): Add function.
3005 * rtl.def (CLOBBER_HIGH): Add expression.
3006 * rtl.h (RTL_CHECKC3): Add macro.
3007 (rtl_check_failed_code3): Add declaration.
3008 (XC3EXP): Add macro.
3010 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
3013 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
3014 cfun->machine->max_used_stack_alignment if needed.
3016 2018-08-04 Martin Sebor <msebor@redhat.com>
3018 PR tree-optimization/86571
3019 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
3022 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
3025 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3028 2018-08-03 Jeff Law <law@redhat.com>
3031 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3032 Define to speculation_safe_value_not_needed.
3034 2018-08-03 David Malcolm <dmalcolm@redhat.com>
3036 * doc/gcov.texi (-x): Remove duplicate "to".
3037 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
3038 (-Wif-not-aligned): Remove duplicate "is".
3039 (-flto): Remove duplicate "the".
3040 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
3041 duplicate "v5.00.b".
3042 (MSP430 Options): Remove duplicate "and" from the description
3043 of "-mgprel-sec=regexp".
3044 (x86 Options): Remove duplicate copies of "vmldLog102" and
3045 vmlsLog104 from description of "-mveclibabi=type".
3047 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
3049 * internal-fn.h (first_commutative_argument): Declare.
3050 * internal-fn.c (first_commutative_argument): New function.
3051 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
3052 restrictions for pattern statements. Use first_commutative_argument
3053 to look for commutative operands in calls to internal functions.
3055 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
3057 * Makefile.in (wide-int-range.o): New.
3058 * tree-vrp.c: Move all the wide_int_* functions to...
3059 * wide-int-range.cc: ...here.
3060 * tree-vrp.h: Move all the wide_int_* prototypes to...
3061 * wide-int-range.h: ...here.
3063 2018-08-03 Tom de Vries <tdevries@suse.de>
3065 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
3067 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
3068 * except.c (output_function_exception_table): Do early exit if
3069 targetm_common.except_unwind_info (&global_options) == UI_NONE.
3071 2018-08-03 Martin Liska <mliska@suse.cz>
3073 * predict.c (dump_prediction): Change to 2 digits
3076 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
3078 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
3080 (falkor_am_1_gtov_gtov): ... a new insn reservation.
3082 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
3084 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
3085 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
3086 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
3087 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
3088 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
3090 2018-08-02 David Malcolm <dmalcolm@redhat.com>
3092 * diagnostic-show-locus.c (diagnostic_show_locus): Use
3093 pp_take_prefix when saving the existing prefix.
3094 * diagnostic.c (diagnostic_append_note): Likewise.
3095 * langhooks.c (lhd_print_error_function): Likewise.
3096 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
3097 param's type. Free the existing prefix.
3098 (pp_take_prefix): New function.
3099 (pretty_printer::pretty_printer): Drop the prefix parameter.
3100 Rename the length parameter to match the comment.
3101 (pretty_printer::~pretty_printer): Free the prefix.
3102 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
3104 (struct pretty_printer): Drop the "const" from "prefix" field's
3105 type and clarify memory management.
3106 (pp_set_prefix): Drop the "const" from the 2nd param.
3107 (pp_take_prefix): New decl.
3109 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
3111 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
3112 for word_mode_ok here instead of passing as argument.
3113 (expand_block_compare): Change select_block_compare_mode() call.
3114 (expand_strncmp_gpr_sequence): New function.
3115 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
3117 2018-08-02 Jeff Law <law@redhat.com>
3120 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3121 Define to speculation_safe_value_not_needed.
3124 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3125 Define to speculation_safe_value_not_needed.
3127 2018-08-02 Tom de Vries <tdevries@suse.de>
3130 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
3131 function. Return UI_TARGET unconditionally.
3132 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
3133 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
3135 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3137 * genemit.c (print_overload_test): Fix typo.
3139 2018-08-02 Richard Biener <rguenther@suse.de>
3141 PR tree-optimization/86816
3142 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
3143 which checks for value availability before querying it.
3144 (gvn_uses_equal): Use it.
3145 (same_succ_hash): Likewise.
3146 (gimple_equal_p): Likewise.
3148 2018-08-02 Nick Clifton <nickc@redhat.com>
3151 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3152 Define to speculation_safe_value_not_needed.
3155 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3156 Define to speculation_safe_value_not_needed.
3159 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3160 Define to speculation_safe_value_not_needed.
3163 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3164 Define to speculation_safe_value_not_needed.
3167 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3168 Define to speculation_safe_value_not_needed.
3171 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3172 Define to speculation_safe_value_not_needed.
3175 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3176 Define to speculation_safe_value_not_needed.
3179 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3180 Define to speculation_safe_value_not_needed.
3183 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
3184 speculation_safe_value_not_needed.
3187 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
3188 to speculation_safe_value_not_needed.
3190 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3192 * doc/md.texi: Expand the documentation of instruction names
3193 to mention port-local uses. Document '@' in pattern names.
3194 * read-md.h (overloaded_instance, overloaded_name): New structs.
3196 (md_reader::handle_overloaded_name): New member function.
3197 (md_reader::get_overloads): Likewise.
3198 (md_reader::m_first_overload): New member variable.
3199 (md_reader::m_next_overload_ptr): Likewise.
3200 (md_reader::m_overloads_htab): Likewise.
3201 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
3202 m_next_overload_ptr and m_overloads_htab.
3203 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
3204 (get_mode_token, get_code_token, get_int_token): New functions.
3205 (map_attr_string): Add an optional argument that passes back
3206 the associated iterator.
3207 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
3208 (md_reader::handle_overloaded_name, add_overload_instance): New
3210 (apply_iterators): Handle '@' names. Report an error if '@'
3211 is used without iterators.
3212 (initialize_iterators): Initialize the new iterator_group fields.
3213 * genopinit.c (handle_overloaded_code_for)
3214 (handle_overloaded_gen): New functions.
3215 (main): Use them to print declarations of maybe_code_for_* and
3216 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
3217 * genemit.c (print_overload_arguments, print_overload_test)
3218 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
3219 (main): Use it to print definitions of maybe_code_for_* and
3220 maybe_gen_* functions.
3221 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
3222 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
3223 instead of explicit mode checks.
3224 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
3225 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
3226 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
3227 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
3228 (aarch64_expand_compare_and_swap): Likewise
3229 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
3230 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
3231 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
3232 (aarch64_constant_pool_reload_icode): Delete.
3233 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
3234 instead of aarch64_constant_pool_reload_icode. Use
3235 code_for_aarch64_reload_mov instead of explicit mode checks.
3236 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
3237 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
3238 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
3239 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
3240 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
3241 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
3242 (aarch64_atomic_load_op_code): Delete.
3243 (aarch64_emit_atomic_load_op): Likewise.
3244 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
3245 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
3246 instead of aarch64_emit_atomic_load_op.
3247 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
3248 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
3249 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
3250 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
3251 character before the pattern name.
3252 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
3253 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
3254 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
3255 (aarch64_frecps<mode>): Likewise.
3256 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
3257 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
3258 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
3259 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
3260 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
3262 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3264 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
3265 Allow HFmode constants if TARGET_FP_F16INST.
3267 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
3270 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
3271 No longer check last store for clobber of address register.
3273 2018-08-02 Martin Liska <mliska@suse.cz>
3275 PR gcov-profile/86817
3276 * gcov.c (process_all_functions): New function.
3278 (process_file): Move functions processing to
3279 process_all_functions.
3281 2018-08-02 David Malcolm <dmalcolm@redhat.com>
3283 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
3284 "const" to the "gimple *" and "rtx_insn *" parameters.
3285 * dumpfile.h (dump_user_location_t::dump_user_location_t):
3287 (dump_location_t::dump_location_t): Likewise.
3289 2018-08-01 Martin Sebor <msebor@redhat.com>
3291 PR tree-optimization/86650
3292 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
3293 rather than a "gcall *". Directly pass the data of interest
3294 to percent_K_format, rather than building a temporary CALL_EXPR
3296 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
3297 (gimple_fold_builtin_strncat): Adjust.
3298 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
3299 gcall* argument with gimple*.
3300 * gimple-ssa-warn-restrict.c (check_call): Same.
3301 (wrestrict_dom_walker::before_dom_children): Same.
3302 (builtin_access::builtin_access): Same.
3303 (check_bounds_or_overlap): Same
3304 (maybe_diag_overlap): Same.
3305 (maybe_diag_offset_bounds): Same.
3306 * tree-diagnostic.c (default_tree_printer): Move usage of
3307 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
3309 * tree-pretty-print.c (percent_K_format): Add argument.
3310 * tree-pretty-print.h: Add argument.
3311 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
3312 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
3313 (maybe_diag_stxncpy_trunc): Same.
3314 (handle_builtin_stxncpy): Same.
3315 (handle_builtin_strcat): Same.
3317 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3319 * match.pd: Optimise pointer range checks.
3321 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3323 PR tree-optimization/86758
3324 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
3325 to remove pattern statements.
3327 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3329 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
3330 result of dfs_enumerate_from when constructing stmt_vec_infos,
3331 instead of additionally calling get_loop_body.
3333 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3335 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
3337 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
3338 When creating an iv, assert that the step is not known to be zero.
3339 (vect_setup_realignment): Update call accordingly.
3340 * tree-vect-stmts.c (vectorizable_store): Likewise.
3341 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
3343 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3345 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
3346 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
3347 (vectorizable_reduction): Likewise.
3348 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
3349 (vect_detect_hybrid_slp_stmts): Likewise.
3350 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
3352 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
3354 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
3355 (wide_int_set_zero_nonzero_bits): ...this.
3356 (zero_nonzero_bits_from_vr): Rename to...
3357 (vrp_set_zero_nonzero_bits): ...this.
3358 (extract_range_from_multiplicative_op_1): Abstract wide int
3360 (wide_int_range_multiplicative_op): ...here.
3361 (extract_range_from_binary_expr_1): Extract wide int binary
3362 operations into their own functions.
3363 (wide_int_range_lshift): New.
3364 (wide_int_range_can_optimize_bit_op): New.
3365 (wide_int_range_shift_undefined_p): New.
3366 (wide_int_range_bit_xor): New.
3367 (wide_int_range_bit_ior): New.
3368 (wide_int_range_bit_and): New.
3369 (wide_int_range_trunc_mod): New.
3370 (extract_range_into_wide_ints): New.
3371 (vrp_shift_undefined_p): New.
3372 (extract_range_from_multiplicative_op): New.
3373 (vrp_can_optimize_bit_op): New.
3374 * tree-vrp.h (value_range::dump): New.
3375 (wide_int_range_multiplicative_op): New.
3376 (wide_int_range_lshift):New.
3377 (wide_int_range_shift_undefined_p): New.
3378 (wide_int_range_bit_xor): New.
3379 (wide_int_range_bit_ior): New.
3380 (wide_int_range_bit_and): New.
3381 (wide_int_range_trunc_mod): New.
3382 (zero_nonzero_bits_from_bounds): Rename to...
3383 (wide_int_set_zero_nonzero_bits): ...this.
3384 (zero_nonzero_bits_from_vr): Rename to...
3385 (vrp_set_zero_nonzero_bits): ...this.
3386 (range_easy_mask_min_max): Rename to...
3387 (wide_int_range_can_optimize_bit_op): this.
3388 * vr-values.c (simplify_bit_ops_using_ranges): Rename
3389 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
3391 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3393 * tree-vectorizer.h (vect_orig_stmt): New function.
3394 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
3395 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
3396 (vect_create_epilog_for_reduction): Likewise.
3397 (vectorizable_live_operation): Likewise.
3398 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
3399 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
3400 * tree-vect-stmts.c (vectorizable_call): Likewise.
3401 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
3403 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3405 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
3407 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
3408 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
3409 (vect_transform_loop): Likewise.
3410 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3412 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3414 * tree-vectorizer.h (vect_schedule_slp): Return void.
3415 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3416 (vect_schedule_slp): Likewise.
3418 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3420 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
3422 (vect_transform_loop): Update calls accordingly. Schedule SLP
3423 instances before the main loop, if any exist.
3425 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3427 PR tree-optimization/86749
3428 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
3429 If the lhs is used in a COND_EXPR, check that it is being used
3430 as the "then" or "else" value.
3432 2018-08-01 Tom de Vries <tdevries@suse.de>
3435 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
3436 speculation_safe_value_not_needed.
3438 2018-08-01 Richard Biener <rguenther@suse.de>
3440 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
3443 2018-08-01 Martin Liska <mliska@suse.cz>
3445 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
3446 format how successful transformation is dumped.
3447 (gimple_mod_pow2_value_transform): Likewise.
3448 (gimple_mod_subtract_transform): Likewise.
3449 (gimple_stringops_transform): Likewise.
3451 2018-08-01 Martin Liska <mliska@suse.cz>
3454 * value-prof.c (interesting_stringop_to_profile_p):
3455 Simplify the code and add BUILT_IN_MEMMOVE.
3456 (gimple_stringops_transform): Likewise.
3458 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
3460 * config/aarch64/aarch64-simd.md
3461 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
3462 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
3463 use GPI iterator instead of SI mode.
3465 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
3467 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
3468 rs6000_speculation_barrier.
3469 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
3470 new barrier pattern name.
3472 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
3474 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
3475 (speculation_barrier): New insn.
3477 2018-08-01 Richard Biener <rguenther@suse.de>
3480 * graphite.h: Include isl/id.h and isl/space.h to allow build
3483 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
3486 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
3487 mode for COFF targets.
3488 * defaults.h (TARGET_COFF): Define.
3489 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
3490 TARGET_COFF): Define.
3491 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
3492 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
3493 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
3495 2018-07-31 Alexandre Oliva <oliva@adacore.com>
3497 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
3498 * gimple-streamer-out.c (output_bb): Save it.
3499 * lto-streamer-in.c (input_struct_function_base): Restore
3500 instance discriminator if available. Create map on demand.
3501 * lto-streamer-out.c (output_struct_function_base): Save it if
3503 * final.c (decl_to_instance_map): Document LTO strategy.
3505 2018-07-31 Alexandre Oliva <oliva@adacore.com>
3506 Olivier Hainque <hainque@adacore.com>
3508 * debug.h (decl_to_instance_map_t): New type.
3509 (decl_to_instance_map): Declare.
3510 (maybe_create_decl_to_instance_map): New inline function.
3511 * final.c (bb_discriminator, last_bb_discriminator): New statics,
3512 to track basic block discriminators.
3513 (final_start_function_1): Initialize them.
3514 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
3516 (decl_to_instance_map): New variable.
3517 (map_decl_to_instance, maybe_set_discriminator): New functions.
3518 (notice_source_line): Set discriminator.
3520 2018-07-31 Ian Lance Taylor <iant@golang.org>
3522 * targhooks.c (default_have_speculation_safe_value): Add
3525 2018-07-31 David Malcolm <dmalcolm@redhat.com>
3527 * dump-context.h: Include "pretty-print.h".
3528 (dump_context::refresh_dumps_are_enabled): New decl.
3529 (dump_context::emit_item): New decl.
3530 (class dump_context): Add fields "m_test_pp" and
3532 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
3533 (temp_dump_context::get_dumped_text): New decl.
3534 (class temp_dump_context): Add field "m_pp".
3535 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
3536 (dump_context::refresh_dumps_are_enabled): ...and add a test for
3538 (set_dump_file): Update for above change.
3539 (set_alt_dump_file): Likewise.
3540 (dump_loc): New overload, taking a pretty_printer *.
3541 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
3542 to any test pretty-printer.
3543 (make_item_for_dump_gimple_stmt): New function, adapted from
3544 optinfo::add_gimple_stmt.
3545 (dump_context::dump_gimple_stmt): Call it, and use the result,
3546 eliminating the direct usage of dump_file and alt_dump_file in
3547 favor of indirectly using them via emit_item.
3548 (make_item_for_dump_gimple_expr): New function, adapted from
3549 optinfo::add_gimple_expr.
3550 (dump_context::dump_gimple_expr): Call it, and use the result,
3551 eliminating the direct usage of dump_file and alt_dump_file in
3552 favor of indirectly using them via emit_item.
3553 (make_item_for_dump_generic_expr): New function, adapted from
3555 (dump_context::dump_generic_expr): Call it, and use the result,
3556 eliminating the direct usage of dump_file and alt_dump_file in
3557 favor of indirectly using them via emit_item.
3558 (make_item_for_dump_printf_va): New function, adapted from
3559 optinfo::add_printf_va.
3560 (make_item_for_dump_printf): New function.
3561 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
3562 and use the result, eliminating the direct usage of dump_file and
3563 alt_dump_file in favor of indirectly using them via emit_item.
3564 (make_item_for_dump_dec): New function.
3565 (dump_context::dump_dec): Call it, and use the result,
3566 eliminating the direct usage of dump_file and alt_dump_file in
3567 favor of indirectly using them via emit_item.
3568 (make_item_for_dump_symtab_node): New function, adapted from
3569 optinfo::add_symtab_node.
3570 (dump_context::dump_symtab_node): Call it, and use the result,
3571 eliminating the direct usage of dump_file and alt_dump_file in
3572 favor of indirectly using them via emit_item.
3573 (dump_context::begin_scope): Reimplement, avoiding direct usage
3574 of dump_file and alt_dump_file in favor of indirectly using them
3576 (dump_context::emit_item): New member function.
3577 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
3578 Set up test pretty-printer on the underlying context. Call
3579 refresh_dumps_are_enabled.
3580 (temp_dump_context::~temp_dump_context): Call
3581 refresh_dumps_are_enabled.
3582 (temp_dump_context::get_dumped_text): New member function.
3583 (selftest::verify_dumped_text): New function.
3584 (ASSERT_DUMPED_TEXT_EQ): New macro.
3585 (selftest::test_capture_of_dump_calls): Run all tests twice, with
3586 and then without optinfo enabled. Add uses of
3587 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
3588 * dumpfile.h: Update comment for the dump_* API.
3589 * optinfo-emit-json.cc
3590 (selftest::test_building_json_from_dump_calls): Update for new
3591 param for temp_dump_context ctor.
3592 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
3593 and "m_owned" field.
3594 (optinfo_item::~optinfo_item): Likewise.
3595 (optinfo::add_item): New member function.
3596 (optinfo::emit): Update comment.
3597 (optinfo::add_string): Delete.
3598 (optinfo::add_printf): Delete.
3599 (optinfo::add_printf_va): Delete.
3600 (optinfo::add_gimple_stmt): Delete.
3601 (optinfo::add_gimple_expr): Delete.
3602 (optinfo::add_tree): Delete.
3603 (optinfo::add_symtab_node): Delete.
3604 (optinfo::add_dec): Delete.
3605 * optinfo.h (class dump_context): New forward decl.
3606 (optinfo::add_item): New decl.
3607 (optinfo::add_string): Delete.
3608 (optinfo::add_printf): Delete.
3609 (optinfo::add_printf_va): Delete.
3610 (optinfo::add_gimple_stmt): Delete.
3611 (optinfo::add_gimple_expr): Delete.
3612 (optinfo::add_tree): Delete.
3613 (optinfo::add_symtab_node): Delete.
3614 (optinfo::add_dec): Delete.
3615 (optinfo::add_poly_int): Delete.
3616 (optinfo_item::optinfo_item): Remove "owned" param.
3617 (class optinfo_item): Remove field "m_owned".
3619 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3622 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
3623 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
3624 requested variable alignment.
3625 (expand_one_ssa_partition): Likewise.
3626 (expand_one_var): Likewise.
3628 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3630 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
3631 to speculation_safe_value_not_needed.
3633 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3635 * targhooks.h (speculation_safe_value_not_needed): New prototype.
3636 * targhooks.c (speculation_safe_value_not_needed): New function.
3637 * target.def (have_speculation_safe_value): Update documentation.
3638 * doc/tm.texi: Regenerated.
3640 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3642 * config/aarch64/iterators.md (ALLI_TI): New iterator.
3643 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
3645 (despeculate_copy<ALLI:mode>_insn): New insn.
3646 (despeculate_copyti_insn): New insn.
3647 (despeculate_simple<ALLI:mode>): New insn
3648 (despeculate_simpleti): New insn.
3649 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
3651 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
3652 aarch64_speculation_safe_value.
3653 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
3655 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3657 * config/aarch64/aarch64-speculation.cc: New file.
3658 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
3659 before pass_reorder_blocks.
3660 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
3662 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
3663 X14 and X15 when tracking speculation.
3664 * config/aarch64/aarch64.md (register name constants): Add
3665 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
3666 (unspec): Add UNSPEC_SPECULATION_TRACKER.
3667 (speculation_barrier): New insn attribute.
3668 (cmp<mode>): Allow SP in comparisons.
3669 (speculation_tracker): New insn.
3670 (speculation_barrier): Add speculation_barrier attribute.
3671 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
3672 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
3673 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
3675 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3677 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
3678 aarch64_track_speculation is true.
3679 (tb<optab><mode>1): Likewise.
3680 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
3681 generate CB[N]Z when tracking speculation.
3682 (aarch64_split_compare_and_swap): Likewise.
3683 (aarch64_split_atomic_op): Likewise.
3685 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3687 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
3689 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3691 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
3692 (speculation_barrier): New insn.
3694 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3696 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
3697 * config/arm/arm.md (speculation_barrier): New expand.
3698 (speculation_barrier_insn): New pattern.
3700 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3702 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
3703 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
3704 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
3705 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
3707 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
3708 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
3709 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
3710 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
3711 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
3712 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
3713 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
3714 * builtins.c (expand_speculation_safe_value): New function.
3715 (expand_builtin): Call it.
3716 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
3717 * doc/extend.texi: Document __builtin_speculation_safe_value.
3718 * doc/md.texi: Document "speculation_barrier" pattern.
3719 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
3720 TARGET_HAVE_SPECULATION_SAFE_VALUE.
3721 * doc/tm.texi: Regenerated.
3722 * target.def (have_speculation_safe_value, speculation_safe_value): New
3724 * targhooks.c (default_have_speculation_safe_value): New function.
3725 (default_speculation_safe_value): New function.
3726 * targhooks.h (default_have_speculation_safe_value): Add prototype.
3727 (default_speculation_safe_value): Add prototype.
3729 2018-07-31 David Malcolm <dmalcolm@redhat.com>
3731 * dump-context.h (dump_context::dump_loc): New decl.
3732 * dumpfile.c (dump_context::dump_loc): New member function.
3733 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
3734 and dump_gimple_stmt.
3735 (dump_context::dump_gimple_expr_loc): Likewise, using
3737 (dump_context::dump_generic_expr_loc): Likewise, using
3739 (dump_context::dump_printf_loc_va): Likewise, using
3741 (dump_context::begin_scope): Explicitly using the global function
3742 "dump_loc", rather than the member function.
3744 2018-07-31 Martin Sebor <msebor@redhat.com>
3746 PR tree-optimization/86741
3747 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
3749 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
3751 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
3752 depend on whether prefetch instructions will be emitted or not.
3753 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
3754 will be emitted or not.
3755 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
3756 (TARGET_SETMEM_PFD): New macros.
3758 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3760 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
3761 (NULL_STMT_VEC_INFO): Delete.
3762 (stmt_vec_info::operator*): Likewise.
3763 (stmt_vec_info::operator gimple *): Likewise.
3764 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
3765 of NULL_STMT_VEC_INFO.
3766 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
3767 (vect_reassociating_reduction_p): Likewise.
3768 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
3769 (vectorizable_store): Likewise.
3770 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
3771 (vec_info::free_stmt_vec_infos): Likewise.
3773 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3775 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
3776 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
3778 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3780 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
3781 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
3782 (vec_info::free_stmt_vec_info): New private member functions.
3783 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
3784 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
3785 * tree-parloops.c (gather_scalar_reductions): Remove calls to
3786 set_stmt_vec_info_vec and free_stmt_vec_infos.
3787 * tree-vect-loop.c (_loop_vec_info): Remove call to
3788 set_stmt_vec_info_vec.
3789 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
3790 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
3791 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
3792 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
3793 (vec_info::free_stmt_vec_info): ...these new functions. Remove
3794 assignments in {vec_info::,}new_stmt_vec_info that are redundant
3795 with the clearing in the xcalloc.
3796 (stmt_vec_info_vec): Delete.
3797 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
3798 (vectorize_loops): Likewise.
3799 (vec_info::~vec_info): Remove argument from call to
3800 free_stmt_vec_infos.
3801 (vec_info::add_stmt): Remove vinfo argument from call to
3804 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3806 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
3807 rather than a gimple stmt.
3808 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
3809 information for pattern statements when passed the original
3810 statement; instead wait to be passed the pattern statement itself.
3811 Don't call set_vinfo_for_stmt here.
3812 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
3813 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
3814 stmt_vec_infos here.
3815 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
3816 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
3817 stmt_vec_infos entry.
3819 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3821 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
3822 * tree-vectorizer.c (vec_info::replace_stmt): New function.
3823 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
3824 * tree-vect-stmts.c (vectorizable_call): Likewise.
3825 (vectorizable_simd_clone_call): Likewise.
3827 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3829 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
3830 * tree-vectorizer.c (vec_info::remove_stmt): New function.
3831 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
3832 * tree-vect-loop.c (vect_transform_loop): Likewise.
3833 * tree-vect-slp.c (vect_schedule_slp): Likewise.
3834 * tree-vect-stmts.c (vect_remove_stores): Likewise.
3836 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3838 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
3839 (vect_dr_stmt): Delete.
3840 * tree-vectorizer.c (vec_info::lookup_dr): New function.
3841 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
3843 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
3844 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
3845 (vect_verify_datarefs_alignment, vect_peeling_supportable)
3846 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
3847 (vect_analyze_data_refs): Likewise.
3848 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
3850 (vect_find_same_alignment_drs): Likewise.
3851 (vect_slp_analyze_node_dependences): Update calls accordingly.
3852 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
3853 instead of DR_VECT_AUX.
3854 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
3855 of a vector data references. Use vec_info::lookup_dr instead of
3857 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
3858 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
3859 instead of DR_VECT_AUX.
3861 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3863 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
3865 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
3867 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
3868 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
3869 (vect_gen_prolog_loop_niters): Likewise.
3871 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3873 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
3874 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
3875 (vect_known_alignment_in_bytes, vect_dr_behavior)
3876 (vect_get_scalar_dr_size): Take references as dr_vec_infos
3877 instead of data_references. Update calls to other routines for
3878 which the same change has been made.
3879 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
3880 dr_vec_infos instead of stmt_vec_infos.
3881 (vect_analyze_data_ref_dependence): Update call accordingly.
3882 (vect_slp_analyze_data_ref_dependence)
3883 (vect_record_base_alignments): Use DR_VECT_AUX.
3884 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
3885 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
3886 (vector_alignment_reachable_p, vect_get_data_access_cost)
3887 (vect_peeling_supportable, vect_analyze_group_access_1)
3888 (vect_analyze_group_access, vect_analyze_data_ref_access)
3889 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
3890 (vect_compile_time_alias, vect_small_gap_p)
3891 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
3892 (vect_supportable_dr_alignment): Take references as dr_vec_infos
3893 instead of data_references. Update calls to other routines for
3894 which the same change has been made.
3895 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
3896 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
3897 (vect_slp_analyze_and_verify_node_alignment)
3898 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
3899 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
3900 (vect_setup_realignment): Use dr_vec_infos. Update calls after
3902 (_vect_peel_info::dr): Replace with...
3903 (_vect_peel_info::dr_info): ...this new field.
3904 (vect_peeling_hash_get_most_frequent)
3905 (vect_peeling_hash_choose_best_peeling): Update accordingly.
3906 (vect_peeling_hash_get_lowest_cost):
3907 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
3908 routines for which the same change has been made.
3909 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
3911 * tree-vect-loop-manip.c (get_misalign_in_elems)
3912 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
3914 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
3915 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
3916 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
3917 (get_group_load_store_type, get_negative_load_store_type)
3918 (vect_get_data_ptr_increment, vectorizable_store)
3919 (vectorizable_load): Likewise.
3920 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
3921 Update calls to other routines for which the same change has been made.
3923 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3925 * tree-vectorizer.h (vec_info::move_dr): New member function.
3926 (dataref_aux): Rename to...
3927 (dr_vec_info): ...this and add "dr" and "stmt" fields.
3928 (_stmt_vec_info::dr_aux): Update accordingly.
3929 (_stmt_vec_info::data_ref_info): Delete.
3930 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
3931 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
3932 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
3933 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
3935 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
3936 (STMT_VINFO_DR_INFO): New macro.
3937 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
3938 (set_dr_misalignment): Update after rename of dataref_aux.
3939 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
3940 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
3941 initialization of STMT_VINFO_DATA_REF.
3942 * tree-vectorizer.c (vec_info::move_dr): New function.
3943 * tree-vect-patterns.c (vect_recog_bool_pattern)
3944 (vect_recog_mask_conversion_pattern)
3945 (vect_recog_gather_scatter_pattern): Use it.
3946 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
3947 the "dr" and "stmt" fields of dr_vec_info instead of
3948 STMT_VINFO_DATA_REF.
3950 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3952 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
3953 (is_pattern_stmt_p): Use it.
3954 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
3955 on pattern statements.
3957 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3959 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
3960 original stmt as a stmt_vec_info rather than a gimple stmt.
3961 (vect_pattern_recog_1): Take the statement directly as a
3962 stmt_vec_info, rather than via a gimple_stmt_iterator.
3963 Update call to vect_mark_pattern_stmts.
3964 (vect_pattern_recog): Update calls accordingly.
3966 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3968 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
3969 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
3970 a vect_def_type for the first argument.
3971 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
3972 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
3973 operand if it isn't defined by a vectorized statement.
3974 (vect_build_gather_load_calls): Remove the mask_dt argument and
3975 update calls to vect_get_vec_def_for_stmt_copy.
3976 (vectorizable_bswap): Likewise the dt argument.
3977 (vectorizable_call): Update calls to vectorizable_bswap and
3978 vect_get_vec_def_for_stmt_copy.
3979 (vectorizable_simd_clone_call, vectorizable_assignment)
3980 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
3981 (vectorizable_comparison): Update calls to
3982 vect_get_vec_def_for_stmt_copy.
3983 (vectorizable_store): Likewise. Remove now-unnecessary calls to
3985 (vect_get_loop_based_defs): Remove dt argument and update call
3986 to vect_get_vec_def_for_stmt_copy.
3987 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
3988 and vect_get_vec_def_for_stmt_copy.
3989 (vectorizable_load): Update calls to vect_build_gather_load_calls
3990 and vect_get_vec_def_for_stmt_copy.
3991 * tree-vect-loop.c (vect_create_epilog_for_reduction)
3992 (vectorizable_reduction, vectorizable_live_operation): Update calls
3993 to vect_get_vec_def_for_stmt_copy.
3995 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3997 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
3998 and gimple stmt arguments with a stmt_vec_info.
3999 (vect_record_base_alignments): Update calls accordingly.
4000 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
4001 and gimple stmt arguments with a stmt_vec_info.
4002 (vect_build_slp_tree_1): Remove vinfo argument and update call
4003 to vect_record_max_nunits.
4004 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
4005 and vect_record_max_nunits.
4007 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4009 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
4010 file and take a stmt_vec_info instead of a gimple stmt.
4011 (supportable_widening_operation, vect_finish_replace_stmt)
4012 (vect_finish_stmt_generation, vect_get_store_rhs)
4013 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
4014 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
4015 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
4016 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
4017 (vect_create_data_ref_ptr, bump_vector_ptr)
4018 (vect_permute_store_chain, vect_setup_realignment)
4019 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
4020 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
4021 (vectorizable_reduction, vectorizable_induction)
4022 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
4023 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
4024 than gimple stmts as arguments.
4025 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
4026 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
4027 (can_group_stmts_p, vect_check_gather_scatter)
4028 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
4029 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
4030 (vect_permute_load_chain, vect_shift_permute_load_chain)
4031 (vect_transform_grouped_load)
4032 (vect_record_grouped_load_vectors): Likewise.
4033 * tree-vect-loop.c (vect_fixup_reduc_chain)
4034 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
4035 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
4036 (vectorizable_reduction, vectorizable_induction)
4037 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
4038 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
4039 (vect_get_load_store_mask): Likewise.
4040 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
4041 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
4042 * tree-vect-stmts.c (vect_mark_relevant)
4043 (is_simple_and_all_uses_invariant)
4044 (exist_non_indexing_operands_for_use_p, process_use)
4045 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
4046 (vect_get_vec_def_for_operand, vect_get_vec_defs)
4047 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
4048 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
4049 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
4050 (get_negative_load_store_type, get_load_store_type)
4051 (vect_check_load_store_mask, vect_check_store_rhs)
4052 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
4053 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4054 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4055 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4056 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
4057 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
4058 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
4059 (supportable_widening_operation): Likewise.
4061 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4063 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
4064 a stmt_vec_info instead of a gcall.
4065 (vect_check_gather_scatter): Update call accordingly.
4066 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
4068 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
4069 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
4070 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
4071 instead of a gimple stmt.
4072 (vect_transform_loop): Update calls accordingly.
4073 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
4074 stmt_vec_infos instead of gimple stmts.
4075 (vect_analyze_slp_instance): Update use accordingly.
4076 * tree-vect-stmts.c (read_vector_array, write_vector_array)
4077 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
4078 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
4079 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
4080 (vect_gen_widened_results_half, vect_get_loop_based_defs)
4081 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
4082 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
4083 down to subroutines.
4085 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4087 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
4088 of the worklist from a vector of gimple stmts to a vector of
4090 * tree-vect-stmts.c (vect_mark_relevant, process_use)
4091 (vect_mark_stmts_to_be_vectorized): Likewise
4093 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4095 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
4096 statement before passing it to vect_analyze_stmt.
4097 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
4098 the chain of phi vector definitions. Track the exit phi via its
4100 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
4101 STMT_VINFO_REDUC_DEF.
4102 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
4103 stmt_vec_infos to handle the statement chains.
4104 (vect_get_slp_defs): Record the first statement in the node
4105 using a stmt_vec_info.
4106 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
4107 statements here and pass their stmt_vec_info down to subroutines.
4108 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
4109 down to vect_finish_stmt_generation.
4110 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
4111 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
4112 stmt_vec_infos to subroutines.
4113 (vect_remove_stores): Use stmt_vec_infos to handle the statement
4116 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4118 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
4119 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
4120 (vect_permute_store_chain, vect_setup_realignment)
4121 (vect_permute_load_chain, vect_shift_permute_load_chain)
4122 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
4123 stmts internally, and when passing values to other vectorizer routines.
4124 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
4125 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
4126 (vect_analyze_loop_operations, get_initial_def_for_reduction)
4127 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
4128 (vectorizable_reduction, vectorizable_induction)
4129 (vectorizable_live_operation, vect_transform_loop_stmt)
4130 (vect_transform_loop): Likewise.
4131 * tree-vect-patterns.c (vect_reassociating_reduction_p)
4132 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
4133 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
4134 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4135 (vect_slp_analyze_node_operations_1): Likewise.
4136 * tree-vect-stmts.c (vect_mark_relevant, process_use)
4137 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
4138 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
4139 (vect_finish_stmt_generation_1, get_group_load_store_type)
4140 (get_load_store_type, vect_build_gather_load_calls)
4141 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4142 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4143 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4144 (vectorizable_store, vectorizable_load, vectorizable_condition)
4145 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
4146 (supportable_widening_operation): Likewise.
4147 (vect_get_vector_types_for_stmt): Likewise.
4148 * tree-vectorizer.h (vect_dr_behavior): Likewise.
4150 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4152 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
4153 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
4154 (vect_permute_store_chain, vect_permute_load_chain)
4155 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
4156 repeated stmt_vec_info lookups.
4157 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
4158 (vect_update_ivs_after_vectorizer): Likewise.
4159 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
4160 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
4161 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
4162 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4163 (vect_bb_slp_scalar_cost): Likewise.
4164 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
4166 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4168 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
4169 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
4170 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
4171 to get gassigns and gcalls, rather than operating on generc gimple
4173 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
4174 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
4175 (vectorizable_load, vect_analyze_stmt): Likewise.
4176 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
4178 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4180 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
4181 return stmt_vec_infos rather than gimple stmts. Do not accept
4183 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
4185 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
4186 Update use of get_later_stmt.
4187 (vect_get_constant_vectors): Update call accordingly.
4188 (vect_schedule_slp_instance): Likewise
4189 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
4190 (vect_slp_analyze_instance_dependence): Likewise.
4191 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
4193 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4195 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
4196 (stmt_info_for_cost::stmt_info): ...this new field.
4197 (add_stmt_costs): Update accordingly.
4198 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
4199 (vect_get_known_peeling_cost): Likewise.
4200 (vect_estimate_min_profitable_iters): Likewise.
4201 * tree-vect-stmts.c (record_stmt_cost): Likewise.
4203 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4205 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
4206 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4207 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
4209 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
4211 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4213 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
4214 a gimple stmt to a stmt_vec_info.
4215 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
4217 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4219 * tree-vectorizer.h (vec_info::grouped_stores): Change from
4220 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4221 (_loop_vec_info::reduction_chains): Likewise.
4222 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
4224 * tree-vect-slp.c (vect_analyze_slp): Likewise.
4226 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4228 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
4229 a gimple stmt to a stmt_vec_info.
4230 (_stmt_vec_info::next_element): Likewise.
4231 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
4232 (vect_slp_analyze_and_verify_node_alignment)
4233 (vect_analyze_group_access_1, vect_analyze_group_access)
4234 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
4235 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
4236 (vect_supportable_dr_alignment): Update accordingly.
4237 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
4238 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
4239 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
4240 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
4241 * tree-vect-slp.c (vect_build_slp_tree_1)
4242 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
4243 (vect_split_slp_store_group, vect_analyze_slp_instance)
4244 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
4245 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
4246 (get_group_load_store_type, get_load_store_type)
4247 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
4248 (vect_transform_stmt, vect_remove_stores): Likewise.
4250 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4252 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
4254 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
4255 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
4256 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
4257 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
4258 (vector_alignment_reachable_p, vect_get_data_access_cost)
4259 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
4260 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
4261 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
4262 (vect_analyze_group_access_1, vect_analyze_group_access)
4263 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
4264 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
4265 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
4266 result of vect_dr_stmt and use the stmt_vec_info instead of
4267 the associated gimple stmt.
4268 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
4269 (vect_gen_prolog_loop_niters): Likewise.
4270 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
4272 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4274 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
4275 to a vec<stmt_vec_info>.
4276 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
4277 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
4279 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
4280 to a vec<stmt_vec_info>.
4281 (bst_traits::value_type, bst_traits::value_type): Likewise.
4282 (bst_traits::hash): Update accordingly.
4283 (vect_get_and_check_slp_defs): Change the stmts parameter from
4284 a vec<gimple *> to a vec<stmt_vec_info>.
4285 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
4286 (vect_build_slp_tree): Likewise.
4287 (vect_build_slp_tree_2): Likewise. Update uses of
4288 SLP_TREE_SCALAR_STMTS.
4289 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
4290 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
4291 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
4292 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
4293 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
4294 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
4295 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
4296 (vect_get_constant_vectors, vect_get_slp_defs)
4297 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
4298 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
4299 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
4300 instead of gimple stmts.
4301 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
4302 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
4303 (vect_slp_analyze_instance_dependence): Update uses of
4304 SLP_TREE_SCALAR_STMTS.
4305 (vect_slp_analyze_and_verify_node_alignment): Likewise.
4306 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
4307 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
4308 (get_initial_defs_for_reduction): Likewise.
4309 (vect_create_epilog_for_reduction): Likewise.
4310 (vectorize_fold_left_reduction): Likewise.
4311 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
4312 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
4313 (can_vectorize_live_stmts): Likewise.
4315 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4317 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
4318 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4319 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
4321 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
4323 (gather_scalar_reductions): Update after above interface changes.
4324 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
4325 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
4327 (vect_force_simple_reduction): Likewise.
4328 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
4329 LOOP_VINFO_REDUCTIONS.
4330 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4332 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4334 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
4335 a gimple stmt to a stmt_vec_info.
4336 * tree-vect-loop.c (vect_active_double_reduction_p)
4337 (vect_force_simple_reduction, vectorizable_reduction): Update
4340 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4342 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
4343 vec<gimple *> to a vec<stmt_vec_info>.
4344 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
4345 the reduction_phis argument from a vec<gimple *> to a
4347 (vectorizable_reduction): Likewise the phis local variable that
4348 is passed to vect_create_epilog_for_reduction. Update for new type
4349 of SLP_TREE_VEC_STMTS.
4350 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
4351 (vectorizable_live_operation): Likewise.
4352 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
4353 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
4355 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4357 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
4358 a gimple stmt to a stmt_vec_info.
4359 (vectorizable_condition, vectorizable_live_operation)
4360 (vectorizable_reduction, vectorizable_induction): Pass back the
4361 vectorized statement as a stmt_vec_info.
4362 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
4363 use of STMT_VINFO_VEC_STMT.
4364 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
4365 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
4366 as stmt_vec_infos rather than gimple stmts.
4367 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
4369 (vectorizable_live_operation): Likewise.
4370 (vectorizable_reduction, vectorizable_induction): Likewise,
4371 updating use of STMT_VINFO_VEC_STMT.
4372 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
4373 of STMT_VINFO_VEC_STMT.
4374 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
4375 (vectorizable_simd_clone_call, vectorizable_conversion)
4376 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4377 (vectorizable_store, vectorizable_load, vectorizable_condition)
4378 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
4379 from a gimple stmt to a stmt_vec_info.
4380 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
4381 pointer to a stmt_vec_info to the vectorizable_* routines.
4383 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4385 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
4386 a gimple stmt to a stmt_vec_info.
4387 (is_pattern_stmt_p): Update accordingly.
4388 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
4389 (vect_record_grouped_load_vectors): Likewise.
4390 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
4391 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
4392 (vect_model_reduction_cost): Likewise.
4393 (vect_create_epilog_for_reduction): Likewise.
4394 (vectorizable_reduction, vectorizable_induction): Likewise.
4395 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4396 Return the stmt_vec_info for the pattern statement.
4397 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
4398 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
4399 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
4400 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
4401 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
4402 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
4403 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
4404 (free_stmt_vec_info, vect_is_simple_use): Likewise.
4406 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4408 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
4409 (vect_finish_stmt_generation): Likewise.
4410 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
4411 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
4412 (vect_build_gather_load_calls): Use the return value of the above
4413 functions instead of a separate call to vinfo_for_stmt. Use narrow
4414 scopes for the input gimple stmt and wider scopes for the associated
4415 stmt_vec_info. Use vec_info::lookup_def when setting these
4416 stmt_vec_infos from an SSA_NAME definition.
4417 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4418 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4419 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4420 (vectorizable_store, vectorizable_load, vectorizable_condition)
4421 (vectorizable_comparison): Likewise.
4422 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
4423 (vectorizable_reduction): Likewise.
4425 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4427 * tree-vectorizer.h (vect_is_simple_use): Add an optional
4428 stmt_vec_info * parameter before the optional gimple **.
4429 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
4430 (process_use, vect_get_vec_def_for_operand_1): Update callers.
4431 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
4432 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4433 (vectorizable_live_operation): Likewise.
4434 * tree-vect-patterns.c (type_conversion_p): Likewise.
4435 (vect_look_through_possible_promotion): Likewise.
4436 (vect_recog_rotate_pattern): Likewise.
4437 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
4439 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4441 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
4442 a typedef to a wrapper class.
4443 (NULL_STMT_VEC_INFO): New macro.
4444 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
4445 (stmt_vec_info::operator*): New function.
4446 (stmt_vec_info::operator gimple *): Likewise.
4447 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
4448 (add_stmt_costs): Likewise.
4449 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
4450 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
4451 (vect_get_known_peeling_cost): Likewise.
4452 (vect_estimate_min_profitable_iters): Likewise.
4453 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4454 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
4455 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
4456 (vectorizable_store, free_stmt_vec_infos): Likewise.
4457 (new_stmt_vec_info): Change return type of xcalloc to
4460 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4462 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
4463 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
4464 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
4465 a single_imm_use-based sequence.
4466 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
4468 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4470 * tree-vectorizer.h (vec_info::lookup_def): Declare.
4471 * tree-vectorizer.c (vec_info::lookup_def): New function.
4472 * tree-vect-patterns.c (vect_get_internal_def): Use it.
4473 (vect_widened_op_tree): Likewise.
4474 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
4475 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
4476 (vectorizable_reduction): Likewise.
4477 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
4479 (vect_is_slp_reduction): Update calls accordingly. Use
4480 vec_info::lookup_def.
4481 (vect_is_simple_reduction): Likewise
4482 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
4484 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4486 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
4487 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
4488 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
4490 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
4491 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
4492 (vect_update_vf_for_slp, vect_analyze_loop_operations)
4493 (vect_is_slp_reduction, vectorizable_induction)
4494 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
4495 * tree-vect-patterns.c (vect_init_pattern_stmt):
4496 (vect_determine_min_output_precision_1, vect_determine_precisions)
4497 (vect_pattern_recog): Likewise.
4498 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
4499 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
4500 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
4501 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
4502 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
4503 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
4504 info field from a loop to a loop_vec_info.
4506 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4508 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
4509 (vec_info::add_stmt): Declare.
4510 * tree-vectorizer.c (vec_info::add_stmt): New function.
4511 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
4512 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
4513 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
4514 (vectorizable_induction): Likewise.
4515 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
4516 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
4517 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
4518 (vectorizable_load): Likewise.
4519 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4520 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
4521 (vect_recog_gather_scatter_pattern): Likewise.
4522 (append_pattern_def_seq): Likewise. Remove a check that is
4523 performed by add_stmt itself.
4525 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4527 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
4528 which make_ssa_name was called with new_stmt before new_stmt
4531 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4533 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
4535 (vect_is_slp_reduction): ...here...
4536 (vect_is_simple_reduction): ...and here. Remove repetition of tests
4537 that are already known to be false.
4539 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4541 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
4542 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
4543 STMT_VINFO_NUM_SLP_USES when it's true.
4544 (vect_free_slp_instance): Add a final_p parameter and pass it to
4546 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
4547 (vect_analyze_slp_instance): Likewise.
4548 (vect_slp_analyze_operations): Likewise.
4549 (vect_slp_analyze_bb_1): Likewise.
4550 * tree-vectorizer.c (vec_info): Likewise.
4551 * tree-vect-loop.c (vect_transform_loop): Likewise.
4553 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4555 * tree-vect-loop.c (vectorizable_reduction): Assert that the
4556 function is not called for second and subsequent members of
4559 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4561 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
4562 cases for nested loops from here to ...
4563 (vect_create_epilog_for_reduction): ...here. Only call
4564 vect_is_simple_use for inner-loop reductions.
4566 2018-07-31 Martin Liska <mliska@suse.cz>
4568 PR gcov-profile/85338
4569 PR gcov-profile/85350
4570 PR gcov-profile/85372
4571 * profile.c (struct location_triplet): New.
4572 (struct location_triplet_hash): Likewise.
4573 (output_location): Do not output a BB that
4574 is already recorded for a line.
4575 (branch_prob): Use streamed_locations.
4577 2018-07-31 Martin Liska <mliska@suse.cz>
4579 PR gcov-profile/85370
4580 * coverage.c (coverage_begin_function): Do not mark target
4581 clones as artificial functions.
4583 2018-07-31 Martin Liska <mliska@suse.cz>
4585 PR gcov-profile/83813
4586 PR gcov-profile/84758
4587 PR gcov-profile/85217
4588 PR gcov-profile/85332
4589 * profile.c (branch_prob): Do not record GOTO expressions
4590 for GIMPLE statements which locations are already streamed.
4592 2018-07-31 Olivier Hainque <hainque@adacore.com>
4594 * gcc.c (handle_spec_function): Accept a soft_matched_part
4595 argument, as do_spec_1. Pass it down to ...
4596 (eval_spec_function): Accept a soft_matched_part argument,
4597 and pass it down to ...
4598 (do_spec_2): Accept a soft_matched_part argument, and pass
4599 it down to do_spec_1.
4600 (do_spec_1): Pass soft_matched_part to handle_spec_function.
4601 (handle_braces): Update call to handle_spec_function.
4602 (driver::set_up_specs): Update calls to do_spec_2.
4603 (compare_debug_dump_opt_spec_function): Likewise.
4604 (compare_debug_self_opt_spec_function): Likewise.
4606 2018-07-31 Olivier Hainque <hainque@adacore.com>
4608 * common.opt (nolibc): New option.
4609 * doc/invoke.texi (Link Options): Document it.
4610 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
4611 * config/alpha/linux.h: Likewise.
4612 * config/arc/elf.h: Likewise.
4613 * config/arm/uclinux-elf.h: Likewise.
4614 * config/arm/unknown-elf.h: Likewise.
4615 * config/avr/avrlibc.h: Likewise.
4616 * config/bfin/bfin.h: Likewise.
4617 * config/bfin/linux.h: Likewise.
4618 * config/bfin/uclinux.h: Likewise.
4619 * config/darwin.h: Likewise.
4620 * config/darwin10.h: Likewise.
4621 * config/darwin12.h: Likewise.
4622 * config/gnu-user.h: Likewise.
4623 * config/lm32/uclinux-elf.h: Likewise.
4624 * config/pa/pa-hpux11.h: Likewise.
4625 * config/pa/pa64-hpux.h: Likewise.
4626 * config/sparc/sparc.h: Likewise.
4628 2018-07-31 Olivier Hainque <hainque@adacore.com>
4630 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
4631 undefined variables.
4633 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
4636 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
4639 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
4641 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
4642 terminated string literal.
4644 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
4646 PR rtl-optimization/85160
4647 * combine.c (is_just_move): New function.
4648 (try_combine): Allow combining two instructions into two if neither of
4649 the original instructions was a move.
4651 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
4654 * doc/extend.texi (Global Register Variables): Discourage use of type
4656 (Local Register Variables): Likewise.
4658 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
4660 PR tree-optimization/86506
4661 * hwint.h (ceil_log2): Resync with hwint.c implementation.
4663 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
4666 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
4667 hard_regno, make sure no insn between `from` and `to` clobbers it.
4669 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
4670 Tom de Vries <tdevries@suse.de>
4672 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
4673 (PTX_DEFAULT_RUNTIME_DIM): ... this.
4674 (nvptx_goacc_validate_dims): Set default worker and gang dims to
4675 PTX_DEFAULT_RUNTIME_DIM.
4676 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
4678 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
4680 * config/pa/pa.c (pa_output_addr_vec): Align address table.
4681 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
4682 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
4684 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
4686 * config/rs6000/constraints.md (wG constraint): Delete, no longer
4688 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
4689 predicate to reflect toc fusion has been deleted.
4690 (toc_fusion_mem_raw): Delete, no longer used.
4691 (toc_fusion_mem_wrapped): Likewise.
4692 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
4694 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
4695 Delete, no longer used.
4696 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
4697 meant to be used for toc fusion.
4698 (rs6000_debug_print_mode): Delete toc fusion debugging.
4699 (rs6000_debug_reg_global): Likewise.
4700 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
4701 fusion and secondary reload support that were never used.
4702 (rs6000_option_override_internal): Delete TOC fusion, that was only
4703 partially defined, and it did not work unless you also used the
4705 (rs6000_legitimate_address_p): Delete TOC fusion support.
4706 (rs6000_opt_masks): Likewise.
4707 (fusion_wrap_memory_address): Delete function, no longer used.
4708 (fusion_split_address); Delete TOC fusion support.
4709 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
4710 longer used with toc fusion being deleted.
4711 (TARGET_TOC_FUSION_FP): Likewise.
4712 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
4714 (toc fusion spliter): Delete TOC fusion support.
4715 (toc_fusionload_<mode>): Likewise.
4716 (toc_fusionload_di): Likewise.
4717 (fusion_gpr_load_<mode>): Delete generator function, this insn no
4718 longer needs to be named. Rename predicate to delete TOC fusion.
4719 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
4720 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
4721 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
4722 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
4723 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
4725 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
4727 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
4728 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
4729 __int128 in built-in function prototypes.
4730 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
4731 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
4733 2018-07-27 Martin Sebor <msebor@redhat.com>
4735 PR tree-optimization/86696
4736 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
4737 types, including enums.
4738 (handle_char_store): Be prepared for the above function to fail.
4740 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
4742 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
4743 when optimization level is lower than 2 or optimize for size.
4745 2018-07-26 Martin Sebor <msebor@redhat.com>
4747 PR tree-optimization/86043
4748 PR tree-optimization/86042
4749 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
4750 (get_string_cst_length): Rename...
4751 (get_min_string_length): ...to this. Add argument.
4752 (handle_char_store): Extend to handle multi-character stores by
4754 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
4755 * tree.h (initializer_zerop): Add argument.
4757 2018-07-26 Jakub Jelinek <jakub@redhat.com>
4760 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
4761 declare target to variables if they have always,{to,from,tofrom} map
4764 2018-07-26 Martin Liska <mliska@suse.cz>
4767 * lto-wrapper.c: Add linker_output as prefix
4768 for ltrans_output_file.
4770 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
4772 PR rtl-optimization/85805
4773 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
4774 value for hard registers if that was written in the same mode.
4776 2018-07-26 Martin Liska <mliska@suse.cz>
4778 PR gcov-profile/86536
4779 * gcov.c (format_gcov): Use printf format %.*f directly
4780 and do not handle special values.
4782 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4784 * common/config/arc/arc-common.c (arc_option_optimization_table):
4785 Update default optimizations for size.
4787 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4789 * config/arc/arc.md (movsf_insn): Add short instruction selection.
4790 * config/arc/constraints.md (CfZ): New constraint.
4791 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
4792 (subsf3_fpu): Likewise.
4793 (cmpsf_fpu): Likewise.
4794 (cmpsf_fpu_uneq): Likewise.
4796 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4798 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
4800 (arc_is_uncached_mem_p): uncached applies to both the variable and
4803 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4805 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
4808 2018-07-25 David Malcolm <dmalcolm@redhat.com>
4810 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
4811 field "m_scopes" from vec to auto_vec.
4813 2018-07-25 Martin Liska <mliska@suse.cz>
4815 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
4818 2018-07-25 Richard Biener <rguenther@suse.de>
4821 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
4822 special wrt context_die late.
4823 (gen_subprogram_die): Re-use DIEs in local scope.
4825 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
4827 PR tree-optimization/86644
4828 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
4830 2018-07-25 Martin Liska <mliska@suse.cz>
4833 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
4834 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
4836 2018-07-25 Martin Liska <mliska@suse.cz>
4839 * params.def: Explain ASan abbreviation and provide
4840 a documentation link.
4842 2018-07-24 Martin Sebor <msebor@redhat.com>
4844 PR tree-optimization/86622
4845 PR tree-optimization/86532
4846 * builtins.h (string_length): Declare.
4847 * builtins.c (c_strlen): Correct handling of non-constant offsets.
4848 (check_access): Be prepared for non-constant length ranges.
4849 (string_length): Make extern.
4850 * expr.c (string_constant): Only handle the minor non-constant
4851 array index. Use string_constant to compute the length of
4852 a generic string constant.
4854 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
4856 PR tree-optimization/86618
4857 * tree-vect-stmts.c (vectorizable_call): Don't take the address
4858 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
4860 2018-07-24 David Malcolm <dmalcolm@redhat.com>
4862 PR tree-optimization/86636
4863 * json.cc (json::object::set): Fix comment. Add assertions.
4864 (json::array::append): Move here from json.h. Add comment and an
4866 (json::string::string): Likewise.
4867 * json.h (json::array::append): Move to json.cc.
4868 (json::string::string): Likewise.
4869 * optinfo-emit-json.cc
4870 (optrecord_json_writer::impl_location_to_json): Assert that we
4871 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
4872 wrapper around it. Expand the location once, rather than three
4874 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
4875 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
4877 (optrecord_json_writer::optinfo_to_json): Likewise, in four
4878 places. Fix some overlong lines.
4880 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
4882 * config/aarch64/aarch64-simd.md
4883 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
4884 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
4885 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
4886 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
4887 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
4888 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
4889 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
4890 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
4891 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
4893 2018-07-24 Jakub Jelinek <jakub@redhat.com>
4896 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
4897 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
4898 and abs_d == d, do the power of two handling if profitable.
4900 2018-07-24 Richard Biener <rguenther@suse.de>
4902 * match.pd: Add BIT_FIELD_REF canonicalizations.
4904 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4907 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
4909 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4911 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
4912 terminated STRING_CST object.
4914 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4916 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
4918 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
4920 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
4921 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
4922 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
4923 the elements into a register.
4924 (rs6000_split_v4si_init_di_reg): Delete.
4925 (rs6000_split_v4si_init): Delete.
4926 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
4927 (vsx_init_v4si): Rewrite as a define_expand.
4929 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
4931 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
4932 zero_extend argument from memory): New.
4934 2018-07-22 Martin Sebor <msebor@redhat.com>
4937 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
4938 diagnosing calls with unknown arguments unless -Walloca-larger-than
4939 is restricted to less than PTRDIFF_MAX bytes.
4941 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
4943 * doc/gcov.texi (Invoking Gcov): Editorial changes.
4945 2018-07-20 David Malcolm <dmalcolm@redhat.com>
4947 * pretty-print.c (text_info::set_location): Remove redundant
4948 "line_table" parameter from call to rich_location::set_range.
4950 2018-07-20 Martin Sebor <msebor@redhat.com>
4953 * builtins.c (expand_builtin_alloca): Adjust.
4954 * calls.c (alloc_max_size): Simplify.
4955 * cgraphunit.c (cgraph_node::expand): Adjust.
4956 * common.opt (larger_than_size, warn_frame_larger_than): Remove
4958 (frame_larger_than_size): Same.
4959 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
4960 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
4962 * doc/invoke.texi (GCC Command Options): Document option arguments.
4963 Explain byte-size arguments and suffixes.
4964 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
4965 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
4966 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
4967 * doc/options.texi (UInteger): Expand.
4968 (Host_Wide_Int, ByteSize): Document new properties.
4969 * final.c (final_start_function_1): Include sizes in an error message.
4970 * function.c (frame_offset_overflow): Same.
4971 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
4972 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
4973 Diagnose unbounded alloca calls only for limits of less than
4975 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
4976 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
4978 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
4979 only for limits of less than PTRDIFF_MAX.
4980 * langhooks-def.h (lhd_handle_option): Change function argument
4982 * langhooks.c (lhd_handle_option): Same.
4983 * langhooks.h (handle_option): Same.
4984 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
4986 (var_type, var_type_struct): Same.
4987 (var_set): Handle ByteSize flag.
4988 * optc-gen.awk: Add comments to output to ease debugging. Make
4989 use of HOST_WIDE_INT where appropriate.
4990 * opts-gen-save.awk: Use %lx to format unsigned long.
4991 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
4992 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
4993 arguments. Parse bytes-size suffixes.
4994 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
4995 (enum_value_to_arg): Same.
4996 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
4997 (handle_option): Adjust.
4998 (generate_option): Change function argument to HOST_WIDE_INT.
4999 (cmdline_handle_error): Adjust.
5000 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
5001 (set_option): Change function argument to HOST_WIDE_INT.
5002 (option_enabled): Handle cl_host_wide_int.
5003 (get_option_state): Handle CLVC_SIZE.
5004 (control_warning_option): Same.
5005 * opts.c (common_handle_option): Change function argument to
5006 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
5007 OPT_Wvla_larger_than_.
5008 * opts.h (enum cl_var_type): Add an enumerator.
5009 * stor-layout.c (layout_decl): Print a more meaningful warning.
5010 * toplev.c (output_stack_usage): Adjust.
5012 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
5014 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
5015 call to inline_expand_builtin_string_cmp.
5016 (expand_builtin_strcmp): Likewise.
5017 (expand_builtin_strncmp): Likewise.
5018 (inline_string_cmp): Delete the last parameter, change char_type_node
5019 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
5021 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
5022 the inlining expansion on target where the type of the call has same or
5023 narrower precision than unsigned char.
5025 2018-07-20 David Malcolm <dmalcolm@redhat.com>
5027 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
5028 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
5029 * common.opt (fsave-optimization-record): New option.
5030 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
5031 * doc/invoke.texi (-fsave-optimization-record): New option.
5032 * dumpfile.c: Include "optinfo-emit-json.h".
5033 (struct kv_pair): Move to coretypes.h.
5034 (optgroup_options): Make non-static.
5035 (dump_context::end_scope): Call
5036 optimization_records_maybe_pop_dump_scope.
5037 * dumpfile.h (optgroup_options): New decl.
5038 * json.cc: New file.
5040 * optinfo-emit-json.cc: New file.
5041 * optinfo-emit-json.h: New file.
5042 * optinfo.cc: Include "optinfo-emit-json.h".
5043 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
5044 (optinfo_enabled_p): Check optimization_records_enabled_p.
5045 (optinfo_wants_inlining_info_p): Likewise.
5046 * optinfo.h: Update comment.
5047 * profile-count.c (profile_quality_as_string): New function.
5048 * profile-count.h (profile_quality_as_string): New decl.
5049 (profile_count::quality): New accessor.
5050 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
5051 and optinfo_emit_json_cc_tests.
5052 * selftest.h (selftest::json_cc_tests): New decl.
5053 (selftest::optinfo_emit_json_cc_tests): New decl.
5054 * toplev.c: Include "optinfo-emit-json.h".
5055 (compile_file): Call optimization_records_finish.
5056 (do_compile): Call optimization_records_start.
5057 * tree-ssa-live.c: Include optinfo.h.
5058 (remove_unused_scope_block_p): Retain inlining information if
5059 optinfo_wants_inlining_info_p returns true.
5061 2018-07-20 Richard Biener <rguenther@suse.de>
5064 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
5065 to cover -flto-partition=none.
5067 2018-07-20 Martin Liska <mliska@suse.cz>
5069 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
5070 (get_decl_source_range): Remove unused function.
5072 2018-07-20 Richard Biener <rguenther@suse.de>
5074 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
5075 (struct vn_phi_s): Likewise.
5076 (struct vn_reference_s): Likewise.
5077 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
5078 for searching the slot of an entry known to be in the hash itself.
5079 (vn_phi_hasher::equal): Likewise.
5080 (vn_reference_hasher::equal): Likewise.
5081 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
5083 (optimistic_info, current_info): Remove, keeping only valid_info.
5084 (vn_reference_lookup_1): Remove fallback lookup.
5085 (vn_reference_lookup_2): Likewise.
5086 (vn_nary_op_lookup_1): Likewise.
5087 (vn_phi_lookup): Likewise.
5088 (vn_nary_build_or_lookup_1): Make sure to not chain the built
5090 (vn_reference_insert): Adjust, chain the inserted hash element
5091 at last_inserted_ref.
5092 (vn_reference_insert_pieces): Likewise.
5093 (visit_reference_op_call): Likewise.
5094 (vn_nary_op_insert_into): Chain the inserted hash element at
5096 (vn_nary_op_insert_pieces): Adjust.
5097 (vn_nary_op_insert): Likewise.
5098 (vn_nary_op_insert_stmt): Likewise.
5099 (vn_phi_insert): Adjust, chain the inserted hash element at
5101 (process_scc): Remove clearing and copying the optimistic
5102 table. Instead remove elements inserted during an optimistic
5103 iteration from the single table we maintain.
5104 (init_scc_vn): Adjust.
5105 (free_scc_vn): Likewise.
5106 (sccvn_dom_walker::record_cond): Likewise.
5107 (sccvn_dom_walker::after_dom_children): Likewise.
5109 2018-07-19 Martin Sebor <msebor@redhat.com>
5111 PR tree-optimization/84047
5112 PR tree-optimization/83776
5113 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
5114 (check_array_bounds): Call it.
5116 2018-07-19 Martin Sebor <msebor@redhat.com>
5118 * align.h (align_flags): Use member initialization.
5120 2018-07-19 David Malcolm <dmalcolm@redhat.com>
5122 * Makefile.in (OBJS): Add optinfo.o.
5123 * coretypes.h (class symtab_node): New forward decl.
5124 (struct cgraph_node): New forward decl.
5125 (class varpool_node): New forward decl.
5126 * dump-context.h: New file.
5127 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
5129 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
5130 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
5131 (set_alt_dump_file): Likewise.
5132 (dump_context::~dump_context): New dtor.
5133 (dump_gimple_stmt): Move implementation to...
5134 (dump_context::dump_gimple_stmt): ...this new member function.
5135 Add the stmt to any pending optinfo, creating one if need be.
5136 (dump_gimple_stmt_loc): Move implementation to...
5137 (dump_context::dump_gimple_stmt_loc): ...this new member function.
5138 Start a new optinfo and add the stmt to it.
5139 (dump_gimple_expr): Move implementation to...
5140 (dump_context::dump_gimple_expr): ...this new member function.
5141 Add the stmt to any pending optinfo, creating one if need be.
5142 (dump_gimple_expr_loc): Move implementation to...
5143 (dump_context::dump_gimple_expr_loc): ...this new member function.
5144 Start a new optinfo and add the stmt to it.
5145 (dump_generic_expr): Move implementation to...
5146 (dump_context::dump_generic_expr): ...this new member function.
5147 Add the tree to any pending optinfo, creating one if need be.
5148 (dump_generic_expr_loc): Move implementation to...
5149 (dump_context::dump_generic_expr_loc): ...this new member
5150 function. Add the tree to any pending optinfo, creating one if
5152 (dump_printf): Move implementation to...
5153 (dump_context::dump_printf_va): ...this new member function. Add
5154 the text to any pending optinfo, creating one if need be.
5155 (dump_printf_loc): Move implementation to...
5156 (dump_context::dump_printf_loc_va): ...this new member function.
5157 Start a new optinfo and add the stmt to it.
5158 (dump_dec): Move implementation to...
5159 (dump_context::dump_dec): ...this new member function. Add the
5160 value to any pending optinfo, creating one if need be.
5161 (dump_context::dump_symtab_node): New member function.
5162 (dump_context::get_scope_depth): New member function.
5163 (dump_context::begin_scope): New member function.
5164 (dump_context::end_scope): New member function.
5165 (dump_context::ensure_pending_optinfo): New member function.
5166 (dump_context::begin_next_optinfo): New member function.
5167 (dump_context::end_any_optinfo): New member function.
5168 (dump_context::s_current): New global.
5169 (dump_context::s_default): New global.
5170 (dump_scope_depth): Delete global.
5171 (dumpfile_ensure_any_optinfo_are_flushed): New function.
5172 (dump_symtab_node): New function.
5173 (get_dump_scope_depth): Reimplement in terms of dump_context.
5174 (dump_begin_scope): Likewise.
5175 (dump_end_scope): Likewise.
5176 (selftest::temp_dump_context::temp_dump_context): New ctor.
5177 (selftest::temp_dump_context::~temp_dump_context): New dtor.
5178 (selftest::verify_item): New function.
5179 (ASSERT_IS_TEXT): New macro.
5180 (ASSERT_IS_TREE): New macro.
5181 (ASSERT_IS_GIMPLE): New macro.
5182 (selftest::test_capture_of_dump_calls): New test.
5183 (selftest::dumpfile_c_tests): Call it.
5184 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
5185 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
5186 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
5187 descriptive comment.
5188 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
5189 (dump_node, dump_bb): Move these unrelated decls.
5190 (class dump_manager): Add leading comment.
5191 * optinfo.cc: New file.
5192 * optinfo.h: New file.
5194 2018-07-19 Michael Collison <michael.collison@arm.com>
5195 Richard Henderson <rth@redhat.com>
5197 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
5198 (subti): Handle op1 zero.
5199 (subvti4, usub4ti4): New.
5200 (*sub<GPI>3_compare1_imm): New.
5201 (sub<GPI>3_carryinCV): New.
5202 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
5203 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
5205 2018-07-19 Michael Collison <michael.collison@arm.com>
5206 Richard Henderson <rth@redhat.com>
5208 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
5209 (addti3): Create simpler code if low part is already known to be 0.
5210 (addvti4, uaddvti4): New.
5211 (*add<GPI>3_compareC_cconly_imm): New.
5212 (*add<GPI>3_compareC_cconly): New.
5213 (*add<GPI>3_compareC_imm): New.
5214 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
5215 handle constants within this pattern..
5216 (*add<GPI>3_compareV_cconly_imm): New.
5217 (*add<GPI>3_compareV_cconly): New.
5218 (*add<GPI>3_compareV_imm): New.
5219 (add<GPI>3_compareV): New.
5220 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
5221 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
5222 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
5223 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
5224 with 'comparison' operator.
5225 (*add<GPI>3_compareV_cconly_imm): Ditto.
5226 (*add<GPI>3_compareV_cconly): Ditto.
5227 (*add<GPI>3_compareV_imm): Ditto.
5228 (add<GPI>3_compareV): Ditto.
5229 (add<mode>3_carryinC): Ditto.
5230 (*add<mode>3_carryinC_zero): Ditto.
5231 (*add<mode>3_carryinC): Ditto.
5232 (add<mode>3_carryinV): Ditto.
5233 (*add<mode>3_carryinV_zero): Ditto.
5234 (*add<mode>3_carryinV): Ditto.
5236 2018-07-19 Michael Collison <michael.collison@arm.com>
5237 Richard Henderson <rth@redhat.com>
5239 * config/aarch64/aarch64-modes.def (CC_V): New.
5240 * config/aarch64/aarch64-protos.h
5241 (aarch64_addti_scratch_regs): Declare
5242 (aarch64_subvti_scratch_regs): Declare.
5243 (aarch64_expand_subvti): Declare.
5244 (aarch64_gen_unlikely_cbranch): Declare
5245 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
5246 for signed overflow using CC_Vmode.
5247 (aarch64_get_condition_code_1): Handle CC_Vmode.
5248 (aarch64_gen_unlikely_cbranch): New function.
5249 (aarch64_addti_scratch_regs): New function.
5250 (aarch64_subvti_scratch_regs): New function.
5251 (aarch64_expand_subvti): New function.
5253 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5255 * config/aarch64/aarch64-option-extensions.def: New entry for profile
5257 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
5258 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
5261 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5264 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
5265 address check not strict.
5267 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5269 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
5271 (store_pair_lanes<mode>): Likewise.
5272 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
5273 enum value 'ADDR_QUERY_LDP_STP_N'.
5274 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
5275 (aarch64_print_address_internal): Add declaration.
5276 (aarch64_print_ldpstp_address): Remove.
5277 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
5278 (aarch64_print_operand): Change printing of 'y'.
5279 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
5280 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
5281 'true' rather than '1'.
5282 * config/aarch64/constraints.md (Uml): Likewise.
5283 (Uml): Rename to Umn.
5286 2018-07-19 Richard Biener <rguenther@suse.de>
5288 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
5290 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
5291 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
5292 (vn_reference_hasher): Likewise.
5293 (struct vn_tables_s): Remove obstack and alloc-pool members.
5294 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
5295 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
5296 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
5297 (vn_reference_insert_pieces): Likewise.
5298 (alloc_vn_nary_op_noinit): Adjust.
5299 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
5300 (vn_phi_eq): Adjust.
5301 (shared_lookup_phiargs): Remove.
5302 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
5303 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
5304 (visit_reference_op_call): Likewise.
5305 (copy_nary, copy_phi, copy_reference): Remove.
5306 (process_scc): Rewind the obstack when iterating. Do not
5307 copy the elements to valid_info but just move them from one
5308 hashtable to the other.
5309 (allocate_vn_table): Adjust.
5310 (free_vn_table): Likewise.
5311 (init_scc_vn): Likewise.
5312 (free_scc_vn): Likewise.
5314 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
5317 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
5318 indirect_return as function type attribute.
5319 (ix86_attribute_table): Change indirect_return to function
5321 * doc/extend.texi: Update indirect_return attribute.
5323 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
5325 * wide-int.h (widest2_int): New.
5326 * gimple-fold.c (arith_overflowed_p): Use it.
5327 * tree.h (widest2_int_cst): New.
5328 * tree-vrp.c (wide_int_binop_overflow): Rename from
5329 vrp_int_const_binop.
5330 Rewrite to work on trees.
5331 (extract_range_from_multiplicative_op_1): Abstract code to...
5332 (wide_int_range_min_max): ...here.
5333 (wide_int_range_cross_product): ...and here.
5334 (extract_range_from_binary_expr_1): Abstract overflow code to...
5335 (wide_int_range_mult_wrapping): ...here.
5336 * tree-vrp.h (wide_int_range_cross_product): New.
5337 (wide_int_range_mult_wrapping): New.
5339 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
5340 Julia Koval <julia.koval@intel.com>
5342 * config/i386/x86-tune-costs.h (skylake_memcpy,
5343 skylake_memset): Replace rep_prefix with unrolling for size 512.
5345 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
5348 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
5349 comparision with EQ_EXPR in last stmt.
5351 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
5353 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
5354 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
5355 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
5356 previously known as "PowerPC AltiVec Built-in Functions". Move
5357 some material to new subsubsections "PowerPC AltiVec Built-in
5358 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
5360 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
5361 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
5362 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
5363 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
5365 2018-07-18 Richard Biener <rguenther@suse.de>
5367 PR tree-optimization/86557
5368 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
5371 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
5373 * config/s390/s390.c (s390_function_profiler): Generate CFI.
5375 2018-07-17 Jeff Law <law@redhat.com>
5377 * config/arm/arm.c (get_label_padding): Update for recent
5378 changes to label_to_alignment.
5380 PR tree-optimization/86010
5381 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
5383 * config/mips/mips.c (vr4130_align_insns): Update for recent
5384 changes to label_to_alignment.
5386 * config/frv/frv.c (frv_label_align): Update for recent changes
5387 to label_to_alignment.
5389 * config/nios2/nios2.c (nios2_label_align): Update for recent
5390 changes which dropped ALIGN_LABELS_LOG.
5392 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
5394 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
5395 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
5397 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5399 * config/arc/arc.c (arc_label_align): Use align_labels instead of
5400 deprecated align_labels_log.
5402 2018-07-17 Richard Biener <rguenther@suse.de>
5405 * dwarf2out.c (init_sections_and_labels): Always generate
5406 a debug_line_str_section for early LTO debug.
5407 (dwarf2out_finish): Reset debug_line_str_hash output early.
5408 Bump counter for extra dwarf5 .debug_loc labels to not conflict
5410 (dwarf2out_early_finish): Output debug_line_str.
5412 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
5414 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
5415 index register on z196 or later.
5417 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
5419 * config/s390/s390.c (s390_default_align): Set default function
5421 (s390_override_options_after_change): Call s390_default align.
5422 (s390_option_override_internal): Call s390_default_align.
5423 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
5425 2018-07-17 Jakub Jelinek <jakub@redhat.com>
5428 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
5429 to _looptemp_ clauses, other than the first two.
5431 2018-07-17 Martin Liska <mliska@suse.cz>
5433 * opts.c: Do not enable OPT_falign_* for -Os.
5435 2018-07-17 Martin Liska <mliska@suse.cz>
5437 * align.h (MAX_CODE_ALIGN): New.
5438 (MAX_CODE_ALIGN_VALUE): New.
5439 * common/config/i386/i386-common.c (ix86_handle_option):
5440 (MAX_CODE_ALIGN): Moved to align.h.
5441 * final.c (MAX_CODE_ALIGN): Likewise.
5442 * opts.c (parse_and_check_align_values):
5443 (MAX_CODE_ALIGN): Likewise.
5444 (MAX_CODE_ALIGN_VALUE): Likewise.
5446 2018-07-17 Martin Liska <mliska@suse.cz>
5448 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
5449 in order to fulfil coding style.
5450 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
5451 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
5452 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
5453 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
5454 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
5455 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
5456 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
5457 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
5458 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
5460 2018-07-17 Martin Liska <mliska@suse.cz>
5462 * align.h: New file.
5463 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
5465 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
5466 align_flags of label_to_alignment.
5467 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
5469 * config/m68k/m68k.c: Do not use removed align_labels_value and
5471 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
5472 (LOOP_ALIGN): Likewise.
5473 (LABEL_ALIGN): Likewise.
5474 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
5475 Remove not used macro.
5476 (rs6000_loop_align): Change return type to align_flags.
5477 (rs6000_loop_align_max_skip): Remove.
5478 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
5479 Change return type to align_flags.
5480 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
5481 Remove not used macro.
5482 (rs6000_loop_align): Change return type to align_flags.
5483 (rs6000_loop_align_max_skip): Remove.
5484 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
5485 * config/rx/rx-protos.h (rx_align_for_label): Make it
5487 * config/rx/rx.c (rx_align_for_label): Change return type
5489 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
5491 into align_flags class.
5492 (LABEL_ALIGN): Likewise.
5493 (LOOP_ALIGN): Likewise.
5494 * config/s390/s390.c (s390_label_align): Use align_flags
5496 (s390_asm_output_function_label): Likewise.
5497 * config/sh/sh.c (sh_override_options_after_change):
5498 Use align_flags class directly without macros.
5499 (find_barrier): Likewise.
5500 (barrier_align): Likewise.
5501 (sh_loop_align): Likewise.
5502 * config/spu/spu.c (spu_option_override):
5503 Use align_flags_tuple::get_value instead of removed macros.
5504 (spu_sched_init): Likewise.
5505 * config/spu/spu.h (GTY): Likewise.
5506 * config/visium/visium.c (visium_option_override):
5507 Set "8" as default secondary alignment.
5508 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
5509 in order to guarantee secondary alignment of 8.
5510 * coretypes.h: Include align.h header file.
5511 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
5512 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
5513 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
5514 * doc/tm.texi.in: Likewise.
5515 * final.c (struct label_alignment): Remove not used structure.
5516 (LABEL_ALIGN): Change type to align_flags.
5517 (LOOP_ALIGN): Likewise.
5518 (JUMP_ALIGN): Likewise.
5519 (default_loop_align_max_skip): Remove.
5520 (default_label_align_max_skip): Likewise.
5521 (default_jump_align_max_skip): Likewise.
5522 (default_label_align_after_barrier_max_skip):
5523 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
5524 (LABEL_TO_MAX_SKIP): Remove.
5525 (label_to_alignment): Return align_flags type instead of integer.
5526 (label_to_max_skip): Remove.
5527 (align_fuzz): Use align_flags type.
5528 (compute_alignments): Use align_flags type and use align_flags::max
5529 to combine multiple alignments.
5530 (grow_label_align): Grow vec instead of C array.
5531 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
5532 (shorten_branches): Use align_flags type and use align_flags::max
5533 to combine multiple alignments.
5534 (final_scan_insn_1): Remove usage of secondary alignment that comes
5535 from label alignment, but instead use proper secondary alignment
5536 which is computed in grow_label_align.
5537 * flags.h (struct align_flags_tuple): Move to align.h.
5538 (struct align_flags): Likewise.
5539 (state_align_loops): Rename to align_loops.
5540 (state_align_jumps): Rename to align_jumps.
5541 (state_align_labels): Rename to align_labels.
5542 (state_align_functions): Rename to align_functions.
5543 (align_loops_log): Remove.
5544 (align_jumps_log): Remove.
5545 (align_labels_log): Remove.
5546 (align_functions_log): Remove.
5547 (align_loops_max_skip): Remove.
5548 (align_jumps_max_skip): Remove.
5549 (align_labels_max_skip): Remove.
5550 (align_functions_max_skip): Remove.
5551 (align_loops_value): Remove.
5552 (align_jumps_value): Remove.
5553 (align_labels_value): Remove.
5554 (align_functions_value): Remove.
5555 * output.h (label_to_alignment): Change return type to align_flags.
5556 (label_to_max_skip): Remove.
5557 * target.def: Remove loop_align_max_skip, label_align_max_skip,
5558 jump_align_max_skip macros.
5559 * targhooks.h (default_loop_align_max_skip): Remove.
5560 (default_label_align_max_skip): Likewise.
5561 (default_jump_align_max_skip): Likewise.
5562 (default_label_align_after_barrier_max_skip): Remove.
5563 * toplev.c (read_log_maxskip): Use ::normalize function.
5564 (parse_N_M): Remove not used argument and also call ::normalize.
5565 (parse_alignment_opts): Do not pass unused arguments.
5566 * varasm.c (assemble_start_function): Use directly align_functions
5567 instead of removed macros.
5568 * system.h: Do not poison removed macros.
5570 2018-07-17 Jakub Jelinek <jakub@redhat.com>
5573 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
5574 and cond temporaries don't have reference type if iterator has
5575 pointer type. For init use &for_pre_body instead of pre_p if
5576 for_pre_body is non-empty.
5578 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
5580 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
5581 double-double modes to SFmode directly directly.
5582 (trunc<mode>sf2_fprs): Delete.
5584 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
5586 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
5587 for conversions between IFmode and the decimal floating point modes.
5588 (init_float128_ieee): Use the correct names for conversions between
5589 KFmode and the decimal floating point modes.
5591 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
5593 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
5594 for the conversions between TDmode and IFmode.
5595 (init_float128_ieee): Use more correct names for the conversions
5596 between TDmode and KFmode.
5598 2018-07-16 Jakub Jelinek <jakub@redhat.com>
5600 PR tree-optimization/86526
5601 * builtins.c (expand_builtin_memcmp): Formatting fixes.
5602 (inline_expand_builtin_string_cmp): Likewise.
5603 (inline_string_cmp): Likewise. Use c_readstr instead of
5604 builtin_memcpy_read_str. Add unit_mode temporary.
5606 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
5609 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
5610 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
5612 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
5614 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
5615 Alphabetize prototypes of built-in functions, separating out
5616 built-in functions that are listed in this section but should be
5617 described elsewhere.
5619 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
5622 * expmed.c (emit_store_flag): Do not emit setcc followed by a
5623 conditional move when trapping comparison was split to a
5624 non-trapping one (and vice versa).
5626 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
5628 * config/s390/s390.c (s390_function_profiler): Generate nops
5629 instead of profiler call sequences.
5630 * config/s390/s390.opt: Add the new option.
5632 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
5634 * config/s390/s390.c (s390_function_profiler): Generate
5635 __mcount_loc section.
5636 * config/s390/s390.opt: Add the new option.
5638 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
5640 * common.opt: Add the new warning.
5641 * config/s390/s390.c (s390_function_profiler): Emit "brasl
5642 %r0,__fentry__" when -mfentry is specified.
5643 (s390_option_override_internal): Disallow -mfentry for 31-bit
5645 * config/s390/s390.opt: Add the new option.
5647 2018-07-16 Richard Biener <rguenther@suse.de>
5650 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
5651 for function-local FUNCTION_DECL and RESULT_DECL immediately.
5653 2018-07-16 Martin Liska <mliska@suse.cz>
5656 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
5659 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
5661 * config/arc/arcHS.md: Update ARCHS scheduling rules.
5663 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
5665 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
5667 * config/arc/arc-cpus.def (hs4x): New cpu.
5669 * config/arc/arc-tables.opt: Regenerate.
5670 * config/arc/arc.c (arc_sched_issue_rate): New function.
5671 (TARGET_SCHED_ISSUE_RATE): Define.
5672 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
5673 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
5675 (attr tune): Add ARCHS4x tune values.
5676 (attr tune_dspmpy): Define.
5677 (*tst): Correct instruction type.
5678 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
5679 * config/arc/arcHS4x.md: New file.
5680 * config/arc/fpu.md: Update instruction type attributes.
5681 * config/arc/t-multilib: Regenerate.
5683 2018-07-16 Tom de Vries <tdevries@suse.de>
5686 * var-tracking.c (vt_initialize): Fix pre_dec handling.
5688 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5690 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
5693 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
5695 PR tree-optimization/86514
5696 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
5697 conversion to a boolean type from a type with greater precision.
5699 2018-07-16 Tom de Vries <tdevries@suse.de>
5701 * var-tracking.c (vt_initialize): Print adjusted insn slim if
5702 dump_flags request TDF_SLIM.
5704 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
5706 * fold-const.c (int_const_binop_1): Abstract...
5707 (wide_int_binop): ...wide int code here.
5708 (poly_int_binop): ...poly int code here.
5709 Abstract the rest of int_const_binop_1 into int_const_binop.
5710 * fold-const.h (wide_int_binop): New.
5711 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
5712 Remove useless PLUS/MINUS_EXPR case.
5713 (zero_nonzero_bits_from_vr): Move wide int code...
5714 (zero_nonzero_bits_from_bounds): ...here.
5715 (extract_range_from_binary_expr_1): Move mask optimization code...
5716 (range_easy_mask_min_max): ...here.
5717 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
5718 (range_easy_mask_min_max): New.
5720 2018-07-15 Jeff Law <law@redhat.com>
5723 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
5726 2018-07-14 Jim Wilson <jimw@sifive.com>
5728 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
5730 2018-07-14 Paul Koning <ni1d@arrl.net>
5732 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
5734 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
5736 * lto-streamer-out.c (copy_function_or_variable): Dump info about
5739 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
5740 Steve Munroe <munroesj52@gmail.com>
5742 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
5743 (_mm_andnot_si128): Likewise.
5744 (_mm_or_si128): Likewise.
5745 (_mm_xor_si128): Likewise.
5747 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
5750 * builtins.c (expand_builtin_memcmp): Inline the calls first
5751 when result_eq is false.
5752 (expand_builtin_strcmp): Inline the calls first.
5753 (expand_builtin_strncmp): Likewise.
5754 (inline_string_cmp): New routine. Expand a string compare
5755 call by using a sequence of char comparison.
5756 (inline_expand_builtin_string_cmp): New routine. Inline expansion
5757 a call to str(n)cmp/memcmp.
5758 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
5760 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
5762 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
5764 * config/arm/driver-arm.c: Include arm-native.h.
5765 (host_detect_local_cpu): Use auto-generated data tables.
5766 (vendors, arm_cpu_table): Delete. Move part information to ...
5767 * config/arm/arm-cpus.in: ... here.
5768 * config/arm/parsecpu.awk (gen_native): New function.
5769 (vendor, part): New CPU fields.
5770 (END): Add support for building the native CPU detection tables.
5771 * config/arm/t-arm (arm-native.h): Add build rule.
5772 (driver-arm.o): Add dependency on arm-native.h.
5774 2018-07-13 Richard Biener <rguenther@suse.de>
5777 * match.pd (addr1 - addr2): Allow either of the operand to
5780 2018-07-13 Tom de Vries <tdevries@suse.de>
5782 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
5785 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
5787 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
5788 arrays instead of numbered variables.
5790 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
5792 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
5793 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
5795 2018-07-13 Richard Biener <rguenther@suse.de>
5798 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
5799 instead of get_context_die.
5801 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
5802 Richard Biener <rguenther@suse.de>
5805 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
5806 that the loop latch destination where phi is defined.
5808 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
5810 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
5811 (riscv_expand_epilogue): Add assertion to check interrupt mode.
5812 (riscv_set_current_function): Extract getting interrupt type to new
5814 (riscv_get_interrupt_type): New function.
5815 (riscv_merge_decl_attributes): New function, checking interrupt type is
5817 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
5819 2018-07-12 Paul Koning <ni1d@arrl.net>
5821 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
5824 2018-07-12 Paul Koning <ni1d@arrl.net>
5826 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
5827 zero reference, add doloop_end instead.
5828 * doc/md.texi (decrement_and_branch_until_zero): Remove.
5829 (Looping patterns): Remove decrement_and_branch_until_zero. Add
5830 detail for doloop_end.
5832 2018-07-12 Martin Sebor <msebor@redhat.com>
5835 * attribs.c (decl_attributes): Reject conflicting attributes before
5836 calling attribute handlers.
5838 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
5840 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
5842 (gcc::dump_manager::get_dump_file_name): likewise.
5843 (dump_begin): Likewise.
5844 * dumpfile.h (dump_begin): Update prototype.
5845 (gcc::dump_manager::get_dump_file_name,
5846 gcc::dump_manager::get_dump_file_name): Update prototype.
5848 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
5850 * internal-fn.h (vectorizable_internal_fn_p): New function.
5851 * tree-vect-slp.c (compatible_calls_p): Likewise.
5852 (vect_build_slp_tree_1): Remove nops argument. Handle calls
5853 to internal functions.
5854 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
5856 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
5858 * fold-const.h (inverse_conditions_p): Declare.
5859 * fold-const.c (inverse_conditions_p): New function.
5860 * match.pd: Use inverse_conditions_p. Add folds of view_converts
5861 that test the inverse condition of a conditional internal function.
5862 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
5863 * internal-fn.c (internal_fn_mask_index): Handle conditional
5865 (vectorized_internal_fn_supported_p): New function.
5866 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
5867 (any_pred_load_store): Replace with...
5868 (need_to_predicate): ...this new variable.
5869 (redundant_ssa_names): New variable.
5870 (ifcvt_can_use_mask_load_store): Move initial checks to...
5871 (ifcvt_can_predicate): ...this new function. Handle tree codes
5872 for which a conditional internal function exists.
5873 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
5874 instead of ifcvt_can_use_mask_load_store. Update after variable
5876 (predicate_load_or_store): New function, split out from
5877 predicate_mem_writes.
5878 (check_redundant_cond_expr): New function.
5879 (value_available_p): Likewise.
5880 (predicate_rhs_code): Likewise.
5881 (predicate_mem_writes): Rename to...
5882 (predicate_statements): ...this. Use predicate_load_or_store
5883 and predicate_rhs_code.
5884 (combine_blocks, tree_if_conversion): Update after above name changes.
5885 (ifcvt_local_dce): Handle redundant_ssa_names.
5886 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
5887 general conditional functions.
5888 * tree-vect-stmts.c (vectorizable_call): Likewise.
5890 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
5891 Alan Hayward <alan.hayward@arm.com>
5892 David Sherwood <david.sherwood@arm.com>
5894 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
5895 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
5896 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
5897 plus and minus and convert them into IFN_COND_FMA-based sequences.
5898 (convert_mult_to_fma): Handle conditional plus and minus.
5900 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
5902 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
5903 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
5904 (cond_fnms_optab): New optabs.
5905 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
5907 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
5908 * internal-fn.h (get_conditional_internal_fn): Declare.
5909 (get_unconditional_internal_fn): Likewise.
5910 * internal-fn.c (cond_ternary_direct): New macro.
5911 (expand_cond_ternary_optab_fn): Likewise.
5912 (direct_cond_ternary_optab_supported_p): Likewise.
5913 (FOR_EACH_COND_FN_PAIR): Likewise.
5914 (get_conditional_internal_fn): New function.
5915 (get_unconditional_internal_fn): Likewise.
5916 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
5917 (gimple_match_op::gimple_match_op): Add a new overload for 5
5919 (gimple_match_op::set_op): Likewise.
5920 (gimple_resimplify5): Declare.
5921 * genmatch.c (decision_tree::gen): Generate simplifications for
5923 * gimple-match-head.c (gimple_simplify): Define an overload for
5924 5 operands. Handle calls with 5 arguments in the top-level overload.
5925 (convert_conditional_op): Handle conversions from unconditional
5926 internal functions to conditional ones.
5927 (gimple_resimplify5): New function.
5928 (build_call_internal): Pass a fifth operand.
5929 (maybe_push_res_to_seq): Likewise.
5930 (try_conditional_simplification): Try converting conditional
5931 internal functions to unconditional internal functions.
5932 Handle 3-operand unconditional forms.
5933 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
5934 Define ternary equivalents of the current rules for binary conditional
5936 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
5938 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
5939 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
5940 (optab): Handle them.
5941 (SVE_COND_FP_TERNARY): New int iterator.
5942 (sve_fmla_op, sve_fmad_op): New int attributes.
5943 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
5944 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
5945 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
5947 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
5949 * target.def (preferred_else_value): New target hook.
5950 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
5951 * doc/tm.texi: Regenerate.
5952 * targhooks.h (default_preferred_else_value): Declare.
5953 * targhooks.c (default_preferred_else_value): New function.
5954 * internal-fn.h (conditional_internal_fn_code): Declare.
5955 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
5956 (get_conditional_internal_fn): Use it.
5957 (conditional_internal_fn_code): New function.
5958 * gimple-match.h (gimple_match_cond): New struct.
5959 (gimple_match_op): Add a cond member function.
5960 (gimple_match_op::gimple_match_op): Update all forms to take a
5962 * genmatch.c (expr::gen_transform): Use the same condition as res_op
5963 for the suboperation, but don't specify a particular else_value.
5964 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
5965 (visit_nary_op, visit_reference_op_load): Pass
5966 gimple_match_cond::UNCOND to the gimple_match_op constructor.
5967 * gimple-match-head.c: Include tree-eh.h
5968 (convert_conditional_op): New function.
5969 (maybe_resimplify_conditional_op): Likewise.
5970 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
5971 (gimple_resimplify2): Likewise.
5972 (gimple_resimplify3): Likewise.
5973 (gimple_resimplify4): Likewise.
5974 (maybe_push_res_to_seq): Return null for conditional operations.
5975 (try_conditional_simplification): New function.
5976 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
5978 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
5980 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
5982 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
5984 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
5986 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
5988 (hash_tree): Do not hash DECL_FCONTEXT
5989 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
5990 Do not stream DECL_FCONTEXT.
5991 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
5992 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
5994 2018-07-12 Richard Biener <rguenther@suse.de>
5997 * dwarf2out.c (gen_block_die): Only output blocks when they have
5998 at least one !DECL_IGNORED_P variable.
6000 2018-07-12 Richard Biener <rguenther@suse.de>
6003 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
6004 Remove -mieee-fp handling.
6006 2018-07-12 Richard Biener <rguenther@suse.de>
6008 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
6009 left-over from last patch.
6011 2018-07-12 Jakub Jelinek <jakub@redhat.com>
6013 PR tree-optimization/86492
6014 * gimple-ssa-store-merging.c
6015 (imm_store_chain_info::coalesce_immediate_stores): Call
6016 check_no_overlap even for the merge_overlapping case. Formatting fix.
6018 2018-07-12 Richard Biener <rguenther@suse.de>
6021 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
6022 move possibly trapping operations into the conditional.
6024 2018-07-12 Richard Biener <rguenther@suse.de>
6026 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
6027 (vn_lookup_simplify_result): Remove recursion limit applied
6029 (vn_nary_build_or_lookup_1): Adjust.
6030 (try_to_simplify): Likewise.
6031 * gimple-match-head.c (gimple_resimplify1): Instead apply one
6033 (gimple_resimplify2): Likewise.
6034 (gimple_resimplify3): Likewise.
6035 (gimple_resimplify4): Likewise.
6037 2018-07-11 Jakub Jelinek <jakub@redhat.com>
6039 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
6040 Use __mmask64 type instead of __mmask8 for __M argument.
6041 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
6042 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
6044 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
6045 __mmask16 for __M argument.
6046 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
6047 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
6048 to __mmask16 instead of __mmask8.
6049 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
6050 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
6051 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
6052 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
6053 instead of __mmask16 for __U argument.
6054 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
6055 __mmask16 instead of __mmask8 for __U argument.
6056 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
6058 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
6060 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
6062 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
6064 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
6066 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
6068 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
6070 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
6072 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
6074 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
6076 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
6077 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
6078 return type as well as __M argument type and all casts from __mmask8
6080 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
6081 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
6082 return type as well as __M argument type and all casts from __mmask8
6084 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
6085 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
6086 return type as well as __M argument type and all casts from __mmask8
6088 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
6089 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
6090 return type as well as __M argument type and all casts from __mmask8
6092 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
6093 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
6096 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
6098 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
6099 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
6102 2018-07-11 Paul Koning <ni1d@arrl.net>
6104 * doc/md.texi (define_subst): Document how multiple occurrences of
6105 the same argument in the replacement pattern are handled.
6107 2018-07-11 Paul Koning <ni1d@arrl.net>
6109 * doc/extend.texi (Common Variable Attributes): Move "mode" into
6111 (Common Type Attributes): Add "mode" attribute.
6113 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
6115 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
6116 stream DECL_ORIGINAL_TYPE.
6117 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
6118 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
6119 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
6120 Do not walk original type.
6121 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
6123 (write_ts_decl_non_common_tree_pointers): Do not stream
6125 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
6126 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
6128 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
6130 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
6131 thread twice from the same starting edge.
6133 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
6135 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
6136 * gimple.c (gimple_call_nonnull_result_p): ...here...
6137 (gimple_call_nonnull_arg): ...and here.
6138 * gimple.h (gimple_call_nonnull_result_p): New.
6139 (gimple_call_nonnull_arg): New.
6141 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
6143 * config/arm/arm-cpus.in: Move information from fpu field of each
6144 cpu definition to the isa field.
6145 * config/arm/parsecpu.awk (fpu): Delete match rule.
6146 (gen_comm_data): Don't add bits from the CPU's FPU entry.
6148 2018-07-11 Richard Biener <rguenther@suse.de>
6151 * dwarf2out.c (init_sections_and_labels): Use
6152 output_asm_line_debug_info consistently.
6153 (dwarf2out_early_finish): Likewise.
6154 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
6157 2018-07-11 Richard Biener <rguenther@suse.de>
6159 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
6160 Rework father_bb setting in a way to avoid propagating constants
6161 multiple times on a loop body.
6163 2018-07-10 Mark Wielaard <mark@klomp.org>
6166 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
6168 2018-07-10 Richard Biener <rguenther@suse.de>
6170 * hash-map.h (hash_map::iterator::operator*): Return
6171 references to key and value.
6173 2018-07-10 Jakub Jelinek <jakub@redhat.com>
6176 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
6177 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
6178 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
6179 (gimplify_omp_for): For composite loops, move outer
6180 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
6181 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
6182 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
6183 TREE_LIST for both the original class iterator and the "last" helper
6184 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
6185 loop, remember has_decl_expr from outer composite loops for the
6186 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
6188 2018-07-09 Martin Sebor <msebor@redhat.com>
6192 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
6193 accessing implicitly initialized array elements.
6194 * expr.c (string_constant): Handle string initializers of
6195 character arrays within aggregates.
6196 * gimple-fold.c (fold_array_ctor_reference): Add argument.
6197 Store element offset. As a special case, handle zero size.
6198 (fold_nonarray_ctor_reference): Same.
6199 (fold_ctor_reference): Add argument. Store subobject offset.
6200 * gimple-fold.h (fold_ctor_reference): Add argument.
6202 2018-07-09 Paul Koning <ni1d@arrl.net>
6204 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
6205 (pdp11_insn_cost): New function.
6206 (pdp11_md_asm_adjust): New function.
6207 (TARGET_INVALID_WITHIN_DOLOOP): Define.
6208 (pdp11_rtx_costs): Update to match machine better.
6209 (output_addr_const_pdp11): Correct format mismatch warnings.
6210 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
6211 * config/pdp11/pdp11.md: General change to add base_cost and/or
6212 length attributes for use by new pdp11_insn_cost function.
6213 (MIN_BRANCH): Correct definition.
6215 (doloop_end): Use standard pattern name for looping pattern.
6216 (doloop_end_nocc): New.
6217 (movsf): Add another constraint alternative.
6218 (zero_extendqihi2): Add constraint alternatives for not in place
6220 (zero_extendhisi2): Remove.
6221 (shift patterns): Add CC handling variants.
6225 (define_peephole2): New peephole to recognize mov that sets CC for
6228 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6230 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
6231 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
6234 2018-07-09 Jakub Jelinek <jakub@redhat.com>
6237 * real.c (real_nextafter): Return true if result is denormal.
6239 2018-07-09 Martin Liska <mliska@suse.cz>
6241 * common.opt: Add back wrongly removed attribute.
6243 2018-07-09 Richard Biener <rguenther@suse.de>
6246 * dwarf2out.c (gen_block_die): For an early generated DIE
6247 always output high/low PC attributes.
6249 2018-07-09 Tom de Vries <tdevries@suse.de>
6251 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
6252 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
6253 onto VAR_DECL with abstract origin.
6255 2018-07-07 Jim Wilson <jimw@sifive.com>
6257 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
6259 2018-07-07 Tom de Vries <tdevries@suse.de>
6261 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
6263 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
6265 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
6267 (combine_bound): Use wide-int overflow calculation instead of
6269 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
6271 * fold-const.c (int_const_binop_2): Same.
6272 (extract_muldiv_1): Same.
6273 (fold_div_compare): Same.
6274 (fold_abs_const): Same.
6276 * poly-int.h (add): Same.
6280 * predict.c (predict_iv_comparison): Same.
6281 * profile-count.c (slow_safe_scale_64bit): Same.
6282 * simplify-rtx.c (simplify_const_binary_operation): Same.
6283 * tree-chrec.c (tree_fold_binomial): Same.
6284 * tree-data-ref.c (split_constant_offset_1): Same.
6285 * tree-if-conv.c (idx_within_array_bound): Same.
6286 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
6287 * tree-ssa-phiopt.c (minmax_replacement): Same.
6288 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
6289 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
6290 * vr-values.c (vr_values::adjust_range_with_scev): Same.
6291 * wide-int.cc (wi::add_large): Same.
6292 (wi::mul_internal): Same.
6293 (wi::sub_large): Same.
6294 (wi::divmod_internal): Same.
6295 * wide-int.h: Change overflow type to overflow_type for neg, add,
6296 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
6297 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
6298 mul_internal, divmod_internal.
6299 (overflow_type): New enum.
6300 (accumulate_overflow): New.
6302 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
6304 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
6305 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
6307 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
6309 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
6310 argument is checked for zero before entering loop, avoid checking again.
6312 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
6314 * gimplify.h (generic_expr_could_trap_p): Set as global function.
6315 * gimplify.h (generic_expr_could_trap_p): Likwise.
6316 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
6318 2018-07-06 Jakub Jelinek <jakub@redhat.com>
6320 PR tree-optimization/86401
6321 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
6322 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
6323 (fold_bit_and_mask): ... here. New helper function for match.pd.
6324 * fold-const.h (fold_bit_and_mask): Declare.
6325 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
6327 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
6330 * target.def (translate_mode_attribute): New hook.
6331 * targhooks.h (default_translate_mode_attribute): Declare.
6332 * targhooks.c (default_translate_mode_attribute): New function.
6333 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
6334 * doc/tm.texi: Regenerate.
6335 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
6336 (rs6000_translate_mode_attribute): New function.
6338 2018-07-06 Paul Koning <ni1d@arrl.net>
6340 * doc/md.texi (define_split): Document DONE and FAIL.
6341 (define_peephole2): Ditto.
6343 2018-07-05 Jeff Law <law@redhat.com>
6345 PR tree-optimization/86010
6346 * tree-ssa-dse.c (compute_trims): More aggressively trim at
6347 both the head and tail of mem* and str* calls.
6349 2018-07-05 Jim Wilson <jimw@sifive.com>
6351 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
6353 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
6355 * config/aarch64/aarch64-simd.md: correct flags text for
6356 MIN_EXPR replacement.
6358 2018-07-05 James Clarke <jrtc27@jrtc27.com>
6360 * configure: Regenerated.
6362 2018-07-05 Carl Love <cel@us.ibm.com>
6364 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
6365 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
6366 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
6367 VSX_BUILTIN_DOUBLEL_V4SF.
6369 2018-07-05 Martin Sebor <msebor@redhat.com>
6372 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
6373 than its domain to compute its the upper bound of a char array.
6375 2018-07-05 Nathan Sidwell <nathan@acm.org>
6377 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
6378 * doc/cpp.texi: Update comment.
6379 * doc/tm.texi: Rebuilt.
6380 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
6381 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
6382 * doc/extend.texi (Backwards Compatibility): Clarify it is system
6383 headers affected by extern "C".
6384 * system.h: Poison NO_IMPLICIT_EXTERN_C.
6385 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
6386 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
6387 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
6388 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
6389 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
6390 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
6391 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
6392 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
6393 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
6394 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
6395 config/sparc/sp64-elf.h, config/spu/spu.h,
6396 config/stormy16/stormy16.h, config/v850/v850.h,
6397 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
6398 define NO_IMPLICIT_EXTERN_C.
6399 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
6401 2018-07-05 Tamar Christina <tamar.christina@arm.com>
6404 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
6405 instead of GET_MODE_SIZE when comparing Units.
6407 2018-07-05 Tamar Christina <tamar.christina@arm.com>
6410 * rtlanal.c (set_noop_p): Constrain on mode change,
6411 include hard-reg-set.h
6413 2018-07-05 Tamar Christina <tamar.christina@arm.com>
6415 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
6417 2018-07-05 Jakub Jelinek <jakub@redhat.com>
6420 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
6423 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
6425 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
6427 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
6430 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
6432 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
6434 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
6437 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
6439 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
6441 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
6442 neg_*_op* variables.
6444 2018-07-04 Martin Liska <mliska@suse.cz>
6446 * tree-switch-conversion.c: Define
6447 max_ratio_for_speed and max_ratio_for_size constants.
6449 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
6450 Martin Liska <mliska@suse.cz>
6455 * common.opt: Rename align options with 'str_' prefix.
6456 * common/config/i386/i386-common.c (set_malign_value): New
6458 (ix86_handle_option): Use it to set -falign-* options/
6459 * config/aarch64/aarch64-protos.h (struct tune_params): Change
6460 type from int to string.
6461 * config/aarch64/aarch64.c: Update default values from int
6463 * config/alpha/alpha.c (alpha_override_options_after_change):
6465 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
6466 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6467 max skip conditionally.
6468 * config/i386/freebsd.h (SUBALIGN_LOG): New.
6469 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6470 max skip conditionally.
6471 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6472 max skip conditionally.
6473 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
6474 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6475 max skip conditionally.
6476 * config/i386/i386.c (struct ptt): Change type from int to
6478 (ix86_default_align): Set default values.
6479 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
6480 max skip conditionally.
6481 * config/i386/iamcu.h (SUBALIGN_LOG): New.
6482 (ASM_OUTPUT_MAX_SKIP_ALIGN):
6483 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
6484 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6485 max skip conditionally.
6486 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
6487 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
6488 * config/i386/x86-64.h (SUBALIGN_LOG): New.
6489 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6490 max skip conditionally.
6491 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
6492 * config/ia64/ia64.c (ia64_option_override): Set default values
6493 for alignment options.
6494 * config/m68k/m68k.c: Handle new str_align_* options.
6495 * config/mips/mips.c (mips_set_compression_mode): Change
6497 (mips_option_override): Set default values for options.
6498 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
6500 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6502 * config/rx/rx.c (rx_option_override): Likewise.
6503 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
6504 (LABEL_ALIGN): Use align_labels_log.
6505 (LOOP_ALIGN): Use align_loops_align.
6506 * config/s390/s390.c (s390_asm_output_function_label): Use new
6508 * config/sh/sh.c (sh_override_options_after_change):
6509 Change type of constants.
6510 * config/spu/spu.c (spu_sched_init): Likewise.
6511 * config/sparc/sparc.c (sparc_option_override): Set default
6513 * config/visium/visium.c (visium_option_override): Likewise.
6514 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
6515 emit p2align format with last argument if it's not needed.
6516 * doc/invoke.texi: Document extended format of -falign-*.
6517 * final.c: Use align_labels alignment.
6518 * flags.h (struct target_flag_state): Change type to use
6520 (struct align_flags_tuple): New.
6521 (struct align_flags): Likewise.
6522 (align_loops_log): Redefine macro to use new types.
6523 (align_loops_max_skip): Redefine macro to use new types.
6524 (align_jumps_log): Redefine macro to use new types.
6525 (align_jumps_max_skip): Redefine macro to use new types.
6526 (align_labels_log): Redefine macro to use new types.
6527 (align_labels_max_skip): Redefine macro to use new types.
6528 (align_functions_log): Redefine macro to use new types.
6529 (align_loops): Redefine macro to use new types.
6530 (align_jumps): Redefine macro to use new types.
6531 (align_labels): Redefine macro to use new types.
6532 (align_functions): Redefine macro to use new types.
6533 (align_functions_max_skip): Redefine macro to use new types.
6534 (align_loops_value): New macro.
6535 (align_jumps_value): New macro.
6536 (align_labels_value): New macro.
6537 (align_functions_value): New macro.
6538 * function.c (invoke_set_current_function_hook): Propagate
6539 alignment values from flags to global variables default in
6541 * ipa-icf.c (sem_function::equals_wpa): Use
6542 cl_optimization_option_eq instead of memcmp.
6543 * lto-streamer.h (cl_optimization_stream_out): Support streaming
6545 (cl_optimization_stream_in): Likewise.
6546 * optc-save-gen.awk: Support strings in cl_optimization.
6547 * opth-gen.awk: Likewise.
6548 * opts.c (finish_options): Remove error checking of invalid
6550 (MAX_CODE_ALIGN): Remove.
6551 (MAX_CODE_ALIGN_VALUE): Likewise.
6552 (parse_and_check_align_values): New function.
6553 (check_alignment_argument): Likewise.
6554 (common_handle_option): Use check_alignment_argument.
6555 * opts.h (parse_and_check_align_values): Declare.
6556 * toplev.c (init_alignments): Remove.
6557 (read_log_maxskip): New.
6558 (parse_N_M): Likewise.
6559 (parse_alignment_opts): Likewise.
6560 (backend_init_target): Remove usage of init_alignments.
6561 * toplev.h (parse_alignment_opts): Declare.
6562 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
6564 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
6565 * tree.c (cl_option_hasher::equal): New.
6566 * varasm.c: Use new global macros.
6568 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
6570 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
6571 Use a simpler align directive also if MAXSKIP = ALIGN-1.
6572 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6573 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6574 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6575 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
6576 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
6577 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
6578 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
6579 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6580 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6581 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6582 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6584 2018-07-04 Martin Liska <mliska@suse.cz>
6585 Jonathan Wakely <jwakely@redhat.com>
6587 * coverage.c: Use correct type.
6588 * doc/invoke.texi: Language correction.
6590 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
6593 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
6594 ENDBRANCH for non-tail call which may return via indirect branch.
6595 * doc/extend.texi: Document indirect_return attribute.
6597 2018-07-03 Martin Sebor <msebor@redhat.com>
6599 PR tree-optimization/86274
6600 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
6602 (format_floating): Correct handling of infinities and NaNs.
6604 2018-07-03 Martin Sebor <msebor@redhat.com>
6606 * print-tree.c (print_real_cst): New function.
6607 (print_node_brief): Call it.
6608 (print_node): Ditto.
6610 2018-07-03 Jeff Law <law@redhat.com>
6612 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
6613 into a single pattern.
6615 * config/h8300/h8300.md (ors code_iterator): New.
6616 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
6617 a single pattern and single splitter.
6618 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
6619 (iorqi3_1, xorqi3_1): Likewise.
6620 (iorqi3, xorqi3 expanders): Similarly.
6622 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
6623 (movmd_internal) into a single pattern using the P mode iterator.
6624 (movmd splitters): Similarly.
6625 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
6626 (movsd splitters): Similarly.
6628 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
6629 ADDB, ADDW and ADDL into a single ADD attribute which selects the
6630 right table based on the size of the operand.
6631 * config/h8300/h8300.md (length_table): Corresponding changes. All
6632 references to "addb", "addw" and "addl" changed to "add".
6633 (btst patterns): Merge two variants into a single pattern.
6634 (tstqi, tsthi): Likewise.
6635 (addhi3_incdec, addsi3_incdec): Likewise.
6636 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
6637 (mulhi3, mulsi3): Likewise.
6638 (udivhi3, udivsi3): Likewise.
6639 (divhi3, divsi3): Likewise.
6640 (andorqi3, andorhi3, andorsi3): Likewise.
6642 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
6645 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
6646 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
6648 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6650 PR tree-optimization/85694
6651 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
6652 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
6654 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
6655 (<u>avg<mode>3_ceil): New patterns.
6657 2018-07-03 David Malcolm <dmalcolm@redhat.com>
6659 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
6660 scan-tree-dump directive.
6661 * gcc.dg/vect/slp-perm-2.c: Likewise.
6662 * gcc.dg/vect/slp-perm-3.c: Likewise.
6663 * gcc.dg/vect/slp-perm-5.c: Likewise.
6664 * gcc.dg/vect/slp-perm-6.c: Likewise.
6665 * gcc.dg/vect/slp-perm-7.c: Likewise.
6666 * gcc.dg/vect/slp-perm-8.c: Likewise.
6668 2018-07-03 Marek Polacek <polacek@redhat.com>
6671 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
6673 2018-07-03 Richard Biener <rguenther@suse.de>
6676 * tree-ssa-structalias.c (find_func_clobbers): Properly
6677 handle indirect calls.
6679 2018-07-03 Jeff Law <law@redhat.com>
6681 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
6682 (shifts): New code iterator.
6683 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
6684 expander. Fix HImode handling on H8/SX.
6685 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
6686 (subqi3, subhi3, subsi3 expanders): Likewise.
6687 (andqi3, andhi3, andsi3 expanders): Likewise.
6688 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
6689 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
6690 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
6691 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
6692 (zero_extendqihi2, zero_extendqisi2): Likewise.
6693 (extendqihi2, extendqisi2): Likewise.
6694 (rotlqi3, rotlhi3, rotlsi3): Likewise.
6695 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
6696 (rotlqi3_1, rotlhi3_1): Likewise.
6697 (logicalhi3_sn, logicalsi3_sn): Likewise.
6698 (logicalhi3, logicalsi3): Likewise.
6700 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6702 * tree-vect-patterns.c (vect_recog_rotate_pattern)
6703 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
6704 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
6705 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
6706 type to append_pattern_def_seq instead of creating a stmt_vec_info
6708 (build_mask_conversion): Likewise. Remove vinfo argument.
6709 (vect_add_conversion_to_patterm): Likewise, renaming to...
6710 (vect_add_conversion_to_pattern): ...this.
6711 (vect_recog_mask_conversion_pattern): Update call to
6712 build_mask_conversion. Pass the vector type to
6713 append_pattern_def_seq here too.
6714 (vect_recog_gather_scatter_pattern): Update call to
6715 vect_add_conversion_to_pattern.
6717 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6719 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
6720 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
6721 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
6722 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
6723 STMT_VINFO_PATTERN_DEF_SEQ to null here.
6724 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
6725 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
6726 append_pattern_def_seq instead of new_pattern_def_seq.
6727 (vect_recog_divmod_pattern): Do both of the above.
6728 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
6731 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6733 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
6734 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
6735 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
6736 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
6737 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
6738 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
6739 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
6740 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
6741 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
6742 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
6743 parameter with a single stmt_vec_info.
6744 (vect_recog_func_ptr): Likewise.
6745 (vect_recog_gather_scatter_pattern): Likewise, folding in...
6746 (vect_try_gather_scatter_pattern): ...this.
6747 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
6748 the stmt_vec_info of the statement to be matched. Don't clear
6749 STMT_VINFO_RELATED_STMT.
6750 (vect_pattern_recog): Update call accordingly.
6752 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6754 PR tree-optimization/85694
6755 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
6756 (uavgM3_ceil): Document new optabs.
6757 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
6758 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
6760 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
6761 (savg_ceil_optab): New optabs.
6762 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
6763 (vect_vect_recog_func_ptrs): Add it.
6764 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
6765 constant directly from the associated lhs.
6767 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6769 * tree-vect-patterns.c (vect_split_statement): New function.
6770 (vect_convert_input): Use it to try to split an existing cast.
6772 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6774 * poly-int.h (print_hex): New function.
6775 * dumpfile.h (dump_dec, dump_hex): Declare.
6776 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
6777 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
6778 min_input_precision, operation_precision and operation_sign.
6779 * tree-vect-patterns.c (vect_get_range_info): New function.
6780 (vect_same_loop_or_bb_p, vect_single_imm_use)
6781 (vect_operation_fits_smaller_type): Delete.
6782 (vect_look_through_possible_promotion): Add an optional
6783 single_use_p parameter.
6784 (vect_recog_over_widening_pattern): Rewrite to use new
6785 stmt_vec_info infomration. Handle one operation at a time.
6786 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
6787 (vect_truncatable_operation_p, vect_set_operation_type)
6788 (vect_set_min_input_precision): New functions.
6789 (vect_determine_min_output_precision_1): Likewise.
6790 (vect_determine_min_output_precision): Likewise.
6791 (vect_determine_precisions_from_range): Likewise.
6792 (vect_determine_precisions_from_users): Likewise.
6793 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
6794 (vect_vect_recog_func_ptrs): Put over_widening first.
6796 (vect_pattern_recog): Call vect_determine_precisions.
6798 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6800 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
6801 statements that have been replaced by further pattern statements.
6802 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
6804 2018-07-03 Richard Biener <rguenther@suse.de>
6806 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
6807 always set *dt. Dump vectype in vectype overload.
6808 * dumpfile.h (dump_gimple_expr): New function.
6809 (dump_gimple_expr_loc): Likewise.
6810 * dumpfile.c (dump_gimple_expr): New function.
6811 (dump_gimple_expr_loc): Likewise.
6813 2018-07-02 Jeff Law <law@redhat.com>
6815 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
6816 the H8/300, H8/300H and H8/S variants into a single pattern.
6817 (movhi_h8300, movqi_h8300hs): Similarly.
6818 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
6819 (QHI mode iterator): New.
6821 * config/h8300/h8300.md: Remove trailing whitespace.
6823 2018-07-02 Jim Wilson <jimw@sifive.com>
6825 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
6826 instead of emit_insn for interrupt returns.
6827 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
6828 (riscv_sret, riscv_uret): Likewise.
6830 2018-07-02 David Malcolm <dmalcolm@redhat.com>
6832 * pretty-print.c (selftest::test_pp_format): Move save and restore
6833 of quotes to class auto_fix_quotes, and add an instance.
6834 * selftest.c: Include "intl.h".
6835 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
6836 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
6837 * selftest.h (selftest::auto_fix_quotes): New class.
6839 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
6841 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
6842 (aarch64_sve_prepare_conditional_op): Remove.
6843 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
6844 Allow aarch64_simd_reg_or_zero as select operand; remove
6845 the aarch64_sve_prepare_conditional_op call.
6846 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
6847 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
6848 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
6849 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
6850 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
6851 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
6852 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
6853 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
6854 and a splitters to match all of the *_any patterns.
6855 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
6857 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
6858 (SVE_COND_FP_BINARY_REV): Remove.
6859 (sve_int_op_rev, sve_fp_op_rev): New.
6860 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
6861 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
6862 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
6863 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
6864 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
6865 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
6866 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
6867 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
6868 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
6870 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
6871 Remove match_dup 1 from the inner unspec.
6872 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
6874 * config/aarch64/aarch64.md (movprfx): New attr.
6875 (length): Default movprfx to 8.
6876 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
6877 (*madd<SVE_I>, *msub<SVE_I): Likewise.
6878 (*<su>mul<SVE_I>3_highpart): Likewise.
6879 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
6880 (*v<ASHIFT><SVE_I>3): Likewise.
6881 (*<su><MAXMIN><SVE_I>3): Likewise.
6882 (*<su><MAXMIN><SVE_F>3): Likewise.
6883 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
6884 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
6885 (*div<SVE_F>4): Likewise.
6887 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
6889 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
6892 2018-07-02 Richard Biener <rguenther@suse.de>
6894 PR tree-optimization/86363
6895 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
6896 memset argument refers to a non-variable address.
6898 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
6900 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
6901 {PLUS,MINUS}_EXPR code to...
6902 (adjust_symbolic_bound): ...here,
6903 (combine_bound): ...here,
6904 (set_value_range_with_overflow): ...and here.
6906 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
6908 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
6910 (extract_range_from_abs_expr): ...here.
6912 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
6914 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
6915 -fno-omit-frame-pointer when not optimizing.
6917 2018-07-02 Martin Liska <mliska@suse.cz>
6920 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
6921 (propagate_nothrow): Likewise.
6923 2018-07-02 Martin Liska <mliska@suse.cz>
6926 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
6928 2018-07-02 David Malcolm <dmalcolm@redhat.com>
6930 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
6931 function in r262149, changing "loc" param from source_location to
6932 const dump_location_t &.
6933 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
6934 declaration, as above.
6936 2018-07-01 Paul Koning <ni1d@arrl.net>
6938 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
6939 -munit-asm, -mgnu-asm, -mdec-asm.
6940 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
6941 (pdp11_output_labelref): New.
6942 (pdp11_output_def): New.
6943 (pdp11_output_addr_vec_elt): New.
6944 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
6945 %# and %@ format codes.
6946 (pdp11_option_override): New.
6947 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
6948 (pdp11_output_ident): New.
6949 (pdp11_asm_named_section): New.
6950 (pdp11_asm_init_sections): New.
6951 (pdp11_file_start): New.
6952 (pdp11_file_end): New.
6953 (output_ascii): Use .ascii/.asciz for -mdec-asm.
6954 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
6955 %o, like %c but octal.
6956 (pdp11_option_override): New.
6957 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
6959 (DATA_SECTION_ASM_OP): Ditto.
6960 (READONLY_DATA_SECTION_ASM_OP): New.
6961 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
6962 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
6963 (ASM_OUTPUT_LABELREF): Ditto.
6964 (ASM_OUTPUT_DEF): Ditto.
6965 (ASM_OUTPUT_EXTERNAL): New.
6966 (ASM_OUTPUT_SOURCE_FILENAME): New.
6967 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
6968 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
6969 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
6970 %# and %@ format codes.
6971 * config/pdp11/pdp11.opt (mgnu-asm): New.
6972 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
6973 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
6974 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
6976 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
6978 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
6979 dereferencing path[] beyond its length.
6981 (debug_all_paths): New.
6982 (rewire_first_differing_edge): New.
6983 (adjust_paths_after_duplication): New.
6984 (duplicate_thread_path): Call adjust_paths_after_duplication.
6986 (thread_through_all_blocks): Add new argument to
6987 duplicate_thread_path.
6989 2018-06-30 Jim Wilson <jimw@sifive.com>
6991 * config/riscv/predicates.md (p2m1_shift_operand): New.
6992 (high_mask_shift_operand): New.
6993 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
6994 pattern using p2m1_shift_operand.
6995 (lshsi3_zero_extend_3+2): New combiner pattern using
6996 high_mask_shift_operand.
6998 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7000 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
7002 (vect_recog_rotate_pattern): ...here.
7003 (vect_convert_input): Try to insert casts of invariants in the
7005 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
7006 preheader to be empty.
7008 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7010 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
7011 vector type. If given, install it in the new statement's
7013 (vect_element_precision): New function.
7014 (vect_unpromoted_value): New struct.
7015 (vect_unpromoted_value::vect_unpromoted_value): New function.
7016 (vect_unpromoted_value::set_op): Likewise.
7017 (vect_look_through_possible_promotion): Likewise.
7018 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
7019 (vect_widened_op_tree, vect_convert_input): Likewise.
7020 (vect_convert_inputs, vect_convert_output): Likewise.
7021 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
7022 to handle the optional cast of the multiplication result and
7023 vect_widened_op_tree to detect the widened multiplication itself.
7024 Do not require the input and output of promotion casts to have
7025 the same sign, but base the signedness of the operation on the
7026 input rather than the result. If the pattern includes two
7027 promotions, check that those promotions have the same sign.
7028 Do not restrict the MULT_EXPR handling to a double-width result;
7029 handle quadruple-width results and wider. Use vect_convert_inputs
7030 to convert the inputs to the common type.
7031 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
7032 to handle the optional cast of the ABS result. Also allow a sign
7033 change or a sign extension between the ABS and MINUS.
7034 Use vect_widened_op_tree to detect the widened subtraction and use
7035 vect_convert_inputs to convert the inputs to the common type.
7036 (vect_handle_widen_op_by_const): Delete.
7037 (vect_recog_widen_op_pattern): New function.
7038 (vect_recog_widen_mult_pattern): Use it.
7039 (vect_recog_widen_shift_pattern): Likewise.
7040 (vect_recog_widen_sum_pattern): Use
7041 vect_look_through_possible_promotion to handle the promoted
7044 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7046 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
7047 the containing gimple_seq *.
7048 * gimple-iterator.h (gsi_for_stmt): Declare it.
7049 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
7050 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
7051 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
7052 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
7053 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
7055 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
7057 (vect_mark_pattern_stmts): ...here. Handle cases in which the
7058 statement being replaced is part of an existing pattern
7059 definition sequence, inserting the new pattern statements before
7061 (vect_pattern_recog_1): Don't return a bool. If the statement
7062 is already part of a pattern, instead apply pattern matching
7063 to the pattern definition statements. Don't clear the
7064 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
7065 (vect_pattern_recog): Don't break after the first match;
7066 continue processing the pattern definition statements instead.
7067 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
7069 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7071 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
7072 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
7073 (vect_recog_widen_sum_pattern): Use it.
7075 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7077 * tree-vect-loop.c (vectorizable_reduction): Assert that the
7078 phi is not a pattern statement and has not been replaced by
7079 a pattern statement.
7080 * tree-vect-patterns.c (type_conversion_p): Don't check
7081 STMT_VINFO_IN_PATTERN_P.
7082 (vect_recog_vector_vector_shift_pattern): Likewise.
7083 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
7084 the pattern statement rather than the original statement; check
7085 directly for a WIDEN_MULT_EXPR here.
7086 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
7087 vect_is_simple_use to return the pattern statement rather
7088 than the original statement; use is_pattern_stmt_p to check
7089 for such a pattern statement.
7090 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
7091 to return the pattern statement rather than the original statement;
7092 don't do the same transformation here.
7093 (vect_is_simple_use): If the defining statement has been replaced
7094 by a pattern statement, return the pattern statement instead.
7095 Remove the corresponding (local) transformation from the vectype
7098 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7100 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
7101 end and default to null.
7102 * tree-vect-loop.c (vect_create_epilog_for_reduction)
7103 (vectorizable_reduction): Update calls accordingly, dropping the
7104 gimple ** argument if the passed-back statement isn't needed.
7105 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
7106 (vect_recog_rotate_pattern): Likewise.
7107 (vect_recog_mask_conversion_pattern): Likewise.
7108 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
7109 (vect_mask_constant_operand_p): Likewise.
7110 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
7111 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
7112 (get_group_load_store_type, get_load_store_type): Likewise.
7113 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
7114 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
7115 (vectorizable_conversion, vectorizable_assignment): Likewise.
7116 (vectorizable_shift, vectorizable_operation): Likewise.
7117 (vectorizable_store, vect_is_simple_cond): Likewise.
7118 (vectorizable_condition, vectorizable_comparison): Likewise.
7119 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
7120 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
7121 and move it to the end. Cope with null def_stmt_outs.
7123 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
7125 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
7127 2018-06-29 Jeff Law <law@redhat.com>
7129 * config/v850/v850.c (v850_legitimate_address_p): Handle large
7130 displacements for TARGET_V850E2V3 and newer.
7131 (TARGET_LRA_P): Remove. Defaults to LRA now.
7132 * config/v850/v850.md (sign23byte_load): Remove.
7133 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
7134 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
7136 2018-06-29 Martin Liska <mliska@suse.cz>
7139 * coverage.c (coverage_init): Mangle full path name.
7140 * doc/invoke.texi: Document the change.
7141 * gcov-io.c (mangle_path): New.
7142 * gcov-io.h (mangle_path): Likewise.
7143 * gcov.c (mangle_name): Use mangle_path for path mangling.
7145 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7147 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
7148 if starting source register is not even.
7150 2018-06-29 Martin Liska <mliska@suse.cz>
7152 PR tree-optimization/86263
7153 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
7154 Make edge redirection.
7156 2018-06-29 David Malcolm <dmalcolm@redhat.com>
7158 * dumpfile.c (dump_loc): Add indentation based on scope depth.
7159 (dump_scope_depth): New variable.
7160 (get_dump_scope_depth): New function.
7161 (dump_begin_scope): New function.
7162 (dump_end_scope): New function.
7163 * dumpfile.h (get_dump_scope_depth): New declaration.
7164 (dump_begin_scope): New declaration.
7165 (dump_end_scope): New declaration.
7166 (class auto_dump_scope): New class.
7167 (AUTO_DUMP_SCOPE): New macro.
7168 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
7171 2018-06-29 Richard Biener <rguenther@suse.de>
7173 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
7174 compute_all_dependences succeeds.
7175 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
7176 exceed --param loop-max-datarefs-for-datadeps.
7178 2018-06-29 Jakub Jelinek <jakub@redhat.com>
7180 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
7182 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
7185 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
7186 alternative 0 in preferred_for_speed attribute.
7188 2018-06-28 Paul Koning <ni1d@arrl.net>
7190 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
7191 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
7192 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
7193 * config/pdp11/pdp11.md: Correct "length" attribute calculation
7194 for shift insn patterns.
7196 2018-06-28 David Malcolm <dmalcolm@redhat.com>
7198 * cgraph.c (cgraph_node::get_body): Replace assignments to
7199 "dump_file" with calls to set_dump_file.
7200 * dumpfile.c (alt_dump_file): Make static, and group with...
7201 (alt_flags): ...this definition.
7202 (dumps_are_enabled): New variable.
7203 (refresh_dumps_are_enabled): New function.
7204 (set_dump_file): New function.
7205 (set_alt_dump_file): New function.
7206 (gcc::dump_manager::dump_start): Replace assignments to
7207 "dump_file" and "alt_dump_file" with calls to set_dump_file and
7209 (gcc::dump_manager::dump_finish): Likewise.
7210 * dumpfile.h (alt_dump_file): Delete decl.
7211 (dumps_are_enabled): New variable decl.
7212 (set_dump_file): New function decl.
7213 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
7215 * tree-nested.c (lower_nested_functions): Replace assignments to
7216 "dump_file" with calls to set_dump_file.
7218 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
7220 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
7221 goto_locus of each outgoing edge of each basic block.
7223 2018-06-28 Richard Biener <rguenther@suse.de>
7225 * dwarf2out.c (decl_scope_table): Remove.
7226 (push_decl_scope): Likewise.
7227 (pop_decl_scope): Likewise.
7228 (gen_type_die_for_member): Do not call push/pop_decl_scope.
7229 (gen_struct_or_union_type_die): Likewise.
7230 (gen_tagged_type_die): Likewise.
7231 (dwarf2out_init): Do not initialize decl_scope_table.
7232 (dwarf2out_c_finalize): Do not free it.
7234 2018-06-28 Richard Biener <rguenther@suse.de>
7236 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
7237 deciding whether to not re-use a DIE.
7239 2018-06-28 Richard Biener <rguenther@suse.de>
7241 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
7242 DW_AT_abstract_origin attribute.
7244 2018-06-28 Martin Liska <mliska@suse.cz>
7246 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
7247 Use newly introduced constants.
7248 * tree-switch-conversion.h (struct jump_table_cluster):
7249 Define max_ratio_for_size and max_ratio_for_speed.
7251 2018-06-28 Martin Liska <mliska@suse.cz>
7253 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
7254 Add new checking assert to catch invalid state.
7255 (jump_table_cluster::can_be_handled): Handle single case
7257 (jump_table_cluster::is_beneficial): Bail out for such case.
7258 (bit_test_cluster::find_bit_tests):
7259 Add new checking assert to catch invalid state.
7260 (bit_test_cluster::can_be_handled): Handle single case
7262 (bit_test_cluster::is_beneficial): Bail out for such case.
7263 (switch_decision_tree::analyze_switch_statement):
7266 2018-06-28 Martin Liska <mliska@suse.cz>
7268 * common.opt: Introduce -completion option.
7269 * gcc.c (driver_handle_option): Handle it.
7270 (driver::main): Print completions if completion
7272 * opt-suggestions.c (option_proposer::get_completions):
7274 (option_proposer::suggest_completion): Likewise.
7275 (option_proposer::find_param_completions): Likewise.
7276 (verify_autocompletions): Likewise.
7277 (test_completion_valid_options): Likewise.
7278 (test_completion_valid_params): Likewise.
7279 (in_completion_p): Likewise.
7280 (empty_completion_p): Likewise.
7281 (test_completion_partial_match): Likewise.
7282 (test_completion_garbage): Likewise.
7283 (opt_proposer_c_tests): Likewise.
7284 * opt-suggestions.h: Declare new functions.
7285 * opts.c (common_handle_option): Handle OPT__completion_.
7286 * selftest-run-tests.c (selftest::run_tests): Add
7287 opt_proposer_c_tests.
7288 * selftest.c (assert_str_startswith): New.
7289 * selftest.h (assert_str_startswith): Likewise.
7290 (opt_proposer_c_tests): New.
7291 (ASSERT_STR_STARTSWITH): Likewise.
7293 2018-06-28 Martin Liska <mliska@suse.cz>
7295 * Makefile.in: Add opt-suggestions.o.
7296 * gcc-main.c: Include opt-suggestions.h.
7297 * gcc.c (driver::driver): Likewise.
7298 (driver::~driver): Remove m_option_suggestions.
7299 (driver::build_option_suggestions): Moved to option_proposer.
7300 (driver::suggest_option): Likewise.
7301 (driver::handle_unrecognized_options): Use option_proposer.
7302 * gcc.h (class driver): Add new memver m_option_proposer.
7303 * opt-suggestions.c: New file.
7304 * opt-suggestions.h: New file.
7306 2018-06-28 Martin Liska <mliska@suse.cz>
7308 * vec.h (class auto_string_vec): New (moved from auto_argvec).
7309 (auto_string_vec::~auto_string_vec): Likewise.
7311 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
7313 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
7314 prevent_decl_creation_for_types fields up and add reset_location field.
7315 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
7316 statement if id->reset_location is true.
7317 (copy_edges_for_bb): Do not set goto_locus on the new edges if
7318 id->reset_location is true.
7319 (copy_phis_for_bb): Force input_location on the arguments if
7320 id->reset_location is true.
7321 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
7322 is set on the function to be inlined.
7324 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
7326 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
7328 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
7330 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
7331 registers for Pmode.
7332 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
7333 hard registers for the clobbered pseudo.
7335 2018-06-27 Paul Koning <ni1d@arrl.net>
7337 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
7338 mutually exclusive options.
7339 * config/pdp11/constraints.md (h): New constraint.
7340 (O): Update definition to match shift code generation.
7341 (D): New constraint.
7342 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
7344 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
7346 (output_jump): Change arguments.
7347 (pdp11_fixed_cc_regs): New function.
7348 (pdp11_cc_mode): Ditto.
7349 (pdp11_expand_shift): Ditto.
7350 (pdp11_assemble_shift): Ditto.
7351 (pdp11_small_shift): Ditto.
7352 (pdp11_branch_cost): Remove.
7353 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
7355 (pdp11_register_move_cost): Update for CC registers.
7356 (pdp11_rtx_costs): Add case for LSHIFTRT.
7357 (pdp11_output_jump): Add CCNZ mode conditional branches.
7358 (notice_update_cc_on_set): Remove.
7359 (pdp11_cc_mode): New function.
7360 (simple_memory_operand): Correct pre/post decrement case.
7361 (no_side_effect_operand): New function.
7362 (pdp11_regno_reg_class): Add CC_REGS class.
7363 (pdp11_fixed_cc_regs): New function.
7364 (pdp11_small_shift): New function.
7365 (pdp11_expand_shift): New function to expand shift insns.
7366 (pdp11_assemble_shift): New function to output shifts.
7367 (pdp11_branch_cost): Remove.
7368 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
7369 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
7370 (WCHAR_TYPE): Ditto.
7371 (PTRDIFF_TYPE): Ditto.
7372 (ADJUST_INSN_LENGTH): New macro.
7373 (FIXED_REGISTERS): Add CC registers.
7374 (CALL_USED_REGISTERS): Ditto.
7376 (REG_CLASS_NAMES): Ditto.
7377 (REG_CLASS_CONTENTS): Ditto.
7378 (SELECT_CC_MODE): Use new function.
7379 (TARGET_FLAGS_REGNUM): New macro.
7380 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
7381 (cc0_reg_rtx): Remove.
7382 (CC_STATUS_MDEP): Remove.
7383 (CC_STATUS_MDEFP_INIT): Remove.
7384 (CC_IN_FPU): Remove.
7385 (NOTICE_UPDATE_CC): Remove.
7386 (REGISTER_NAMES): Add CC registers.
7387 (BRANCH_COST): Change to constant 1.
7388 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
7390 * config/pdp11/pdp11.opt (mbcopy): Remove.
7391 (mbcopy-builtin): Remove.
7392 (mbranch-cheap): Remove.
7393 (mbranch-expensive): Remove.
7394 * config/pdp11/predicates.md (expand_shift_operand): Update to
7395 match shift code generation.
7396 (ccnz_operator): New predicate.
7397 * doc/invoke.texi (PDP-11 Options): Remove deleted options
7398 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
7399 Remove non-existent option -mabshi, -mno-abshi. Document mutually
7401 * doc/md.texi (PDP-11): Document new D and h constraints. Update
7402 description of O constraint.
7404 2018-06-27 Jeff Law <law@redhat.com>
7405 Austin Law <austinklaw@gmail.com>
7407 * config/v850/v850.md (addsi3_set_flags): New pattern.
7408 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
7409 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
7410 (zero_extendhisi2_v850_set_flags): Likewise.
7411 (zero_extendqisi2_v850_set_flags): Likewise.
7412 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
7413 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
7414 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
7416 * config/v850/v850-protos.h (notice_update_cc): Remove.
7417 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
7418 (v850_print_operand): Handle 'D' and "d".
7419 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
7420 Add handling of arithmetic/logical operations compared against zero.
7421 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
7422 Do not look at v850_compare_op, instead get mode from last argument.
7423 (v850_gen_compare): Remove
7424 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
7425 after reload for prologue insns.
7426 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
7428 (construct_save_jarl): Likewise.
7429 (TARGET_FLAGS_REGNUM): Define.
7430 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
7431 (NOTICE_UPDATE_CC): Remove.
7432 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
7433 than cc0. Conditionalize on reload_completed.
7434 (cmpsi_insn, setfcc_insn): Likewise.
7435 (tst1 splitter): Turn into define_and_split which sets the flags
7437 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
7438 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
7439 (cstoresf4, cstoredf4): Clobber the flags.
7440 (cmpsi, cmpsf, cmpdf): Remove expanders.
7441 (setf_insn): Remove pattern.
7442 (addsi3): Turn into define_and_split which clobbers the flags after
7443 reload and a suitable pattern (addsi3_clobber_flags) for use after
7445 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
7446 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
7447 (ashrsi3, ashrsi3_v850e2): Likewise.
7448 (bins): Clobber the flags.
7449 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
7450 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
7451 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
7452 (call_value_internal_short, call_value_internal_long): Likewise.
7453 (callt_save_interrupt, callt_return_interrupt): Likewise.
7454 (save_interrupt, return_interrupt): Likewise.
7455 (callt_save_all_interrupt, save_all_interrupt): Likewise.
7456 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
7457 (restore_all_interrupt, _restore_all_interrupt): Likewise.
7458 (All FP comparisons): Only allow after reload has completed.
7460 (divh, divhu): Tweak output template.
7461 (branch_z_normal, branch_z_invert): Remove
7462 (branch_nz_normal, branch_nz_invert): Likewise.
7463 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
7465 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
7466 * config/v850/v850.c (notice_update_cc): Remove.
7467 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
7468 (CC_NO_CARRY): Likewise.
7469 (NOTICE_UPDATE_CC): Define to nothing.
7470 * config/v850/v850.md: Remove block comment on cc0 handling
7471 Remove "cc" attribute from all patterns. Remove cc_status handling
7472 from all patterns. Minor formatting fixes.
7474 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7476 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
7477 (cortex-a76.cortex-a55): Likewise.
7478 * config/aarch64/aarch64-tune.md: Regenerate.
7479 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
7480 cortex-a76.cortex-a55.
7482 2018-06-27 Jeff Law <law@redhat.com>
7484 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
7485 (MULTILIB_DIRNAMES): Similarly.
7487 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
7489 * gimple.h (gimple_return_retbnd): Delete.
7490 (gimple_return_set_retbnd): Likewise.
7491 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
7492 gimple_return_set_retbnd.
7493 * gimple-pretty-print.c (dump_gimple_return): Remove call to
7494 gimple_return_retbnd and adjust.
7495 * tree-inline.h (struct copy_body_data): Remove retbnd field.
7496 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
7497 Explicitly return NULL in a couple more cases. Move assertion
7498 on debug statements and remove unreachable code.
7499 (reset_debug_binding): Do not test id->retbnd.
7500 (expand_call_inline): Do not set it.
7502 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
7504 * configure.ac: Add --disable-gcov option.
7505 * configure: Regenerate.
7506 * Makefile.in: Honour @enable_gcov@.
7507 * doc/install.texi: Document --disable-gcov.
7509 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7511 * config/arm/arm-cpus.in (cortex-a76): New entry.
7512 (cortex-a76.cortex-a55): Likewise.
7513 * config/arm/arm-tables.opt: Regenerate.
7514 * config/arm/arm-tune.md: Likewise.
7515 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
7516 * doc/invoke.texi (ARM Options): Document cortex-a76 and
7517 cortex-a76.cortex-a55.
7519 2018-06-27 Tamar Christina <tamar.christina@arm.com>
7522 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
7524 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
7526 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
7528 (EPILOGUE_USES): Likewise.
7530 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
7532 * tree-inline.c (remap_location): New function extracted from...
7533 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
7534 (copy_phis_for_bb): ...here. Call remap_location.
7535 (copy_cfg_body): Adjust call to copy_edges_for_bb.
7537 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
7539 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
7540 unaligned vsx for 16B memset.
7542 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
7545 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
7546 ieee128_float_type_node to long_double_type_node unless
7547 TARGET_LONG_DOUBLE_128 is set.
7549 2018-06-26 David Malcolm <dmalcolm@redhat.com>
7551 * cfgloop.c (get_loop_location): Convert return type from
7552 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
7553 by implicit construction from rtx_insn *, and using
7554 dump_user_location_t::from_function_decl for the fallback case.
7555 * cfgloop.h (get_loop_location): Convert return type from
7556 location_t to dump_user_location_t.
7557 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
7558 dump_printf_loc to pass in a dump_location_t rather than a
7559 location_t, via the gimple stmt.
7560 * coverage.c (get_coverage_counts): Update calls to
7561 dump_printf_loc to pass in dump_location_t rather than a
7563 * doc/optinfo.texi (Dump types): Convert example of
7564 dump_printf_loc from taking "locus" to taking "insn". Update
7565 description of the "_loc" calls to cover dump_location_t.
7566 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
7568 (dump_user_location_t::dump_user_location_t): New constructors,
7569 from gimple *stmt and rtx_insn *.
7570 (dump_user_location_t::from_function_decl): New function.
7571 (dump_loc): Make static.
7572 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
7573 const dump_location_t &.
7574 (dump_generic_expr_loc): Delete.
7575 (dump_printf_loc): Convert param "loc" from location_t to
7576 const dump_location_t &.
7577 (selftest::test_impl_location): New function.
7578 (selftest::dumpfile_c_tests): New function.
7579 * dumpfile.h: Include "profile-count.h".
7580 (class dump_user_location_t): New class.
7581 (struct dump_impl_location_t): New struct.
7582 (class dump_location_t): New class.
7583 (dump_printf_loc): Convert 2nd param from source_location to
7584 const dump_location_t &.
7585 (dump_generic_expr_loc): Delete.
7586 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
7587 const dump_location_t &.
7588 * gimple-fold.c (fold_gimple_assign): Update call to
7589 dump_printf_loc to pass in a dump_location_t rather than a
7590 location_t, via the gimple stmt.
7591 (gimple_fold_call): Likewise.
7592 * gimple-loop-interchange.cc
7593 (loop_cand::analyze_iloop_reduction_var): Update for change to
7594 check_reduction_path.
7595 (tree_loop_interchange::interchange): Update for change to
7597 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
7598 change in return-type of find_loop_location.
7599 (graphite_regenerate_ast_isl): Likewise.
7600 * graphite-optimize-isl.c (optimize_isl): Likewise.
7601 * graphite.c (graphite_transform_loops): Likewise.
7602 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
7603 pass in a dump_location_t rather than a location_t, via the
7605 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
7606 * ipa.c (walk_polymorphic_call_targets): Likewise.
7607 * loop-unroll.c (report_unroll): Convert "locus" param from
7608 location_t to dump_location_t.
7609 (decide_unrolling): Update for change to get_loop_location's
7611 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
7612 location_t to dump_user_location_t.
7613 (grid_find_single_omp_among_assignments_1): Updates calls to
7614 dump_printf_loc to pass in a dump_location_t rather than a
7615 location_t, via the gimple stmt.
7616 (grid_parallel_clauses_gridifiable): Convert "tloc" from
7617 location_t to dump_location_t. Updates calls to dump_printf_loc
7618 to pass in a dump_location_t rather than a location_t, via the
7620 (grid_inner_loop_gridifiable_p): Likewise.
7621 (grid_dist_follows_simple_pattern): Likewise.
7622 (grid_gfor_follows_tiling_pattern): Likewise.
7623 (grid_target_follows_gridifiable_pattern): Likewise.
7624 (grid_attempt_target_gridification): Convert initialization
7625 of local "grid" from memset to zero-initialization; FIXME: does
7626 this require C++11? Update call to dump_printf_loc to pass in a
7627 optinfo_location rather than a location_t, via the gimple stmt.
7628 * profile.c (read_profile_edge_counts): Updates call to
7629 dump_printf_loc to pass in a dump_location_t rather than a
7631 (compute_branch_probabilities): Likewise.
7632 * selftest-run-tests.c (selftest::run_tests): Call
7634 * selftest.h (dumpfile_c_tests): New decl.
7635 * tree-loop-distribution.c (pass_loop_distribution::execute):
7636 Update for change in return type of find_loop_location.
7637 * tree-parloops.c (parallelize_loops): Likewise.
7638 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
7639 "locus" from location_t to dump_user_location_t.
7640 (canonicalize_loop_induction_variables): Likewise.
7641 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
7642 for change in return type of find_loop_location.
7643 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
7644 to dump_printf_loc to pass in a dump_location_t rather than a
7645 location_t, via the stmt.
7646 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
7648 * tree-vect-loop-manip.c (find_loop_location): Convert return
7649 type from source_location to dump_user_location_t.
7650 (vect_do_peeling): Update for above change.
7651 (vect_loop_versioning): Update for change in type of
7653 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
7654 from location_t to dump_user_location_t.
7655 (vect_estimate_min_profitable_iters): Update for change in type
7657 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
7658 location_t to dump_location_t.
7659 (vect_slp_bb): Update for change in type of vect_location.
7660 * tree-vectorizer.c (vect_location): Convert from source_location
7661 to dump_user_location_t.
7662 (try_vectorize_loop_1): Update for change in vect_location's type.
7663 (vectorize_loops): Likewise.
7664 (increase_alignment): Likewise.
7665 * tree-vectorizer.h (vect_location): Convert from source_location
7666 to dump_user_location_t.
7667 (find_loop_location): Convert return type from source_location to
7668 dump_user_location_t.
7669 (check_reduction_path): Convert 1st param from location_t to
7670 dump_user_location_t.
7671 * value-prof.c (check_counter): Update call to dump_printf_loc to
7672 pass in a dump_user_location_t rather than a location_t; update
7673 call to error_at for change in type of "locus".
7674 (check_ic_target): Update call to dump_printf_loc to
7675 pass in a dump_user_location_t rather than a location_t, via the
7678 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
7680 * config/s390/s390.h (enum processor_flags): Do not use
7681 default tune parameter when -march was specified.
7683 2018-06-26 Jakub Jelinek <jakub@redhat.com>
7686 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
7687 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
7690 2018-06-26 Richard Biener <rguenther@suse.de>
7692 PR tree-optimization/86287
7694 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
7695 (vect_analyze_loop): Initialize n_stmts.
7697 2018-06-26 Richard Biener <rguenther@suse.de>
7700 * fold-const.c (fold_convertible_p): Pointer extension
7703 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
7706 * dwarf2out.c (loc_list_has_views): Adjust comments.
7707 (dw_loc_list): Split single cross-partition range with
7710 2018-06-25 Jeff Law <law@redhat.com>
7712 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
7713 on -mbig-switch by default.
7715 * config/v850/predicates.md (const_float_1_operand): Fix match_code
7717 (const_float_0_operand): Remove unused predicate.
7718 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
7719 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
7720 (recipsf2): New expander. Original pattern now called
7722 (recipdf2, recipdf2_insn): Similarly.
7723 (rsqrtsf2, rsqrtsf2_insn): Similarly
7724 (rsqrtdf2, rsqrtdf2_insn): Similarly
7726 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
7728 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
7729 Simplify logic for FreeBSD (twice).
7731 2018-06-25 Martin Sebor <msebor@redhat.com>
7733 PR tree-optimization/86204
7734 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
7735 a strnlen result if it's less than the length of the string.
7737 2018-06-25 Martin Sebor <msebor@redhat.com>
7739 PR tree-optimization/85700
7740 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
7741 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
7742 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
7744 2018-06-25 Martin Sebor <msebor@redhat.com>
7746 * doc/extend.texi (Zero-length arrays): Update and clarify.
7748 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
7750 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
7751 added IEEE/IBM long double multilib support on PowerPC little
7752 endian Linux systems.
7753 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
7754 (MULTILIB_DEFAULTS): Likewise.
7755 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7757 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
7758 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
7759 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
7761 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
7764 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
7765 (REORDER_45): Likewise.
7767 2018-06-25 Jeff Law <law@redhat.com>
7769 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
7770 dividend to 32 bits. Adjust length.
7771 (udivmodhi4): Cleanup output template. Fix length.
7773 2018-06-25 Carl Love <cel@us.ibm.com>
7775 * config/rs6000/vsx.md: Change word selector to prefered location.
7777 2018-06-25 Richard Biener <rguenther@suse.de>
7779 PR tree-optimization/86304
7780 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
7781 epilogue-if-converted loops as well.
7783 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
7785 * lto-section-out.c (lto_begin_section): Do not print section
7786 name for noaddr and unnumbered dumps.
7788 2018-06-25 Richard Biener <rguenther@suse.de>
7790 * tree-vectorizer.h (struct vec_info_shared): New structure
7791 with parts split out from struct vec_info and loop_nest from
7792 struct _loop_vec_info.
7793 (struct vec_info): Adjust accordingly.
7794 (struct _loop_vec_info): Likewise.
7795 (LOOP_VINFO_LOOP_NEST): Adjust.
7796 (LOOP_VINFO_DATAREFS): Likewise.
7797 (LOOP_VINFO_DDRS): Likewise.
7798 (struct _bb_vec_info): Likewise.
7799 (BB_VINFO_DATAREFS): Likewise.
7800 (BB_VINFO_DDRS): Likewise.
7801 (struct _stmt_vec_info): Add dr_aux member.
7802 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
7803 (DR_MISALIGNMENT_UNINITIALIZED): New.
7804 (set_dr_misalignment): Adjust.
7805 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
7806 (vect_analyze_loop): Adjust prototype.
7807 (vect_analyze_loop_form): Likewise.
7808 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
7809 Compute dependences lazily.
7810 (vect_record_base_alignments): Use shared datarefs/ddrs.
7811 (vect_verify_datarefs_alignment): Likewise.
7812 (vect_analyze_data_refs_alignment): Likewise.
7813 (vect_analyze_data_ref_accesses): Likewise.
7814 (vect_analyze_data_refs): Likewise.
7815 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
7816 constructor parameter for shared part.
7817 (vect_analyze_loop_form): Pass in shared part and adjust.
7818 (vect_analyze_loop_2): Pass in storage for the number of
7819 stmts. Move loop nest finding to the caller. Compute
7821 (vect_analyze_loop): Pass in shared part.
7822 (vect_transform_loop): Verify shared datarefs are unchanged.
7823 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
7824 constructor parameter for shared part.
7825 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
7826 (vect_slp_bb): Verify shared datarefs are unchanged before
7828 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
7830 (new_stmt_vec_info): Initialize DR_AUX misalignment to
7831 DR_MISALIGNMENT_UNINITIALIZED.
7832 * tree-vectorizer.c (vec_info::vec_info): Add constructor
7833 parameter for shared part.
7834 (vec_info::~vec_info): Adjust.
7835 (vec_info_shared::vec_info_shared): New.
7836 (vec_info_shared::~vec_info_shared): Likewise.
7837 (vec_info_shared::save_datarefs): Likewise.
7838 (vec_info_shared::check_datarefs): Likewise.
7839 (try_vectorize_loop_1): Construct shared part live for analyses
7840 of a single loop for multiple vector sizes.
7841 * tree-parloops.c (gather_scalar_reductions): Adjust.
7843 2018-06-25 Richard Biener <rguenther@suse.de>
7845 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
7846 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
7847 (vect_analyze_data_refs): Remove similar code from here and
7848 simplify accordingly.
7850 2018-06-25 Richard Biener <rguenther@suse.de>
7852 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
7853 for reverse storage order accesses rather than asserting
7854 they cannot happen here.
7856 2018-06-25 Tom de Vries <tdevries@suse.de>
7859 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
7860 Use data16 instead of .byte for insn prefix.
7862 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
7865 * parser.c (make_char_string_pack): Pass this literal chars
7866 through cpp_interpret_string.
7867 (cp_parser_userdef_numeric_literal): Check the result of
7868 make_char_string_pack.
7870 2018-06-24 Maya Rashish <coypu@sdf.org>
7872 * ginclude/stddef.h: Simplify conditions around avoiding
7873 re-definition of __size_t.
7875 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
7877 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
7878 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
7880 2018-06-22 Maya Rashish <coypu@sdf.org>
7882 * doc/invoke.texi (mno-fancy-math-387): Update for changes
7883 made to OpenBSD and NetBSD through the years.
7885 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
7887 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
7888 behavior of vec_pack (vector double, vector double) to match
7889 behavior of vec_float2 (vector double, vector double).
7891 2018-06-22 Olivier Hainque <hainque@adacore.com>
7893 * gimplify.c (gimplify_function_tree): Prevent creation
7894 of a trampoline for the address of the current function
7895 passed to entry/exit instrumentation hooks.
7897 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
7900 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
7903 2018-06-22 Martin Liska <mliska@suse.cz>
7905 PR tree-optimization/86263
7906 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
7907 Bail out if is_enabled is false.
7908 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
7910 (jump_table_cluster::is_enabled): New function.
7912 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
7914 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
7915 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
7916 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
7917 (lto_input_ts_binfo_tree_pointers): Likewise.
7918 * tree-streamer-out.c (streamer_write_tree_bitfields,
7919 write_ts_binfo_tree_pointers): Likewise.
7920 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
7922 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
7924 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
7926 2018-06-22 Martin Liska <mliska@suse.cz>
7928 * symbol-summary.h (get): Make it pure and inline move
7929 functionality from ::get function.
7930 (get): Remove and inline into ::get and ::get_create.
7931 (get_create): Move code from ::get function.
7933 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7936 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
7937 -x assembler-with-cpp.
7939 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7941 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
7942 _FILE_OFFSET_BITS=64 for C++.
7944 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
7946 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
7947 conversion insn that shows up when pr85657-3.c is compiled using
7948 IEEE 128-bit long double.
7949 (neg<mode>2_internal): Use the correct mode to check whether the
7950 mode is IBM extended.
7951 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
7952 multiply and divide external functions from being created more
7955 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
7957 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
7959 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
7960 the edge can be forwarded.
7961 (cfg_layout_merge_blocks): Likewise.
7963 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
7965 * except.c (finish_eh_generation): Commit edge insertions only after
7966 the EH edges have been redirected from post-landing to landing pads.
7968 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
7970 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
7971 create_tmp_var_for to create the FRAME decl.
7972 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
7974 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
7976 * tree-inline.c (copy_edges_for_bb): Minor tweak.
7977 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
7978 debug statement when resetting its value.
7979 (expand_call_inline): Copy the locus of the call onto the assignment
7980 of the return value, if any. Use local variable in more cases.
7982 2018-06-21 Martin Liska <mliska@suse.cz>
7984 * ipa-pure-const.c (propagate_nothrow): Use
7985 funct_state_summaries->get.
7986 (dump_malloc_lattice): Likewise.
7987 (propagate_malloc): Likewise.
7989 2018-06-21 Richard Biener <rguenther@suse.de>
7991 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
7992 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
7993 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
7995 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
7996 BLOCK_ABSTRACT_ORIGIN unconditionally.
7998 2018-06-21 David Malcolm <dmalcolm@redhat.com>
8000 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
8002 * ipa-reference.c (ipa_reference_c_finalize): Delete
8003 ipa_ref_opt_sum_summaries and set it to NULL.
8005 2018-06-21 Tom de Vries <tdevries@suse.de>
8007 PR tree-optimization/85859
8008 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
8009 test with comment from bb_no_side_effects_p.
8011 2018-06-21 Richard Biener <rguenther@suse.de>
8013 PR tree-optimization/86232
8014 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
8015 max for constant niter.
8017 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
8019 * config/aarch64/aarch64-simd.md
8020 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
8022 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
8024 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
8025 Make opernads of the unspec commutative.
8027 2018-06-21 Richard Biener <rguenther@suse.de>
8029 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
8030 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
8031 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
8032 (vect_analyze_data_ref_dependence): Re-order checks to deal with
8034 (vect_record_base_alignments): Do not record base alignment
8035 for gathers or scatters.
8036 (vect_compute_data_ref_alignment): Drop return value that is always
8037 true. Bail out early for gathers or scatters.
8038 (vect_enhance_data_refs_alignment): Bail out early for gathers
8040 (vect_find_same_alignment_drs): Likewise.
8041 (vect_analyze_data_refs_alignment): Remove dead code.
8042 (vect_slp_analyze_and_verify_node_alignment): Likewise.
8043 (vect_analyze_data_refs): For possible gathers or scatters do
8044 not create an alternate DR, just check their possible validity
8045 and mark them. Adjust DECL_NONALIASED handling to not rely
8047 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
8048 update inits of gathers or scatters.
8049 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
8050 Also copy gather/scatter flag to pattern vinfo.
8052 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
8054 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
8055 behavior of vec_packsu (vector unsigned long long, vector unsigned
8056 long long) to match behavior of vec_packs with same signature.
8058 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
8059 Thomas Schwinge <thomas@codesourcery.com>
8060 Cesar Philippidis <cesar@codesourcery.com>
8062 * gimplify.c (gimplify_scan_omp_clauses): Add support for
8063 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
8064 (gimplify_adjust_omp_clauses): Likewise.
8065 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
8066 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
8067 (gimplify_omp_target_update): Update handling of acc update and
8069 * omp-low.c (install_var_field): Remove unused parameter
8070 base_pointers_restrict.
8071 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
8072 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
8074 (omp_target_base_pointers_restrict_p): Delete.
8075 (scan_omp_target): Update call to scan_sharing_clauses.
8076 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
8078 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
8079 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
8080 (convert_local_omp_clauses): Likewise.
8081 * tree-pretty-print.c (dump_omp_clause): Likewise.
8082 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
8084 (omp_clause_code_name): Likewise.
8086 2018-06-20 Jakub Jelinek <jakub@redhat.com>
8089 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
8092 PR tree-optimization/86231
8093 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
8094 anti-range don't overwrite *vr0min before using it to compute *vr0max.
8096 2018-06-20 Tom de Vries <tdevries@suse.de>
8098 PR tree-optimization/86097
8099 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
8100 iv type if signedness of iv type is not the same as that of *nit.
8102 2018-06-20 Jakub Jelinek <jakub@redhat.com>
8104 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
8105 EDGE_EH edges, verify they are all EDGE_EH.
8107 2018-06-20 Maya Rashish <coypu@sdf.org>
8109 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
8111 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8113 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
8114 * config/aarch64/aarch64.c (xgene1_tunings): Add
8115 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
8116 (aarch64_mode_valid_for_sched_fusion_p):
8117 Allow 16-byte modes.
8118 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
8119 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
8121 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
8123 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
8124 * config/aarch64/iterators.md (VQ2): New mode iterator.
8126 2018-06-20 Martin Liska <mliska@suse.cz>
8128 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
8129 Change default ratio from 10 to 8.
8131 2018-06-20 Martin Liska <mliska@suse.cz>
8133 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
8135 (bit_test_cluster::find_bit_tests): Likewise.
8136 (switch_decision_tree::analyze_switch_statement): Find clusters.
8137 * tree-switch-conversion.h (struct jump_table_cluster): Document
8140 2018-06-20 Martin Liska <mliska@suse.cz>
8142 * tree-switch-conversion.c (switch_conversion::collect):
8143 Record m_uniq property.
8144 (switch_conversion::expand): Bail out for special conditions.
8145 (group_cluster::~group_cluster): New.
8146 (group_cluster::group_cluster): Likewise.
8147 (group_cluster::dump): Likewise.
8148 (jump_table_cluster::emit): New.
8149 (switch_decision_tree::fix_phi_operands_for_edges): New.
8150 (struct case_node): Remove struct.
8151 (jump_table_cluster::can_be_handled): New.
8152 (case_values_threshold): Moved to header.
8153 (reset_out_edges_aux): Likewise.
8154 (jump_table_cluster::is_beneficial): New.
8155 (bit_test_cluster::can_be_handled): Likewise.
8156 (add_case_node): Remove.
8157 (bit_test_cluster::is_beneficial): New.
8158 (case_bit_test::cmp): New.
8159 (bit_test_cluster::emit): New.
8160 (expand_switch_as_decision_tree_p): Remove.
8161 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
8162 (fix_phi_operands_for_edge): Likewise.
8163 (switch_decision_tree::analyze_switch_statement): New.
8164 (compute_cases_per_edge): Move ...
8165 (switch_decision_tree::compute_cases_per_edge): ... here.
8166 (try_switch_expansion): Likewise.
8167 (switch_decision_tree::try_switch_expansion): Likewise.
8168 (record_phi_operand_mapping): Likewise.
8169 (switch_decision_tree::record_phi_operand_mapping): Likewise.
8170 (emit_case_decision_tree): Likewise.
8171 (switch_decision_tree::emit): Likewise.
8172 (balance_case_nodes): Likewise.
8173 (switch_decision_tree::balance_case_nodes): Likewise.
8174 (dump_case_nodes): Likewise.
8175 (switch_decision_tree::dump_case_nodes): Likewise.
8176 (emit_jump): Likewise.
8177 (switch_decision_tree::emit_jump): Likewise.
8178 (emit_cmp_and_jump_insns): Likewise.
8179 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
8180 (emit_case_nodes): Likewise.
8181 (switch_decision_tree::emit_case_nodes): Likewise.
8182 (conditional_probability): Remove.
8183 * tree-switch-conversion.h (enum cluster_type): New.
8185 (struct cluster): Likewise.
8186 (cluster::cluster): Likewise.
8187 (struct simple_cluster): Likewise.
8188 (simple_cluster::simple_cluster): Likewise.
8189 (struct group_cluster): Likewise.
8190 (struct jump_table_cluster): Likewise.
8191 (struct bit_test_cluster): Likewise.
8192 (struct min_cluster_item): Likewise.
8193 (struct case_tree_node): Likewise.
8194 (case_tree_node::case_tree_node): Likewise.
8195 (jump_table_cluster::case_values_threshold): Likewise.
8196 (struct case_bit_test): Likewise.
8197 (struct switch_decision_tree): Likewise.
8198 (struct switch_conversion): Likewise.
8199 (switch_decision_tree::reset_out_edges_aux): Likewise.
8201 2018-06-20 Martin Liska <mliska@suse.cz>
8203 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
8204 (hoist_edge_and_branch_if_true): Likewise.
8205 (expand_switch_using_bit_tests_p): Likewise.
8206 (struct case_bit_test): Likewise.
8207 (case_bit_test_cmp): Likewise.
8208 (emit_case_bit_tests): Likewise.
8209 (switch_conversion::switch_conversion): New class.
8210 (struct switch_conv_info): Remove old struct.
8211 (collect_switch_conv_info): More to ...
8212 (switch_conversion::collect): ... this.
8213 (check_range): Likewise.
8214 (switch_conversion::check_range): Likewise.
8215 (check_all_empty_except_final): Likewise.
8216 (switch_conversion::check_all_empty_except_final): Likewise.
8217 (check_final_bb): Likewise.
8218 (switch_conversion::check_final_bb): Likewise.
8219 (create_temp_arrays): Likewise.
8220 (switch_conversion::create_temp_arrays): Likewise.
8221 (free_temp_arrays): Likewise.
8222 (gather_default_values): Likewise.
8223 (switch_conversion::gather_default_values): Likewise.
8224 (build_constructors): Likewise.
8225 (switch_conversion::build_constructors): Likewise.
8226 (constructor_contains_same_values_p): Likewise.
8227 (switch_conversion::contains_same_values_p): Likewise.
8228 (array_value_type): Likewise.
8229 (switch_conversion::array_value_type): Likewise.
8230 (build_one_array): Likewise.
8231 (switch_conversion::build_one_array): Likewise.
8232 (build_arrays): Likewise.
8233 (switch_conversion::build_arrays): Likewise.
8234 (gen_def_assigns): Likewise.
8235 (switch_conversion::gen_def_assigns): Likewise.
8236 (prune_bbs): Likewise.
8237 (switch_conversion::prune_bbs): Likewise.
8238 (fix_phi_nodes): Likewise.
8239 (switch_conversion::fix_phi_nodes): Likewise.
8240 (gen_inbound_check): Likewise.
8241 (switch_conversion::gen_inbound_check): Likewise.
8242 (process_switch): Use the newly created class.
8243 (switch_conversion::expand): New.
8244 (switch_conversion::~switch_conversion): New.
8245 * tree-switch-conversion.h: New file.
8247 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8249 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
8250 tree-vect-patterns.c.
8251 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
8252 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
8253 (vect_recog_sad_pattern): Likewise.
8254 (vect_recog_widen_sum_pattern): Likewise.
8255 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
8256 (vect_recog_widen_shift_pattern): Remove the type_in argument.
8257 (vect_recog_rotate_pattern): Likewise.
8258 (vect_recog_mult_pattern): Likewise.
8259 (vect_recog_vector_vector_shift_pattern): Likewise.
8260 (vect_recog_divmod_pattern): Likewise.
8261 (vect_recog_mixed_size_cond_pattern): Likewise.
8262 (vect_recog_bool_pattern): Likewise.
8263 (vect_recog_mask_conversion_pattern): Likewise.
8264 (vect_try_gather_scatter_pattern): Likewise.
8265 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
8266 (vect_recog_over_widening_pattern): Likewise.
8267 (vect_recog_gather_scatter_pattern): Likewise.
8268 (vect_recog_func_ptr): Move from tree-vectorizer.h
8269 (vect_vect_recog_func_ptrs): Move further down the file.
8270 (vect_recog_func): Likewise. Remove the third argument.
8271 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
8272 (vect_pattern_recog_1): Expect the pattern function to do any
8273 necessary target tests. Also expect it to provide a vector type.
8274 Remove the type_in handling.
8276 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8278 * tree-vect-patterns.c (vect_pattern_detected): New function.
8279 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
8280 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
8281 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
8282 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
8283 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
8284 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
8285 (vect_recog_mask_conversion_pattern)
8286 (vect_try_gather_scatter_pattern): Likewise.
8288 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8290 * tree-vect-patterns.c (vect_get_internal_def): New function.
8291 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
8292 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
8293 (search_type_for_mask_1): Use it.
8295 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8297 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
8298 redundant WIDEN_SUM_EXPR handling.
8299 (vect_recog_sad_pattern): Likewise.
8301 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8303 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
8304 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
8305 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
8306 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
8307 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
8309 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8311 * tree-vect-stmts.c (vectorizable_call): Make sure that we
8312 use the stmt_vec_info of the original bb statement for the
8313 new zero assignment, even if the call is part of a pattern.
8315 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8317 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
8318 that the sequence is attached to the original statement rather
8319 than the pattern statement.
8320 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
8321 PATTERN_DEF_SEQ from the original statement rather than
8322 the main pattern statement.
8323 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
8324 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
8325 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
8327 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8329 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
8330 definition statements before the early exit for statements that aren't
8332 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
8334 (vect_transform_loop): ...here. Process pattern definition
8335 statements without first checking whether the main pattern
8336 statement is live or relevant.
8338 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
8340 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
8341 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
8343 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
8345 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
8346 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
8347 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
8348 (expand_block_compare): Change select_block_compare_mode call.
8349 (expand_strncmp_align_check): Use new functions, fix comment.
8350 (emit_final_str_compare_gpr): New function.
8351 (expand_strn_compare): Refactor and clean up code.
8352 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
8354 2018-06-19 Tony Reix <tony.reix@atos.com>
8355 Damien Bergamini <damien.bergamini@atos.com>
8356 David Edelsohn <dje.gcc@gmail.com>
8358 * collect2.c (static_obj): New variable.
8359 (static_libs): New variable.
8360 (is_in_list): Uncomment declaration.
8361 (main): Track AIX libraries linked statically.
8362 (is_in_list): Uncomment definition.
8363 (scan_prog_file): Don't add AIX shared libraries initializer
8364 to constructor list if linking statically.
8366 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
8368 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
8370 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
8372 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
8374 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
8377 2018-06-19 Martin Liska <mliska@suse.cz>
8379 * config/i386/i386.c (ix86_can_inline_p): Do not use
8380 ipa_fn_summaries::get_create.
8381 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
8383 (devirtualization_time_bonus): Likewise.
8384 (ipcp_propagate_stage): Likewise.
8385 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
8386 (edge_set_predicate): Likewise.
8387 (evaluate_conditions_for_known_args): Likewise.
8388 (evaluate_properties_for_edge): Likewise.
8389 (ipa_call_summary::reset): Tranform to ...
8390 (ipa_call_summary::~ipa_call_summary): ... this.
8391 (ipa_fn_summary::reset): Transform to ...
8392 (ipa_fn_summary::~ipa_fn_summary): ... this.
8393 (ipa_fn_summary_t::remove): Rename to ...
8394 (ipa_fn_summary_t::remove_callees): ... this.
8395 (ipa_fn_summary_t::duplicate): Use placement new
8396 instead of memory copy.
8397 (ipa_call_summary_t::duplicate): Likewise.
8398 (ipa_call_summary_t::remove): Remove.
8399 (dump_ipa_call_summary): Change get_create to get.
8400 (ipa_dump_fn_summary): Dump only when summary exists.
8401 (analyze_function_body): Use symbol_summary::get instead
8403 (compute_fn_summary): Likewise.
8404 (estimate_edge_devirt_benefit): Likewise.
8405 (estimate_edge_size_and_time): Likewise.
8406 (inline_update_callee_summaries): Likewise.
8407 (remap_edge_change_prob): Likewise.
8408 (remap_edge_summaries): Likewise.
8409 (ipa_merge_fn_summary_after_inlining): Likewise.
8410 (write_ipa_call_summary): Likewise.
8411 (ipa_fn_summary_write): Likewise.
8412 (ipa_free_fn_summary): Likewise.
8413 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
8414 (struct ipa_call_summary): Likewise.
8415 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
8417 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
8418 (estimate_size_after_inlining): Likewise.
8419 (estimate_growth): Likewise.
8420 (growth_likely_positive): Likewise.
8421 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
8422 (inline_call): Likewise.
8423 * ipa-inline.c (caller_growth_limits): Likewise.
8424 (can_inline_edge_p): Likewise.
8425 (can_inline_edge_by_limits_p): Likewise.
8426 (compute_uninlined_call_time): Likewise.
8427 (compute_inlined_call_time): Likewise.
8428 (want_inline_small_function_p): Likewise.
8429 (edge_badness): Likewise.
8430 (update_caller_keys): Likewise.
8431 (update_callee_keys): Likewise.
8432 (inline_small_functions): Likewise.
8433 (inline_to_all_callers_1): Likewise.
8434 (dump_overall_stats): Likewise.
8435 (early_inline_small_functions): Likewise.
8436 (early_inliner): Likewise.
8437 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
8438 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
8439 * ipa-pure-const.c (malloc_candidate_p): Likewise.
8440 * ipa-split.c (execute_split_functions): Likewise.
8441 * symbol-summary.h: Likewise.
8442 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
8444 2018-06-19 Richard Biener <rguenther@suse.de>
8446 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
8447 (vectorize_loops): ... here. Fix dbgcnt handling.
8448 (try_vectorize_loop): Wrap try_vectorize_loop_1.
8450 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
8453 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
8454 ieee128 argument takes up only one (vector) register, not two (floating
8457 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
8459 * gimplify.c (gimplify_init_constructor): Really never clear for an
8460 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
8462 2018-06-19 Richard Biener <rguenther@suse.de>
8464 PR tree-optimization/86179
8465 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
8466 after failed recognition.
8468 2018-06-18 Martin Sebor <msebor@redhat.com>
8471 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
8472 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
8473 Handle integer subtraction.
8474 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
8475 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
8477 2018-06-18 David Malcolm <dmalcolm@redhat.com>
8479 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
8480 param from rtx to rtx_insn *.
8481 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
8483 (frv_ifcvt_modify_insn): Likwise.
8484 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
8485 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
8486 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
8487 as_a <rtx_insn *> cast to local "unprotected_region" once
8488 it's been established that it's not NULL or pc_rtx.
8489 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
8490 param "sethi" from rtx to rtx_insn *.
8491 (nds32_group_float_insns): Likewise for param "insn".
8492 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
8494 (vax_output_int_subtract): Likewise.
8495 * config/vax/vax.c (vax_output_int_add): Likewise for param
8497 (vax_output_int_subtract): Likewise.
8498 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
8499 (emit_pattern_after): Likewise for param "after".
8500 (emit_insn_after): Likewise.
8501 (emit_jump_insn_after): Likewise.
8502 (emit_call_insn_after): Likewise.
8503 (emit_debug_insn_after): Likewise.
8504 (emit_pattern_before): Likewise for param "before".
8505 (emit_insn_before): Likewise.
8506 (emit_jump_insn_before): Likewise.
8507 * final.c (get_insn_template): Likewise for param "insn", removing
8509 * output.h (get_insn_template): Likewise for 2nd param.
8510 * rtl.h (emit_insn_before): Likewise.
8511 (emit_jump_insn_before): Likewise.
8512 (emit_debug_insn_before_noloc): Likewise.
8513 (emit_insn_after): Likewise.
8514 (emit_jump_insn_after): Likewise.
8515 (emit_call_insn_after): Likewise.
8516 (emit_debug_insn_after): Likewise.
8517 (set_insn_deleted): Likewise for param.
8519 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
8522 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
8523 floating point modes, so that IFmode is numerically greater than
8524 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
8525 to declare the ordering. This prevents IFmode from being
8526 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
8527 machine. Include rs6000-modes.h to share the fractional values
8528 between genmodes* and the rest of the compiler.
8532 * config/rs6000/rs6000-modes.h: New file.
8533 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
8534 meaning of rs6000_long_double_size so that 126..128 selects an
8535 appropriate 128-bit floating point type.
8536 (rs6000_option_override_internal): Likewise.
8537 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
8538 (TARGET_LONG_DOUBLE_128): Change the meaning of
8539 rs6000_long_double_size so that 126..128 selects an appropriate
8540 128-bit floating point type.
8541 (LONG_DOUBLE_TYPE_SIZE): Update comment.
8542 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
8543 source and destination to match the standard usage.
8544 (truncifkf2): Likewise.
8545 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
8546 ISA 2.07 to use an explicit clobber, instead of passing in a
8548 (copysign<mode>3_soft): Likewise.
8550 2018-06-18 David Malcolm <dmalcolm@redhat.com>
8552 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
8553 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
8554 (vect_slp_analyze_instance_dependence): Likewise.
8555 (vect_enhance_data_refs_alignment): Likewise.
8556 (vect_analyze_data_refs_alignment): Likewise.
8557 (vect_slp_analyze_and_verify_instance_alignment
8558 (vect_analyze_data_ref_accesses): Likewise.
8559 (vect_prune_runtime_alias_test_list): Likewise.
8560 (vect_analyze_data_refs): Likewise.
8561 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
8562 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
8563 (vect_analyze_scalar_cycles_1): Likewise.
8564 (vect_get_loop_niters): Likewise.
8565 (vect_analyze_loop_form_1): Likewise.
8566 (vect_update_vf_for_slp): Likewise.
8567 (vect_analyze_loop_operations): Likewise.
8568 (vect_analyze_loop): Likewise.
8569 (vectorizable_induction): Likewise.
8570 (vect_transform_loop): Likewise.
8571 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
8572 * tree-vect-slp.c (vect_analyze_slp): Likewise.
8573 (vect_make_slp_decision): Likewise.
8574 (vect_detect_hybrid_slp): Likewise.
8575 (vect_slp_analyze_operations): Likewise.
8576 (vect_slp_bb): Likewise.
8577 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
8578 (vectorizable_bswap): Likewise.
8579 (vectorizable_call): Likewise.
8580 (vectorizable_simd_clone_call): Likewise.
8581 (vectorizable_conversion): Likewise.
8582 (vectorizable_assignment): Likewise.
8583 (vectorizable_shift): Likewise.
8584 (vectorizable_operation): Likewise.
8585 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
8587 2018-06-18 Martin Sebor <msebor@redhat.com>
8589 PR tree-optimization/81384
8590 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
8591 * builtins.c (expand_builtin_strnlen): New function.
8592 (expand_builtin): Call it.
8593 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
8594 * builtins.def (BUILT_IN_STRNLEN): New.
8595 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
8596 Warn for bounds in excess of maximum object size.
8597 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
8598 single-value ranges. Handle strnlen.
8599 (handle_builtin_strlen): Handle strnlen.
8600 (strlen_check_and_optimize_stmt): Same.
8601 * doc/extend.texi (Other Builtins): Document strnlen.
8603 2018-06-18 Maya Rashish <coypu@sdf.org>
8605 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
8606 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
8607 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
8609 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
8611 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
8613 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8615 * tree.c (escaped_string::escape): Replace cast to char * by
8616 const_cast<char *> (unescaped).
8618 2018-06-18 Nick Clifton <nickc@redhat.com>
8621 * tree.c (escaped_string): New class. Converts an unescaped
8622 string into its escaped equivalent.
8623 (warn_deprecated_use): Use the new class to convert the
8624 deprecation message, if present.
8625 (test_escaped_strings): New self test.
8626 (test_c_tests): Add test_escaped_strings.
8627 * doc/extend.texi (deprecated): Add a note that the
8628 deprecation message is affected by the -fmessage-length
8629 option, and that control characters will be escaped.
8630 (#pragma GCC error): Document this pragma.
8631 (#pragma GCC warning): Likewise.
8632 * doc/invoke.texi (-fmessage-length): Document this option's
8633 effect on the #warning and #error preprocessor directives and
8634 the deprecated attribute.
8636 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
8638 * tree.c (decl_value_expr_lookup): Revert latest change.
8639 (decl_value_expr_insert): Likewise.
8641 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
8643 * gimplify.c (nonlocal_vlas): Delete.
8644 (nonlocal_vla_vars): Likewise.
8645 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
8647 (gimplify_body): Do not create and destroy nonlocal_vlas.
8648 * tree-nested.c: Include diagnostic.h.
8649 (use_pointer_in_frame): Tweak.
8650 (lookup_field_for_decl): Add assertion and declare the transformation.
8651 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
8652 internal error when the reference is in a wrong context. Do not
8653 create a debug decl by default.
8654 (note_nonlocal_block_vlas): Delete.
8655 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
8656 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
8657 create a debug decl by default.
8658 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
8659 call is in a wrong context.
8660 (fixup_vla_decls): New function.
8661 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
8662 debug variables were created.
8663 * tree.c (decl_value_expr_lookup): Add checking assertion.
8664 (decl_value_expr_insert): Likewise.
8666 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
8669 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
8670 * tree-scalar-evolution.c (interpret_expr): Likewise.
8671 (expression_expensive_p): Likewise.
8672 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
8673 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
8674 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
8675 (ssa_defined_by_minus_one_stmt_p): New.
8677 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
8680 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
8681 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
8682 * dojump.c (do_jump): Likewise.
8683 * expr.c (expand_expr_real_2): Check operand type's sign.
8684 * fold-const.c (const_unop): Handle ABSU_EXPR.
8685 (fold_abs_const): Likewise.
8686 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
8687 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
8688 (strip_sign_op_1): Likesise.
8689 * match.pd: Add new pattern to generate ABSU_EXPR.
8690 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
8691 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
8692 * tree-eh.c (operation_could_trap_helper_p): Likewise.
8693 * tree-inline.c (estimate_operator_cost): Likewise.
8694 * tree-pretty-print.c (dump_generic_node): Likewise.
8695 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
8696 * tree.def (ABSU_EXPR): New.
8698 2018-06-16 Jakub Jelinek <jakub@redhat.com>
8701 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
8702 documented as preserved for backward compatibility only.
8703 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
8705 PR rtl-optimization/86108
8706 * bb-reorder.c (create_forwarder_block): Renamed to ...
8707 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
8708 jump from new landing pad to the second part.
8709 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
8712 2018-06-15 Jakub Jelinek <jakub@redhat.com>
8715 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
8716 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
8717 Only call store_expr for halves if the mode is the same.
8720 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
8721 Fix up comment formatting.
8723 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8725 * typed-splay-tree.h (typed_splay_tree::remove): New function.
8726 (typed_splay_tree::closure,
8727 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
8728 (typed_splay_tree::typed_splay_tree,
8729 typed_splay_tree::operator =): Declared private.
8730 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
8731 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
8732 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
8733 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
8734 typed_splay_tree::splay_tree_splay,
8735 typed_splay_tree::splay_tree_foreach_helper,
8736 typed_splay_tree::splay_tree_insert,
8737 typed_splay_tree::splay_tree_remove,
8738 typed_splay_tree::splay_tree_lookup,
8739 typed_splay_tree::splay_tree_predecessor,
8740 typed_splay_tree::splay_tree_successor,
8741 typed_splay_tree::splay_tree_min,
8742 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
8743 (typed_splay_tree::root, typed_splay_tree::comp,
8744 typed_splay_tree::delete_key,
8745 typed_splay_tree::delete_value): New data members.
8746 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
8747 typed_splay_tree::remove.
8749 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
8751 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
8752 -mginv and -mno-ginv to the assembler.
8753 * config/mips/mips.opt (-mcrc): New option.
8755 * doc/invoke.text (-mcrc): Document.
8758 2018-06-15 Nick Clifton <nickc@redhat.com>
8761 * tree.c (escaped_string): New class. Converts an unescaped
8762 string into its escaped equivalent.
8763 (warn_deprecated_use): Use the new class to convert the
8764 deprecation message, if present.
8765 (test_escaped_strings): New self test.
8766 (test_c_tests): Add test_escaped_strings.
8767 * doc/extend.texi (deprecated): Add a note that the
8768 deprecation message is affected by the -fmessage-length
8769 option, and that control characters will be escaped.
8770 (#pragma GCC error): Document this pragma.
8771 (#pragma GCC warning): Likewise.
8772 * doc/invoke.texi (-fmessage-length): Document this option's
8773 effect on the #warning and #error preprocessor directives and
8774 the deprecated attribute.
8776 2018-06-15 Richard Biener <rguenther@suse.de>
8778 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
8779 here, also noting vector size used.
8780 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
8781 size used in MSG_OPTIMIZED_LOCATIONS dump.
8782 (pass_slp_vectorize::execute): Adjust.
8784 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
8787 * config/arc/arc.c (arc_return_address_register): Fix
8790 2018-06-15 Richard Biener <rguenther@suse.de>
8793 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
8794 leave useless conversion stripping to force_gimple_operand_gsi.
8795 (gimplify_build2): Likewise.
8796 (gimplify_build1): Likewise.
8798 2018-06-15 Richard Biener <rguenther@suse.de>
8801 * tree-cfg.c (move_stmt_op): unshare invariant addresses
8802 before adjusting their block.
8804 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
8806 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
8807 multilibs for *-*-rtems*.
8808 * config/riscv/t-rtems: New file.
8810 2018-06-14 Jakub Jelinek <jakub@redhat.com>
8813 * match.pd ((A +- CST1) +- CST2): Punt if last resort
8814 unsigned_type_for returns NULL.
8817 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
8818 subregs of multi-word pseudos unless the float mode has word size.
8820 2018-06-14 Richard Biener <rguenther@suse.de>
8823 * tree-vect-generic.c (build_word_mode_vector_type): Remove
8824 duplicate and harmful type_hash_canon.
8825 * tree.c (type_hash_canon): Assert we didn't find ourselves.
8827 2018-06-14 Richard Biener <rguenther@suse.de>
8830 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
8833 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
8835 * config/rtems.h (STDINT_LONG32): Define.
8837 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
8838 Prachi Godbole <prachi.godbole@imgtec.com>
8840 * config/mips/mips-cpus.def: Define P6600.
8841 * config/mips/mips-tables.opt: Regenerate.
8842 * config/mips/mips.c (mips_ucbranch_type): New enum.
8843 (mips_rtx_cost_data): Add support for P6600.
8844 (mips_issue_rate): Likewise.
8845 (mips_multipass_dfa_lookahead): Likewise.
8846 (mips_avoid_hazard): Likewise.
8847 (mips_reorg_process_insns): Likewise.
8848 (mips_classify_branch_p6600): New function.
8849 * config/mips/mips.h (TUNE_P6600): New define.
8850 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
8851 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
8852 * config/mips/mips.md: Include p6600.md.
8853 (processor): Add p6600.
8854 * config/mips/p6600.md: New file.
8855 * doc/invoke.texi: Add p6600 to supported architectures.
8857 2018-06-13 Martin Sebor <msebor@redhat.com>
8859 PR tree-optimization/86114
8860 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
8862 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
8864 2018-06-13 Richard Biener <rguenther@suse.de>
8866 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
8867 Properly set vector type of the intermediate stmt.
8868 * tree-vect-stmts.c (vectorizable_operation): The destination
8869 var always has vectype_out type.
8871 2018-06-13 Jeff Law <law@redhat.com>
8873 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
8874 integer 0 for argument to print_rtl_with_bb.
8875 (rl78_reorg): Likewise.
8877 2018-06-13 David Malcolm <dmalcolm@redhat.com>
8879 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
8880 from rtx to rtx_insn *.
8881 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
8883 (add_sched_insns_for_speculation): Likewise for local "target",
8884 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
8885 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
8886 from rtx_insn ** to rtx_code_label **.
8887 (reorg_emit_nops): Likewise.
8888 (c6x_reorg): Likewise for local "call_labels".
8889 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
8891 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
8892 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
8893 the loops over LABEL_REFS.
8894 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
8896 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
8897 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
8898 (split_branches): Strengthen local "olabel" from rtx to
8899 rtx_insn *, adding a safe_as_a cast.
8900 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
8902 (add_insn_after): Likewise for first two params.
8903 (add_insn_before): Likewise.
8904 (remove_insn): Likewise for param.
8905 (emit_pattern_before_noloc): Likewise for second and third params.
8906 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
8907 (emit_call_insn_before_noloc): Likewise.
8908 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
8910 (emit_barrier_before): Likewise.
8911 (emit_label_before): Strengthen "label" param from "rtx" to
8912 "rtx_code_label *". Strengthen "before" param from "rtx" to
8914 (emit_insn_after_1): Strengthen "after" param from "rtx" to
8916 (emit_pattern_after_noloc): Likewise.
8917 (emit_insn_after_noloc): Likewise.
8918 (emit_jump_insn_after_noloc): Likewise.
8919 (emit_call_insn_after_noloc): Likewise.
8920 (emit_debug_insn_after_noloc): Likewise.
8921 (emit_barrier_after): Likewise.
8922 (emit_label_after): Likewise for both params.
8923 (emit_pattern_after_setloc): Likewise for "after" param. Convert
8924 "loc" param from "int" to "location_t".
8925 (emit_insn_after_setloc): Likewise.
8926 (emit_jump_insn_after_setloc): Likewise.
8927 (emit_call_insn_after_setloc): Likewise.
8928 (emit_debug_insn_after_setloc): Likewise.
8929 (emit_pattern_before_setloc): Likewise for "before" param. Convert
8930 "loc" param from "int" to "location_t".
8931 (emit_pattern_before): Convert NULL_RTX to NULL.
8932 (emit_insn_before_setloc): Convert "loc" param from "int" to
8934 (emit_jump_insn_before_setloc): Likewise.
8935 (emit_call_insn_before_setloc): Likewise.
8936 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
8937 rtx_insn *. Convert "loc" param from "int" to "location_t".
8938 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
8939 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
8940 Convert 3rd param from "int" to "location_t".
8941 (emit_barrier_before, emit_barrier_after, next_real_insn):
8942 Strengthen param from rtx to rtx_insn *.
8943 (emit_label_before): Strengthen 1st param from "rtx" to
8944 "rtx_code_label *". Strengthen 2nd param from "rtx" to
8946 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
8947 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
8948 Strengthen 2nd param from "rtx" to "rtx_insn *".
8949 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
8950 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
8951 Likewise. Convert 3rd param from "int" to "location_t".
8952 (emit_label_after): Strengthen 1st param from "rtx" to
8954 (next_real_insn, remove_insn): Strengthen param from "rtx" to
8956 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
8957 from "rtx" to "rtx_insn *".
8959 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
8961 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
8962 bodies streamed in with -Q.
8963 * dumpfile.c (dump_files): Add lto-stream-out dump file.
8964 * dumpfile.h (tree_dump_index): Add lto_stream_out.
8965 * gimple-streamer-out.c: Include gimple-pretty-print.h
8966 (output_bb): Dump stmts streamed.
8967 * lto-section-out.c: Include print-tree.h
8968 (lto_begin_section): Dump sections created.
8969 (lto_output_decl_index): Dump decl encoded.
8970 * lto-streamer-out.c: Include print-tree.h
8971 (create_output_block): Dump output block created.
8972 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
8973 (output_function): Dump function output.
8974 (output_constructor): Dump constructor streamed.
8975 (write_global_stream): Output indexes encoded.
8976 (produce_asm_for_decls): Dump streams encoded.
8977 * lto-streamer.c (streamer_dump_file): New global var.
8978 * lto-streamer.h (streamer_dump_file): Declare.
8979 * passes.c (ipa_write_summaries): Initialize streamer dump.
8980 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
8983 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
8986 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
8987 offsets for register save directives. Emit a second batch of save
8988 directives, if need be, when the function accesses prior frames.
8990 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
8992 * config/arc/fpu.md (fmasf4): Force operand to register.
8993 (fnmasf4): Likewise.
8995 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
8997 * config/arc/arc-protos.h (arc_pad_return): Remove.
8998 * config/arc/arc.c (machine_function): Remove force_short_suffix
9000 (arc_print_operand): Adjust printing of '&'.
9001 (arc_verify_short): Remove conditional printing of short suffix.
9002 (arc_final_prescan_insn): Remove reference to size_reason.
9003 (pad_return): New function.
9004 (arc_reorg): Call pad_return.
9005 (arc_pad_return): Remove.
9006 (arc_init_machine_status): Remove reference to force_short_suffix.
9007 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
9008 (attr length): When attribute iscompact is true force to 2
9009 regardless; in the case of maybe check if we want to force the
9010 instruction to have 4 bytes length.
9011 (nopv): Change it to generate 4 byte long nop as well.
9012 (blockage): New pattern.
9013 (simple_return): Remove call to arc_pad_return.
9014 (p_return_i): Likewise.
9016 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9018 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
9020 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9022 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
9025 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9027 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
9028 for ARC700 and ARCv2.
9030 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
9033 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
9034 operands[2] instead of operands[1].
9037 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9039 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
9040 case, check whether the outer register overlaps an unallocatable
9041 register, not just whether it fits the required class.
9043 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9045 * poly-int.h (can_div_trunc_p): Add new overload in which all values
9047 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
9048 (memrefs_conflict_p): Likewise.
9049 (init_alias_analysis): Likewise.
9050 * cfgexpand.c (expand_debug_expr): Likewise.
9051 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
9052 * cse.c (fold_rtx): Likewise.
9053 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
9054 * expr.c (emit_block_move_hints): Likewise.
9055 (clear_storage_hints, push_block, emit_push_insn): Likewise.
9056 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
9057 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
9058 (emit_group_store): Likewise.
9059 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
9060 to read the PRE/POST_MODIFY increment.
9061 * calls.c (store_one_arg): Use strip_offset.
9062 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
9064 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
9066 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
9067 (simplify_binary_operation_1): Extend CONST_INT handling to
9069 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
9070 than a HOST_WIDE_INT.
9071 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
9073 (adjust_mems, add_stores): Update accodingly.
9074 (vt_canonicalize_addr): Track polynomial offsets.
9075 (emit_note_insn_var_location): Likewise.
9076 (vt_add_function_parameter): Likewise.
9077 (vt_initialize): Likewise.
9079 2018-06-12 Jeff Law <law@redhat.com>
9081 * config.gcc (alpha*-*-freebsd*): Remove.
9082 * config/alpha/freebsd.h: Remove.
9084 2018-06-12 David Malcolm <dmalcolm@redhat.com>
9087 * spellcheck-tree.c (levenshtein_distance): Rename to...
9088 (get_edit_distance): ...this, and update for underlying renaming.
9089 * spellcheck-tree.h (levenshtein_distance): Rename to...
9090 (get_edit_distance): ...this.
9091 * spellcheck.c (levenshtein_distance): Rename to...
9092 (get_edit_distance): ...this. Convert from Levenshtein distance
9093 to Damerau-Levenshtein distance by supporting transpositions of
9094 adjacent characters. Rename "v1" to "v_next" and "v0" to
9096 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
9097 (selftest::test_edit_distance_unit_test_oneway): ...this, and
9098 update for underlying renaming.
9099 (selftest::levenshtein_distance_unit_test): Rename to...
9100 (selftest::test_get_edit_distance_unit): ...this, and update for
9101 underlying renaming.
9102 (selftest::test_find_closest_string): Add example from PR 69968
9103 where transposition helps
9104 (selftest::test_metric_conditions): Update for renaming.
9105 (selftest::test_metric_conditions): Likewise.
9106 (selftest::spellcheck_c_tests): Likewise.
9107 * spellcheck.h (levenshtein_distance): Rename both overloads to...
9108 (get_edit_distance): ...this.
9109 (best_match::consider): Update for renaming.
9111 2018-06-12 Martin Sebor <msebor@redhat.com>
9113 PR tree-optimization/85259
9114 * builtins.c (compute_objsize): Handle constant offsets.
9115 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
9116 true iff a warning has been issued.
9117 * gimple.h (gimple_nonartificial_location): New function.
9118 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
9119 gimple_nonartificial_location and handle -Wno-system-headers.
9120 (handle_builtin_stxncpy): Same.
9122 2018-06-12 Martin Sebor <msebor@redhat.com>
9125 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
9127 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9129 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9130 BUILTIN_VEC_XST entries for pointer to double and long long.
9132 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
9135 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
9137 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
9140 2018-06-12 Martin Liska <mliska@suse.cz>
9142 * doc/options.texi: Document IntegerRange.
9144 2018-06-12 Martin Liska <mliska@suse.cz>
9146 * config/i386/i386.opt: Make MPX-related options as Deprecated.
9147 * opt-functions.awk: Handle Deprecated flag.
9148 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
9150 (read_cmdline_option): Report warning for a deprecated option.
9151 * opts.h (struct cl_option): Add new field cl_deprecated.
9152 (CL_ERR_DEPRECATED): New.
9154 2018-06-12 Martin Liska <mliska@suse.cz>
9156 * doc/options.texi: Document Deprecated option flag.
9158 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9160 * config/arc/arc-arch.h (arc_extras): New enum.
9161 (arc_cpu_t):Add field extra.
9162 (arc_cpu_types): Consider the extras.
9163 * config/arc/arc-cpus.def: Add extras info.
9164 * config/arc/arc-opts.h (processor_type): Consider extra field.
9165 * config/arc/arc.c (arc_override_options): Handle extra field.
9167 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9169 * config/arc/arc-arch.h: Update ARC_OPTX macro.
9170 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
9172 * config/arc/arc.c (arc_init): Update pic warning.
9173 (irq_range): Update irq range parsing warnings.
9174 (arc_override_options): Update various warning messages.
9175 (arc_handle_aux_attribute): Likewise.
9177 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
9179 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
9181 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
9183 * doc/sourcebuild.texi: Document usage of line number 0 in verify
9184 compiler messages directives.
9186 2018-06-12 Matthew Fortune <mfortune@gmail.com>
9188 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
9189 * config/mips/mips-tables.opt: Regenerate.
9190 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
9192 * doc/invoke.texi: Document -march=i6500.
9194 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
9196 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
9197 (i6400_gpmul): Add cpu_unit.
9198 (i6400_gpdiv): Likewise.
9199 (i6400_msa_add_d): Update reservations.
9200 (i6400_msa_int_add) Likewise.
9201 (i6400_msa_short_logic3) Likewise.
9202 (i6400_msa_short_logic2) Likewise.
9203 (i6400_msa_short_logic) Likewise.
9204 (i6400_msa_move) Likewise.
9205 (i6400_msa_cmp) Likewise.
9206 (i6400_msa_short_float2) Likewise.
9207 (i6400_msa_div_d) Likewise.
9208 (i6400_msa_long_logic1) Likewise.
9209 (i6400_msa_long_logic2) Likewise.
9210 (i6400_msa_mult) Likewise.
9211 (i6400_msa_long_float2) Likewise.
9212 (i6400_msa_long_float4) Likewise.
9213 (i6400_msa_long_float5) Likewise.
9214 (i6400_msa_long_float8) Likewise.
9215 (i6400_fpu_fadd): Include frint type.
9216 (i6400_fpu_store): New define_insn_reservation.
9217 (i6400_fpu_load): Likewise.
9218 (i6400_fpu_move): Likewise.
9219 (i6400_fpu_fcmp): Likewise.
9220 (i6400_fpu_fmadd): Likewise.
9221 (i6400_int_mult): Include imul3nc type and update reservation.
9222 (i6400_int_div): Include idiv3 type and update reservation.
9223 (i6400_int_load): Update to check type not move_type.
9224 (i6400_int_store): Likewise.
9225 (i6400_int_prefetch): Set zero latency.
9227 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
9229 * gcc.c: Document new %@{...} sequence.
9230 (LINK_COMMAND_SPEC): Use it for the -L switches.
9231 (cpp_unique_options): Use it for the -I switches.
9232 (at_file_argbuf): New global variable.
9233 (in_at_file): Likewise.
9234 (alloc_args): Create at_file_argbuf.
9235 (clear_args): Truncate at_file_argbuf.
9236 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
9237 (open_at_file): New function.
9238 (close_at_file): Likewise.
9239 (create_at_file): Delete.
9240 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
9243 (validate_switches_from_spec): Deal with %@{...} sequence.
9244 (validate_switches): Likewise.
9245 (driver::finalize): Call clear_args.
9247 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
9249 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
9251 2018-06-11 Martin Sebor <msebor@redhat.com>
9253 * doc/invoke.texi (-Wall): List -Wc++17-compat.
9254 (Wno-class-memaccess): Add @opindex.
9255 (Wno-templates, Wno-multiple-inheritance): Same.
9256 (Wno-virtual-inheritance, Wno-namespaces): Same.
9257 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
9258 (Wno-format-overflow, Wno-format-truncation): Same.
9259 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
9260 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
9261 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
9262 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
9263 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
9264 (Wno-misspelled-isr): Same.
9266 2018-06-11 Martin Sebor <msebor@redhat.com>
9268 * PR tree-optimization/86083
9269 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
9271 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
9273 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
9275 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
9278 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
9279 on the correct operand.
9280 (*movdi_internal64): Ditto.
9282 2018-06-11 Martin Liska <mliska@suse.cz>
9284 PR tree-optimization/86089
9285 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
9287 2018-06-11 Julia Koval <julia.koval@intel.com>
9289 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
9290 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
9291 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
9293 2018-06-11 Olivier Hainque <hainque@adacore.com>
9295 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
9296 for Ada with strict dwarf2.
9298 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
9301 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
9304 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
9306 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
9308 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
9310 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
9313 2018-06-08 Martin Liska <mliska@suse.cz>
9315 * tree-cfg.h (debug_function): Fix argument type to match
9318 2018-06-08 Martin Liska <mliska@suse.cz>
9320 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
9321 Remove usage of MPX-related (and removed) fields.
9322 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
9324 2018-06-08 David Malcolm <dmalcolm@redhat.com>
9326 * cfg.c (debug): Use TDF_NONE rather than 0.
9327 * cfghooks.c (debug): Likewise.
9328 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
9329 (struct dump_option_value_info): Convert to...
9330 (struct kv_pair): ...this template type.
9331 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
9333 (optinfo_verbosity_options): Likewise.
9334 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
9336 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
9337 than int for "optgroup_flags" param.
9338 (dump_generic_expr_loc): Use dump_flags_t rather than int for
9340 (dump_dec): Likewise.
9341 (dump_finish): Use TDF_NONE rather than 0.
9342 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
9343 rather than int for "optgroup_flags" param. Use TDF_NONE rather
9344 than 0. Update for change to option_ptr.
9345 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
9346 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
9347 0. Update for changes to optinfo_verbosity_options and
9349 (opt_info_switch_p): Convert optgroup_flags from int to
9351 (dump_basic_block): Use dump_flags_t rather than int
9352 for "dump_kind" param.
9353 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
9354 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
9355 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
9356 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
9357 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
9358 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
9359 TDF_NONE): Convert from macros to...
9360 (enum dump_flag): ...this new enum.
9361 (dump_flags_t): Update to use enum.
9362 (operator|, operator&, operator~, operator|=, operator&=):
9363 Implement for dump_flags_t.
9364 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
9365 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
9366 Convert from macros to...
9367 (enum optgroup_flag): ...this new enum.
9368 (optgroup_flags_t): New typedef.
9369 (operator|, operator|=): Implement for optgroup_flags_t.
9370 (struct dump_file_info): Convert field "alt_flags" to
9371 dump_flags_t. Convert field "optgroup_flags" to
9373 (dump_basic_block): Use dump_flags_t rather than int for param.
9374 (dump_generic_expr_loc): Likewise.
9375 (dump_dec): Likewise.
9376 (dump_register): Convert param "optgroup_flags" to
9378 (opt_info_enable_passes): Likewise.
9379 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
9381 * gimple-pretty-print.c (debug): Likewise.
9382 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
9383 (merged_store_group::apply_stores): Likewise.
9384 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
9385 * gimple.c (verify_gimple_pp): Likewise.
9386 * graphite-poly.c (print_pbb_body): Likewise.
9387 * passes.c (pass_manager::register_one_dump_file): Convert
9388 local "optgroup_flags" to optgroup_flags_t.
9389 * print-tree.c (print_node): Use TDF_NONE rather than 0.
9391 (debug_body): Likewise.
9392 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
9393 to optgroup_flags_t.
9394 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
9396 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
9397 (convert_mult_to_fma): Likewise.
9398 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
9399 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
9400 * tree-vect-data-refs.c (dump_lower_bound): Convert param
9401 "dump_kind" to dump_flags_t.
9403 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
9405 * config/rs6000/rs6000.c (min, max): Delete.
9407 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
9409 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
9410 -mabi=spe and -mabi=no-spe.
9412 2018-06-08 Martin Liska <mliska@suse.cz>
9414 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
9415 where we expect an existing summary.
9417 2018-06-08 Martin Liska <mliska@suse.cz>
9419 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
9420 * ipa-inline.h (estimate_edge_growth): Likewise.
9422 2018-06-08 Martin Liska <mliska@suse.cz>
9424 * cgraph.c (function_version_hasher::hash): Use
9425 cgraph_node::get_uid ().
9426 (function_version_hasher::equal):
9427 * cgraph.h (cgraph_node::get_uid): New method.
9428 * ipa-inline.c (update_caller_keys): Use
9429 cgraph_node::get_uid ().
9430 (update_callee_keys): Likewise.
9431 * ipa-utils.c (searchc): Likewise.
9432 (ipa_reduced_postorder): Likewise.
9433 * lto-cgraph.c (input_node): Likewise.
9434 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
9435 * symbol-summary.h (symtab_insertion): Likewise.
9436 (symtab_removal): Likewise.
9437 (symtab_duplication): Likewise.
9438 * tree-pretty-print.c (dump_function_header): Likewise.
9439 * tree-sra.c (convert_callers_for_node): Likewise.
9441 2018-06-08 Martin Liska <mliska@suse.cz>
9443 * cgraph.c (symbol_table::create_edge): Always assign a new
9445 (symbol_table::free_edge): Do not recycle numbers.
9446 * cgraph.h (cgraph_edge::get): New method.
9447 * symbol-summary.h (symtab_removal): Use it.
9448 (symtab_duplication): Likewise.
9449 (call_summary::hashable_uid): Remove.
9451 2018-06-08 Martin Liska <mliska@suse.cz>
9453 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
9454 (initialize_growth_caches): Remove.
9455 (free_growth_caches): Likewise.
9456 (do_estimate_edge_time): Use edge_growth_cache.
9457 (do_estimate_edge_size): Likewise.
9458 (do_estimate_edge_hints): Likewise.
9459 * ipa-inline.c (reset_edge_caches): Likewise.
9460 (recursive_inlining): Likewise.
9461 (inline_small_functions): Likewise.
9462 * ipa-inline.h (initialize_growth_caches): Remove.
9463 (estimate_edge_size): Likewise.
9464 (estimate_edge_time): Likewise.
9465 (estimate_edge_hints): Likewise.
9466 (reset_edge_growth_cache): Likewise.
9467 * symbol-summary.h (call_summary::remove): New method.
9469 2018-06-08 Martin Liska <mliska@suse.cz>
9471 * ipa-cp.c (class edge_clone_summary): New summary.
9472 (grow_edge_clone_vectors): Remove.
9473 (ipcp_edge_duplication_hook): Remove.
9474 (class edge_clone_summary_t): New call_summary class.
9475 (ipcp_edge_removal_hook): Remove.
9476 (edge_clone_summary_t::duplicate): New function.
9477 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
9478 (create_specialized_node): Likewise.
9479 (ipcp_driver): Initialize edge_clone_summaries and do not
9482 2018-06-08 Martin Liska <mliska@suse.cz>
9484 * symbol-summary.h (get): New function.
9485 (call_summary::m_initialize_when_cloning): New class member.
9487 2018-06-08 Martin Liska <mliska@suse.cz>
9489 * cgraph.c (cgraph_node::remove): Do not recycle uid.
9490 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
9491 (symbol_table::allocate_cgraph_symbol): Do not set uid.
9492 * passes.c (uid_hash_t): Record removed_nodes by their uids.
9493 (remove_cgraph_node_from_order): Use the removed_nodes set.
9494 (do_per_function_toporder): Likwise.
9495 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
9496 instead of summary_uid.
9497 (symtab_removal): Likewise.
9498 (symtab_duplication): Likewise.
9500 2018-06-08 Martin Liska <mliska@suse.cz>
9502 * ipa-cp.c (ipcp_store_bits_results): Use
9503 ipcp_transformation_sum.
9504 (ipcp_store_vr_results): Likewise.
9505 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
9507 (ipcp_transformation_initialize): ... this.
9508 (ipa_set_node_agg_value_chain):
9509 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
9510 (write_ipcp_transformation_info): Likewise.
9511 (read_ipcp_transformation_info): Likewise.
9512 (ipcp_update_bits): Likewise.
9513 (ipcp_update_vr): Likewise.
9514 (ipcp_transform_function): Likewise.
9515 * ipa-prop.h: Rename ipcp_transformation_summary to
9516 ipcp_transformation.
9517 (class ipcp_transformation_t): New function summary.
9518 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
9519 (ipa_get_agg_replacements_for_node): Likewise.
9521 2018-06-08 Martin Liska <mliska@suse.cz>
9523 * ipa-pure-const.c (struct funct_state_d): Do it class instead
9525 (class funct_state_summary_t): New function_summary class.
9526 (has_function_state): Remove.
9527 (get_function_state): Likewise.
9528 (set_function_state): Likewise.
9529 (add_new_function): Likewise.
9530 (funct_state_summary_t::insert): New function.
9531 (duplicate_node_data): Remove.
9532 (remove_node_data): Remove.
9533 (funct_state_summary_t::duplicate): New function.
9534 (register_hooks): Create new funct_state_summaries.
9535 (pure_const_generate_summary): Use it.
9536 (pure_const_write_summary): Likewise.
9537 (pure_const_read_summary): Likewise.
9538 (propagate_pure_const): Likewise.
9539 (propagate_nothrow): Likewise.
9540 (dump_malloc_lattice): Likewise.
9541 (propagate_malloc): Likewise.
9542 (execute): Do not register hooks, just remove summary
9544 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
9547 2018-06-08 Martin Liska <mliska@suse.cz>
9549 * ipa-reference.c (remove_node_data): Remove.
9550 (duplicate_node_data): Likewise.
9551 (class ipa_ref_var_info_summary_t): New class.
9552 (class ipa_ref_opt_summary_t): Likewise.
9553 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
9554 (get_reference_optimization_summary): Use
9555 ipa_ref_opt_sum_summaries.
9556 (set_reference_vars_info): Remove.
9557 (set_reference_optimization_summary): Likewise.
9558 (ipa_init): Create summaries.
9559 (init_function_info): Use function summary.
9560 (ipa_ref_opt_summary_t::duplicate): New function.
9561 (ipa_ref_opt_summary_t::remove): New function.
9562 (get_read_write_all_from_node): Fix GNU coding style.
9563 (propagate): Use function summary.
9564 (write_node_summary_p): Fix GNU coding style.
9565 (stream_out_bitmap): Likewise.
9566 (ipa_reference_read_optimization_summary): Use function summary.
9567 (ipa_reference_c_finalize): Do not release hooks.
9569 2018-06-08 Martin Liska <mliska@suse.cz>
9571 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
9572 (analyze_function_body): Extract multiple calls of get_create.
9573 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
9574 * ipa-inline.c (recursive_inlining): Use ::get method.
9575 * ipa-inline.h (estimate_edge_growth): Likewise.
9577 2018-06-08 Martin Liska <mliska@suse.cz>
9579 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
9581 (hsa_function_summary::hsa_function_summary): Use the new enum
9583 (hsa_gpu_implementation_p): Use hsa_summaries::get.
9584 * hsa-gen.c (hsa_get_host_function): Likewise.
9585 (get_brig_function_name): Likewise.
9586 * ipa-hsa.c (process_hsa_functions): Likewise.
9587 (ipa_hsa_write_summary): Likewise.
9588 * symbol-summary.h (symtab_duplication): Use ::get function/
9589 (get): New function.
9591 2018-06-08 Martin Liska <mliska@suse.cz>
9593 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
9595 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
9596 (hsa_register_kernel): Likewise.
9597 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
9598 * hsa-gen.c (hsa_get_host_function): Likewise.
9599 (get_brig_function_name): Likewise.
9600 (generate_hsa): Likewise.
9601 (pass_gen_hsail::execute): Likewise.
9602 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
9603 (devirtualization_time_bonus): Likewise.
9604 (ipcp_propagate_stage): Likewise.
9605 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
9606 (edge_set_predicate): Likewise.
9607 (evaluate_conditions_for_known_args): Likewise.
9608 (evaluate_properties_for_edge): Likewise.
9609 (ipa_fn_summary::reset): Likewise.
9610 (ipa_fn_summary_t::duplicate): Likewise.
9611 (dump_ipa_call_summary): Likewise.
9612 (ipa_dump_fn_summary): Likewise.
9613 (analyze_function_body): Likewise.
9614 (compute_fn_summary): Likewise.
9615 (estimate_edge_devirt_benefit): Likewise.
9616 (estimate_edge_size_and_time): Likewise.
9617 (estimate_calls_size_and_time): Likewise.
9618 (estimate_node_size_and_time): Likewise.
9619 (inline_update_callee_summaries): Likewise.
9620 (remap_edge_change_prob): Likewise.
9621 (remap_edge_summaries): Likewise.
9622 (ipa_merge_fn_summary_after_inlining): Likewise.
9623 (ipa_update_overall_fn_summary): Likewise.
9624 (read_ipa_call_summary): Likewise.
9625 (inline_read_section): Likewise.
9626 (write_ipa_call_summary): Likewise.
9627 (ipa_fn_summary_write): Likewise.
9628 (ipa_free_fn_summary): Likewise.
9629 * ipa-hsa.c (process_hsa_functions): Likewise.
9630 (ipa_hsa_write_summary): Likewise.
9631 (ipa_hsa_read_section): Likewise.
9632 * ipa-icf.c (sem_function::merge): Likewise.
9633 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
9634 (do_estimate_edge_time): Likewise.
9635 (estimate_size_after_inlining): Likewise.
9636 (estimate_growth): Likewise.
9637 (growth_likely_positive): Likewise.
9638 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
9639 (inline_call): Likewise.
9640 * ipa-inline.c (caller_growth_limits): Likewise.
9641 (can_inline_edge_p): Likewise.
9642 (can_inline_edge_by_limits_p): Likewise.
9643 (compute_uninlined_call_time): Likewise.
9644 (compute_inlined_call_time): Likewise.
9645 (want_inline_small_function_p): Likewise.
9646 (edge_badness): Likewise.
9647 (update_caller_keys): Likewise.
9648 (update_callee_keys): Likewise.
9649 (recursive_inlining): Likewise.
9650 (inline_small_functions): Likewise.
9651 (inline_to_all_callers_1): Likewise.
9652 (dump_overall_stats): Likewise.
9653 (early_inline_small_functions): Likewise.
9654 (early_inliner): Likewise.
9655 * ipa-inline.h (estimate_edge_growth): Likewise.
9656 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
9657 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
9658 * ipa-prop.h (IPA_NODE_REF): Likewise.
9659 (IPA_EDGE_REF): Likewise.
9660 * ipa-pure-const.c (malloc_candidate_p): Likewise.
9661 (propagate_malloc): Likewise.
9662 * ipa-split.c (execute_split_functions): Likewise.
9663 * symbol-summary.h: Rename get to get_create.
9665 (get_create): Likewise.
9666 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
9668 2018-06-08 Martin Liska <mliska@suse.cz>
9670 * symbol-summary.h (release): Move definition out of class
9672 (symtab_removal): Likewise.
9673 (symtab_duplication): Likewise.
9675 2018-06-08 Martin Liska <mliska@suse.cz>
9677 * symbol-summary.h (function_summary): Move constructor
9678 implementation out of class declaration.
9679 (release): Likewise.
9680 (symtab_insertion): Likewise.
9681 (symtab_removal): Likewise.
9682 (symtab_duplication): Likewise.
9685 2018-06-08 Martin Liska <mliska@suse.cz>
9687 * Makefile.in: Remove support for MPX (macros, related functions,
9688 fields in cgraph_node, ...).
9689 * builtin-types.def (BT_BND): Likewise.
9690 (BT_FN_BND_CONST_PTR): Likewise.
9691 (BT_FN_CONST_PTR_BND): Likewise.
9692 (BT_FN_VOID_PTR_BND): Likewise.
9693 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
9694 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
9695 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
9696 (expand_builtin_mempcpy_with_bounds): Likewise.
9697 (expand_builtin_memset_with_bounds): Likewise.
9698 (expand_builtin_memset_args): Likewise.
9699 (std_expand_builtin_va_start): Likewise.
9700 (expand_builtin): Likewise.
9701 (expand_builtin_with_bounds): Likewise.
9702 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
9703 (DEF_LIB_BUILTIN_CHKP): Likewise.
9704 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
9705 (DEF_CHKP_BUILTIN): Likewise.
9706 (BUILT_IN_MEMCPY): Likewise.
9707 (BUILT_IN_MEMMOVE): Likewise.
9708 (BUILT_IN_MEMPCPY): Likewise.
9709 (BUILT_IN_MEMSET): Likewise.
9710 (BUILT_IN_STPCPY): Likewise.
9711 (BUILT_IN_STRCAT): Likewise.
9712 (BUILT_IN_STRCHR): Likewise.
9713 (BUILT_IN_STRCPY): Likewise.
9714 (BUILT_IN_STRLEN): Likewise.
9715 (BUILT_IN_MEMCPY_CHK): Likewise.
9716 (BUILT_IN_MEMMOVE_CHK): Likewise.
9717 (BUILT_IN_MEMPCPY_CHK): Likewise.
9718 (BUILT_IN_MEMSET_CHK): Likewise.
9719 (BUILT_IN_STPCPY_CHK): Likewise.
9720 (BUILT_IN_STRCAT_CHK): Likewise.
9721 (BUILT_IN_STRCPY_CHK): Likewise.
9722 * calls.c (store_bounds): Likewise.
9723 (emit_call_1): Likewise.
9724 (special_function_p): Likewise.
9725 (maybe_warn_nonstring_arg): Likewise.
9726 (initialize_argument_information): Likewise.
9727 (finalize_must_preallocate): Likewise.
9728 (compute_argument_addresses): Likewise.
9729 (expand_call): Likewise.
9730 * cfgexpand.c (expand_call_stmt): Likewise.
9731 (expand_return): Likewise.
9732 (expand_gimple_stmt_1): Likewise.
9733 (pass_expand::execute): Likewise.
9734 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
9735 (cgraph_node::remove): Likewise.
9736 (cgraph_node::dump): Likewise.
9737 (cgraph_node::verify_node): Likewise.
9738 * cgraph.h (chkp_function_instrumented_p): Likewise.
9739 (symtab_node::get_alias_target): Likewise.
9740 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
9741 (cgraph_local_p): Likewise.
9742 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
9743 (cgraph_edge::rebuild_references): Likewise.
9744 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
9745 (walk_polymorphic_call_targets): Likewise.
9746 (cgraph_node::expand_thunk): Likewise.
9747 (symbol_table::output_weakrefs): Likewise.
9748 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
9749 (ix86_handle_option): Likewise.
9750 * config/i386/constraints.md: Likewise.
9751 * config/i386/i386-builtin-types.def (BND): Likewise.
9755 * config/i386/i386-builtin.def (BDESC_END): Likewise.
9756 (BDESC_FIRST): Likewise.
9758 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
9759 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
9760 * config/i386/i386.c (enum reg_class): Likewise.
9761 (ix86_target_string): Likewise.
9762 (ix86_option_override_internal): Likewise.
9763 (ix86_conditional_register_usage): Likewise.
9764 (ix86_valid_target_attribute_inner_p): Likewise.
9765 (ix86_set_indirect_branch_type): Likewise.
9766 (ix86_set_current_function): Likewise.
9767 (ix86_function_arg_regno_p): Likewise.
9768 (init_cumulative_args): Likewise.
9769 (ix86_function_arg_advance): Likewise.
9770 (ix86_function_arg): Likewise.
9771 (ix86_pass_by_reference): Likewise.
9772 (ix86_function_value_regno_p): Likewise.
9773 (ix86_function_value_1): Likewise.
9774 (ix86_function_value_bounds): Likewise.
9775 (ix86_return_in_memory): Likewise.
9776 (ix86_setup_incoming_vararg_bounds): Likewise.
9777 (ix86_va_start): Likewise.
9778 (indirect_thunk_need_prefix): Likewise.
9779 (print_reg): Likewise.
9780 (ix86_print_operand): Likewise.
9781 (ix86_expand_call): Likewise.
9782 (ix86_output_function_return): Likewise.
9783 (reg_encoded_number): Likewise.
9784 (BDESC_VERIFYS): Likewise.
9785 (ix86_init_mpx_builtins): Likewise.
9786 (ix86_init_builtins): Likewise.
9787 (ix86_emit_cmove): Likewise.
9788 (ix86_emit_move_max): Likewise.
9789 (ix86_expand_builtin): Likewise.
9790 (ix86_builtin_mpx_function): Likewise.
9791 (ix86_get_arg_address_for_bt): Likewise.
9792 (ix86_load_bounds): Likewise.
9793 (ix86_store_bounds): Likewise.
9794 (ix86_load_returned_bounds): Likewise.
9795 (ix86_store_returned_bounds): Likewise.
9796 (ix86_class_likely_spilled_p): Likewise.
9797 (ix86_hard_regno_mode_ok): Likewise.
9798 (x86_order_regs_for_local_alloc): Likewise.
9799 (ix86_mitigate_rop): Likewise.
9800 (ix86_bnd_prefixed_insn_p): Likewise.
9801 (ix86_mpx_bound_mode): Likewise.
9802 (ix86_make_bounds_constant): Likewise.
9803 (ix86_initialize_bounds): Likewise.
9804 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
9805 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
9806 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
9807 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
9808 (TARGET_CHKP_BOUND_MODE): Likewise.
9809 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
9810 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
9811 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
9812 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
9813 * config/i386/i386.h (TARGET_MPX): Likewise.
9814 (TARGET_MPX_P): Likewise.
9815 (VALID_BND_REG_MODE): Likewise.
9816 (FIRST_BND_REG): Likewise.
9817 (LAST_BND_REG): Likewise.
9818 (enum reg_class): Likewise.
9819 (BND_REG_P): Likewise.
9820 (BND_REGNO_P): Likewise.
9821 (BNDmode): Likewise.
9822 (ADJUST_INSN_LENGTH): Likewise.
9823 * config/i386/i386.md: Likewise.
9824 * config/i386/i386.opt: Likewise.
9825 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
9826 (defined): Likewise.
9827 (LINK_MPX): Likewise.
9828 (MPX_SPEC): Likewise.
9829 (LIBMPX_SPEC): Likewise.
9830 (LIBMPXWRAPPERS_SPEC): Likewise.
9831 (CHKP_SPEC): Likewise.
9832 * config/i386/predicates.md: Likewise.
9833 * dbxout.c (dbxout_type): Likewise.
9834 * doc/extend.texi: Likewise.
9835 * doc/invoke.texi: Likewise.
9836 * doc/md.texi: Likewise.
9837 * doc/tm.texi: Likewise.
9838 * doc/tm.texi.in: Likewise.
9839 * dwarf2out.c (is_base_type): Likewise.
9840 (gen_formal_types_die): Likewise.
9841 (gen_subprogram_die): Likewise.
9842 (gen_type_die_with_usage): Likewise.
9843 (gen_decl_die): Likewise.
9844 (dwarf2out_late_global_decl): Likewise.
9845 * expr.c (expand_assignment): Likewise.
9846 (emit_storent_insn): Likewise.
9847 (store_expr_with_bounds): Likewise.
9848 (store_expr): Likewise.
9849 (expand_expr_real_1): Likewise.
9850 * expr.h (store_expr_with_bounds): Likewise.
9851 * function.c (use_register_for_decl): Likewise.
9852 (struct bounds_parm_data): Likewise.
9853 (assign_parms_augmented_arg_list): Likewise.
9854 (assign_parm_find_entry_rtl): Likewise.
9855 (assign_parm_is_stack_parm): Likewise.
9856 (assign_parm_load_bounds): Likewise.
9857 (assign_bounds): Likewise.
9858 (assign_parms): Likewise.
9859 (expand_function_start): Likewise.
9860 * gcc.c (CHKP_SPEC): Likewise.
9861 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9862 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
9863 (wrestrict_dom_walker::check_call): Likewise.
9864 * gimple.c (gimple_build_call_from_tree): Likewise.
9865 * gimple.h (enum gf_mask): Likewise.
9866 (gimple_call_with_bounds_p): Likewise.
9867 (gimple_call_set_with_bounds): Likewise.
9868 * gimplify.c (gimplify_init_constructor): Likewise.
9869 * ipa-cp.c (initialize_node_lattices): Likewise.
9870 (propagate_constants_across_call): Likewise.
9871 (find_more_scalar_values_for_callers_subset): Likewise.
9872 * ipa-hsa.c (process_hsa_functions): Likewise.
9873 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
9874 * ipa-icf.c (sem_function::merge): Likewise.
9875 * ipa-inline.c (early_inliner): Likewise.
9876 * ipa-pure-const.c (warn_function_noreturn): Likewise.
9877 (warn_function_cold): Likewise.
9878 (propagate_pure_const): Likewise.
9879 * ipa-ref.h (enum GTY): Likewise.
9880 * ipa-split.c (find_retbnd): Likewise.
9881 (consider_split): Likewise.
9882 (split_function): Likewise.
9883 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
9884 * ipa.c (walk_polymorphic_call_targets): Likewise.
9885 (symbol_table::remove_unreachable_nodes): Likewise.
9886 (process_references): Likewise.
9887 (cgraph_build_static_cdtor_1): Likewise.
9888 * lto-cgraph.c (lto_output_node): Likewise.
9889 (output_refs): Likewise.
9890 (compute_ltrans_boundary): Likewise.
9891 (input_overwrite_node): Likewise.
9892 (input_node): Likewise.
9893 (input_cgraph_1): Likewise.
9894 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
9895 * passes.c (pass_manager::execute_early_local_passes): Likewise.
9896 (class pass_chkp_instrumentation_passes): Likewise.
9897 (make_pass_chkp_instrumentation_passes): Likewise.
9898 * passes.def: Likewise.
9899 * rtl.h (struct GTY): Likewise.
9900 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
9901 * stor-layout.c (layout_type): Likewise.
9902 * symtab.c: Likewise.
9903 * target.def: Likewise.
9904 * targhooks.c (default_chkp_bound_type): Likewise.
9905 (default_chkp_bound_mode): Likewise.
9906 (default_builtin_chkp_function): Likewise.
9907 (default_chkp_function_value_bounds): Likewise.
9908 (default_chkp_make_bounds_constant): Likewise.
9909 (default_chkp_initialize_bounds): Likewise.
9910 * targhooks.h (default_chkp_bound_type): Likewise.
9911 (default_chkp_bound_mode): Likewise.
9912 (default_builtin_chkp_function): Likewise.
9913 (default_chkp_function_value_bounds): Likewise.
9914 (default_chkp_make_bounds_constant): Likewise.
9915 (default_chkp_initialize_bounds): Likewise.
9916 * toplev.c (compile_file): Likewise.
9917 (process_options): Likewise.
9918 * tree-core.h (DEF_BUILTIN): Likewise.
9919 (DEF_BUILTIN_CHKP): Likewise.
9920 * tree-inline.c (declare_return_variable): Likewise.
9921 (remap_gimple_stmt): Likewise.
9922 (copy_bb): Likewise.
9923 (initialize_inlined_parameters): Likewise.
9924 (expand_call_inline): Likewise.
9925 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
9926 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
9927 (make_pass_ipa_chkp_produce_thunks): Likewise.
9928 (make_pass_chkp): Likewise.
9929 (make_pass_chkp_opt): Likewise.
9930 (make_pass_chkp_instrumentation_passes): Likewise.
9931 * tree-pretty-print.c (dump_generic_node): Likewise.
9932 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
9933 * tree-ssa-dce.c (propagate_necessity): Likewise.
9934 (eliminate_unnecessary_stmts): Likewise.
9935 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
9936 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
9937 * tree-ssa-sccvn.h: Likewise.
9938 * tree-ssa-strlen.c (get_string_length): Likewise.
9939 (valid_builtin_call): Likewise.
9940 (adjust_last_stmt): Likewise.
9941 (handle_builtin_strchr): Likewise.
9942 (handle_builtin_strcpy): Likewise.
9943 (handle_builtin_stxncpy): Likewise.
9944 (handle_builtin_memcpy): Likewise.
9945 (handle_builtin_strcat): Likewise.
9946 (strlen_check_and_optimize_stmt): Likewise.
9947 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
9948 * tree-streamer-in.c: Likewise.
9949 * tree-streamer.c (record_common_node): Likewise.
9950 * tree.c (tree_code_size): Likewise.
9951 (wide_int_to_tree_1): Likewise.
9952 (type_contains_placeholder_1): Likewise.
9953 (build_common_tree_nodes): Likewise.
9954 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
9955 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
9956 (POINTER_BOUNDS_P): Likewise.
9957 (BOUNDED_TYPE_P): Likewise.
9958 (BOUNDED_P): Likewise.
9959 (CALL_WITH_BOUNDS_P): Likewise.
9960 (pointer_bounds_type_node): Likewise.
9961 * value-prof.c (gimple_ic): Likewise.
9962 * var-tracking.c (vt_add_function_parameters): Likewise.
9963 * varasm.c (make_decl_rtl): Likewise.
9964 (assemble_start_function): Likewise.
9965 (output_constant): Likewise.
9966 (maybe_assemble_visibility): Likewise.
9967 * varpool.c (ctor_for_folding): Likewise.
9968 * chkp-builtins.def: Remove.
9969 * ipa-chkp.c: Remove.
9970 * ipa-chkp.h: Remove.
9971 * rtl-chkp.c: Remove.
9972 * rtl-chkp.h: Remove.
9973 * tree-chkp-opt.c: Remove.
9974 * tree-chkp.c: Remove.
9975 * tree-chkp.h: Remove.
9977 2018-06-07 Carl Love <cel@us.ibm.com>
9979 * config/rs6000/vsx.md (vextract_fp_from_shorth,
9980 vextract_fp_from_shortl): Add BE support.
9982 2018-06-07 Paul Koning <ni1d@arrl.net>
9984 * compare-elim.c (try_merge_compare): Don't merge compare if
9985 address contains a side effect.
9986 (try_eliminate_compare): Likewise.
9988 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
9990 * config.gcc: Support "tremont".
9991 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
9992 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
9994 * config/i386/i386.c (m_TREMONT): Define.
9995 (processor_target_table): Add "tremont".
9996 (PTA_TREMONT): Define.
9997 (ix86_lea_outperforms): Add TARGET_TREMONT.
9998 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
9999 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
10000 and M_INTEL_GOLDMONT_PLUS.
10001 (fold_builtin_cpu): Add "tremont".
10002 (ix86_add_stmt_cost): Add TARGET_TREMONT.
10003 (ix86_option_override_internal): Add "tremont".
10004 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
10005 (processor_type): Add PROCESSOR_TREMONT.
10006 * config/i386/x86-tune.def: Add m_TREMONT.
10007 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
10009 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10011 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
10012 symbol defined for msp430i* devices to be lower case.
10014 2018-06-07 Richard Biener <rguenther@suse.de>
10016 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
10017 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
10018 Properly wrap signed arithmetic if overflow wraps.
10020 2018-06-07 Jakub Jelinek <jakub@redhat.com>
10022 PR tree-optimization/69615
10023 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
10024 of a cast from a same precision integral SSA_NAME in a bb dominated
10025 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
10026 cast to utype if rhs2 has already a compatible type.
10028 2018-06-07 Richard Biener <rguenther@suse.de>
10030 PR tree-optimization/85935
10031 * graphite-scop-detection.c (find_params_in_bb): Analyze
10032 condition operands with respect to the correct loop. Assert
10033 the analysis doesn't fail.
10035 2018-06-04 Carl Love <cel@us.ibm.com>
10037 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
10038 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
10039 as it is slightly cheaper.
10040 (first_match_or_eos_index_<mode>):
10041 Calculate index using natural element order.
10042 (first_match_index_<mode>):
10043 Calculate index using natural element order.
10044 (first_match_or_eos_index_<mode>):
10045 Calculate index using natural order.
10046 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
10047 for BE and LE modes.
10048 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
10049 P9V_BUILTIN_VCLZLSBB_V16QI.
10050 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
10053 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
10055 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
10056 indentation and line wrap for many prototypes. Add missing
10057 @smallexample directives around block of prototypes for vec_xl and
10060 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
10062 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
10063 track if we pass or return IEEE 128-bit floating point.
10064 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
10065 C++ mangling that is compatible with GCC 8.1.
10066 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
10067 (init_cumulative_args): Note if we pass or return IEEE 128-bit
10068 floating point types.
10069 (rs6000_function_arg_advance_1): Likewise.
10070 (rs6000_mangle_type): Optionally generate mangled names that match
10071 what GCC 8.1 generated for IEEE 128-bit floating point types.
10072 (rs6000_globalize_decl_name): If we have an external function that
10073 passes or returns IEEE 128-bit types, generate a weak reference
10074 from the mangled name used in GCC 8.1 to the current mangled
10076 (rs6000_init_builtins): Make __ibm128 use the long double type if
10077 long double is IBM extended double. Make __float128 use the long
10078 double type if long double is IEEE 128-bit.
10081 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
10082 macro for __ibm128 built-in functions.
10083 (PACK_IF): Add __ibm128 pack/unpack functions.
10084 (UNPACK_IF): Likewise.
10085 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
10086 enable long double built-in functions if long double is IEEE
10087 128-bit floating point.
10088 (rs6000_invalid_builtin): Update long double built-in function
10090 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
10091 functions, adjust the built-in function to use the long double
10092 built-in function if __ibm128 and long double are the same type.
10093 * doc/extend.texi (PowerPC builtins): Update documention for
10094 __builtin_{,un}pack_longdouble. Add documentation for
10095 __builtin_{,un}pack_ibm128.
10097 2018-06-06 Jim Wilson <jimw@sifive.com>
10099 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
10100 (struct machine_function): New field interrupt_mode.
10101 (riscv_handle_type_attribute): New function. Add forward declaration.
10102 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
10103 (riscv_expand_epilogue): Check interrupt_mode field.
10104 (riscv_set_current_function): Check interrupt attribute args and
10105 set interrupt_mode field.
10106 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
10107 (riscv_sret, riscv_uret): New.
10108 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
10109 new arguments to interrupt attribute.
10111 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
10114 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
10115 Don't handle -mcpu=power8 if -mpower9-vector is also used.
10117 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
10119 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
10120 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
10121 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
10122 several redundant entries.
10124 2018-06-06 David Malcolm <dmalcolm@redhat.com>
10126 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
10127 type from "rtx" to "rtx_insn *".
10128 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
10129 for local "call_insn", removing cast.
10130 (ix86_expand_call): Likewise, introducing a "call_insn" local.
10132 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
10134 PR tree-optimization/86066
10135 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
10136 for BIT_INSERT_EXPR stores.
10138 2018-06-06 Richard Biener <rguenther@suse.de>
10140 PR tree-optimization/86062
10141 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
10142 component refs ontop
10143 of to be offsetted base.
10145 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10147 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
10148 to be static and remove check on interrupt attribute name.
10150 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
10152 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
10153 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
10155 2018-06-05 Steve Ellcey <sellcey@cavium.com>
10158 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
10160 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
10161 Remove second argument, change how error is called.
10162 (aarch64_layout_arg): Remove second argument from
10163 aarch64_err_no_fpadvsimd call.
10164 (aarch64_init_cumulative_args): Ditto.
10165 (aarch64_gimplify_va_arg_expr): Ditto.
10166 * config/aarch64/aarch64.md (mov<mode>): Ditto.
10168 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
10170 * config/i386/i386.md (simple_return_indirect_internal): New expander.
10171 (*simple_return_indirect_internal<mode>): Rename from
10172 simple_return_indirect_internal. Use W mode iterator.
10173 (rstorssp): New expander.
10174 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
10175 (clrssbsy): New expander.
10176 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
10178 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
10180 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
10182 (cmse_check_pointed_object): Likewise.
10184 2018-06-05 Martin Liska <mliska@suse.cz>
10186 PR gcov-profile/47618
10187 * doc/invoke.texi: Document how -fprofile-dir format
10190 2018-06-05 Richard Biener <rguenther@suse.de>
10192 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
10193 removal pretend DOM info isn't available so we do not update
10194 it and only remove edges, not dominated blocks. Actually free
10195 DOM info in case we removed something. Remove unreachable blocks.
10196 (mfb_keep_latches): Work with either DOM info or marked backedges.
10197 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
10198 first. Mark backedges if DOM info isn't available.
10199 (Re-)compute DOM info after cleanup_control_flow_pre.
10201 2018-06-05 Richard Biener <rguenther@suse.de>
10203 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
10204 (locus_discrim_hasher::hash): Adjust.
10205 (locus_discrim_hasher::equal): Likewise.
10206 (next_discriminator_for_locus): Work on line directly.
10207 (same_line_p): Pass in expanded locus1 as well.
10208 (assign_discriminators): Avoid redundant location expansions.
10210 2018-06-05 Richard Biener <rguenther@suse.de>
10212 PR tree-optimization/86046
10213 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
10214 if required after clearing TREE_ADDRESSABLE.
10216 2018-06-05 Richard Biener <rguenther@suse.de>
10218 PR tree-optimization/86047
10219 * tree-ssa-loop.c (for_each_index): Glob handling of all
10220 decls and constants and really handle all of them.
10222 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10225 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
10226 qualifier_void_pointer and qualifier_const_void_pointer.
10227 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
10228 (arm_init_builtins): Handle the above.
10229 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
10230 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
10231 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
10234 2018-06-05 Martin Liska <mliska@suse.cz>
10236 * auto-profile.c (read_autofdo_file): Do not use
10237 gcov_ctr_summary struct.
10238 (afdo_callsite_hot_enough_for_early_inline): Likewise.
10239 * coverage.c (struct counts_entry): Likewise.
10240 (read_counts_file): Read just single summary entry.
10241 (get_coverage_counts): Use gcov_summary struct.
10242 * coverage.h (get_coverage_counts): Likewise.
10243 * gcov-dump.c (dump_working_sets): Likewise.
10244 (tag_summary): Dump just single summary.
10245 * gcov-io.c (gcov_write_summary): Write just histogram
10247 (gcov_read_summary): Read just single summary.
10248 (compute_working_sets): Use gcov_summary struct.
10249 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
10250 of GCOV_COUNTERS_SUMMABLE.
10251 (GCOV_COUNTERS_SUMMABLE): Remove.
10252 (GCOV_FIRST_VALUE_COUNTER): Replace with
10253 GCOV_COUNTER_V_INTERVAL.
10254 (struct gcov_ctr_summary): Remove.
10255 (struct gcov_summary): Directly use fields of former
10257 (compute_working_sets): Use gcov_summary struct.
10258 * gcov.c (read_count_file): Do not use ctrs fields.
10259 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
10261 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
10263 * profile.c: Likewise.
10264 * profile.h: Likewise.
10266 2018-06-05 Martin Liska <mliska@suse.cz>
10268 PR gcov-profile/84846
10269 * gcov.c (output_lines): Print working directory only
10270 in intermediate format.
10272 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
10274 * config/s390/s390-builtin-types.def: Add void function type.
10275 * config/s390/s390-builtins.def: Use the function type for the
10278 2018-06-04 Jim Wilson <jimw@sifive.com>
10280 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
10282 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
10283 and maybe_eh_return. Change regno to unsigned int. Use new args to
10284 handle EH_RETURN_DATA_REGNO registers properly.
10285 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
10286 (riscv_expand_epilogue): Update comment. Change argument name and
10287 type. Update code to use new name and type. Pass new args to
10288 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
10290 * config/riscv/riscv.md (NORMAL_RETURN): New.
10291 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
10292 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
10293 (eh_return): Call gen_eh_return_internal and emit barrier.
10294 (eh_return_internal): Call riscv_expand_epilogue.
10296 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
10298 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
10299 bit_insertion field and declare can_be_merged_into method.
10300 (merged_store_group::can_be_merged_into): New method.
10301 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
10302 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
10303 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
10305 2018-06-04 Richard Biener <rguenther@suse.de>
10307 PR tree-optimization/85955
10308 * builtins.c (fold_builtin_sincos): Convert pointers to
10309 destination to appropriate type before dereferencing.
10311 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
10313 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
10315 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
10317 * expr.c (expand_expr_real_1): Force the operand into memory if
10318 its TYPE_MODE is BLKmode and if there is no integer mode for
10319 the number of bits being extracted.
10321 2018-06-04 Jakub Jelinek <jakub@redhat.com>
10325 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
10326 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
10328 2018-06-04 Richard Biener <rguenther@suse.de>
10330 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
10331 (cleanup_tree_cfg_noloop): ... single caller. Do
10332 start_recording_case_labels later.
10334 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
10336 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
10337 to _IMMINTRIN_H_INCLUDED.
10338 * config/i386/pconfigintrin.h: Ditto.
10339 * config/i386/waitpkgintrin.h: Ditto.
10340 * config/i386/immintrin.h: Add includes for sgxintrin.h,
10341 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
10342 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
10343 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
10344 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
10345 waitpkgintrin.h and cldemoteintrin.h.
10347 2018-06-04 Richard Biener <rguenther@suse.de>
10349 PR tree-optimization/86038
10350 * tracer.c (find_best_successor): Check probability for
10351 being initialized, bail out if not.
10353 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
10356 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
10357 of bits to ignore when comparing architectures.
10359 2018-06-04 Jakub Jelinek <jakub@redhat.com>
10361 PR tree-optimization/69615
10362 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
10363 maximum or minimum of the type, try to merge it also as if
10364 range1 is + [-, x - 1] or + [x + 1, -].
10367 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
10369 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
10371 PR tree-optimization/86034
10372 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
10373 the unsigned bitfield type in a bit insertion sequence if it does not
10374 have a larger precision than the bitfield size.
10375 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
10377 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
10379 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
10381 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
10383 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
10384 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
10385 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
10386 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
10388 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
10390 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
10391 Disable -fdelete-null-pointer-checks for ELF toolchain.
10393 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
10394 Kito Cheng <kito.cheng@gmail.com>
10396 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
10397 (nds32le-*-*, nds32be-*-*): Integrate checking process.
10398 (nds32*-*-*): Add glibc and uclibc conditions.
10399 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
10400 (TARGET_EXCEPT_UNWIND_INFO): Define.
10401 * config/nds32/elf.h: New file.
10402 * config/nds32/linux.h: New file.
10403 * config/nds32/nds32-elf.opt: New file.
10404 * config/nds32/nds32-linux.opt: New file.
10405 * config/nds32/nds32-fp-as-gp.c
10406 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
10407 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
10409 (nds32_asm_file_end): Ditto.
10410 (nds32_print_operand): Ditto.
10411 (nds32_insert_attributes): Ditto.
10412 (nds32_init_libfuncs): New function.
10413 (TARGET_HAVE_TLS): Define.
10414 (TARGET_INIT_LIBFUNCS): Define.
10415 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
10417 (TARGET_ELF): Apply different mcmodel setting.
10418 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
10419 been migrated into elf.h and linux.h files.
10420 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
10421 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
10422 (mcmodel): The content has been migrated into nds32-elf.opt and
10423 nds32-linux.opt files.
10424 * config/nds32/t-elf: New file.
10425 * config/nds32/t-linux: New file.
10427 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
10428 Shiva Chen <shiva0217@gmail.com>
10430 * config/nds32/constants.md (unspec_volatile_element): Add
10431 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
10432 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
10434 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
10435 (make_pass_nds32_fp_as_gp): Declare.
10436 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
10438 (nds32_asm_function_end_prologue): Remove unused asm output.
10439 (nds32_asm_function_begin_epilogue): Remove unused asm output.
10440 (nds32_asm_file_start): Output necessary fp_as_gp information.
10441 (nds32_option_override): Adjust register usage.
10442 (nds32_expand_prologue): Consider fp_as_gp situation.
10443 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
10444 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
10447 (simple_return): Ditto.
10448 (omit_fp_begin): Output special directive for fp_as_gp.
10449 (omit_fp_end): Output special directive for fp_as_gp.
10450 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
10451 mforbid-fp-as-gp): New options.
10453 2018-06-01 Mark Wielaard <mark@klomp.org>
10455 * dwarf2out.c (dwarf2out_finish): Remove generation of
10456 DW_AT_loclists_base.
10458 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
10460 * gimple-ssa-store-merging.c: Include gimple-fold.h.
10461 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
10462 (struct merged_store_group): Add bit_insertion field.
10463 (dump_char_array): Use standard hexadecimal format.
10464 (merged_store_group::merged_store_group): Set bit_insertion to false.
10465 (merged_store_group::apply_stores): Use optimal buffer size. Deal
10466 with BIT_INSERT_EXPR stores. Move up code updating the mask and
10467 also print the mask in the dump file.
10468 (pass_store_merging::gate): Minor tweak.
10469 (imm_store_chain_info::coalesce_immediate): Fix wrong association
10470 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
10471 stores with INTEGER_CST stores.
10472 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
10473 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
10474 and use it throughout. Generate bit insertion sequences if need be.
10475 (pass_store_merging::process_store): Remove redundant condition.
10476 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
10478 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
10480 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
10481 the 128-bit floating point types. Fix function comment.
10483 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10485 * config/aarch64/aarch64-simd.md
10486 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
10488 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
10491 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
10493 PR tree-optimization/85989
10494 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
10496 (backprop::intersect_uses): Check it when deciding whether this
10497 is a backedge reference.
10498 (backprop::process_block): Add each phi to m_visited_phis
10499 after visiting it, then clear it at the end.
10501 2018-06-01 Richard Biener <rguenther@suse.de>
10503 * tree-vectorizer.h (vect_dr_stmt): New function.
10504 (vect_get_load_cost): Adjust.
10505 (vect_get_store_cost): Likewise.
10506 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
10507 Use vect_dr_stmt instead of DR_SMTT.
10508 (vect_record_base_alignments): Likewise.
10509 (vect_calculate_target_alignment): Likewise.
10510 (vect_compute_data_ref_alignment): Likewise and make static.
10511 (vect_update_misalignment_for_peel): Likewise.
10512 (vect_verify_datarefs_alignment): Likewise.
10513 (vector_alignment_reachable_p): Likewise.
10514 (vect_get_data_access_cost): Likewise. Pass down
10515 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
10516 (vect_get_peeling_costs_all_drs): Likewise.
10517 (vect_peeling_hash_get_lowest_cost): Likewise.
10518 (vect_enhance_data_refs_alignment): Likewise.
10519 (vect_find_same_alignment_drs): Likewise.
10520 (vect_analyze_data_refs_alignment): Likewise.
10521 (vect_analyze_group_access_1): Likewise.
10522 (vect_analyze_group_access): Likewise.
10523 (vect_analyze_data_ref_access): Likewise.
10524 (vect_analyze_data_ref_accesses): Likewise.
10525 (vect_vfa_segment_size): Likewise.
10526 (vect_small_gap_p): Likewise.
10527 (vectorizable_with_step_bound_p): Likewise.
10528 (vect_prune_runtime_alias_test_list): Likewise.
10529 (vect_analyze_data_refs): Likewise.
10530 (vect_supportable_dr_alignment): Likewise.
10531 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
10532 (vect_gen_prolog_loop_niters): Likewise.
10533 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
10534 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
10536 (vect_recog_mask_conversion_pattern): Likewise.
10537 (vect_try_gather_scatter_pattern): Likewise.
10538 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
10539 to vect_get_store_cost.
10540 (vect_get_store_cost): Get stmt_info instead of DR.
10541 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
10542 (vect_get_load_cost): Get stmt_info instead of DR.
10544 2018-06-01 Richard Biener <rguenther@suse.de>
10546 PR middle-end/86017
10547 * gimple-fold.c (var_decl_component_p): Also allow offsetted
10548 vars wrapped in MEM_REFs.
10550 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
10552 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10553 Fix subreg tests so that we only return a choice between
10554 GENERAL_REGS and FP_REGS if the original classes included both.
10556 2018-06-01 Richard Biener <rguenther@suse.de>
10559 * tree-ssa-structalias.c (get_function_part_constraint):
10560 Handle NULL fi->decl.
10561 (find_func_aliases_for_call): Properly handle indirect
10562 fi from direct call.
10563 (find_func_clobbers): Likewise.
10564 (ipa_pta_execute): Likewise.
10565 (create_variable_info_for): For functions that are ifunc_resolver
10566 resolve to a varinfo that contains the result of the resolver call.
10567 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
10570 2018-05-31 Michael Collison <michael.collison@arm.com>
10572 * config/aarch64/aarch64.md:
10573 (*fix_to_zero_extenddfdi2): New pattern.
10574 * gcc.target/aarch64/fix_extend1.c: New testcase.
10576 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
10578 PR middle-end/78809
10579 PR middle-end/83026
10580 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
10581 and BUILT_IN_STRNCMP_EQ.
10582 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
10583 BUILT_IN_STRNCMP_EQ.
10584 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
10585 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
10586 (gimple_fold_builtin): Likewise.
10587 * tree-ssa-strlen.c (compute_string_length): New function.
10588 (determine_min_obsize): New function.
10589 (handle_builtin_string_cmp): New function to handle calls to
10590 string compare functions.
10591 (strlen_optimize_stmt): Add handling to builtin string compare
10593 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
10594 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
10595 * tree.c (build_common_builtin_nodes): Add new defines of
10596 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
10598 2018-05-31 Jakub Jelinek <jakub@redhat.com>
10601 * bb-reorder.c (pass_partition_blocks::gate): Return false for
10602 functions with naked attribute.
10604 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
10606 * config/i386/sse.md (avx_vec_concat<mode>):
10607 Substitute concat_tg_mode mode attribute with xtg_mode.
10608 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
10609 (concat_tg_mode): Remove mode attribute.
10611 2018-05-31 Martin Sebor <msebor@redhat.com>
10614 * calls.c (alloc_max_size): Correct a logic error/typo.
10615 Treat excessive arguments as infinite. Warn for invalid arguments.
10616 * doc/invoke.texi (-Walloc-size-larger-than): Update.
10618 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
10621 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
10622 and movx for Haswell.
10624 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
10625 Cesar Philippidis <cesar@codesourcery.com>
10627 PR middle-end/85879
10628 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
10629 when emitting error on private/firstprivate reductions.
10630 * omp-low.c (lower_omp_target): Avoid reference-type processing
10631 on pointers for firstprivate clause.
10633 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
10635 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
10638 * config/aarch64/aarch64-simd.md
10639 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
10640 (aarch64_ld1_x3_<mode>): Likewise
10641 (aarch64_st1x2<VALLDIF:mode>): Likewise
10642 (aarch64_st1_x2_<mode>): Likewise
10643 (aarch64_st1x3<VALLDIF:mode>): Likewise
10644 (aarch64_st1_x3_<mode>): Likewise
10645 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
10646 (vld1_s8_x3): Likewise.
10647 (vld1_u16_x3): Likewise.
10648 (vld1_s16_x3): Likewise.
10649 (vld1_u32_x3): Likewise.
10650 (vld1_s32_x3): Likewise.
10651 (vld1_u64_x3): Likewise.
10652 (vld1_s64_x3): Likewise.
10653 (vld1_f16_x3): Likewise.
10654 (vld1_f32_x3): Likewise.
10655 (vld1_f64_x3): Likewise.
10656 (vld1_p8_x3): Likewise.
10657 (vld1_p16_x3): Likewise.
10658 (vld1_p64_x3): Likewise.
10659 (vld1q_u8_x3): Likewise.
10660 (vld1q_s8_x3): Likewise.
10661 (vld1q_u16_x3): Likewise.
10662 (vld1q_s16_x3): Likewise.
10663 (vld1q_u32_x3): Likewise.
10664 (vld1q_s32_x3): Likewise.
10665 (vld1q_u64_x3): Likewise.
10666 (vld1q_s64_x3): Likewise.
10667 (vld1q_f16_x3): Likewise.
10668 (vld1q_f32_x3): Likewise.
10669 (vld1q_f64_x3): Likewise.
10670 (vld1q_p8_x3): Likewise.
10671 (vld1q_p16_x3): Likewise.
10672 (vld1q_p64_x3): Likewise.
10673 (vst1_s64_x2): Likewise.
10674 (vst1_u64_x2): Likewise.
10675 (vst1_f64_x2): Likewise.
10676 (vst1_s8_x2): Likewise.
10677 (vst1_p8_x2): Likewise.
10678 (vst1_s16_x2): Likewise.
10679 (vst1_p16_x2): Likewise.
10680 (vst1_s32_x2): Likewise.
10681 (vst1_u8_x2): Likewise.
10682 (vst1_u16_x2): Likewise.
10683 (vst1_u32_x2): Likewise.
10684 (vst1_f16_x2): Likewise.
10685 (vst1_f32_x2): Likewise.
10686 (vst1_p64_x2): Likewise.
10687 (vst1q_s8_x2): Likewise.
10688 (vst1q_p8_x2): Likewise.
10689 (vst1q_s16_x2): Likewise.
10690 (vst1q_p16_x2): Likewise.
10691 (vst1q_s32_x2): Likewise.
10692 (vst1q_s64_x2): Likewise.
10693 (vst1q_u8_x2): Likewise.
10694 (vst1q_u16_x2): Likewise.
10695 (vst1q_u32_x2): Likewise.
10696 (vst1q_u64_x2): Likewise.
10697 (vst1q_f16_x2): Likewise.
10698 (vst1q_f32_x2): Likewise.
10699 (vst1q_f64_x2): Likewise.
10700 (vst1q_p64_x2): Likewise.
10701 (vst1_s64_x3): Likewise.
10702 (vst1_u64_x3): Likewise.
10703 (vst1_f64_x3): Likewise.
10704 (vst1_s8_x3): Likewise.
10705 (vst1_p8_x3): Likewise.
10706 (vst1_s16_x3): Likewise.
10707 (vst1_p16_x3): Likewise.
10708 (vst1_s32_x3): Likewise.
10709 (vst1_u8_x3): Likewise.
10710 (vst1_u16_x3): Likewise.
10711 (vst1_u32_x3): Likewise.
10712 (vst1_f16_x3): Likewise.
10713 (vst1_f32_x3): Likewise.
10714 (vst1_p64_x3): Likewise.
10715 (vst1q_s8_x3): Likewise.
10716 (vst1q_p8_x3): Likewise.
10717 (vst1q_s16_x3): Likewise.
10718 (vst1q_p16_x3): Likewise.
10719 (vst1q_s32_x3): Likewise.
10720 (vst1q_s64_x3): Likewise.
10721 (vst1q_u8_x3): Likewise.
10722 (vst1q_u16_x3): Likewise.
10723 (vst1q_u32_x3): Likewise.
10724 (vst1q_u64_x3): Likewise.
10725 (vst1q_f16_x3): Likewise.
10726 (vst1q_f32_x3): Likewise.
10727 (vst1q_f64_x3): Likewise.
10728 (vst1q_p64_x3): Likewise.
10730 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10732 * config/msp430/msp430.c (msp430_output_labelref): Prepend
10733 user_label_prefix to name.
10735 * tree-core.h: Update comment about the format of NAME string
10736 passed to handler in attribute_spec.
10738 * config/msp430/msp430.md: Remove erroneous subreg expression from
10739 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
10740 zero_extend{q,h}isi2.
10742 2018-05-30 Borislav Petkov <bp@suse.de>
10744 * doc/extend.texi: Document some architecture specific
10745 constraints and sort entries.
10747 2018-05-30 Martin Sebor <msebor@redhat.com>
10749 PR middle-end/85369
10750 * builtins.c (expand_builtin_stpcpy_1): New function.
10751 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
10752 only if the former succeeds.
10754 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
10756 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
10759 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10761 * doc/invoke.texi (-flinker-output): Document
10763 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10765 * passes.c (ipa_write_summaries): Only modify statements if body
10767 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
10768 incrementally linking.
10769 (ipa_passes): Likewise.
10770 * lto-cgraph.c (lto_output_node): When incrementally linking do not
10771 pass down resolution info.
10772 * common.opt (flag_incremental_link): Update info.
10773 * gcc.c (plugin specs): Turn flinker-output=* to
10774 -plugin-opt=-linker-output-known
10775 * toplev.c (compile_file): Also cut compilation when doing incremental
10777 * flag-types. (enum lto_partition_model): Add
10778 LTO_LINKER_OUTPUT_NOLTOREL.
10779 (invoke.texi): Add -flinker-output docs.
10780 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
10781 link same way as WPA; do not stream in dead initializers.
10783 * dwarf2out.c (dwarf2out_die_ref_for_decl,
10784 darf2out_register_external_decl): Support incremental link.
10786 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10788 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
10790 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10792 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
10793 it down to simple_object_copy_lto_debug_sections.
10794 (run_gcc): Determine incremental LTO link time and configure
10795 lto1 into non-wpa mode, disable renaming of debug sections.
10797 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
10799 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
10800 descriptions of various incorrectly documented functions.
10802 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
10805 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
10806 address check not strict.
10808 2018-05-30 Richard Biener <rguenther@suse.de>
10810 PR tree-optimization/85964
10811 * tracer.c (better_p): Drop initialized count check, we only
10812 call the function with initialized counts now.
10813 (find_best_successor): Do find a best edge if one
10814 has uninitialized count.
10815 (find_best_predecessor): Likewise. Do BB frequency check only
10816 if count is initialized.
10818 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
10820 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
10821 (aarch64_ldrstr_offset_compare): New.
10822 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
10823 load/store orderings.
10824 (aarch64_gen_adjusted_ldpstp): Likewise.
10826 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
10828 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10829 Check for subset of GENERAL_REGS and FP_REGS.
10830 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
10833 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
10835 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
10836 and wi::to_poly_offset. Add the current offset and then check
10837 whether the sum fits, rather than using an unchecked addition of
10838 a checked term. Check for a shwi rather than a uhwi.
10839 * expr.c (get_bit_range): Use tree_to_poly_uint64.
10840 (store_constructor): Use poly_int_tree_p.
10841 (expand_expr_real_1): Likewise.
10842 * function.c (assign_temp): Likewise.
10843 * fold-const.c (const_binop): Use poly_int_tree_p and
10844 wi::to_poly_offset.
10845 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
10847 * ipa-icf-gimple.c (func_checker::compare_operand): Use
10848 to_poly_offset for MEM offsets.
10849 * ipa-icf.c (sem_variable::equals): Likewise.
10850 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
10851 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
10852 wi::to_poly_offset for BIT_FIELD_REF offsets.
10853 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
10854 wi::to_poly_offset.
10855 * var-tracking.c (emit_note_insn_var_location): Use
10856 tree_to_poly_uint64.
10858 2018-05-29 Jim Wilson <jimw@sifive.com>
10860 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
10862 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
10865 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
10866 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
10868 (sse4_1_round<mode>2): Use nonimmediate_operand
10869 for operand 1 predicate.
10871 2018-05-29 Martin Sebor <msebor@redhat.com>
10872 Richard Biener <rguenther@suse.de>
10875 * calls.c (get_size_range): Call determine_value_range instead
10876 of get_value_range..
10877 * tree-vrp.h (determine_value_range): Declared new function.
10878 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
10880 2018-05-29 Richard Biener <rguenther@suse.de>
10882 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
10883 sure to use non-pattern stmts for get_earlier_stmt arguments.
10884 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
10885 called on pattern stmts.
10886 (get_later_stmt): Likewise.
10888 2018-05-29 Martin Liska <mliska@suse.cz>
10890 PR gcov-profile/85759
10891 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
10894 2018-05-29 Jakub Jelinek <jakub@redhat.com>
10896 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
10898 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
10902 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
10903 VEC_PACK_FLOAT_EXPR): New tree codes.
10904 * tree-pretty-print.c (op_code_prio): Handle
10905 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
10906 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
10907 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
10908 * tree-inline.c (estimate_operator_cost): Likewise.
10909 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
10910 * fold-const.c (const_binop): Likewise.
10911 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
10912 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
10913 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
10914 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
10915 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
10916 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
10917 * expr.c (expand_expr_real_2): Likewise.
10918 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
10919 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
10920 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
10922 * optabs.c (expand_widen_pattern_expr): For
10923 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
10924 sign from result type rather than operand's type.
10925 (expand_binop_directly): For vec_packu_float_optab and
10926 vec_packs_float_optab allow result type to be different from operand's
10928 * optabs-tree.c (optab_for_tree_code): Handle
10929 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
10930 VEC_PACK_FLOAT_EXPR. Formatting fixes.
10931 * tree-vect-generic.c (expand_vector_operations_1): Handle
10932 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
10933 VEC_PACK_FLOAT_EXPR.
10934 * tree-vect-stmts.c (supportable_widening_operation): Handle
10936 (supportable_narrowing_operation): Handle FLOAT_EXPR.
10937 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
10938 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
10939 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
10940 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
10942 (vec_pack<floatprefix>_float_<mode>): New expander.
10943 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
10945 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
10946 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
10947 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
10948 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
10949 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
10951 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
10952 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
10953 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
10954 VEC_PACK_FLOAT_EXPR): Document.
10956 2018-05-29 Richard Biener <rguenther@suse.de>
10958 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
10960 (stmt_vec_info_vec): Make pointer.
10961 (init_stmt_vec_info_vec): Remove.
10962 (free_stmt_vec_info_vec): Likewise.
10963 (set_stmt_vec_info_vec): New function.
10964 (free_stmt_vec_infos): Likewise.
10965 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
10966 (set_vinfo_for_stmt): Likewise.
10967 (get_earlier_stmt): Likewise.
10968 (get_later_stmt): Likewise.
10969 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
10970 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
10971 (vec_info::~vec_info): Free stmt_vec_infos.
10972 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
10973 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
10974 (pass_slp_vectorize::execute): Likewise.
10975 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
10976 (free_stmt_vec_info_vec): Likewise.
10977 (set_stmt_vec_info_vec): New function.
10978 (free_stmt_vec_infos): Likewise.
10979 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
10980 the global stmt_vec_info_vec.
10981 * tree-parloops.c (gather_scalar_reductions): Use
10982 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
10985 2018-05-29 Richard Biener <rguenther@suse.de>
10987 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
10989 2018-05-29 Martin Liska <mliska@suse.cz>
10990 David Malcolm <dmalcolm@redhat.com>
10992 * vec.c (test_reverse): New.
10993 (vec_c_tests): Add new test.
10994 * vec.h (vl_ptr>::reverse): New function.
10996 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
10998 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
11000 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
11003 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
11005 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
11007 2018-05-28 Richard Biener <rguenther@suse.de>
11009 PR tree-optimization/85933
11010 * tree-vect-data-refs.c (vect_record_base_alignments): Only
11011 look at stmts marked as vectorizable.
11013 2018-05-28 Richard Biener <rguenther@suse.de>
11015 PR tree-optimization/85934
11016 * tree-vect-generic.c (expand_vector_operations_1): Hoist
11017 vector boolean check before scalar optimization.
11019 2018-05-28 Jakub Jelinek <jakub@redhat.com>
11021 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
11024 2018-05-28 Mark Wielaard <mark@klomp.org>
11026 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
11027 if it is an expression containing a minus sign.
11029 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
11031 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
11033 2018-05-27 Paul Koning <ni1d@arrl.net>
11035 * config/pdp11/pdp11.md (truncsihi2): Remove.
11037 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
11038 Chung-Ju Wu <jasonwucj@gmail.com>
11040 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
11042 (unaligned_store_dw): Ditto.
11043 * config/nds32/nds32-memory-manipulation.c
11044 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
11045 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
11046 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
11047 (emit_setmem_word_loop): Rename to ...
11048 (emit_setmem_doubleword_loop): ... this.
11049 (nds32_gen_dup_4_byte_to_word_value): New function.
11050 (nds32_gen_dup_8_byte_to_double_word_value): New function.
11051 (nds32_expand_setmem_loop): Refine implementation.
11052 (nds32_expand_setmem_loop_v3m): Ditto.
11053 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
11056 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
11058 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
11060 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
11062 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
11063 (nds32_init_machine_status): Initialize machine->attr_naked_p and
11064 machine->attr_no_prologue_p.
11065 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
11066 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
11067 (nds32_expand_epilogue): Consider attr_naked_p.
11068 (nds32_expand_epilogue_v3pop): Likewise.
11069 (nds32_can_use_return_insn): Likewise.
11070 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
11071 attr_no_prologue_p fields.
11072 * config/nds32/nds32.opt (mret-in-naked-func): New option.
11074 2018-05-27 Jakub Jelinek <jakub@redhat.com>
11077 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
11079 * config/i386/sse.md
11080 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
11082 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
11084 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
11086 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
11088 (*<floatsuffix>floatv2div2sf2): Rename to ...
11089 (*float<floatunssuffix>v2div2sf2): ... this.
11090 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
11091 (float<floatunssuffix>v2div2sf2_mask): ... this.
11092 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
11093 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
11094 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
11096 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
11098 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
11100 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
11102 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
11104 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
11106 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
11107 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
11108 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
11109 gen_ufix_truncv8dfv8si2.
11110 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
11111 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
11112 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
11113 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
11114 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
11115 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
11116 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
11117 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
11119 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
11123 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
11125 2018-05-25 Jim Wilson <jimw@sifive.com>
11127 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
11128 * config/riscv/riscv.c (struct machine_function): Add
11129 interrupt_handler_p and attribute_checked_p fields.
11130 (riscv_attribute_table): Add interrupt.
11131 (riscv_interrupt_type_p): New.
11132 (riscv_save_reg_p): Save extra regs for interrupt handler.
11133 (riscv_use_save_libcall): Return false for interrupt handler.
11134 (riscv_first_stack_step): Add forward declaration.
11135 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
11136 for interrupt handler with large frame. Use it for saved reg list.
11137 (riscv_expand_prologue): Move flag_stack_usage_info support to
11138 eliminate duplication.
11139 (riscv_expand_epilogue): Generate mret for interrupt handler.
11140 (riscv_epilogue_uses): New.
11141 (riscv_can_use_return_insn): Return false for interrupt handler.
11142 (riscv_function_ok_for_sibcall): Likewise.
11143 (riscv_set_current_function): Add interrupt handler support.
11144 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
11145 * config/riscv/riscv.md (UNSPECV_MRET): New.
11147 (riscv_frflags, riscv_fsflags): Use tab after opcode.
11149 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
11151 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
11153 PR tree-optimization/85712
11154 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
11155 this candidate has already been replaced in-situ by a copy.
11157 2018-05-25 Jason Merrill <jason@redhat.com>
11159 PR c++/80485 - inline function non-zero address.
11160 * symtab.c (nonzero_address): Check DECL_COMDAT.
11162 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
11165 * config/alpha/alpha.md (ashlsi3): New insn pattern.
11166 (*ashlsi_se): Rename from *ashldi_se. Define as sign
11167 extension of SImode operation. Use const123_operand predicate.
11168 (*saddsi_1): Remove.
11169 (*saddl_se_1): Ditto.
11170 (*ssubsi_1): Ditto.
11171 (*ssubl_se_1): Ditto.
11172 * config/alpha/predicates.md (const123_operand): New predicate.
11173 * config/alpha/constraints.md (P): Use IN_RANGE.
11175 2018-05-25 Richard Biener <rguenther@suse.de>
11177 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
11179 (ref_maybe_used_by_stmt_p): Likewise.
11180 (stmt_may_clobber_ref_p): Likewise.
11181 (stmt_may_clobber_ref_p_1): Likewise.
11182 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
11184 (ref_maybe_used_by_stmt_p): Likewise.
11185 (stmt_may_clobber_ref_p): Likewise.
11186 (stmt_may_clobber_ref_p_1): Likewise.
11187 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
11188 the alias oracle to disambiguate DRs with stmts DR analysis
11190 (vect_analyze_data_refs): Do not give up on not analyzable
11191 DRs for BB vectorization. Remove code truncating the dataref
11194 2018-05-25 Jakub Jelinek <jakub@redhat.com>
11197 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
11198 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
11199 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
11201 2018-05-25 Richard Biener <rguenther@suse.de>
11203 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
11204 function, combining stmt data ref gathering and fatal analysis
11206 (vect_analyze_data_refs): Remove now redudnant code and simplify.
11207 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
11208 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
11209 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
11210 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
11212 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11214 PR tree-optimization/85720
11215 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
11216 SCC if all partitions are builtins.
11217 (version_loop_by_alias_check): New parameter. Generate cancelable
11218 runtime alias check if all partitions are builtins.
11219 (distribute_loop): Update call to above function.
11221 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11223 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
11224 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
11225 (parm_default_def_partition_arg): Ditto.
11226 (set_parm_default_def_partition): Ditto.
11227 (get_parm_default_def_partitions): Ditto and make it static.
11228 (get_undefined_value_partitions): Ditto and make it static.
11229 (remove_ssa_form): Refactor call to init_var_map here.
11230 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
11231 computation for loop region.
11232 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
11233 (register_default_def): Delete.
11234 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
11235 (parm_default_def_partition_arg): Ditto.
11236 (set_parm_default_def_partition): Ditto.
11237 (get_parm_default_def_partitions): Ditto and make it static.
11238 (get_undefined_value_partitions): Ditto and make it static.
11239 (coalesce_with_default, coalesce_with_default): Update comment.
11240 (create_coalesce_list_for_region): New func factored out from
11241 create_outofssa_var_map.
11242 (populate_coalesce_list_for_outofssa): New func factored out from
11243 create_outofssa_var_map and coalesce_ssa_name.
11244 (create_outofssa_var_map): Delete.
11245 (coalesce_ssa_name): Refactor to support live range computation.
11246 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
11247 (get_parm_default_def_partitions): Delete.
11248 (get_undefined_value_partitions): Ditto.
11249 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
11250 computation for loop region.
11251 (new_tree_live_info, loe_visit_block): Ditto.
11252 (live_worklist, set_var_live_on_entry): Ditto.
11253 (calculate_live_on_exit, verify_live_on_entry): Ditto.
11254 * tree-ssa-live.h (struct _var_map): New fields.
11255 (init_var_map): Change decl.
11256 (region_contains_p): New.
11258 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11260 * tree-ssa-live.h (live_merge_and_clear): Delete.
11262 2018-05-25 Richard Biener <rguenther@suse.de>
11265 * tree-dump.c (dequeue_and_dump): Remove access to removed
11266 operand 2 of a SWITCH_EXPR.
11268 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11270 * doc/sourcebuild.texi (vect_double_cond_arith): Include
11271 multiplication and division.
11272 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
11273 (cond_udiv@var{m}, cond_umod@var{m}): Document.
11274 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
11275 (cond_udiv_optab, cond_umod_optab): New optabs.
11276 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
11277 (IFN_COND_RDIV): New internal functions.
11278 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
11279 TRUNC_MOD_EXPR and RDIV_EXPR.
11280 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
11281 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
11283 (SVE_INT_BINARY): Include mult.
11284 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
11285 (optab, sve_int_op): Handle mult.
11286 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
11288 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
11289 for SVE_INT_BINARY_SD.
11291 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11293 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
11294 (optab, sve_int_op): Handle div and udiv.
11295 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
11296 for SVE_INT_BINARY_SD.
11297 (*<optab><mode>3): New insn for the same.
11299 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11301 * tree-vect-patterns.c: Include predict.h.
11302 (vect_recog_divmod_pattern): Restrict check for division support
11303 to when optimizing for size.
11305 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11307 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
11308 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
11309 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
11310 (gimple_match_op::set_op): Likewise.
11311 (gimple_resimplify4): Declare.
11312 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
11313 (expr::gen_transform): Likewise.
11314 (decision_tree::gen): Generate a simplification routine for 4 operands.
11315 * gimple-match-head.c (gimple_simplify): Add an overload for
11316 4 operands. In the top-level function, handle up to 4 call
11317 arguments and call gimple_resimplify4.
11318 (gimple_resimplify4): New function.
11319 (build_call_internal): Pass a fourth operand.
11320 (maybe_push_to_seq): Likewise.
11321 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
11322 Fold VEC_COND_EXPRs of an operation and a default value into
11323 an IFN_COND_* function if possible.
11324 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
11326 (SVE_COND_FP_BINARY): Include them.
11327 (optab, sve_fp_op): Handle them.
11328 (SVE_INT_BINARY_REV): New code iterator.
11329 (SVE_COND_FP_BINARY_REV): New int iterator.
11330 (commutative): New int attribute.
11331 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
11333 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
11335 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
11336 (*cond_<optab><mode>): New patterns for reversed operands.
11338 2018-05-25 Richard Biener <rguenther@suse.de>
11340 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
11341 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
11342 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
11343 (STMT_VINFO_GROUPED_ACCESS): Adjust.
11344 * tree-vect-data-refs.c (everywhere): Adjust users.
11345 * tree-vect-loop.c (everywhere): Likewise.
11346 * tree-vect-slp.c (everywhere): Likewise.
11347 * tree-vect-stmts.c (everywhere): Likewise.
11348 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
11350 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11352 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
11354 (gcc_cv_as_section_exclude): ... this.
11355 Try Solaris as #exclude syntax.
11356 * configure: Regenerate.
11357 * config.in: Regenerate.
11358 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
11360 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
11361 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
11363 * varasm.c (default_elf_asm_named_section): Don't check if
11364 HAVE_GAS_SECTION_EXCLUDE is defined.
11366 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11368 * doc/md.texi: Update the documentation of the cond_* optabs
11369 to mention the new final operand. Fix GET_MODE_NUNITS call.
11370 Describe the scalar case too.
11371 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
11372 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
11374 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
11375 (get_conditional_internal_fn): Update comment.
11376 * tree-vect-loop.c (vectorizable_reduction): Pass the original
11377 accumulator value as a final argument to conditional functions.
11378 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
11379 a define_expand and add an "else" operand. Assert for now that
11380 the else operand is equal to operand 2. Use SVE_INT_BINARY and
11381 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
11382 (*cond_<optab><mode>): New patterns.
11383 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
11384 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
11385 (UNSPEC_COND_EOR): Delete.
11386 (optab): Remove associated mappings.
11387 (SVE_INT_BINARY): New code iterator.
11388 (sve_int_op): Remove int attribute and add "minus" to the code
11390 (SVE_COND_INT_OP): Delete.
11391 (SVE_COND_FP_OP): Rename to...
11392 (SVE_COND_FP_BINARY): ...this.
11394 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11396 * optabs.c (can_reuse_operands_p): New function.
11397 (maybe_legitimize_operands): Try to reuse the results for
11400 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
11402 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
11403 Add {q} suffix to insn mnemonic.
11405 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
11407 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
11408 (msp430_warn_func_return): New.
11410 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
11412 * fold-const.c (tree_nonzero_bits): New function.
11413 * fold-const.h (tree_nonzero_bits): Likewise.
11414 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
11415 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
11417 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
11421 * varasm.c (assemble_alias): Check ifunc_resolver only on
11424 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
11427 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
11428 when memory input operand is handled.
11430 2018-05-24 Luis Machado <luis.machado@linaro.org>
11432 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
11434 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
11436 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
11438 * match.pd: Delay FMA folds until after vectorization.
11440 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
11443 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
11444 address check not strict.
11446 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
11448 * gimple-match.h (gimple_match_op): New class.
11449 (mprts_hook): Replace parameters with a gimple_match_op *.
11450 (maybe_build_generic_op): Likewise.
11451 (gimple_simplified_result_is_gimple_val): Replace parameters with
11452 a const gimple_match_op *.
11453 (gimple_simplify): Replace code_helper * and tree * parameters with
11454 a gimple_match_op * parameter.
11455 (gimple_resimplify1): Replace code_helper *, tree and tree *
11456 parameters with a gimple_match_op * parameter.
11457 (gimple_resimplify2): Likewise.
11458 (gimple_resimplify3): Likewise.
11459 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
11460 parameters with a gimple_match_op * parameter.
11461 * gimple-match-head.c (gimple_simplify): Change prototypes of
11462 auto-generated functions to take a gimple_match_op * instead of
11463 separate code_helper * and tree * parameters. Make the same
11464 change in the top-level overload and update calls to the
11465 gimple_resimplify routines. Update calls to the auto-generated
11466 functions and to maybe_push_res_to_seq in the publicly-facing
11467 operation-specific gimple_simplify overloads.
11468 (gimple_match_op::MAX_NUM_OPS): Define.
11469 (gimple_resimplify1): Replace rcode and ops with a single res_op
11470 parameter. Update call to gimple_simplify.
11471 (gimple_resimplify2): Likewise.
11472 (gimple_resimplify3): Likewise.
11473 (mprts_hook): Replace parameters with a gimple_match_op *.
11474 (maybe_build_generic_op): Likewise.
11475 (build_call_internal): Replace type, nargs and ops with
11476 a gimple_match_op *.
11477 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
11478 with a single gimple_match_op *. Update calls to mprts_hook,
11479 build_call_internal and gimple_simplified_result_is_gimple_val.
11480 Factor out code that is common to the tree_code and combined_fn cases.
11481 * genmatch.c (expr::gen_transform): Replace tem_code and
11482 tem_ops with a gimple_match_op called tem_op. Update calls
11483 to the gimple_resimplify functions and maybe_push_res_to_seq.
11484 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
11485 res_ops. Update call to the gimple_resimplify functions.
11486 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
11487 (decision_tree::gen): Make the functions take a gimple_match_op *
11488 called res_op instead of separate res_code and res_ops parameters.
11489 Update call accordingly.
11490 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
11491 and ops with a single res_op parameter. Update calls to
11492 maybe_build_generic_op and maybe_push_res_to_seq.
11493 (fold_stmt_1): Update calls to gimple_simplify and
11494 replace_stmt_with_simplification.
11495 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
11496 and gimple_simplified_result_is_gimple_val.
11497 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
11499 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
11500 with a gimple_match_op *.
11501 (vn_nary_build_or_lookup): Likewise. Update call to
11502 vn_nary_build_or_lookup_1.
11503 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
11504 gimple_match_op *. Update calls to the gimple_resimplify routines
11505 and to gimple_simplified_result_is_gimple_val.
11506 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
11507 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
11508 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
11509 (visit_nary_op): Likewise.
11510 (visit_reference_op_load): Likewise.
11512 2018-05-23 Luis Machado <luis.machado@linaro.org>
11514 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
11515 modifier for printing the step amount.
11517 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11520 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
11523 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
11525 * doc/sourcebuild.texi (Endianness): New subsubsection.
11527 2018-05-23 Luis Machado <luis.machado@linaro.org>
11529 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
11530 <prefetch_dynamic_strides>: New const bool field.
11531 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
11532 prefetch_dynamic_strides.
11533 (exynosm1_prefetch_tune): Likewise.
11534 (thunderxt88_prefetch_tune): Likewise.
11535 (thunderx_prefetch_tune): Likewise.
11536 (thunderx2t99_prefetch_tune): Likewise.
11537 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
11539 (aarch64_override_options_internal): Update to set
11540 PARAM_PREFETCH_DYNAMIC_STRIDES.
11541 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
11542 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
11543 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
11544 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
11545 prefetch-dynamic-strides setting.
11547 2018-05-23 Luis Machado <luis.machado@linaro.org>
11549 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
11550 <minimum_stride>: New const int field.
11551 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
11552 minimum_stride field defaulting to -1.
11553 (exynosm1_prefetch_tune): Likewise.
11554 (thunderxt88_prefetch_tune): Likewise.
11555 (thunderx_prefetch_tune): Likewise.
11556 (thunderx2t99_prefetch_tune): Likewise.
11557 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
11558 <default_opt_level>: Set to 3.
11559 (aarch64_override_options_internal): Update to set
11560 PARAM_PREFETCH_MINIMUM_STRIDE.
11561 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
11562 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
11563 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
11564 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
11565 stride is constant and is below the minimum stride threshold.
11567 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11569 * config/arm/arm-cpus.in (mode26): Delete.
11570 (armv4): Delete mode26 reference.
11571 * config/arm/arm.c (arm_configure_build_target): Delete use of
11574 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
11576 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
11578 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
11579 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
11581 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
11582 (floatunsdidf2): Ditto.
11584 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
11586 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
11587 (fixuns_trunc<mode>si2_avx512f): Ditto.
11588 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
11589 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
11590 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
11592 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
11594 PR rtl-optimization/79985
11595 * df-scan.c (df_insn_refs_collect): Remove special case for
11596 global registers and asm statements.
11598 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
11600 * extend.texi (Global Register Variables): Rewrite the bullet list.
11601 Note that the register is available for allocation. Note that access
11602 via inline asm must use constraints. Add note about async-signal
11603 handlers. Remove paragraph about automagic register selection.
11605 2018-05-23 Richard Biener <rguenther@suse.de>
11607 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
11608 of fixed offset from memset VN.
11610 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
11612 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
11613 first_interp field.
11614 (alloc_cand_and_find_basis): Initialize first_interp field.
11615 (slsr_process_mul): Modify first_interp field.
11616 (slsr_process_add): Likewise.
11617 (slsr_process_cast): Modify first_interp field for each new
11619 (slsr_process_copy): Likewise.
11620 (dump_candidate): Dump first_interp field.
11621 (replace_mult_candidate): Process all interpretations, not just
11623 (replace_rhs_if_not_dup): Likewise.
11624 (replace_one_candidate): Likewise.
11626 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
11628 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
11630 (aarch64_needs_frame_chain): New function.
11631 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
11633 2018-05-23 Sudakshina Das <sudi.das@arm.com>
11636 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
11637 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
11638 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
11639 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
11640 as true for strict-align.
11641 (aarch64_can_inline_p): Perform checks even when callee has no
11642 attributes to check for strict alignment.
11643 * doc/extend.texi (AArch64 Function Attributes): Document
11645 * doc/invoke.texi: (AArch64 Options): Likewise.
11647 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
11649 PR tree-optimization/85853
11650 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
11651 the handling of the root of the node to...
11652 (vect_slp_analyze_node_operations_1): ...this new function,
11653 and run the whole thing with the child nodes' def types
11654 set according to their SLP node's def type.
11656 2018-05-23 Richard Biener <rguenther@suse.de>
11658 PR middle-end/85874
11659 * tree-data-ref.c (create_runtime_alias_checks): Defer
11660 and ignore overflow warnings.
11662 2018-05-23 Yury Gribov <tetra2005@gmail.com>
11664 PR tree-optimization/85822
11665 * tree-vrp.c (is_masked_range_test): Fix handling of negative
11668 2018-05-23 Richard Biener <rguenther@suse.de>
11670 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
11671 memset constants via native_interpret_expr.
11673 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
11676 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
11678 (cgraph_node::create_alias): Likewise.
11679 (cgraph_node::get_availability): Check ifunc_resolver instead
11680 of looking up ifunc attribute.
11681 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
11682 * varasm.c (do_assemble_alias): Likewise.
11683 (assemble_alias): Likewise.
11684 (default_binds_local_p_3): Likewise.
11685 * cgraph.h (cgraph_node): Add ifunc_resolver.
11686 (cgraph_node::only_called_directly_or_aliased_p): Return false
11687 for IFUNC resolver.
11688 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
11690 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
11691 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
11692 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
11693 instead of looking up ifunc attribute.
11695 2018-05-22 Luis Machado <luis.machado@linaro.org>
11697 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
11699 2018-05-22 Martin Sebor <msebor@redhat.com>
11701 PR middle-end/85359
11702 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
11703 only when expasion succeeds.
11704 (expand_builtin_strcmp): Same.
11705 (expand_builtin_strncmp): Same.
11707 2018-05-22 Martin Sebor <msebor@redhat.com>
11709 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
11711 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
11712 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11714 * config/aarch64/aarch64-ldpstp.md: Replace uses of
11715 aarch64_mem_pair_operand with memory_operand and delete operand swapping
11717 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
11718 Add check for legitimate_address.
11719 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
11720 (aarch64_swap_ldrstr_operands): New.
11721 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
11724 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
11725 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11727 * config/aarch64/aarch64.md: New patterns to generate stp
11729 (store_pair_sw, store_pair_dw): New patterns to generate stp for
11730 single words and double words.
11731 (load_pair_sw, load_pair_dw): Likewise.
11732 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
11734 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
11736 * config/aarch64/aarch64-ldpstp.md: Modify peephole
11737 for different mode ldpstp and add peephole for merged zero stores.
11738 Likewise for loads.
11739 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
11741 (aarch64_gen_store_pair): Rename calls to match new patterns.
11742 (aarch64_gen_load_pair): Rename calls to match new patterns.
11743 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
11744 (load_pair<DREG:mode><DREG2:mode>): ... This.
11745 (store_pair<mode>): Rename to...
11746 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
11747 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
11748 New mode iterators.
11749 (V_INT_EQUIV): Handle SImode.
11750 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
11753 2018-05-22 Martin Sebor <msebor@redhat.com>
11756 * calls.c (maybe_warn_nonstring_arg): Use string length to set
11757 or ajust the presumed bound on an operation to avoid unnecessary
11760 2018-05-22 Martin Sebor <msebor@redhat.com>
11762 PR tree-optimization/85826
11763 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
11764 assuming that a DECL necesarily has a constant size.
11766 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
11768 PR middle-end/85862
11769 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
11771 2018-05-22 Richard Biener <rguenther@suse.de>
11773 PR tree-optimization/85834
11774 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
11775 non-constant and non-zero memset arguments.
11777 2018-05-22 Martin Liska <mliska@suse.cz>
11780 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
11782 2018-05-22 Richard Biener <rguenther@suse.de>
11784 PR tree-optimization/85863
11785 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
11786 comparisons when vectype is specified.
11787 (vectorizable_condition): Do not specify vectype for
11788 vect_is_simple_cond when SLP vectorizing.
11790 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
11793 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
11794 define __ibm128 as long double.
11795 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
11796 as a distinct type when IEEE 128-bit support is enabled.
11797 (init_float128_ieee): Fix up conversions between IFmode and IEEE
11798 128-bit types to use the correct functions.
11799 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
11800 convert between 128-bit floating point types that have different
11801 modes but the same representation, instead of using gen_lowpart to
11803 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
11805 (IFKF_reg): New attributes to give the register constraints for
11807 (extend<mode>tf2_internal): New insns to mark an explicit
11808 conversion between 128-bit floating point types that have a
11809 different mode but share the same representation.
11811 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
11813 PR tree-optimization/85814
11814 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
11815 a null return from get_strinfo when unsharing the next
11816 strinfo in the chain.
11818 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
11821 * varasm.c (weak_finish): Clean up weak_decls.
11823 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11825 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
11826 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
11827 UNSPEC_UADALP values.
11828 * config/aarch64/iterators.md (ABAL): New int iterator.
11831 (sur): Add mappings for the above.
11832 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
11834 (aarch64_<sur>abal<mode>_4): Likewise.
11835 (aarch64_<sur>adalp<mode>_3): Likewise.
11836 (<sur>sadv16qi): New define_expand.
11838 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
11840 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
11841 (*movdf_internal): Ditto.
11842 (*rcpsf2_sse): Ditto.
11843 (*rsqrtsf2_sse): Ditto.
11844 (*sqrt<mode>2_sse): Ditto.
11846 2018-05-21 Tamar Christina <tamar.christina@arm.com>
11848 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
11850 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
11851 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
11852 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
11853 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
11855 * config/aarch64/arm_neon.h: Likewise.
11856 * config/aarch64/iterators.md (VQ_I): New.
11858 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
11860 * config.gcc: Add arc/t-multilib-linux to tmake_file for
11862 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
11865 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
11867 * config/nds32/constraints.md (S): New constraint.
11868 * config/nds32/nds32.md (call_internal): Use constraint S.
11869 (call_value_internal): Likewise.
11870 (sibcall_internal): Likewise.
11871 (sibcall_value_internal): Likewise.
11873 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
11874 Chung-Ju Wu <jasonwucj@gmail.com>
11876 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
11877 into consideration.
11879 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
11880 Chung-Ju Wu <jasonwucj@gmail.com>
11882 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
11883 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
11884 (nds32_rtx_costs_impl): Simplify.
11885 (nds32_address_cost_impl): Simplify.
11886 (nds32_init_rtx_costs): New function.
11887 (nds32_rtx_costs_speed_prefer): Likewise.
11888 (nds32_rtx_costs_size_prefer): Likewise.
11889 (nds32_address_cost_speed_prefer): Likewise.
11890 (nds32_address_cost_speed_fwprop): Likewise.
11891 (nds32_address_cost_size_prefer): Likewise.
11892 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
11893 * config/nds32/nds32.c (nds32_option_override): Use
11894 nds32_init_rtx_costs function.
11896 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
11898 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
11899 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
11900 (TARGET_PIPELINE_N8): Likewise.
11901 (TARGET_PIPELINE_N10): Likewise.
11902 (TARGET_PIPELINE_N13): Likewise.
11903 (TARGET_PIPELINE_GRAYWOLF): Likewise.
11905 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
11907 * config/nds32/nds32-fpu.md: Update copyright year.
11909 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
11911 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
11913 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
11915 * config/nds32/nds32.c
11916 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
11917 * config/nds32/nds32.opt (minline-asm-r15): New option.
11919 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
11921 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
11923 * config/nds32/nds32.md (abssi2): New pattern.
11925 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
11927 * config/i386/i386.md (rex64namesuffix): New mode attribute.
11928 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
11929 Merge insn pattern from sse_cvtsi2ss<round_name> and
11930 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
11931 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
11932 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
11933 using SWI48 mode iterator.
11934 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
11935 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
11936 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
11937 pattern from sse_cvttss2si<round_saeonly_name>
11938 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
11939 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
11940 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
11941 using SWI48 mode iterator.
11942 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
11943 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
11944 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
11945 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
11946 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
11947 using SWI48 mode iterator.
11948 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
11949 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
11950 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
11951 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
11952 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
11953 SWI48 mode iterator.
11954 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
11955 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
11956 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
11957 pattern from sse_cvttsd2si<round_saeonly_name>
11958 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
11960 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
11962 * config/nds32/nds32-md-auxiliary.c
11963 (nds32_valid_smw_lwm_base_p): Refine.
11964 (nds32_output_smw_single_word): Refine.
11965 (nds32_output_smw_double_word): New.
11966 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
11968 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
11970 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
11971 (nds32_output_stack_pop): Refine.
11972 (nds32_expand_unaligned_load): Refine.
11973 (nds32_expand_unaligned_store): Refine.
11975 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
11976 Chung-Ju Wu <jasonwucj@gmail.com>
11978 * config/nds32/constants.md: Add TP_REGNUM constant.
11979 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
11980 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
11982 * config/nds32/nds32-doubleword.md: Consider flag_pic.
11983 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
11984 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
11985 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
11986 and PIC code generation.
11987 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
11989 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
11991 * config/nds32/nds32.md: Support TLS and PIC.
11992 * config/nds32/nds32.c: Support TLS and PIC.
11993 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
11994 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
11997 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
11999 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
12000 mode with E_ prefix.
12002 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
12003 Chung-Ju Wu <jasonwucj@gmail.com>
12005 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
12006 * config/nds32/nds32-md-auxiliary.c
12007 (symbolic_reference_mentioned_p): New.
12008 (nds32_legitimize_ict_address): New.
12009 (nds32_expand_ict_move): New.
12010 (nds32_indirect_call_referenced_p): New.
12011 (nds32_symbol_binds_local_p): Delete.
12012 (nds32_long_call_p): Modify.
12013 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
12014 * config/nds32/nds32-protos.h
12015 (symbolic_reference_mentioned_p): Declare.
12016 (nds32_legitimize_ict_address): Declare.
12017 (nds32_expand_ict_move): Declare.
12018 (nds32_indirect_call_referenced_p): Declare.
12019 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
12020 (nds32_relax_group): Use nds32_ict_const_p as condition.
12021 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
12022 (nds32_asm_file_start): Output ict_model directive in asm code.
12023 (nds32_legitimate_address_p): Consider indirect call.
12024 (nds32_print_operand): Consider indirect call.
12025 (nds32_print_operand_address): Consider indirect call.
12026 (nds32_insert_attributes): Handle "indirect_call" attribute.
12027 (TARGET_LEGITIMATE_ADDRESS_P): Define.
12028 (TARGET_LEGITIMATE_CONSTANT_P): Define.
12029 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
12030 (TARGET_DELEGITIMIZE_ADDRESS): Define.
12031 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
12032 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
12033 (TARGET_ICT_MODEL_SMALL): Define.
12034 (TARGET_ICT_MODEL_LARGE): Define.
12035 * config/nds32/nds32.md (movsi): Consider ict model.
12036 (call, call_value): Consider ict model.
12037 (sibcall, sibcall_value): Consider ict model.
12038 * config/nds32/nds32.opt (mict-model): New option.
12039 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
12042 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
12043 Monk Chiang <sh.chiang04@gmail.com>
12044 Jim Wilson <jimw@sifive.com>
12046 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
12047 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
12048 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
12049 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
12050 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
12051 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
12052 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
12053 compute save_libcall_adjustment properly.
12054 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
12055 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
12056 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
12057 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
12058 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
12059 (ABI_SPEC): Handle mabi=ilp32e.
12060 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
12061 (RVE): Add RVE mask.
12062 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
12063 <-march>: Add rv32e as an example.
12065 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
12068 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
12069 (intra_create_variable_infos): Handle C++ constructors.
12071 2018-05-18 Martin Liska <mliska@suse.cz>
12073 * passes.def: Remove a redundant pass.
12075 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
12078 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
12080 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12082 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
12084 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
12086 (armv2, armv2a, armv3, armv3m): Delete architectures.
12087 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
12088 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
12089 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
12091 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
12092 (*mulsidi3adddi): Likewise.
12093 (mulsidi3): Likewise.
12094 (*mulsidi3_nov6): Likewise.
12095 (umulsidi3): Likewise.
12096 (umulsidi3_nov6): Likewise.
12097 (umaddsidi4): Likewise.
12098 (*umulsidi3adddi): Likewise.
12099 (smulsi3_highpart): Likewise.
12100 (*smulsi3_highpart_nov6): Likewise.
12101 (umulsi3_highpart): Likewise.
12102 (*umulsi3_highpart_nov6): Likewise.
12103 * config/arm/arm.h (arm_arch3m): Delete.
12104 * config/arm/arm.c (arm_arch3m): Delete.
12105 (arm_option_override_internal): Update armv3-related comment.
12106 (arm_configure_build_target): Delete use of isa_bit_mode32.
12107 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
12108 (arm_rtx_costs_internal): Delete check of arm_arch3m.
12109 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
12110 (mulsa3): Likewise.
12111 (mulusa3): Likewise.
12112 * config/arm/arm-protos.h (arm_arch3m): Delete.
12113 * config/arm/arm-tables.opt: Regenerate.
12114 * config/arm/arm-tune.md: Likewise.
12115 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
12116 deleted architectures.
12118 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12120 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
12121 (armv5t, armv5te): New features.
12122 (ARMv5, ARMv5e): Delete fgroups.
12123 (ARMv5t, ARMv5te): Adjust for above changes.
12124 (ARMv6m): Likewise.
12125 (armv5, armv5e): Delete arches.
12126 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
12128 (*call_reg_arm): Likewise.
12129 (*call_value_reg_armv5): Likewise.
12130 (*call_value_reg_arm): Likewise.
12131 (*call_symbol): Likewise.
12132 (*call_value_symbol): Likewise.
12133 (*sibcall_insn): Likewise.
12134 (*sibcall_value_insn): Likewise.
12135 (clzsi2): Likewise.
12136 (prefetch): Likewise.
12137 (define_split and define_peephole2 dependent on arm_arch5):
12139 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
12141 (TARGET_ARM_QBIT): Likewise.
12142 (TARGET_DSP_MULTIPLY): Likewise.
12143 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
12144 (arm_arch5, arm_arch5e): Delete.
12145 (arm_arch5t, arm_arch5te): Declare.
12146 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
12147 (arm_arch5t): Declare.
12148 (arm_option_reconfigure_globals): Update for the above.
12149 (arm_options_perform_arch_sanity_checks): Update comment, replace
12150 use of arm_arch5 with arm_arch5t.
12151 (use_return_insn): Likewise.
12152 (arm_emit_call_insn): Likewise.
12153 (output_return_instruction): Likewise.
12154 (arm_final_prescan_insn): Likewise.
12155 (arm_coproc_builtin_available): Likewise.
12156 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
12157 arm_arch5e with arm_arch5t and arm_arch5te.
12158 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
12159 (arm_arch5t, arm_arch5te): Declare.
12160 * config/arm/arm-tables.opt: Regenerate.
12161 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
12162 * config/arm/t-multilib: Likewise.
12163 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
12164 instead of arm_arch5.
12165 (*call_reg_thumb1): Likewise.
12166 (*call_value_reg_thumb1_v5): Likewise.
12167 (*call_value_reg_thumb1): Likewise.
12168 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
12170 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
12172 2018-05-18 Martin Liska <mliska@suse.cz>
12174 PR gcov-profile/84846
12175 * doc/gcov.texi: Document -t option of gcov tool.
12177 2018-05-18 Martin Liska <mliska@suse.cz>
12179 PR gcov-profile/84846
12180 * gcov.c (print_usage): Add new -t option.
12181 (process_args): Handle the option.
12182 (generate_results): Use stdout as output when requested by
12185 2018-05-18 Martin Liska <mliska@suse.cz>
12187 PR gcov-profile/84846
12188 * coverage.c (coverage_init): Write PWD to .gcno file.
12189 * doc/gcov.texi: Document how working directory is printed.
12190 * gcov-dump.c (dump_gcov_file): Print PWD.
12191 * gcov.c (output_intermediate_file): Likewise.
12192 (read_graph_file): Read PWD string.
12193 (output_lines): Print PWD.
12195 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12197 PR middle-end/85817
12198 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
12199 for retval and return false if all args to phi are zero.
12201 2018-05-18 Richard Biener <rguenther@suse.de>
12203 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
12205 (evrp_dom_walker::before_dom_children): Call it.
12207 2018-05-18 Richard Biener <rguenther@suse.de>
12209 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
12210 results when processing array refs with variable index.
12212 2018-05-18 Toon Moene <toon@moene.org>
12214 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
12215 directly after that of -floop-interchange. Indicate that both
12216 options are enabled by default when specifying -O3.
12218 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12220 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
12221 iterator. Delete separate integer-mode vec_set<mode> expander.
12222 (aarch64_simd_vec_setv2di): Delete.
12223 (vec_setv2di): Delete.
12224 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
12225 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
12226 the "w, r" alternative.
12228 2018-05-18 Martin Liska <mliska@suse.cz>
12230 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
12231 * tree-pass.h (make_pass_lower_switch_O0): New function.
12232 * tree-switch-conversion.c (node_has_low_bound): Remove.
12233 (node_has_high_bound): Likewise.
12234 (node_is_bounded): Likewise.
12235 (class pass_lower_switch): Make it a template type and create
12237 (pass_lower_switch::execute): Add template argument.
12238 (make_pass_lower_switch): New function.
12239 (make_pass_lower_switch_O0): New function.
12240 (do_jump_if_equal): Remove.
12241 (emit_case_nodes): Simplify to just handle all 3 cases and leave
12242 all the hard work to tree optimization passes.
12244 2018-05-18 Martin Liska <mliska@suse.cz>
12246 * dbgcnt.c (limit_low): Renamed from limit.
12247 (limit_high): New variable.
12248 (dbg_cnt_is_enabled): Check for upper limit.
12249 (dbg_cnt): Adjust dumping.
12250 (dbg_cnt_set_limit_by_index): Add new argument for high
12252 (dbg_cnt_set_limit_by_name): Likewise.
12253 (dbg_cnt_process_single_pair): Parse new format.
12254 (dbg_cnt_process_opt): Use strtok.
12255 (dbg_cnt_list_all_counters): Remove 'value' and add
12257 * doc/invoke.texi: Document changes.
12259 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
12261 * doc/sourcebuild.texi (scalar_all_fma): Document.
12262 * tree.def (FMA_EXPR): Delete.
12263 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
12264 * internal-fn.c (ternary_direct): New macro.
12265 (expand_ternary_optab_fn): Likewise.
12266 (direct_ternary_optab_supported_p): Likewise.
12267 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
12268 * builtins.c (fold_builtin_fma): Delete.
12269 (fold_builtin_3): Don't call it.
12270 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
12271 * expr.c (expand_expr_real_2): Likewise.
12272 * fold-const.c (operand_equal_p): Likewise.
12273 (fold_ternary_loc): Likewise.
12274 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
12275 * gimple.c (DEFTREECODE): Likewise.
12276 * gimplify.c (gimplify_expr): Likewise.
12277 * optabs-tree.c (optab_for_tree_code): Likewise.
12278 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12279 * tree-eh.c (operation_could_trap_p): Likewise.
12280 (stmt_could_throw_1_p): Likewise.
12281 * tree-inline.c (estimate_operator_cost): Likewise.
12282 * tree-pretty-print.c (dump_generic_node): Likewise.
12283 (op_code_prio): Likewise.
12284 * tree-ssa-loop-im.c (stmt_cost): Likewise.
12285 * tree-ssa-operands.c (get_expr_operands): Likewise.
12286 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
12287 * fold-const-call.h (fold_fma): Delete.
12288 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
12289 CFN_FNMA and CFN_FNMS.
12290 (fold_fma): Delete.
12291 * genmatch.c (combined_fn): New enum.
12292 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
12293 (commutative_op): New function.
12294 (commutate): Use it. Handle more than 2 operands.
12295 (dt_operand::gen_gimple_expr): Use commutative_op.
12296 (parser::parse_expr): Allow :c to be used with non-binary
12297 operators if the commutative operand is known.
12298 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
12299 CFN_FMS, CFN_FNMA and CFN_FNMS.
12300 (backprop::process_assign_use): Remove FMA_EXPR handling.
12301 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
12302 (gen_hsa_fma): New function.
12303 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
12304 IFN_FNMA and IFN_FNMS.
12305 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
12306 * gimple-fold.h (follow_all_ssa_edges): Declare.
12307 * gimple-fold.c (follow_all_ssa_edges): New function.
12308 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
12309 gimple_build interface and use follow_all_ssa_edges to fold the result.
12310 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
12311 instead of checking for optabs directly.
12312 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
12313 rather than FMA_EXPRs.
12314 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
12315 call to IFN_FMA instead of an FMA_EXPR.
12317 2018-05-17 Jim Wilson <jimw@sifive.com>
12319 * expr.c (do_tablejump): When converting index to Pmode, if we have a
12320 sign extended promoted subreg, and the range does not have the sign bit
12321 set, then do a sign extend.
12323 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
12324 test, check for sign extended subreg and/or constant operands, and
12325 do a sign extend in that case.
12327 2018-05-17 Steve Ellcey <sellcey@cavium.com>
12329 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
12330 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
12332 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
12333 Change logics_shift_reg to logics_shift_imm.
12334 (thunderx2t99_fp_loadpair_basic): Delete.
12335 (thunderx2t99_fp_storepair_basic): Delete.
12336 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
12337 (thunderx2t99_asimd_polynomial): Delete.
12338 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
12339 and neon_fp_mul_d_scalar_q.
12340 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
12341 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
12342 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
12343 (thunderx2t99_asimd_lut): Add missing tbl types.
12344 (thunderx2t99_asimd_ext): Delete.
12345 (thunderx2t99_asimd_load1_1_mult): Delete.
12346 (thunderx2t99_asimd_load1_2_mult): Delete.
12347 (thunderx2t99_asimd_load1_ldp): New.
12348 (thunderx2t99_asimd_load1): New.
12349 (thunderx2t99_asimd_load2): Add missing *load2* types.
12350 (thunderx2t99_asimd_load3): New.
12351 (thunderx2t99_asimd_load4): New.
12352 (thunderx2t99_asimd_store1_1_mult): Delete.
12353 (thunderx2t99_asimd_store1_2_mult): Delete.
12354 (thunderx2t99_asimd_store2_mult): Delete.
12355 (thunderx2t99_asimd_store2_onelane): Delete.
12356 (thunderx2t99_asimd_store_stp): New.
12357 (thunderx2t99_asimd_store1): New.
12358 (thunderx2t99_asimd_store2): New.
12359 (thunderx2t99_asimd_store3): New.
12360 (thunderx2t99_asimd_store4): New.
12362 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
12364 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
12365 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
12367 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
12368 Segher Boessenkool <segher@kernel.crashing.org>
12371 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
12374 2018-05-17 Richard Biener <rguenther@suse.de>
12376 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
12377 for pruning loop and prune defs feeding only already visited PHIs.
12379 2018-05-17 Richard Biener <rguenther@suse.de>
12381 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
12383 2018-05-17 Bin Cheng <bin.cheng@arm.com>
12384 Richard Biener <rguenther@suse.de>
12386 PR tree-optimization/85793
12387 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
12388 for VMAT_ELEMENTWISE.
12390 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
12392 * internal-fn.h (lookup_internal_fn): Declare
12393 * internal-fn.c (lookup_internal_fn): New function.
12394 * gimple.c (gimple_build_call_from_tree): Handle calls to
12395 internal functions.
12396 * gimple-pretty-print.c (dump_gimple_call): Print "." before
12397 internal function names.
12398 * tree-pretty-print.c (dump_generic_node): Likewise.
12399 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
12401 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
12403 * gimple-fold.h (gimple_build): Make the function forms take
12404 combined_fn rather than built_in_function.
12405 (gimple_simplify): Likewise.
12406 * gimple-match-head.c (gimple_simplify): Likewise.
12407 * gimple-fold.c (gimple_build): Likewise.
12408 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
12409 rather than gimple_build_call_internal.
12410 (get_initial_defs_for_reduction): Likewise.
12411 (vect_create_epilog_for_reduction): Likewise.
12412 (vectorizable_live_operation): Likewise.
12414 2018-05-17 Martin Liska <mliska@suse.cz>
12416 * gimple-ssa-sprintf.c (format_directive): Do not use
12417 space in between 'G_' and '('.
12419 2018-05-17 Jakub Jelinek <jakub@redhat.com>
12422 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
12423 even if the mask is not all ones.
12426 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
12428 (ix86_gimple_fold_builtin): Likewise.
12431 * config/i386/i386.c: Include tree-vector-builder.h.
12432 (ix86_vector_shift_count): New function.
12433 (ix86_fold_builtin): Fold shift builtins by scalar count.
12434 (ix86_gimple_fold_builtin): Likewise.
12436 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
12437 _mm512_setzero): New intrinsics.
12439 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
12440 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12442 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
12443 code generation for cases where splatting a value is not useful.
12444 * simplify-rtx.c (simplify_ternary_operation): Simplify
12445 vec_merge across a vec_duplicate and a paradoxical subreg forming
12446 a vector mode to a vec_concat.
12448 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
12450 * config.gcc: Support "goldmont-plus".
12451 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
12453 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12454 PROCESSOR_GOLDMONT_PLUS.
12455 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
12456 (processor_target_table): Add "goldmont-plus".
12457 (PTA_GOLDMONT_PLUS): Define.
12458 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
12459 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
12460 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
12461 (fold_builtin_cpu): Add "goldmont-plus".
12462 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
12463 (ix86_option_override_internal): Add "goldmont-plus".
12464 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
12465 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
12466 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
12467 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
12469 2018-05-17 Richard Biener <rguenther@suse.de>
12471 PR tree-optimization/85757
12472 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
12473 remove defs that only feed that PHI from further processing.
12475 2018-05-16 Jim Wilson <jimw@sifive.com>
12477 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
12479 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
12481 2018-05-16 Mark Wielaard <mark@klomp.org>
12483 * dwarf2out.c (count_index_strings): New function.
12484 (output_indirect_strings): Call count_index_strings and generate
12485 header for dwarf_version >= 5.
12487 2018-05-16 Mark Wielaard <mark@klomp.org>
12489 * dwarf2out.c (dwarf_FORM): New function.
12490 (set_indirect_string): Use dwarf_FORM.
12491 (reset_indirect_string): Likewise.
12492 (size_of_die): Likewise.
12493 (value_format): Likewise.
12494 (output_die): Likewise.
12495 (add_skeleton_AT_string): Likewise.
12496 (output_macinfo_op): Likewise.
12497 (index_string): Likewise.
12498 (output_index_string_offset): Likewise.
12499 (output_index_string): Likewise.
12500 (count_index_strings): Likewise.
12502 2018-05-16 Carl Love <cel@us.ibm.com>
12504 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
12505 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
12507 2018-05-16 Martin Jambor <mjambor@suse.cz>
12509 * ipa-prop.c (ipa_free_all_edge_args): Remove.
12510 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
12512 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
12514 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
12515 (fnma<mode>4): Likewise.
12516 (fms<mode>4): Likewise.
12517 (fnms<mode>4): Likewise.
12518 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
12519 (aarch64_fnma<mode>4): Likewise.
12520 (aarch64_fms<mode>4): Likewise.
12521 (aarch64_fnms<mode>4): Likewise.
12522 (aarch64_fnmadd<mode>4): Likewise.
12524 2018-05-16 Jason Merrill <jason@redhat.com>
12526 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
12528 2018-05-16 Richard Biener <rguenther@suse.de>
12530 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
12531 (dump_stmt_cost): Declare.
12532 (add_stmt_cost): Dump cost we add.
12533 (add_stmt_costs): New function.
12534 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
12535 No longer exported.
12536 (vect_analyze_stmt): Adjust prototype.
12537 (vectorizable_condition): Likewise.
12538 (vectorizable_live_operation): Likewise.
12539 (vectorizable_reduction): Likewise.
12540 (vectorizable_induction): Likewise.
12541 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
12542 cost vector to pass to vectorizable_ and record afterwards.
12543 (vect_model_reduction_cost): Take cost vector argument and adjust.
12544 (vect_model_induction_cost): Likewise.
12545 (vectorizable_reduction): Likewise.
12546 (vectorizable_induction): Likewise.
12547 (vectorizable_live_operation): Likewise.
12548 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
12549 SLP_TREE_NUMBER_OF_VEC_STMTS.
12550 (vect_analyze_slp_cost_1): Remove.
12551 (vect_analyze_slp_cost): Likewise.
12552 (vect_slp_analyze_node_operations): Take visited args and
12553 a target cost vector. Avoid processing already visited stmt sets.
12554 (vect_slp_analyze_operations): Use a local cost vector to gather
12555 costs and register those of non-discarded instances.
12556 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
12557 (vect_schedule_slp_instance): Remove copying of
12558 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
12560 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
12561 adding cost. Record cost entry location.
12562 (vect_prologue_cost_for_slp_op): Function to compute cost of
12563 a constant or invariant generated for SLP vect in the prologue,
12564 split out from vect_analyze_slp_cost_1.
12565 (vect_model_simple_cost): Make static. Adjust for SLP costing.
12566 (vect_model_promotion_demotion_cost): Likewise.
12567 (vect_model_store_cost): Likewise, make static.
12568 (vect_model_load_cost): Likewise.
12569 (vectorizable_bswap): Add cost vector arg and adjust.
12570 (vectorizable_call): Likewise.
12571 (vectorizable_simd_clone_call): Likewise.
12572 (vectorizable_conversion): Likewise.
12573 (vectorizable_assignment): Likewise.
12574 (vectorizable_shift): Likewise.
12575 (vectorizable_operation): Likewise.
12576 (vectorizable_store): Likewise.
12577 (vectorizable_load): Likewise.
12578 (vectorizable_condition): Likewise.
12579 (vectorizable_comparison): Likewise.
12580 (can_vectorize_live_stmts): Likewise.
12581 (vect_analyze_stmt): Likewise.
12582 (vect_transform_stmt): Adjust calls to vectorizable_*.
12583 * tree-vectorizer.c: Include gimple-pretty-print.h.
12584 (dump_stmt_cost): New function.
12586 2018-05-16 Richard Biener <rguenther@suse.de>
12588 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
12589 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
12590 * tree-ssa-dse.c: Include tree-ssa-loop.h.
12591 (check_name): New callback.
12592 (dse_classify_store): Track cycles via a visited bitmap of PHI
12593 defs and simplify handling of in-loop and across loop dead stores
12594 and properly fail for loop-variant refs. Handle byte-tracking with
12595 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
12598 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
12600 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
12601 (vect_get_mask_type_for_stmt): Likewise.
12602 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
12604 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
12605 to determine the statement's vector type and the vector type that
12606 should be used for calculating nunits. Deal with cases in which
12607 the type has to be deferred.
12608 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
12609 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
12610 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
12611 (vect_determine_vf_for_stmt): New functions, split out from...
12612 (vect_determine_vectorization_factor): ...here.
12613 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
12614 (vect_get_mask_type_for_stmt): New functions, split out from
12615 vect_determine_vectorization_factor.
12617 2018-05-16 Richard Biener <rguenther@suse.de>
12619 * tree-cfg.c (verify_gimple_assign_ternary): Properly
12620 verify the [VEC_]COND_EXPR embedded comparison.
12622 2018-05-15 Martin Sebor <msebor@redhat.com>
12624 PR tree-optimization/85753
12625 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
12626 RECORD_TYPE in addition to ARRAY_TYPE.
12628 2018-05-15 Martin Sebor <msebor@redhat.com>
12630 PR middle-end/85643
12631 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
12633 2018-05-15 Richard Biener <rguenther@suse.de>
12635 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
12636 add by_clobber_p one. Change algorithm to collect all defs
12637 representing uses we need to walk and try reducing them to
12638 a single one before failing.
12639 (dse_dom_walker::dse_optimize_stmt): Adjust.
12641 2018-05-13 Mark Wielaard <mark@klomp.org>
12643 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
12644 (size_of_loc_descr): Likewise.
12645 (output_loc_operands): Likewise.
12646 (output_loc_operands_raw): Likewise.
12647 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
12648 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
12649 (hash_loc_operands): Likewise.
12650 (compare_loc_operands): Likewise.
12652 2018-05-14 Mark Wielaard <mark@klomp.org>
12654 * dwarf2out.c (count_index_addrs): New function.
12655 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
12657 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12659 PR tree-optimization/83648
12660 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
12661 return value as malloc candidate.
12663 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12666 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
12667 param as true in call to suggest_attribute.
12669 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
12671 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
12672 -mreadonly-in-sdata.
12674 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12676 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
12678 (aarch64_crypto_aesd_fused): Likewise.
12680 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
12682 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
12683 (movsi_aarch64): Likewise.
12684 (load_pairsi): Likewise.
12685 (load_pairdi): Likewise.
12686 (store_pairsi): Likewise.
12687 (store_pairdi): Likewise.
12688 (load_pairsf): Likewise.
12689 (load_pairdf): Likewise.
12690 (store_pairsf): Likewise.
12691 (store_pairdf): Likewise.
12692 (zero_extend): Likewise.
12693 (trunc): Swap alternatives.
12694 (fcvt_target): Add '?' to prefer w over r.
12696 2018-05-14 Jakub Jelinek <jakub@redhat.com>
12699 * config/i386/i386.md: Disallow non-commutative arithmetics in
12700 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
12701 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
12702 in the peephole2 before it.
12704 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
12706 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
12707 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
12708 (ix86_handle_option): Handle -mcldemote.
12709 * config.gcc: New header.
12710 * config/i386/cldemoteintrin.h: New file.
12711 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
12712 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
12714 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12715 OPTION_MASK_ISA_CLDEMOTE.
12716 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
12717 (ix86_valid_target_attribute_inner_p): Ditto.
12718 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
12719 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
12720 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
12721 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
12722 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
12724 * config/i386/i386.opt: Add -mcldemote.
12725 * config/i386/x86intrin.h: New header.
12726 * doc/invoke.texi: Add -mcldemote.
12728 2018-05-14 Richard Biener <rguenther@suse.de>
12730 * doc/match-and-simplify.texi: Adjust :s documentation.
12732 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
12734 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
12735 intended memcpy size.
12736 (REORDER_45): Likewise.
12738 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
12740 * sort.cc: New file.
12741 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
12742 * vec.c (qsort_chk): Use gcc_qsort.
12743 * Makefile.in (OBJS-libcommon): Add sort.o.
12744 (build/sort.o): New target. Use it...
12745 (BUILD_RTL): ... here, and...
12746 (build/gencfn-macros): ... here, and...
12747 (build/genmatch): ... here.
12749 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
12750 Chung-Ju Wu <jasonwucj@gmail.com>
12752 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
12753 * config/nds32/nds32-graywolf.md: New file.
12754 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
12755 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
12757 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
12758 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
12759 * config/nds32/nds32.md (pipeline_model): Add graywolf.
12760 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
12761 * config/nds32/pipelines.md: Include n15 settings.
12763 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
12764 Chung-Ju Wu <jasonwucj@gmail.com>
12766 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
12767 * config/nds32/nds32-n13.md: New file.
12768 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
12769 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
12771 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
12772 * config/nds32/nds32.md (pipeline_model): Add n13.
12773 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
12774 * config/nds32/pipelines.md: Include n13 settings.
12776 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
12777 Chung-Ju Wu <jasonwucj@gmail.com>
12779 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
12780 * config/nds32/nds32-n10.md: New file.
12781 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
12782 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
12784 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
12785 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
12786 * config/nds32/nds32.md (pipeline_model): Add n10.
12787 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
12788 * config/nds32/pipelines.md: Include n10 settings.
12790 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
12791 Kito Cheng <kito.cheng@gmail.com>
12792 Chung-Ju Wu <jasonwucj@gmail.com>
12794 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
12795 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
12796 Add enum values for DSP extension instructions.
12797 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
12799 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
12800 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
12801 New code iterators.
12802 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
12803 * config/nds32/nds32-dspext.md: New file for DSP implementation.
12804 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
12805 * config/nds32/nds32-intrinsic.md: Likewise.
12806 * config/nds32/nds32_intrinsic.h: Likewise.
12807 * config/nds32/nds32-md-auxiliary.c: Likewise.
12808 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
12809 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
12810 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
12811 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
12812 * config/nds32/nds32-protos.h: New declarations for DSP extension.
12813 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
12814 TYPE_DMAC in switch statement.
12815 * config/nds32/nds32.c: New checking and implementation for DSP
12816 extension instructions.
12817 * config/nds32/nds32.h: Likewise.
12818 * config/nds32/nds32.md: Likewise.
12819 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
12820 * config/nds32/predicates.md: Implement new predicates for DSP
12823 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
12825 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
12826 Reformat alternatives and attributes so it is easier to identify
12827 which constraints/attributes go with which instruction.
12828 (mov<mode>_hardfloat32, FMOVE64): Likewise.
12829 (mov<mode>_softfloat32, FMOVE64): Likewise.
12830 (mov<mode>_hardfloat64, FMOVE64): Likewise.
12831 (mov<mode>_softfloat64, FMOVE64): Likewise.
12833 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
12835 * doc/extend.texi (PowerPC Built-in Functions): Rename this
12837 (Basic PowerPC Built-in Functions): The new name of the
12838 subsection previously known as "PowerPC Built-in Functions".
12839 (Basic PowerPC Built-in Functions Available on all Configurations):
12841 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
12842 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
12843 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
12844 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
12846 2018-05-11 Martin Jambor <mjambor@suse.cz>
12849 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
12852 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
12855 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
12857 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
12859 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
12860 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
12861 (ix86_handle_option): Handle -mwaitpkg.
12862 * config.gcc: New header.
12863 * config/i386/cpuid.h (bit_WAITPKG): New bit.
12864 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
12865 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
12867 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12868 OPTION_MASK_ISA_WAITPKG.
12869 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
12870 (ix86_option_override_internal): Add PTA_WAITPKG.
12871 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
12872 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
12873 IX86_BUILTIN_TPAUSE.
12874 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
12875 __builtin_ia32_umwait and __builtin_ia32_tpause.
12876 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
12877 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
12878 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
12879 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
12880 UNSPECV_TPAUSE): New.
12881 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
12882 * config/i386/i386.opt: Add -mwaitpkg.
12883 * config/i386/waitpkgintrin.h: New file.
12884 * config/i386/x86intrin.h: New header.
12885 * doc/invoke.texi: Add -mwaitpkg.
12887 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
12890 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
12892 (cortex-m0): Use armv6s-m isa.
12893 (cortex-m0plus): Likewise.
12894 (cortex-m1): Likewise.
12895 (cortex-m0.small-multiply): Likewise.
12896 (cortex-m0plus.small-multiply): Likewise.
12897 (cortex-m1.small-multiply): Likewise.
12899 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
12900 Jakub Jelinek <jakub@redhat.com>
12902 PR tree-optimization/85692
12903 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
12904 source permute as well.
12906 2018-05-11 Martin Liska <mliska@suse.cz>
12909 * doc/extend.texi: Document LLVM style format for no_sanitize
12912 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
12914 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
12915 mode_supports_vsx_dform_quad to mode_supports_dq_form.
12916 (mode_supports_vsx_dform_quad): Likewise.
12917 (mode_supports_vmx_dform): Move these functions to be next to the
12918 other mode_supports functions.
12919 (mode_supports_dq_form): Likewise.
12920 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
12921 mode_supports_dq_form.
12922 (reg_offset_addressing_ok_p): Likewise.
12923 (offsettable_ok_by_alignment): Likewise.
12924 (rs6000_legitimate_offset_address_p): Likewise.
12925 (legitimate_lo_sum_address_p): Likewise.
12926 (rs6000_legitimize_address): Likewise.
12927 (rs6000_legitimize_reload_address): Likewise.
12928 (rs6000_secondary_reload_inner): Likewise.
12929 (rs6000_preferred_reload_class): Likewise.
12930 (rs6000_output_move_128bit): Likewise.
12932 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
12934 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
12935 Generate SImode target register for null target.
12936 <case IX86_BUILTIN_XGETBV>: Ditto.
12937 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
12938 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
12940 2018-05-10 Carl Love <cel@us.ibm.com>
12942 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
12943 dcbtt and dcbtstt if operands[2] is 0.
12945 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
12948 * config/i386/sse.md (usadv64qi): New expander.
12950 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
12952 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
12953 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
12954 -maltivec=be support.
12955 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
12956 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
12957 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12958 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
12959 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
12960 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
12961 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
12962 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
12963 altivec_vsumsws): Adjust.
12964 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
12965 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
12966 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
12968 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
12969 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
12970 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
12971 (altivec_lve<VI_char>x): Delete expand.
12972 (*altivec_lve<VI_char>x_internal): Rename to...
12973 (altivec_lve<VI_char>x): ... this.
12974 (altivec_lvxl_<mode>): Delete expand.
12975 (*altivec_lvxl_<mode>_internal): Rename to ...
12976 (altivec_lvxl_<mode>): ... this.
12977 (altivec_stvxl_<mode>): Delete expand.
12978 (*altivec_stvxl_<mode>_internal): Rename to ...
12979 (altivec_stvxl_<mode>): ... this.
12980 (altivec_stve<VI_char>x): Delete expand.
12981 (*altivec_stve<VI_char>x_internal): Rename to ...
12982 (altivec_stve<VI_char>x): ... this.
12983 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
12984 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
12985 reduc_plus_scal_<mode>): Adjust.
12986 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
12988 (rs6000_cpu_cpp_builtins): Adjust.
12989 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
12990 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
12991 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
12992 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
12993 -maltivec=be support.
12994 (rs6000_split_vec_extract_var): Adjust.
12995 (rs6000_split_v4si_init): Adjust.
12996 (swap_selector_for_mode): Delete.
12997 (altivec_expand_lvx_be, altivec_expand_stvx_be,
12998 altivec_expand_stvex_be): Delete.
12999 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
13000 -maltivec=be support.
13001 (rs6000_gimple_fold_builtin): Ditto.
13002 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
13004 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
13005 (TARGET_DIRECT_MOVE_64BIT): Adjust.
13006 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
13007 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
13008 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
13009 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
13010 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
13011 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
13012 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
13013 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
13014 anonymous split): Adjust.
13015 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
13016 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
13018 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
13020 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
13021 when --with-gxx-include-dir is also specified.
13022 * configure: Regenerate.
13024 2018-05-09 Jim Wilson <jimw@sifive.com>
13027 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
13028 * config/riscv/t-withmultilib: New.
13029 * config/riscv/withmultilib.h: New.
13030 * doc/install.texi: Document RISC-V --with-multilib-list support.
13032 2018-05-09 Richard Biener <rguenther@suse.de>
13034 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
13036 (vect_bb_vectorization_profitable_p): Adjust. Compute
13037 actual scalar cost using the cost vector and the add_stmt_cost
13040 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13042 PR rtl-optimization/85645
13043 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
13044 in the REG_CFA_REGISTER note for LR, don't leave it empty.
13046 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13048 PR rtl-optimization/85645
13049 * shrink-wrap.c (spread_components): Return a boolean saying if
13050 anything was changed.
13051 (try_shrink_wrapping_separate): Iterate spread_components until
13052 nothing changes anymore.
13054 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13056 PR rtl-optimization/85645
13057 * regrename.c (build_def_use): Also kill the chains that include the
13058 destination of a REG_CFA_REGISTER note.
13060 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13062 PR rtl-optimization/85645
13063 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
13064 insn that has a REG_CFA_REGISTER note.
13066 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
13068 * cfgexpand.c (expand_clobber): New function.
13069 (expand_gimple_stmt_1): Use it.
13070 * tree-vect-stmts.c (vect_clobber_variable): New function,
13072 (vectorizable_simd_clone_call): ...here.
13073 (vectorizable_store): Emit a clobber either side of an
13074 IFN_STORE_LANES sequence.
13075 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
13077 2018-05-09 Tom de Vries <tom@codesourcery.com>
13080 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
13081 (define_insn "trap_if_false"): Add exit after trap.
13083 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
13085 PR rtl-optimization/85638
13086 * bb-reorder.c: Include common/common-target.h.
13087 (create_forwarder_block): New function extracted from...
13088 (fix_up_crossing_landing_pad): ...here. Rename into...
13089 (dw2_fix_up_crossing_landing_pad): ...this.
13090 (sjlj_fix_up_crossing_landing_pad): New function.
13091 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
13092 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
13093 from both partitions and exit the loop after one iteration.
13095 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13098 * doc/extend.texi (PowerPC Built-in Functions): Rename this
13100 (Basic PowerPC Built-in Functions): The new name of the
13101 subsection previously known as "PowerPC Built-in Functions".
13102 (Basic PowerPC Built-in Functions Available on all Configurations):
13104 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
13106 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
13108 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
13110 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
13113 2018-05-08 Jim Wilson <jimw@sifive.com>
13115 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
13116 (LD_EMUL_SUFFIX): New.
13117 (LINK_SPEC): Use it.
13119 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13121 * doc/extend.texi (PowerPC Built-in Functions): Rename this
13123 (Basic PowerPC Built-in Functions): The new name of the
13124 subsection previously known as "PowerPC Built-in Functions".
13125 (Basic PowerPC Built-in Functions Available on all Configurations):
13127 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
13129 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
13131 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
13133 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
13136 2018-05-08 Jakub Jelinek <jakub@redhat.com>
13139 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
13140 after cmpelim optimization.
13142 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
13144 * config.gcc: Support "goldmont".
13145 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
13146 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13147 PROCESSOR_GOLDMONT.
13148 * config/i386/i386.c (m_GOLDMONT): Define.
13149 (processor_target_table): Add "goldmont".
13150 (PTA_GOLDMONT): Define.
13151 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
13152 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
13153 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
13154 (fold_builtin_cpu): Add "goldmont".
13155 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
13156 (ix86_option_override_internal): Add "goldmont".
13157 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
13158 (processor_type): Add PROCESSOR_GOLDMONT.
13159 * config/i386/i386.md: Add CPU "glm".
13160 * config/i386/glm.md: New file.
13161 * config/i386/x86-tune.def: Add m_GOLDMONT.
13162 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
13164 2018-05-08 Jakub Jelinek <jakub@redhat.com>
13167 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
13169 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
13170 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
13171 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
13174 * config/i386/i386.c (ix86_fold_builtin): Handle
13175 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
13178 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
13179 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
13181 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
13184 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
13185 (check_arch): Likewise.
13186 (check_fpu): Return the result rather than printing it.
13187 (end arch): Fix operator precedence.
13188 (end cpu): Likewise.
13189 (END): Print the result from check_fpu.
13191 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
13192 Alan Hayward <alan.hayward@arm.com>
13193 David Sherwood <david.sherwood@arm.com>
13195 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
13196 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
13197 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
13198 (*fcmuo<mode>_and): New patterns.
13200 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
13202 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
13203 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
13204 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
13205 (cmp_op, sve_imm_con): New code attributes.
13206 (SVE_COND_INT_CMP, imm_con): Delete.
13207 (cmp_op): Remove above unspecs from int attribute.
13208 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
13210 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
13211 comparison-specific unspecs.
13212 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
13213 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
13214 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
13215 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
13216 (*vec_fcm<cmp_op><mode>): Rename to...
13217 (*fcm<cmp_op><mode>): ...this and adjust likewise.
13218 (*vec_fcmuo<mode>): Rename to...
13219 (*fcmuo<mode>): ...this and adjust likewise.
13220 (*pred_fcm<cmp_op><mode>): New pattern.
13221 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
13222 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
13224 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
13226 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
13227 (aarch64_emit_sve_predicated_cond): New function.
13228 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
13229 (aarch64_emit_unspec_cond_or): Replace with...
13230 (aarch64_emit_sve_or_conds): ...this new function. Use
13231 aarch64_emit_sve_ptrue_op for the individual comparisons and
13232 aarch64_emit_binop to OR them together.
13233 (aarch64_emit_inverted_unspec_cond): Replace with...
13234 (aarch64_emit_sve_inverted_cond): ...this new function. Use
13235 aarch64_emit_sve_ptrue_op for the comparison and
13236 aarch64_emit_unop to invert the result.
13237 (aarch64_expand_sve_vec_cmp_float): Update after the above
13238 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
13240 2018-05-07 Nathan Sidwell <nathan@acm.org>
13242 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
13243 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
13244 (Backwards Compatibility): Likewise.
13246 2018-05-07 Luis Machado <luis.machado@linaro.org>
13250 2018-05-07 Luis Machado <luis.machado@linaro.org>
13252 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13253 <prefetch_dynamic_strides>: New const bool field.
13254 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13255 prefetch_dynamic_strides.
13256 (exynosm1_prefetch_tune): Likewise.
13257 (thunderxt88_prefetch_tune): Likewise.
13258 (thunderx_prefetch_tune): Likewise.
13259 (thunderx2t99_prefetch_tune): Likewise.
13260 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
13262 (aarch64_override_options_internal): Update to set
13263 PARAM_PREFETCH_DYNAMIC_STRIDES.
13264 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
13265 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
13266 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
13267 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
13268 prefetch-dynamic-strides setting.
13270 2018-05-07 Luis Machado <luis.machado@linaro.org>
13272 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13273 <minimum_stride>: New const int field.
13274 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13275 minimum_stride field.
13276 (exynosm1_prefetch_tune): Likewise.
13277 (thunderxt88_prefetch_tune): Likewise.
13278 (thunderx_prefetch_tune): Likewise.
13279 (thunderx2t99_prefetch_tune): Likewise.
13280 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
13281 (aarch64_override_options_internal): Update to set
13282 PARAM_PREFETCH_MINIMUM_STRIDE.
13283 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
13284 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
13285 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
13286 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
13287 stride is constant and is below the minimum stride threshold.
13289 2018-05-07 Luis Machado <luis.machado@linaro.org>
13291 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
13294 2018-05-07 Luis Machado <luis.machado@linaro.org>
13296 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13297 <prefetch_dynamic_strides>: New const bool field.
13298 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13299 prefetch_dynamic_strides.
13300 (exynosm1_prefetch_tune): Likewise.
13301 (thunderxt88_prefetch_tune): Likewise.
13302 (thunderx_prefetch_tune): Likewise.
13303 (thunderx2t99_prefetch_tune): Likewise.
13304 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
13306 (aarch64_override_options_internal): Update to set
13307 PARAM_PREFETCH_DYNAMIC_STRIDES.
13308 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
13309 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
13310 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
13311 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
13312 prefetch-dynamic-strides setting.
13314 2018-05-07 Luis Machado <luis.machado@linaro.org>
13316 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13317 <minimum_stride>: New const int field.
13318 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13319 minimum_stride field.
13320 (exynosm1_prefetch_tune): Likewise.
13321 (thunderxt88_prefetch_tune): Likewise.
13322 (thunderx_prefetch_tune): Likewise.
13323 (thunderx2t99_prefetch_tune): Likewise.
13324 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
13325 (aarch64_override_options_internal): Update to set
13326 PARAM_PREFETCH_MINIMUM_STRIDE.
13327 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
13328 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
13329 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
13330 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
13331 stride is constant and is below the minimum stride threshold.
13333 2018-05-06 Jakub Jelinek <jakub@redhat.com>
13336 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
13337 the type is addressable. Don't force op into register if it has
13340 2018-05-05 Roland McGrath <mcgrathr@google.com>
13343 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
13344 any section for which we don't know a specific type it should have,
13345 regardless of name. Previously this was done only for the exact
13346 names ".init_array", ".fini_array", and ".preinit_array".
13347 (default_elf_asm_named_section): Add comment about
13348 relationship with default_section_type_flags and SECTION_NOTYPE.
13349 (get_section): Don't consider it a type conflict if one side has
13350 SECTION_NOTYPE and the other doesn't, as long as neither has the
13351 SECTION_BSS et al used in the default_section_type_flags logic.
13353 2018-05-05 Tom de Vries <tom@codesourcery.com>
13356 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
13357 (workaround_barsyncs): New function.
13358 (nvptx_reorg): Use workaround_barsyncs.
13359 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
13360 (define_expand "nvptx_membar_cta"): New define_expand.
13361 (define_insn "*nvptx_membar_cta"): New insn.
13363 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
13365 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
13366 To improve optimization opportunities.
13367 * builtin-types.def: The new needed builtin types for the above.
13369 2018-05-04 Richard Biener <rguenther@suse.de>
13371 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
13372 * gimple-ssa-store-merging.c
13373 (imm_store_chain_info::output_merged_store): Remove redundant create,
13374 release split_store vector contents on failure.
13375 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
13376 scalar stmt vector on cache hit.
13378 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
13380 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
13382 * config.gcc (powerpc-xilinx-eabi*): Remove.
13383 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
13385 (fusion_addis_mem_combo_load): Ditto.
13386 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
13388 (rs6000_cpu_cpp_builtins): Ditto.
13389 * config/rs6000/rs6000-linux.c
13390 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
13391 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
13392 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
13394 (rs6000_setup_reg_addr_masks): Ditto.
13395 (rs6000_init_hard_regno_mode_ok): Ditto.
13396 (rs6000_option_override_internal): Ditto.
13397 (legitimate_lo_sum_address_p): Ditto.
13398 (rs6000_legitimize_address): Ditto.
13399 (rs6000_legitimize_reload_address): Ditto.
13400 (rs6000_legitimate_address_p): Ditto.
13401 (abi_v4_pass_in_fpr): Ditto.
13402 (setup_incoming_varargs): Ditto.
13403 (rs6000_gimplify_va_arg): Ditto.
13404 (rs6000_split_multireg_move): Ditto.
13405 (rs6000_savres_strategy): Ditto.
13406 (rs6000_emit_prologue_components): Ditto.
13407 (rs6000_emit_epilogue_components): Ditto.
13408 (rs6000_emit_prologue): Ditto.
13409 (rs6000_emit_epilogue): Ditto.
13410 (rs6000_elf_file_end): Ditto.
13411 (rs6000_function_value): Ditto.
13412 (rs6000_libcall_value): Ditto.
13413 * config/rs6000/rs6000.h: Ditto.
13414 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
13415 (TARGET_MINMAX): ... this. New.
13416 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
13417 * config/rs6000/rs6000.md: Remove Xilinx FP support.
13418 (*movsi_internal1_single): Delete.
13419 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
13420 mfpu=, mxilinx-fpu): Delete.
13421 * config/rs6000/singlefp.h: Delete.
13422 * config/rs6000/sysv4.h: Remove Xilinx FP support.
13423 * config/rs6000/t-rs6000: Ditto.
13424 * config/rs6000/t-xilinx: Delete.
13425 * config/rs6000/titan.md: Adjust for fp_type removal.
13426 * config/rs6000/vsx.md: Remove Xilinx FP support.
13427 (VStype_simple): Delete.
13428 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
13429 * config/rs6000/xfpu.h: Delete.
13430 * config/rs6000/xfpu.md: Delete.
13431 * config/rs6000/xilinx.h: Delete.
13432 * config/rs6000/xilinx.opt: Delete.
13433 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
13434 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
13436 2018-05-04 Tom de Vries <tom@codesourcery.com>
13439 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
13442 2018-05-04 Richard Biener <rguenther@suse.de>
13444 PR middle-end/85627
13445 * tree-complex.c (update_complex_assignment): We are always in SSA form.
13446 (expand_complex_div_wide): Likewise.
13447 (expand_complex_operations_1): Likewise.
13448 (expand_complex_libcall): Preserve EH info of the original stmt.
13449 (tree_lower_complex): Handle removed blocks.
13450 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
13451 on complex multiplication and division libcall builtins.
13453 2018-05-04 Richard Biener <rguenther@suse.de>
13455 PR middle-end/85574
13456 * fold-const.c (negate_expr_p): Restrict negation of operand
13457 zero of a division to when we know that can happen without
13459 (fold_negate_expr_1): Likewise.
13461 2018-05-04 Jakub Jelinek <jakub@redhat.com>
13464 * real.h (real_nextafter): Declare.
13465 * real.c (real_nextafter): New function.
13466 * fold-const-call.c (fold_const_nextafter): New function.
13467 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
13468 CASE_CFN_NEXTTOWARD.
13469 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
13470 even when arg1_mode is different from arg0_mode.
13472 2018-05-03 Nathan Sidwell <nathan@acm.org>
13474 * doc/extend.texi (Deprecated Features): Remove
13475 -ffriend-injection.
13476 (Backwards Compatibility): Likewise.
13477 * doc/invoke.texi (C++ Language Options): Likewise.
13478 (C++ Dialect Options): Likewise.
13480 2018-05-03 Jakub Jelinek <jakub@redhat.com>
13483 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
13484 _mm512_mask_mullox_epi64): New intrinsics.
13486 2018-05-03 Tom de Vries <tom@codesourcery.com>
13489 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
13490 dump files): Add offload-tree.
13492 2018-05-03 Richard Biener <rguenther@suse.de>
13494 PR tree-optimization/85615
13495 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
13496 to loops not nested in BBs loop father to avoid creating multi-entry
13499 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13501 PR tree-optimization/70291
13502 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
13503 arguments. Change return type to tree. Emit libcall as a new
13504 statement rather than replacing existing one when inplace_p is true.
13505 (expand_complex_multiplication_components): New function.
13506 (expand_complex_multiplication): Expand floating-point complex
13507 multiplication using the above.
13508 (expand_complex_division): Rename inner_type parameter to type.
13509 Update expand_complex_libcall call-site.
13510 (expand_complex_operations_1): Update expand_complex_multiplication
13511 and expand_complex_division call-sites.
13513 2018-05-02 Jakub Jelinek <jakub@redhat.com>
13516 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
13517 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
13518 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
13519 the highest significant bit of the shift count mask is clear. In
13520 check whether and[sq]i3 is needed verify that all significant bits
13521 of the shift count other than the highest are set.
13523 2018-05-02 Tom de Vries <tom@codesourcery.com>
13526 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
13527 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
13528 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
13529 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
13530 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
13531 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
13532 __builtin_goacc_parlevel_size.
13534 2018-05-02 Richard Biener <rguenther@suse.de>
13536 PR tree-optimization/85597
13537 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
13538 do not use split vect_get_vec_defs call but call vect_get_slp_defs
13541 2018-05-02 Tom de Vries <tom@codesourcery.com>
13544 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
13545 dump files): Add ltrans-tree.
13547 2018-05-02 Tom de Vries <tom@codesourcery.com>
13550 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
13551 dump files): Add wpa-ipa.
13553 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
13555 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
13556 powerpc*-*-linux*paired* target.
13557 * config/rs6000/750cl.h: Delete.
13558 * config/rs6000/paired.h: Delete.
13559 * config/rs6000/paired.md: Delete.
13560 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
13562 * config/rs6000/rs6000-builtin.def: Remove paired float support.
13563 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13564 comment. Remove paired float support.
13565 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
13566 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
13568 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
13569 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
13571 * config/rs6000/rs6000.c: Remove paired float support.
13572 (paired_expand_vector_init, paired_expand_vector_move,
13573 paired_emit_vector_compare, paired_emit_vector_cond_expr,
13574 (paired_expand_lv_builtin, paired_expand_stv_builtin,
13575 paired_expand_builtin, paired_expand_predicate_builtin,
13576 paired_init_builtins): Delete.
13577 * config/rs6000/rs6000.h: Remove paired float support.
13578 * config/rs6000/rs6000.md: Remove paired float support.
13579 (move_from_CR_ov_bit): Delete.
13580 * config/rs6000/rs6000.opt (mpaired): Delete.
13581 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
13582 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
13584 2018-05-02 Richard Biener <rguenther@suse.de>
13586 PR middle-end/85567
13587 * gimplify.c (gimplify_save_expr): When in SSA form allow
13588 SAVE_EXPRs to compute to SSA vars.
13590 2018-05-02 Jakub Jelinek <jakub@redhat.com>
13593 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
13594 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
13595 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
13596 clobber operands[2], instead use a new pseudo. Formatting fixes.
13598 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
13600 PR tree-optimization/85586
13601 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
13602 exit early for statements in the same group if the accesses are
13605 2018-05-02 Tom de Vries <tom@codesourcery.com>
13608 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
13611 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
13613 PR tree-optimization/85143
13614 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
13616 2018-05-01 Tom de Vries <tom@codesourcery.com>
13619 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
13620 not found" error message.
13622 2018-05-01 Tom de Vries <tom@codesourcery.com>
13625 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
13626 * vec.c (test_ordered_remove_if): New function.
13627 (vec_c_tests): Call test_ordered_remove_if.
13628 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
13629 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
13630 * tree-vect-patterns.c (vect_pattern_recog_1): Use
13631 VEC_ORDERED_REMOVE_IF.
13633 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13635 PR tree-optimization/82665
13636 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
13637 pointer subtraction where arguments come from a memchr call.
13639 2018-05-01 Jakub Jelinek <jakub@redhat.com>
13641 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
13642 --push-state --as-needed and --pop-state instead of --as-needed and
13643 --no-as-needed if ld supports it.
13644 * configure: Regenerated.
13647 * doc/install.texi2html: Replace _002d with - and _002a with * in
13648 generated html files using sed.
13650 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13653 * gcc-rich-location.c (blank_line_before_p): New function.
13654 (use_new_line): New function.
13655 (gcc_rich_location::add_fixit_insert_formatted): New function.
13656 * gcc-rich-location.h
13657 (gcc_rich_location::add_fixit_insert_formatted): New function.
13659 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13661 * selftest.c (assert_streq): Rename "expected" and "actual" to
13662 "val1" and "val2". Extend NULL-handling to cover both inputs
13663 symmetrically, while still requiring both to be non-NULL for a pass.
13664 * selftest.h (assert_streq): Rename "expected" and "actual" to
13666 (ASSERT_EQ): Likewise.
13667 (ASSERT_EQ_AT): Likewise.
13668 (ASSERT_KNOWN_EQ): Likewise.
13669 (ASSERT_KNOWN_EQ_AT): Likewise.
13670 (ASSERT_NE): Likewise.
13671 (ASSERT_MAYBE_NE): Likewise.
13672 (ASSERT_MAYBE_NE_AT): Likewise.
13673 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
13674 the assertion to pass.
13675 (ASSERT_STREQ_AT): Likewise.
13677 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
13679 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
13680 interaction with -pie.
13682 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13684 * selftest.h: Fix alphabetization of per-source-file selftest
13687 2018-04-30 Jason Merrill <jason@redhat.com>
13689 PR c++/61982 - dead stores to destroyed objects.
13690 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
13693 2018-04-30 Jason Merrill <jason@redhat.com>
13695 * tree.c (build_clobber): New.
13696 * tree.h: Declare it.
13697 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
13699 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13701 * diagnostic-show-locus.c (layout::layout): Update for
13702 location_get_source_line returning a char_span.
13703 (struct char_span): Move to input.h.
13704 (struct correction): Update for fields in char_span becoming
13706 (struct source_line): Update for location_get_source_line
13707 returning a char_span.
13708 (layout::print_line): Likewise.
13709 * edit-context.c (edited_file::print_content): Likewise.
13710 (edited_file::print_diff_hunk): Likewise.
13711 (edited_file::print_run_of_changed_lines): Likewise.
13712 (edited_file::get_num_lines): Likewise.
13713 (edited_line::edited_line): Likewise.
13714 * final.c (asm_show_source): Likewise.
13715 * input.c (location_get_source_line): Convert return type
13716 from const char * to char_span, losing the final "line_len"
13718 (dump_location_info): Update for the above.
13719 (get_substring_ranges_for_loc): Likewise. Use a char_span
13720 when handling the literal within the line.
13721 (test_reading_source_line): Update for location_get_source_line
13722 returning a char_span.
13723 * input.h (class char_span): Move here from
13724 diagnostic-show-locus.c, converting from a struct to a class.
13725 Make data members private.
13726 (char_span::operator bool): New.
13727 (char_span::length): New.
13728 (char_span::get_buffer): New.
13729 (char_span::operator[]): New.
13730 (char_span::subspan): Make const.
13731 (char_span::xstrdup): New.
13732 (location_get_source_line): Convert return type from const char *
13733 to char_span, losing the final "line_size" param.
13735 2018-04-30 Jan Hubicka <jh@suse.cz>
13737 * lto-wrapper.c (ltrans_priorities): New static var.
13738 (cmp_priority): New.
13739 (run_gcc): Read priorities and if doing parallel build order
13740 the Makefile by them.
13742 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13744 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
13746 2018-04-30 Richard Biener <rguenther@suse.de>
13748 * tree-cfg.c (verify_address): Remove base argument, add
13749 flag whether to check TREE_ADDRESSABLE and do that.
13750 (verify_expr): Remove.
13751 (verify_types_in_gimple_reference): Add pieces from verify_expr.
13752 (verify_gimple_assign_single): Likewise.
13753 (verify_gimple_switch): Likewise.
13754 (verify_expr_location_1): Dereference tp once. Add (disabled)
13755 piece from verify_expr.
13756 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
13758 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
13760 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
13762 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
13764 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
13765 (small_data_pattern): Likewise.
13766 (arc_rewrite_small_data): Likewise.
13767 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
13768 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
13769 (get_symbol_alignment): New function.
13770 (legitimate_small_data_address_p): Likewise.
13771 (legitimate_scaled_address): Update, call
13772 legitimate_small_data_address_p.
13773 (output_sdata): New static variable.
13774 (arc_print_operand): Update how we handle small data operands.
13775 (arc_print_operand_address): Likewise.
13776 (arc_legitimate_address_p): Update, use
13777 legitimate_small_data_address_p.
13778 (arc_rewrite_small_data_p): Remove.
13779 (arc_rewrite_small_data_1): Likewise.
13780 (arc_rewrite_small_data): Likewise.
13781 (small_data_pattern): Likewise.
13782 (compact_sda_memory_operand): Update to use
13783 legitimate_small_data_address_p and get_symbol_alignment.
13784 (prepare_move_operands): Don't rewite sdata pattern.
13785 (prepare_extend_operands): Remove.
13786 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
13788 (zero_extendqisi2): Likewise.
13789 (zero_extendhisi2): Likewise.
13790 (extendqihi2): Likewise.
13791 (extendqisi2): Likewise.
13792 (extendhisi2): Likewise.
13793 (addsi3): Likewise.
13794 (subsi3): Likewise.
13795 (andsi3): Likewise.
13796 * config/arc/constraints.md (Usd): Change it to memory constraint.
13798 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
13800 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
13801 as source of std instructions.
13802 * config/arc/arc.md (movsi_insn): Update pattern predicate to
13803 allow 6-bit constants as source for store instructions.
13804 (movdi_insn): Update instruction pattern to allow 6-bit constants
13805 as source for store instructions.
13807 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
13809 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
13811 2018-04-30 Nathan Sidwell <nathan@acm.org>
13812 Sandra Loosemore <sandra@codesourcery.com>
13814 * dumpfile.c (dump_open): Allow '-' for stdout.
13815 * doc/invoke.texi (Developer Options): Document dump filename
13816 determination early. Document stdin/stdout selection.
13818 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
13820 Microblaze Target: PIC data text relative
13822 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
13823 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
13825 * config/microblaze/microblaze.h (microblaze_constant_address_p):
13826 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
13827 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
13828 New addressing mode for data-text relative position indepenedent code.
13829 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
13830 'ADDRESS_SYMBOLIC_TXT_REL'.
13831 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
13832 (microblaze_legitimate_pic_operand): Exclude function calls from
13833 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
13834 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
13836 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
13837 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
13838 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
13839 for 'address + offset'.
13840 (microblaze_expand_prologue): Add new function prologue call for
13842 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
13843 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
13844 table in case of TARGET_PIC_DATA_TEXT_REL.
13845 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
13846 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
13847 Add new macros 'UNSPEC_TEXT',
13848 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
13849 + exclude function calls from 'UNSPEC_PLT' in case of data text
13851 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
13852 new target hook for generating address diff vector tables in case of
13854 * doc/tm.texi : Regenerate.
13855 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
13856 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
13857 of addr diff vector generation.
13858 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
13859 target hook definition.
13860 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
13861 Add default function for generate_pic_addr_diff_vec -> flag_pic.
13862 * doc/invoke.texi (Add new pic option): Add new microblaze pic
13863 option for data text relative.
13865 2018-04-30 Richard Biener <rguenther@suse.de>
13867 * tree-chrec.h (evolution_function_is_constant_p): Remove
13869 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
13871 2018-04-30 Richard Biener <rguenther@suse.de>
13874 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
13876 2018-04-30 Richard Biener <rguenther@suse.de>
13878 PR tree-optimization/28364
13879 PR tree-optimization/85275
13880 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
13881 copying first exit test.
13883 2018-04-28 Mark Wielaard <mark@klomp.org>
13885 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
13886 dwarf_version >= 5.
13887 (dwarf_AT): Handle DW_AT_addr_base.
13888 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
13890 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
13893 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
13894 (*ashl<dwi>3_doubleword_mask_1): Ditto.
13895 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
13896 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
13898 2018-04-28 Richard Biener <rguenther@suse.de>
13900 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
13901 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
13902 to reflect use. Only add interesting stmts.
13904 2018-04-27 Martin Jambor <mjambor@suse.cz>
13907 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
13908 the jump function allows for passing through aggregate values.
13910 2018-04-27 David Malcolm <dmalcolm@redhat.com>
13912 * input.h (in_system_header_at): Convert from macro to inline
13914 (from_macro_expansion_at): Likewise.
13915 (from_macro_definition_at): Likewise.
13917 2018-04-27 Jeff Law <law@redhat.com>
13919 * config.gcc: Mark tile* targets as deprecated/obsolete.
13921 2018-04-27 Richard Biener <rguenther@suse.de>
13923 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
13926 2018-04-27 Richard Biener <rguenther@suse.de>
13928 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
13930 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
13932 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
13933 with Yd constraint. Set "preferred_for_speed" attribute from
13934 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
13935 with Yd constraint.
13936 (*movdi_internal): Ditto.
13937 (movti_interunit splitters): Remove
13938 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
13939 (movdi_interunit splitters): Ditto.
13940 * config/i386/constraints.md (Ye): Remove.
13941 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
13943 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13946 * config/aarch64/constraints.md (Usg): Limit to 31.
13947 (Usj): Limit to 63.
13949 2018-04-27 Jakub Jelinek <jakub@redhat.com>
13951 PR tree-optimization/85529
13952 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
13953 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
13954 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
13955 zero extension or masking of the MSB bit.
13956 (optimize_range_tests): Add FIRST_BB argument, pass it through
13957 to optimize_range_tests_var_bound.
13958 (maybe_optimize_range_tests, reassociate_bb): Adjust
13959 optimize_range_tests callers.
13961 2018-04-26 Richard Biener <rguenther@suse.de>
13962 Jakub Jelinek <jakub@redhat.com>
13964 * cgraph.h (symbol_table): Just declare debug method here.
13965 * symtab.c (symbol_table::debug): Define.
13967 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
13969 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
13971 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
13973 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
13974 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
13975 (*movdi_internal): Substitute Yi and Yj constraint with x
13976 and Ym and Yn constraint with y constraint. Update "isa"
13977 attribute and set "preferred_for_speed" attribute from
13978 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
13979 (*movsi_internal): Ditto.
13980 (*movdf_internal): Ditto.
13981 (*movsf_internal): Ditto.
13982 (*zero_extendsidi2): Ditto.
13983 * config/i386/sse.md (vec_set<mode>_0): Ditto.
13984 (sse2_loadld): Ditto.
13985 (*vec_extract<ssevecmodelower>_0): Ditto.
13986 (*vec_extractv4si_0_zext_sse4): Ditto.
13987 (vec_concatv2di): Ditto.
13988 (*vec_dup<mode>): Ditto.
13989 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
13990 * config/i386/constraints.md (Yi): Remove.
13995 2018-04-26 Nathan Sidwell <nathan@acm.org>
13997 * dumpfile.c (dump_open): New.
13998 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
13999 (dump_finish): Detect stdio/stderr by value not name.
14001 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
14003 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
14005 2018-04-26 Tom de Vries <tom@codesourcery.com>
14008 * config/nvptx/nvptx.c (verify_neutering_jumps)
14009 (verify_neutering_labels): New function
14010 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
14012 2018-04-26 Tom de Vries <tom@codesourcery.com>
14015 * config/nvptx/nvptx.c (needs_neutering_p): New function.
14016 (nvptx_single): Use needs_neutering_p to skip over insns that do not
14019 2018-04-26 Richard Biener <rguenther@suse.de>
14020 Tom de Vries <tom@codesourcery.com>
14023 * lto-streamer-out.c (output_function): Fixup loops if required to match
14024 discovery done in the reader.
14026 2018-04-26 Richard Biener <rguenther@suse.de>
14028 PR tree-optimization/85116
14029 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
14030 have a loop exit from the single latch predecessor. Remove
14031 case of header with just condition.
14032 (ch_base::copy_headers): Exclude infinite loops from any
14034 (pass_ch::execute): Record exits.
14036 2018-04-26 Richard Biener <rguenther@suse.de>
14038 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
14039 prologue cost vector and pass it to vect_get_load_cost.
14040 (vect_get_peeling_costs_all_drs): Likewise.
14041 (vect_peeling_hash_get_lowest_cost): Likewise.
14042 (vect_enhance_data_refs_alignment): Likewise.
14044 2018-04-26 Richard Biener <rguenther@suse.de>
14046 PR middle-end/85450
14047 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
14048 checking of integer<->pointer conversions.
14049 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
14050 sign-/zero-extending pointer types.
14051 (expand_omp_for_static_chunk): Likewise.
14053 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
14054 Jean Lee <xiaoyur347@gmail.com>
14056 * config/mips/mips.c (mips_asan_shadow_offset): New function.
14057 (TARGET_ASAN_SHADOW_OFFSET): Define.
14058 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
14059 true for -fsanitize=address.
14061 2018-04-25 Mark Wielaard <mark@klomp.org>
14063 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
14066 2018-04-25 Jakub Jelinek <jakub@redhat.com>
14068 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
14069 than "alu", remove explicit "memory" and "imm_disp" attributes.
14070 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
14072 PR middle-end/85414
14073 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
14074 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
14075 gen_lowpart_no_emit.
14077 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
14080 * config/i386/i386.c (ix86_expand_builtin): Change memory
14081 operand to XI, extend p0 to Pmode.
14082 * config/i386/i386.md: Change unspec volatile and operand
14083 1 mode to XI, change operand 0 mode to P.
14085 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14087 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
14088 GET_MODE_MASK before any checking.
14089 (nds32_can_use_bset_p): Likewise.
14090 (nds32_can_use_btgl_p): Likewise.
14092 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14094 * config/nds32/nds32-doubleword.md: New define_split pattern for
14095 illegal register number.
14097 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14099 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
14101 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14103 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
14105 2018-04-25 Richard Biener <rguenther@suse.de>
14107 * lto-streamer.h (LTO_major_version): Bump to 8.
14109 2018-04-25 Jakub Jelinek <jakub@redhat.com>
14111 * BASE-VER: Set to 9.0.0.
14113 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
14115 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
14116 in __abskf2 and __powikf2.
14118 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14121 * config/aarch64/constraints.md (Usg, Usj): New constraints.
14122 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
14123 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
14124 Use the above on operand 2. Reindent.
14125 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
14127 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
14130 * common/config/i386/i386-common.c (ix86_handle_option): Don't
14132 * config/i386/i386.opt (mcet): Removed.
14133 * doc/install.texi: Remove -mcet documentation.
14134 * doc/invoke.texi: Likewise.
14136 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
14139 * doc/install.texi: Remove -mcet from bootstrap-cet.
14141 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14144 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
14145 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
14149 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
14150 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
14151 containing a CONST_VECTOR.
14153 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
14155 * doc/install.texi: Update newlib dependency for nvptx.
14157 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14160 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
14161 instead of INTVAL when shifting x left.
14163 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
14165 PR tree-optimization/85478
14166 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
14167 vect_grouped_store_supported for single element vectors.
14169 2018-04-24 Richard Biener <rguenther@suse.de>
14172 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
14173 load cost increase to the case of non-constant step.
14175 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14178 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
14179 destination if any_malformed_asm.
14181 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
14183 PR middle-end/85496
14184 * expr.c (store_field): In the bitfield case, if the value comes from
14185 a function call and is returned in registers by means of a PARALLEL,
14186 do not change the mode of the temporary unless BLKmode and VOIDmode.
14188 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
14190 PR rtl-optimization/85423
14191 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
14192 dependencies to debug insns when the previous insn is non-debug.
14194 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
14196 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
14197 enums into a single definition.
14198 (fls): Fix predicates and printing.
14201 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
14203 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
14204 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
14205 and short u6 immediate.
14206 (check_if_valid_sleep_operand): Remove.
14207 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
14209 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14211 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
14212 flag_always_save_lp condition.
14213 * config/nds32/nds32.opt (malways-save-lp): New option.
14215 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14217 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
14218 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
14219 * config/nds32/nds32.h
14220 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
14221 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
14223 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14225 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
14226 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
14228 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14229 Chung-Ju Wu <jasonwucj@gmail.com>
14231 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
14233 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
14234 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
14236 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14238 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
14240 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14242 * config/nds32/nds32-protos.h (nds32_data_alignment,
14243 nds32_local_alignment): Declare.
14244 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
14245 nds32_local_alignment): New functions.
14246 (TARGET_CONSTANT_ALIGNMENT): Define.
14247 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
14249 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14251 * config/nds32/nds32.c
14252 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
14253 (TARGET_MODES_TIEABLE_P): Likewise.
14255 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14257 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
14258 level Ofast and Og.
14260 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
14261 Chung-Ju Wu <jasonwucj@gmail.com>
14263 * config/nds32/constants.md (unspec_volatile_element): Add enum values
14264 for unaligned access.
14265 * config/nds32/nds32-intrinsic.c: Implementation of expanding
14267 * config/nds32/nds32-intrinsic.md: Likewise.
14268 * config/nds32/nds32_intrinsic.h: Likewise.
14269 * config/nds32/nds32.h (nds32_builtins): Likewise.
14270 * config/nds32/nds32.opt (munaligned-access): New option.
14271 * config/nds32/nds32.c (nds32_asm_file_start): Display
14272 flag_unaligned_access status.
14274 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
14276 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
14277 -mno-relax is present.
14278 * config/riscv/linux.h (LINK_SPEC): Ditto.
14280 2018-04-20 Martin Sebor <msebor@redhat.com>
14283 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
14285 (gimple_fold_builtin_stxcpy_chk): Same.
14286 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
14288 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
14291 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
14292 __powikf2 when long double is IEEE 128-bit.
14294 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
14296 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
14297 step to make sure stack always aligned.
14299 2018-04-20 Carl Love <cel@us.ibm.com>
14302 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
14303 size check for arg0.
14305 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
14306 Tom de Vries <tom@codesourcery.com>
14309 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
14310 Emit insns for calls too.
14311 (nvptx_find_par): Always look for worker-level predecessor insn.
14312 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
14314 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
14315 (nvptx_process_pars): Propagate frames for calls.
14317 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
14320 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
14322 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
14323 (ix86_handle_option): Don't handle OPT_mibt.
14324 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
14326 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
14328 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
14329 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
14330 (ix86_target_macros): Define __CET__ with flag_cf_protection
14331 for -fcf-protection.
14332 * config/i386/i386.c (isa2_opts): Remove -mibt.
14333 * config/i386/i386.h (TARGET_IBT): Removed.
14334 (TARGET_IBT_P): Likewise.
14335 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
14336 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
14337 * config/i386/i386.opt (mcet): Update help message.
14338 (mshstk): Likewise.
14340 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
14341 -mcet as an alias for -mshstk.
14343 2018-04-20 Richard Biener <rguenther@suse.de>
14345 PR middle-end/85475
14346 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
14347 complexity by forcing a single use of the multiply operand.
14349 2018-04-20 Martin Jambor <mjambor@suse.cz>
14352 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
14353 recursion dependency to only apply to non-clones.
14355 2018-04-20 Martin Jambor <mjambor@suse.cz>
14358 * ipa-cp.c (create_specialized_node): Check that clones of
14359 self-recursive edges exist during IPA-CP.
14361 2018-04-19 Toon Moene <toon@moene.org>
14363 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
14366 2018-04-19 Jakub Jelinek <jakub@redhat.com>
14368 PR tree-optimization/85467
14369 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
14370 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
14371 VECTOR_CST element to type.
14373 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
14376 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
14377 * config/i386/i386.md (builtin_setjmp_setup): Removed.
14378 (builtin_longjmp): Likewise.
14379 (save_stack_nonlocal): New pattern.
14380 (restore_stack_nonlocal): Likewise.
14382 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
14385 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
14386 Replace ASM_OUTPUT_LABEL with fprintf.
14388 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
14391 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
14392 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
14393 * config/i386/i386-c.c (ix86_target_macros_internal): Also
14394 define __IBT__ and __SHSTK__ for -fcf-protection.
14395 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
14397 (ix86_trampoline_init): Likewise.
14398 (x86_output_mi_thunk): Likewise.
14399 (ix86_notrack_prefixed_insn_p): Likewise.
14400 (ix86_option_override_internal): Don't disallow -fcf-protection.
14401 * config/i386/i386.md (rdssp<mode>): Also enable for
14403 (incssp<mode>): Likewise.
14404 (nop_endbr): Likewise.
14405 * config/i386/i386.opt (mcet): Change help message to built-in
14408 (mshstk): Likewise.
14409 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
14410 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
14411 enable CET built-in functions.
14413 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
14415 * common/config/i386/i386-common.c
14416 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
14417 OPTION_MASK_ISA_MOVDIRI_UNSET,
14418 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
14419 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
14420 * config.gcc (movdirintrin.h): New header.
14421 * config/i386/cpuid.h (bit_MOVDIRI,
14422 bit_MOVDIR64B): New bits.
14423 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
14425 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
14426 (VOID, PVOID, PCVOID)): New function types.
14427 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
14428 __builtin_ia32_directstoreu_u64,
14429 __builtin_ia32_movdir64b): New builtins.
14430 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
14431 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
14433 (ix86_valid_target_attribute_inner_p): Ditto.
14434 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
14435 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
14436 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
14437 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
14438 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
14439 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
14440 (movdiri<mode>, movdir64b_<mode>): New.
14441 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
14442 * config/i386/immintrin.h: Include movdirintrin.h.
14443 * config/i386/movdirintrin.h: New file.
14444 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
14446 2018-04-19 Richard Biener <rguenther@suse.de>
14448 PR middle-end/85455
14449 * cfg.c (clear_bb_flags): When loop state says we have
14450 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
14452 2018-04-19 Richard Biener <rguenther@suse.de>
14454 PR tree-optimization/84737
14455 * tree-vect-data-refs.c (vect_copy_ref_info): New function
14456 copying restrict info.
14457 (vect_setup_realignment): Use it.
14458 * tree-vectorizer.h (vect_copy_ref_info): Declare.
14459 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
14460 the first DR to all generated stores.
14461 (vectorizable_load): Likewise for loads.
14463 2018-04-19 Jakub Jelinek <jakub@redhat.com>
14465 PR tree-optimization/85446
14466 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
14467 the integral and pointer types to have the same precision.
14469 * doc/install.texi: Document --disable-cet being the default and
14472 2018-04-18 Martin Liska <mliska@suse.cz>
14474 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
14477 2018-04-18 Martin Liska <mliska@suse.cz>
14480 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
14483 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
14484 arguments if they are comparable.
14486 2018-04-18 Martin Liska <mliska@suse.cz>
14489 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
14492 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
14495 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
14498 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
14499 ENDBR after calling __morestack.
14501 2018-04-18 David Malcolm <dmalcolm@redhat.com>
14504 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
14505 by using gcc_base_ver to generate a gcc_driver_version, and use
14506 it when generating GCC_DRIVER_NAME.
14507 * configure: Regenerate.
14509 2018-04-18 Jakub Jelinek <jakub@redhat.com>
14512 * config.gcc: Obsolete powerpc*-*-*spe*.
14514 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14517 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
14518 (stabstr_D): Change type of unum from unsigned int to
14519 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
14522 2018-04-17 Jim Wilson <jimw@sifive.com>
14525 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
14526 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
14527 Set arg_pointer_offset after using pretend_args_size.
14529 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14531 PR rtl-optimization/85431
14532 * dse.c (record_store): Ignore zero width stores.
14535 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
14536 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
14537 __builtin_stack_restore rather than after it.
14538 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
14539 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
14540 argument instead of virtual_dynamic_stack_rtx.
14542 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
14544 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
14546 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14547 Add note to error message to explain internal mapping of overloaded
14548 built-in function name to non-overloaded built-in function name.
14549 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
14552 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
14555 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
14556 where the inputs overlap with the output.
14558 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14560 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
14561 (=v, v) alternative and explicit "memory" attribute.
14562 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
14563 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
14565 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
14566 "sselog1" type instead of "sselog".
14567 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
14568 "sselog". Remove explicit "memory" attribute.
14569 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
14570 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
14572 (vec_extract_hi_v32hi): Merge all alternatives into one, use
14573 "sselog1" type instead of "sselog". Remove explicit "memory"
14575 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
14576 use "sselog1" type instead of "sselog". Remove explicit "memory"
14578 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
14579 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
14581 (vec_extract_hi_v64qi): Merge all alternatives into one, use
14582 "sselog1" type instead of "sselog". Remove explicit "memory"
14584 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
14585 use "sselog1" type instead of "sselog". Remove explicit "memory"
14589 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
14591 PR middle-end/85414
14592 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
14595 2018-04-17 Martin Jambor <mjambor@suse.cz>
14598 * ipa-cp.c (create_specialized_node): Call
14599 expand_all_artificial_thunks if necessary.
14601 2018-04-17 Martin Liska <mliska@suse.cz>
14604 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
14605 in message, remote space in between '_G' and '('.
14607 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14610 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
14611 avx512f_vmcmp<mode>3<round_saeonly_name>,
14612 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
14613 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
14614 avx512f_rndscale<mode><round_saeonly_name>,
14615 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
14616 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
14617 Use %<iptr>2 instead of %2 for -masm=intel.
14618 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
14619 avx512f_vcvttss2usi<round_saeonly_name>,
14620 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
14622 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
14623 avx512f_vcvttsd2usi<round_saeonly_name>,
14624 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
14625 Use %q1 instead of %1 for -masm=intel.
14626 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
14627 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
14628 of %3 for -masm=intel.
14629 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
14631 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
14633 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
14635 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
14636 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
14638 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
14640 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
14641 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
14642 %g1 and one with %0 and %1.
14643 (avx512er_vmrcp28<mode><round_saeonly_name>,
14644 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
14645 %1 for -masm=intel.
14646 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
14647 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
14648 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
14649 of %0 and %{%4%} for -masm=intel.
14650 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
14651 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
14652 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
14653 order of %0 and %{%5%}%{z%} for -masm=intel.
14655 2018-04-17 Jan Hubicka <jh@suse.cz>
14658 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
14660 2018-04-17 Martin Liska <mliska@suse.cz>
14663 * multiple_target.c (create_dispatcher_calls): Set apostrophes
14664 for target_clone error message. Make default implementation
14665 clone to be a local declaration.
14666 (separate_attrs): Add new argument and check for an empty
14668 (expand_target_clones): Handle it.
14669 (ipa_target_clone): Make redirection just for target_clones
14672 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
14673 Tom de Vries <tom@codesourcery.com>
14675 PR middle-end/84955
14676 * omp-expand.c (expand_oacc_for): Add dummy false branch for
14677 tiled basic blocks without omp continue statements.
14679 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
14682 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
14683 vec_extract expression as having side effects to make sure it gets
14686 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
14689 * config/i386/i386.c (get_builtin_code_for_version): Check
14692 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
14695 * config.gcc: Support "skylake".
14696 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14698 * config/i386/i386.c (m_SKYLAKE): Define.
14699 (processor_target_table): Add "skylake".
14700 (ix86_option_override_internal): Add "skylake".
14701 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
14702 PROCESSOR_CANNONLAKE.
14703 (get_builtin_code_for_version): Fix priority for
14704 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
14705 PROCESSOR_SKYLAKE-AVX512.
14706 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
14707 (processor_type): Add PROCESSOR_SKYLAKE.
14709 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
14710 Jason Merrill <jason@redhat.com>
14713 * convert.c (convert_to_integer_1): Use direct recursion for
14714 enumeral types and types with a precision less than the number
14715 of bits in their mode.
14717 2018-04-16 Julia Koval <julia.koval@intel.com>
14720 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
14721 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
14723 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
14726 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
14727 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
14728 and -mno-direct-move.
14730 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
14733 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
14734 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
14735 Ensure negative shifts result in {0}.
14737 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
14739 PR rtl-optimization/79916
14740 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
14741 regs (if any) to define how to gnerate SD moves when LRA is in
14744 2018-04-13 Jakub Jelinek <jakub@redhat.com>
14746 PR rtl-optimization/85393
14747 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
14748 * except.c (expand_dw2_landing_pad_for_region): Make static.
14749 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
14750 a label and unconditional jump to old_bb, rather than
14751 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
14754 PR rtl-optimization/85376
14755 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
14756 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
14757 instead of a specific value.
14759 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
14760 Bin Cheng <bin.cheng@arm.com>
14762 PR tree-optimization/82965
14763 PR tree-optimization/83991
14764 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
14765 by_profile_only parameter.
14766 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
14767 information if the loop was predicted to iterate too many times.
14768 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
14770 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
14773 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
14776 2018-04-13 Martin Liska <mliska@suse.cz>
14777 Jakub Jelinek <jakub@redhat.com>
14779 PR middle-end/81657
14780 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
14781 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
14782 * builtins.c (expand_builtin_memory_copy_args): Use
14783 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
14784 handle dest_addr == pc_rtx.
14786 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
14789 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
14790 asked to not generate direct moves.
14791 (fix_trunc<mode>si2_stfiwx): Similar.
14792 (fix_trunc<mode>si2_internal): Similar.
14794 2018-04-12 Jakub Jelinek <jakub@redhat.com>
14797 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
14798 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
14799 lookup if dest in some wider mode is known to be const0_rtx and
14800 if so, record permanent equivalence for it to be ZERO_EXTEND of
14801 the narrower mode destination.
14803 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
14805 * lto-streamer-out.c (output_function): Revert 259346.
14806 * omp-expand.c (expand_oacc_for): Likewise.
14808 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
14810 PR rtl-optimization/85354
14811 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
14812 * sel-sched.c (sel_global_init): ... here.
14814 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
14817 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
14818 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
14819 mode for PE-COFF targets.
14820 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
14821 (i386_pe_asm_lto_end): Likewise.
14822 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
14823 (TARGET_ASM_LTO_END): Likewise.
14824 * config/i386/winnt.c (saved_debug_info_level): New static variable.
14825 (i386_pe_asm_lto_start): New function.
14826 (i386_pe_asm_lto_end): Likewise.
14828 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
14829 Richard Biener <rguenther@suse.de>
14831 PR middle-end/84955
14832 * lto-streamer-out.c (output_function): Fix CFG loop state before
14834 * omp-expand.c (expand_oacc_for): Handle calls to internal
14835 functions like regular functions.
14837 2018-04-12 Richard Biener <rguenther@suse.de>
14840 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
14841 for the early LTO debug to properly generate references to it
14842 during DIE emission. Do not re-use that for the skeleton for
14844 (dwarf2out_early_finish): Likewise.
14846 2018-04-12 Jakub Jelinek <jakub@redhat.com>
14849 * config/i386/sse.md
14850 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
14851 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
14852 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
14853 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
14854 and output is a reg, avoid creating invalid lowpart subreg, but
14855 instead split into a 512-bit move. Don't split if not AVX512VL,
14856 input is xmm16+ reg and output is a mem.
14857 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
14858 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
14859 xmm16+ reg and output is a mem.
14861 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14863 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
14864 also for flag_dwarf2_cfi_asm.
14866 2018-04-12 Jakub Jelinek <jakub@redhat.com>
14868 PR rtl-optimization/85342
14869 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
14870 a bool scalar var inside of the loop instead. Don't try to update
14871 recog_data.operand after failed apply_change_group.
14873 2018-04-12 Tom de Vries <tom@codesourcery.com>
14876 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
14877 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
14878 array with flexible array member as array without given dimension.
14879 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
14880 argument for undefined param to true.
14882 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
14885 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
14886 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
14887 from PowerPC section.
14888 * config/rs6000/sysv4.opt (mcall-): Improve help text.
14889 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
14890 help text that is too long.
14891 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
14892 help text that is too long.
14893 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
14894 help text that is too long.
14896 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
14898 * config/alpha/alpha.md (stack_probe_internal): Rename
14899 from "probe_stack". Update all callers.
14901 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
14903 PR rtl-optimization/84566
14904 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
14905 sched_macro_fuse_insns.
14907 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
14910 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
14911 (compute_block_dependences): ... from here.
14913 2018-04-11 Jakub Jelinek <jakub@redhat.com>
14915 PR tree-optimization/85331
14916 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
14917 from int to HOST_WIDE_INT.
14919 2018-04-11 Martin Jambor <mjambor@suse.cz>
14922 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
14923 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
14924 not the same as the source val.
14925 (cgraph_edge_brings_value_p): New parameter.
14926 (gather_edges_for_value): Pass destination value to
14927 cgraph_edge_brings_value_p.
14928 (perhaps_add_new_callers): Likewise.
14929 (get_info_about_necessary_edges): Likewise and exclude values brought
14930 only by self-recursive edges.
14931 (create_specialized_node): Redirect only clones of self-calling edges.
14932 (+self_recursive_pass_through_p): New function.
14933 (find_more_scalar_values_for_callers_subset): Use it.
14934 (find_aggregate_values_for_callers_subset): Likewise.
14935 (known_aggs_to_agg_replacement_list): Removed.
14936 (decide_whether_version_node): Re-calculate known constants for all
14937 remaining context clones.
14939 2018-04-11 Richard Biener <rguenther@suse.de>
14942 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
14943 from early DWARF output.
14944 (dwarf2out_early_finish): Output line info unconditionally into
14945 early DWARF and add reference to it.
14947 2018-04-11 Jakub Jelinek <jakub@redhat.com>
14950 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
14951 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
14952 other than V2DFmode using iptr mode attribute.
14953 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
14955 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
14957 PR rtl-optimization/84659
14958 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
14960 2018-04-11 Jakub Jelinek <jakub@redhat.com>
14963 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
14965 (output_loc_list): Pass address of a dummy size variable even in the
14966 locview handling loop.
14967 (index_location_lists): Add comment on why skip_loc_list_entry can't
14970 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
14973 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
14976 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
14979 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
14980 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
14981 and -mstring-compare-inline-limit.
14983 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
14986 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
14987 for stack clash protection in a register whenever we need it to be in
14990 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
14992 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
14993 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
14995 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
14998 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
15000 (mlong-double-): Ditto.
15001 * config/rs6000/sysv4.opt (msdata=): Ditto.
15002 (mtls-size=): Ditto.
15004 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
15006 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
15007 erroneous entries for
15008 "vector int vec_ldl (int, long int *)", and
15009 "vector unsigned int vec_ldl (int, unsigned long int *)".
15010 Add comments and entries for
15011 "vector bool char vec_ldl (int, bool char *)",
15012 "vector bool short vec_ldl (int, bool short *)",
15013 "vector bool int vec_ldl (int, bool int *)",
15014 "vector bool long long vec_ldl (int, bool long long *)",
15015 "vector pixel vec_ldl (int, pixel *)",
15016 "vector long long vec_ldl (int, long long *)",
15017 "vector unsigned long long vec_ldl (int, unsigned long long *)".
15018 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
15019 type tree bool_long_long_type_node and correct definition of
15020 bool_V2DI_type_node to make reference to this new type tree.
15021 (rs6000_mangle_type): Replace erroneous reference to
15022 bool_long_type_node with bool_long_long_type_node.
15023 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
15024 comments to emphasize sign distinctions for char and int types and
15025 replace RS6000_BTI_bool_long constant with
15026 RS6000_BTI_bool_long_long constant. Also add comment to restrict
15027 use of RS6000_BTI_pixel.
15028 (bool_long_type_node): Remove this macro definition.
15029 (bool_long_long_type_node): New macro definition
15031 2018-04-10 Jakub Jelinek <jakub@redhat.com>
15033 PR rtl-optimization/85300
15034 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
15035 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
15036 simplify_unary_operation fails.
15038 2018-04-10 Martin Liska <mliska@suse.cz>
15040 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
15041 cgraph_edge and ipa_ref.
15043 2018-04-10 Jakub Jelinek <jakub@redhat.com>
15047 * config/i386/sse.md
15048 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
15049 computation of the VEC_MERGE selector from mask.
15050 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
15051 Fix decoding of the VEC_MERGE selector into mask.
15053 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
15055 PR tree-optimization/85286
15056 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
15058 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
15060 * final.c (final_1): Set insn_last_address as well as
15061 insn_current_address.
15063 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15066 * explow.c (emit_stack_probe): Call validize_mem on memory location
15067 before passing it to gen_probe_stack. Create address operand and
15068 legitimize it for the probe_stack_address case.
15070 2018-04-09 Jan Hubicka <jh@suse.cz>
15073 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
15074 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
15075 * tree.c (free_lang_data_in_type): Fix handling of binfos;
15077 (free_lang_data): Rebuild type inheritance graph.
15079 2018-04-09 Martin Sebor <msebor@redhat.com>
15081 * invoke.texi (-finline-small-functions): Mention other optimization
15083 (-findirect-inlining, -fpartial-inlining): Same.
15084 (-finline-functions-called-once): Same.
15085 (-freorder-blocks-and-partition): Same.
15087 2018-04-09 Jan Hubicka <jh@suse.cz>
15090 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
15091 jumps; choose last target that matches the criteria (i.e.
15092 no partition changes for non-crossing jumps).
15093 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
15094 support for redirecting crossing jumps to non-crossing.
15096 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
15098 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
15099 also for naked functions.
15101 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
15103 * config/arc/arc.md (add_shift): New pattern.
15104 (add_shift2): Likewise.
15105 (sub_shift): Likewise.
15106 (sub_shift_cmp0_noout): Likewise.
15107 (compare_si_ashiftsi): Likewise.
15108 (xbfu_cmp0_noout): New combine pattern.
15109 (xbfu_cmp0"): Likewise.
15110 (movsi_set_cc_insn): Place the predicable variant first.
15111 (commutative_binary_cmp0_noout): Remove clobber.
15112 (commutative_binary_cmp0): New pattern.
15113 (noncommutative_binary_cmp0): Likewise.
15114 (noncommutative_binary_cmp0_noout): Likewise.
15115 (noncommutative_binary_comparison_result_used): Removed.
15116 (rsub_cmp0): New pattern.
15117 (rsub_cmp0_noout): Likewise.
15118 (extzvsi): Changed, keep only meaningful variants.
15119 (SQH, SEZ): New iterators.
15120 (SQH_postfix): New mode attribute.
15121 (SEZ_prefix): New code attribute.
15122 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
15123 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
15124 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
15125 of numerical value.
15126 (noncommutative_operator): Check the availability of barrel
15129 2018-04-09 Richard Biener <rguenther@suse.de>
15131 PR tree-optimization/85284
15132 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
15133 Only use the niter constraining form of simple_iv when the exit
15134 is always executed.
15136 2018-04-09 Tom de Vries <tom@codesourcery.com>
15139 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
15140 (define_expand "*memory_barrier"): New define_expand.
15141 (define_insn "memory_barrier"): New insn.
15143 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15145 PR rtl-optimization/80463
15146 PR rtl-optimization/83972
15147 PR rtl-optimization/83480
15149 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
15150 correct producer for the insn.
15151 (tidy_control_flow): Fixup seqnos in case of debug insns.
15153 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15155 PR rtl-optimization/83913
15157 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
15158 different sched-times when merging exprs.
15160 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15162 PR rtl-optimization/83962
15164 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
15165 tidy_fallthru_edge and tidy_control_flow.
15167 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15169 PR rtl-optimization/83530
15171 * sel-sched.c (force_next_insn): New global variable.
15172 (remove_insn_for_debug): When force_next_insn is true, also leave only
15173 next insn in the ready list.
15174 (sel_sched_region): When the region wasn't scheduled, make another pass
15175 over it with force_next_insn set to 1.
15177 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
15179 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
15181 * config/nds32/constants.md (unspec_volatile_element): Add enum values
15182 for interrupt control.
15183 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
15184 functions for interrupt control.
15185 * config/nds32/nds32-intrinsic.md: Likewise.
15186 * config/nds32/nds32_intrinsic.h: Likewise.
15187 * config/nds32/nds32.h (nds32_builtins): Likewise.
15189 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
15191 * config/nds32/nds32.c (nds32_init_machine_status,
15192 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
15193 strict_aligned_p field.
15194 (nds32_expand_to_rtl_hook): New function.
15195 (TARGET_EXPAND_TO_RTL_HOOK): Define.
15196 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
15198 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15199 Chung-Ju Wu <jasonwucj@gmail.com>
15201 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
15202 * config/nds32/nds32-n7.md: New file.
15203 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
15204 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
15206 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
15207 * config/nds32/nds32.md (pipeline_model): Add n7.
15208 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
15209 * config/nds32/pipelines.md: Include n7 settings.
15211 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15212 Chung-Ju Wu <jasonwucj@gmail.com>
15214 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
15215 * config/nds32/nds32-e8.md: New file.
15216 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
15217 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
15219 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
15220 * config/nds32/nds32.md (pipeline_model): Add e8.
15221 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
15222 * config/nds32/pipelines.md: Include e8 settings.
15224 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15225 Chung-Ju Wu <jasonwucj@gmail.com>
15227 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
15228 * config/nds32/nds32-n8.md: New file.
15229 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
15230 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
15232 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
15233 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
15234 * config/nds32/nds32.md (pipeline_model): Add n8.
15235 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
15236 * config/nds32/pipelines.md: Include n8 settings.
15238 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15239 Chung-Ju Wu <jasonwucj@gmail.com>
15241 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
15242 * config/nds32/nds32-n9-2r1w.md: New file.
15243 * config/nds32/nds32-n9-3r2w.md: New file.
15244 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
15245 nds32_register_ports): New or modify for cpu n9.
15246 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
15248 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
15249 * config/nds32/nds32-utils.c: New file.
15250 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
15251 TARGET_MUL_SLOW): Define.
15252 * config/nds32/nds32.md (pipeline_model): New attribute.
15253 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
15254 New options that support cpu n9.
15255 * config/nds32/pipelines.md: Include n9 settings.
15256 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
15258 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
15260 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
15261 information if necessary.
15262 (output_cond_branch_compare_zero): Likewise.
15263 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
15264 (nds32_target_alignment): Refine for alignment.
15265 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
15266 (FUNCTION_BOUNDARY): Modify.
15267 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
15269 * config/nds32/nds32.opt (malways-align, malign-functions): New.
15271 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
15273 * config/nds32/constants.md (unspec_volatile_element): Add values for
15274 TLB operation and data prefetch.
15275 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
15276 functions for TLB operation and data prefetch.
15277 * config/nds32/nds32-intrinsic.md: Likewise.
15278 * config/nds32/nds32_intrinsic.h: Likewise.
15279 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
15280 (nds32_print_operand): Likewise.
15281 * config/nds32/nds32.h (nds32_builtins): Likewise.
15283 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15284 Andrew Pinski <pinsika@gcc.gnu.org>
15286 PR middle-end/82976
15287 * match.pd: Use constant_boolean_node of correct type instead of
15288 boolean_true_node or boolean_false_node for simplifying
15289 pointer comparisons to zero.
15291 2018-04-07 Jakub Jelinek <jakub@redhat.com>
15293 PR tree-optimization/80021
15294 * tree.c (verify_type_variant): Make error call in verify_variant_match
15295 translatable and remove final full stop.
15297 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15299 * config/nds32/constants.md (unspec_volatile_element): Add
15300 UNSPEC_VOLATILE_EH_RETURN.
15301 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
15302 nds32_output_stack_pop): Support dwarf exception handling process.
15303 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
15304 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
15305 exception handling process.
15306 (nds32_compute_stack_frame): Likewise.
15307 (nds32_return_addr_rtx): Likewise.
15308 (nds32_initial_elimination_offset): Likewise.
15309 (nds32_expand_prologue): Likewise.
15310 (nds32_expand_epilogue): Likewise.
15311 (nds32_dynamic_chain_address): New function.
15312 * config/nds32/nds32.h (machine_function): Add fields for dwarf
15313 exception handling.
15314 (DYNAMIC_CHAIN_ADDRESS): Define.
15315 (EH_RETURN_DATA_REGNO): Define.
15316 (EH_RETURN_STACKADJ_RTX): Define.
15317 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
15318 patterns for dwarf exception handling.
15320 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15322 * config/nds32/nds32.h: Clean up obsolete macros.
15324 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15326 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
15327 Add enum values for particular instructions.
15328 * config/nds32/nds32-intrinsic.c: Implementation of expanding
15329 particular intrinsic functions.
15330 * config/nds32/nds32-intrinsic.md: Likewise.
15331 * config/nds32/nds32_intrinsic.h: Likewise.
15332 * config/nds32/nds32.h (nds32_builtins): Likewise.
15333 * config/nds32/nds32.md (type): Add pbsad and pbsada.
15334 (btst, ave): New patterns for particular instructions.
15336 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15338 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
15339 Add enum values for atomic load/store and memory sync.
15340 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
15342 * config/nds32/nds32-intrinsic.md: Likewise.
15343 * config/nds32/nds32_intrinsic.h: Likewise.
15344 * config/nds32/nds32.h (nds32_builtins): Likewise.
15346 2018-04-07 Jakub Jelinek <jakub@redhat.com>
15348 PR tree-optimization/85257
15349 * fold-const.c (native_encode_vector): If not all elts could fit
15350 and off is -1, return 0 rather than offset.
15351 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
15352 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
15353 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
15354 adjust buffer in native_interpret_expr call.
15356 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15358 * config/nds32/constants.md (unspec_volatile_element): Add cache
15359 control enum values.
15360 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
15361 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
15362 * config/nds32/nds32.c (nds32_cctl_names): New.
15363 (nds32_print_operand): Handle cache control register names.
15364 * config/nds32/nds32.h (nds32_builtins): New enum values.
15365 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
15367 * config/nds32/nds32.md (type): Add mmu.
15368 * config/nds32/pipelines.md (simple_insn): Add mmu.
15370 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15372 * config/nds32/nds32.md (type): Remove call.
15373 * config/nds32/pipelines.md (simple_insn): Likewise.
15375 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15377 * config/nds32/constants.md (unspec_volatile_element): Add
15378 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
15379 UNSPEC_VOLATILE_FMFCFG.
15380 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
15381 description for fmfcfg and fmfcsr.
15382 (bdesc_1arg): Add fmtcsr.
15383 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
15384 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
15385 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
15386 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
15387 unspec_fmfcfg): New patterns.
15388 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
15389 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
15390 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
15391 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
15392 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
15393 __nds32__fmfcfg): Define.
15395 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15397 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
15398 intrinsic register names.
15399 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
15400 intrinsic register enum values and macros.
15402 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15404 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
15405 for load/store addressing form.
15406 (nds32_print_operand_address): Likewise.
15408 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
15411 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
15412 based on LABEL_REF. Remove useless assertion.
15413 (pic_address_needs_scratch): Fix formatting.
15414 (sparc_legitimize_pic_address): Minor tweaks.
15415 (sparc_delegitimize_address): Adjust assertion accordingly.
15416 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
15417 into symbolic_operand.
15418 (movsi_high_pic_label_ref): Likewise.
15419 (movsi_lo_sum_pic_label_ref): Likewise.
15420 (movdi_pic_label_ref): Likewise.
15421 (movdi_high_pic_label_ref): Likewise.
15422 (movdi_lo_sum_pic_label_ref): Likewise.
15424 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
15426 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
15427 custom LIB_SPEC setup.
15429 2018-04-06 Ruslan Bukin <br@bsdpad.com>
15430 Kito Cheng <kito.cheng@gmail.com>
15432 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
15433 * config/riscv/freebsd.h: New.
15435 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
15437 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
15438 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
15441 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
15442 Kito Cheng <kito.cheng@gmail.com>
15444 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
15445 nds32_output_call, nds32_symbol_binds_local_p): New functions.
15446 * config/nds32/nds32-protos.h (nds32_output_call,
15447 nds32_output_return): Declare.
15448 * config/nds32/nds32.md: Refine all the call and return patterns.
15450 2018-04-06 Jakub Jelinek <jakub@redhat.com>
15453 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
15454 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
15456 PR rtl-optimization/84872
15457 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
15458 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
15459 EDGE_CROSSING edge.
15461 2018-04-06 Tamar Christina <tamar.christina@arm.com>
15463 * expr.c (copy_blkmode_to_reg): Revert 254862.
15464 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
15466 2018-04-06 Richard Biener <rguenther@suse.de>
15468 PR middle-end/85244
15469 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
15470 after seeing a component reference with an adjacent field. Treat
15471 refs to arrays at struct end of external decls similar to
15472 refs to unconstrained commons.
15474 2018-04-06 Jakub Jelinek <jakub@redhat.com>
15477 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
15478 look through SAVE_EXPRs with non-side-effects argument. Adjust
15480 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
15483 2018-04-06 Richard Biener <rguenther@suse.de>
15485 PR middle-end/85180
15486 * alias.c (find_base_term): New wrapper around find_base_term
15487 unwinding CSELIB_VAL_PTR changes.
15488 (find_base_term): Do not restore CSELIB_VAL_PTR during the
15491 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15493 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
15495 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
15496 constant definitions.
15497 ("nop"): lr 0,0 -> nopr r0
15498 ("nop_lr0", "nop_lr1"): New insn definitions.
15500 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
15502 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
15503 NDS32_V3PUSH_AVAILABLE_P macro.
15505 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
15506 Chung-Ju Wu <jasonwucj@gmail.com>
15508 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
15509 (nds32*-*-*): Add float and fpu_config into supported_defaults.
15510 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
15511 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
15512 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
15513 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
15514 * config/nds32/constraints.md: New constraints and checking for hard
15515 float configuration.
15516 * config/nds32/iterators.md: New mode iterator and attribute for hard
15517 float configuration.
15518 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
15520 * config/nds32/nds32-fpu.md: New file.
15521 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
15522 deal with hard float code generation.
15523 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
15525 (abi_type, float_reg_number): New enum type.
15526 * config/nds32/nds32-predicates.c: New predicates for hard float.
15527 * config/nds32/nds32-protos.h: Declare functions for hard float.
15528 * config/nds32/nds32.c: Implementation for hard float configuration.
15529 * config/nds32/nds32.h: Definitions for hard float configuration.
15530 * config/nds32/nds32.md: Include hard float machine description and
15531 modify patterns for hard float configuration.
15532 * config/nds32/nds32.opt: New options for hard float configuration.
15533 * config/nds32/predicates.md: New predicates for hard float
15536 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
15538 * common/config/nds32/nds32-common.c
15539 (nds32_option_optimization_table): Enable -mreleax-hint by default.
15541 2018-04-05 Jakub Jelinek <jakub@redhat.com>
15543 PR middle-end/85195
15544 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
15545 CONSTRUCTOR_ELT (ctor, ...)->value.
15547 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
15550 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
15552 2018-04-05 Tom de Vries <tom@codesourcery.com>
15555 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
15558 2018-04-05 Shiva Chen <shiva0217@gmail.com>
15559 Kito Cheng <kito.cheng@gmail.com>
15561 * config/nds32/constraints.md (U33): Fine-tune checking condition.
15562 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
15563 * config/nds32/nds32.h (nds32_16bit_address_type): Add
15564 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
15566 2018-04-05 Shiva Chen <shiva0217@gmail.com>
15567 Kito Cheng <kito.cheng@gmail.com>
15569 * config/nds32/constraints.md (Ufe): New memory constraint.
15570 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
15571 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
15572 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
15574 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
15575 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
15577 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15579 * config/nds32/nds32.md: Use optimize_size in the condition for
15580 alu-shift instructions.
15582 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15584 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
15586 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15588 * config/nds32/nds32.md (negsi2): Refine pattern.
15590 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
15591 Chung-Ju Wu <jasonwucj@gmail.com>
15593 * config/nds32/iterators.md (shift_rotate): New code iterator.
15594 (shift): New code attribute.
15595 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
15596 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
15597 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
15598 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
15599 bit-wise operations.
15600 (andsi3, *andsi3): Ditto.
15601 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
15602 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
15603 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
15604 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
15605 nds32_ior_operand, nds32_xor_operand): New predicates.
15607 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15609 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
15610 (addsi3, subsi3): ... this.
15612 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15614 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
15616 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15618 * config/nds32/nds32.md: Adjust indention.
15620 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
15622 * config/nds32/nds32.md (feature): New attribute.
15624 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15626 * config/nds32/nds32.md (subtype): New attribute.
15628 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15631 * config/arm/arm-builtins.c (arm_expand_builtin): Change
15632 expansion to perform a bitwise AND of the argument followed by a
15633 boolean negation of the result.
15635 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
15637 PR rtl-optimization/84878
15638 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
15639 the basic block. Assert the use reference is not artificial and that
15640 it has an associated insn.
15642 2018-04-04 Michael Matz <matz@suse.de>
15644 * builtins.c (compute_objsize): Pass correct operand
15645 to array_at_struct_end_p.
15647 2018-04-04 Richard Biener <rguenther@suse.de>
15650 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
15651 from contexts for DINFO_LEVEL_TERSE and below.
15653 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
15655 * config/nds32/nds32-doubleword.md (move_<mode>): Require
15656 resiter_operand condition.
15657 * config/nds32/nds32.md (*move<mode>): Ditto.
15659 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
15660 Monk Chiang <sh.chiang04@gmail.com>
15662 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
15664 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15666 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
15668 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15669 Kito Cheng <kito.cheng@gmail.com>
15671 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
15672 nds32_cond_code_str, output_cond_branch,
15673 output_cond_branch_compare_zero, nds32_expand_cbranch,
15674 nds32_expand_cstore, nds32_expand_movcc,
15675 nds32_output_cbranchsi4_equality_zero,
15676 nds32_output_cbranchsi4_equality_reg,
15677 nds32_output_cbranchsi4_equality_reg_or_const_int,
15678 nds32_output_cbranchsi4_greater_less_zero: New functions.
15679 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
15680 nds32_expand_cstore, nds32_expand_movcc,
15681 nds32_output_cbranchsi4_equality_zero,
15682 nds32_output_cbranchsi4_equality_reg,
15683 nds32_output_cbranchsi4_equality_reg_or_const_int,
15684 nds32_output_cbranchsi4_greater_less_zero): Declare.
15685 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
15686 nds32_rimm11s_operand): New predicates.
15687 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
15688 * config/nds32/nds32.md: Rewrite all the branch and conditional move
15691 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
15693 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
15694 * config/nds32/nds32.md: Ditto.
15695 * config/nds32/pipelines.md: Ditto.
15697 2018-04-04 Richard Biener <rguenther@suse.de>
15699 PR tree-optimization/85168
15700 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
15701 propagating abnormals.
15703 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15705 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
15707 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15708 Kito Cheng <kito.cheng@gmail.com>
15710 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
15711 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
15712 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
15713 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
15714 * config/nds32/nds32.md (sibcall_internal): New.
15715 (sibcall_register): Remove.
15716 (sibcall_immediate): Remove.
15717 (sibcall_value_internal): New.
15718 (sibcall_value_register): Remove.
15719 (sibcall_value_immediate): Remove.
15720 * config/nds32/predicates.md (nds32_general_register_operand): New.
15721 (nds32_call_address_operand): New.
15723 2018-04-03 Jakub Jelinek <jakub@redhat.com>
15725 PR rtl-optimization/85167
15726 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
15727 bb_defs if *split_p, instead preinitialize it to NULL.
15729 PR tree-optimization/85156
15730 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
15731 evaluating the argument multiple times.
15733 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
15735 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
15737 (_mm_cvtpd_ps): Likewise.
15738 (_mm_cvttpd_epi32): Likewise.
15739 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
15740 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
15741 vector, pixel, and bool following altivec.h include.
15743 2018-04-03 Martin Sebor <msebor@redhat.com>
15745 * doc/extend.texi (Common Function Attributes): Clarify.
15746 (const attribute): Likewise.
15747 (pure attribute): Likewise.
15749 2018-04-03 Jakub Jelinek <jakub@redhat.com>
15752 * config/i386/i386.c (ix86_expand_vector_set): Use
15753 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
15755 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
15757 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
15758 instructions when changing rounding bits to preserve precision bits
15759 in the x87 control word.
15761 2018-04-03 Martin Liska <mliska@suse.cz>
15763 PR tree-optimization/82491
15764 * rtl.h (strip_offset_and_add): Replace += suboffset with
15765 poly_uint64 () + suboffset.
15767 2018-03-29 Martin Liska <mliska@suse.cz>
15768 Martin Jambor <mjambor@suse.cz>
15771 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
15772 param_type is not an integral or pointer type.
15774 2018-04-03 Richard Biener <rguenther@suse.de>
15776 * sese.h (recompute_all_dominators): Remove.
15778 2018-04-02 Martin Sebor <msebor@redhat.com>
15780 * doc/invoke.texi (-Wrestrict): Fix typos.
15782 2018-04-02 Jim Wilson <jimw@sifive.com>
15784 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
15785 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
15786 (<optab>di3, <optab>si3_extend): Likewise.
15787 (<optab>si3_mask, <optab>si3_mask_1): New.
15788 (<optab>di3_mask, <optab>di3_mask_1): New.
15789 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
15790 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
15791 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
15793 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
15795 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
15798 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
15800 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
15801 (nds32_canonicalize_comparison): New function.
15803 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
15804 Kito Cheng <kito.cheng@gmail.com>
15805 Kuan-Lin Chen <kuanlinchentw@gmail.com>
15807 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
15808 * config/nds32/constants.md (unspec_volatile_element): Add
15809 UNSPEC_VOLATILE_RELAX_GROUP.
15810 * config/nds32/nds32-relax-opt.c: New file.
15811 * config/nds32/nds32-predicates.c
15812 (nds32_symbol_load_store_p): New function.
15813 * config/nds32/nds32-protos.h
15814 (nds32_symbol_load_store_p): Declare function.
15815 (make_pass_nds32_relax_opt): Declare new rtl pass function.
15816 * config/nds32/nds32.c
15817 (nds32_register_pass): New function to register pass.
15818 (nds32_register_passes): New function to register passes.
15819 * config/nds32/nds32.md (relax_group): New pattern.
15820 * config/nds32/nds32.opt (mrelax-hint): New option.
15821 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
15823 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
15825 * config/nds32/t-nds32: Modify files dependency.
15827 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
15829 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
15830 (PROFILE_HOOK): Define its implementation.
15832 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
15834 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
15835 type and 32-bit size.
15837 2018-04-01 Jakub Jelinek <jakub@redhat.com>
15839 PR middle-end/85090
15840 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
15841 (V_128_256): New mode iterator.
15842 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
15843 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
15844 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
15846 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
15847 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
15849 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
15852 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
15853 NaN inputs correctly.
15855 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
15858 * config/rs6000/vsx.md (??r): New mode attribute.
15859 (*vsx_mov<mode>_64bit): Use it.
15860 (*vsx_mov<mode>_32bit): Likewise.
15862 2018-03-30 Martin Sebor <msebor@redhat.com>
15864 PR tree-optimization/84818
15865 * builtins.c (check_access): Use warning_n.
15867 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15870 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
15872 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
15875 2018-03-30 Julia Koval <julia.koval@intel.com>
15878 * x86-tune.def (movx, partial_reg_dependency): Enable for
15881 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
15883 PR inline-asm/84985
15884 * lra-constraints.c (process_alt_operands): Move setting
15885 this_alternative_matches below.
15887 2018-03-29 Martin Liska <mliska@suse.cz>
15890 * doc/invoke.texi: Document how LTO works with debug info.
15891 Describe auto-load support of binutils. Mention 'x86-64'
15892 as valid option value of -march option.
15894 2018-03-29 Jakub Jelinek <jakub@redhat.com>
15896 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
15899 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
15900 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
15901 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
15904 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
15907 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
15908 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
15909 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
15910 for RS6000_BTM_POWERPC64.
15911 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
15912 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
15913 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
15916 (DIVDEU): Likewise.
15918 2018-03-28 Carl Love <cel@us.ibm.com>
15921 2017-09-27 Carl Love <cel@us.ibm.com>
15923 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
15924 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
15925 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
15928 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15930 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
15931 instead of __vector bool.
15932 (_mm_max_pu8): Likewise.
15933 (_mm_min_pi16): Likewise.
15935 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
15938 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
15939 (DIVWEUO): Likewise.
15940 (DIVDEO): Likewise.
15941 (DIVDEUO): Likewise.
15942 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
15943 DIVWEUO and DIVDEUO.
15944 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
15945 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
15946 (div_extend): Likewise.
15947 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
15949 (__builtin_divweuo): Likewise.
15950 (__builtin_divdeo): Likewise.
15951 (__builtin_divdeuo): Likewise.
15953 2018-03-28 Jakub Jelinek <jakub@redhat.com>
15956 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
15957 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
15959 PR tree-optimization/82004
15960 * gimple-match-head.c (optimize_pow_to_exp): New function.
15961 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
15962 Don't fold to exp if optimize_pow_to_exp is false.
15964 2018-03-28 Martin Liska <mliska@suse.cz>
15967 * calls.c (initialize_argument_information): Fix trailing space.
15968 * common.opt: Fix typo and provide better explanation for
15969 -fsanitize-coverage option.
15970 * config/i386/i386.opt: Fix typo.
15972 2018-03-28 Jakub Jelinek <jakub@redhat.com>
15973 Martin Liska <mliska@suse.cz>
15976 * gimplify.c (asan_poison_variable): Don't do the check for
15977 gimplify_omp_ctxp here.
15978 (gimplify_decl_expr): Do it here.
15979 (gimplify_target_expr): Likewise.
15981 2018-03-28 Martin Liska <mliska@suse.cz>
15984 * config/i386/i386.c (ix86_function_arg_advance): Do not call
15985 chkp_type_bounds_count if MPX is not enabled.
15987 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
15989 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
15991 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
15994 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
15995 function to create the function decl for complex long double
15996 multiply and divide for -mabi=ieeelongdouble.
15997 (init_float128_ieee): Call it.
15999 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
16002 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
16003 -fcf-protection=branch -mibt.
16004 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
16006 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16009 * config/arm/arm.c (arm_valid_symbolic_address): Handle
16010 arm_word_relocations.
16012 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
16015 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
16016 extern array declarations.
16018 2018-03-27 Richard Biener <rguenther@suse.de>
16020 PR middle-end/84067
16021 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
16022 explicit single_use checks.
16024 2018-03-27 Richard Biener <rguenther@suse.de>
16026 PR tree-optimization/85082
16027 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
16030 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16032 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
16033 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
16034 Turn on fasynchronous-unwind-tables and funwind-tables.
16036 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
16039 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
16040 (*bmi_blsr_<mode>_ccz): Ditto.
16042 2018-03-26 Tom de Vries <tom@codesourcery.com>
16044 PR tree-optimization/85063
16045 * omp-general.c (offloading_function_p): New function. Factor out
16047 * omp-offload.c (pass_omp_target_link::gate): ... here.
16048 * omp-general.h (offloading_function_p): Declare.
16049 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
16050 with attribute omp declare target for offloading functions.
16052 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
16054 PR tree-optimization/84005
16055 * tree-data-ref.h (get_base_for_alignment): Declare.
16056 * tree-data-ref.c (get_base_for_alignment_1): New function.
16057 (get_base_for_alignment): Likewise.
16058 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
16059 get_base_for_alignment to find a suitable base object, instead
16060 of always using drb->base_address.
16062 2018-03-23 Jakub Jelinek <jakub@redhat.com>
16064 PR inline-asm/85022
16065 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
16066 known size by default.
16068 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
16070 PR inline-asm/85030
16071 * lra-constraints.c (process_alt_operands): Don't match BLKmode
16072 and non BLKmode operands.
16074 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16077 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
16078 Clean up attributes.
16080 2018-03-23 Richard Biener <rguenther@suse.de>
16083 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
16084 we are going to emit early debug for LTO.
16086 2018-03-23 Jakub Jelinek <jakub@redhat.com>
16088 PR inline-asm/85034
16089 * function.c (match_asm_constraints_1): Don't optimize if input
16090 doesn't satisfy general_operand predicate for output's mode.
16092 PR inline-asm/85022
16093 * alias.c (write_dependence_p): Don't require for x_canonicalized
16094 non-VOIDmode if x has VOIDmode.
16097 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
16098 just don't try to optimize it rather than assert it never happens.
16100 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
16102 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
16103 macro expansions for definition of ST_INTERNAL_<mode> and
16104 LD_INTERNAL_<mode> builtins.
16105 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
16107 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
16109 (altivec_expand_st_builtin): Likewise.
16110 (altivec_expand_builtin): Remove calls to deleted functions.
16111 (rs6000_address_for_altivec): Delete this function.
16112 * config/rs6000/vector.md: Remove expands for
16113 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
16115 2018-03-22 Sudakshina Das <sudi.das@arm.com>
16118 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
16119 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
16120 re-computing once computed.
16121 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
16122 (arm_init_machine_status): Initialize
16123 machine->static_chain_stack_bytes.
16125 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
16128 * doc/extend.texi: Add four new prototypes for vec_ld.
16129 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
16130 definitions for more logical presentation.
16131 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
16132 entries for V1TI variants of __builtin_altivec_ld builtin.
16133 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
16134 handling of V1TI variant of LVX icode pattern.
16135 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
16136 (rs6000_gimple_fold_builtin): Likewise.
16137 (altivec_init_builtins): Add code to define
16138 __builtin_altivec_lvx_v1ti function.
16140 2018-03-22 Jakub Jelinek <jakub@redhat.com>
16142 PR inline-asm/84941
16143 * function.c (match_asm_constraints_1): Don't do the optimization
16144 if input isn't a REG, SUBREG, MEM or constant.
16146 2018-03-22 Tom de Vries <tom@codesourcery.com>
16148 PR tree-optimization/84956
16149 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
16150 bb_has_abnormal_pred.
16152 2018-03-22 Jakub Jelinek <jakub@redhat.com>
16155 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
16156 DECL_INITIAL (decl) to decl at the end.
16157 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
16158 adjust the comment.
16160 2018-03-21 Joseph Myers <joseph@codesourcery.com>
16162 * doc/extend.texi (__builtin_tgmath): Document when complex
16163 integer types are treated as _Complex _Float64.
16165 2018-03-21 Tom de Vries <tom@codesourcery.com>
16167 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
16169 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16171 PR tree-optimization/84960
16172 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
16173 if it is ENTRY block, move them into single succ of ENTRY in that case.
16175 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
16177 PR tree-optimization/84811
16178 * poly-int.h (poly_span_traits): Remove the T3 parameter and
16179 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
16180 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
16181 (known_subrange_p): Update accordingly. Cast each value involved
16182 in the size comparison, rather than casting the result of the
16185 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16187 PR tree-optimization/84982
16188 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
16189 by flipping the least significant bit rather than all bits from
16190 bitpos to bitpos + bitsize - 1.
16192 2018-03-21 Nathan Sidwell <nathan@acm.org>
16194 * doc/extend.texi (Deprecated Features): Remove mention of
16195 long-deleted deprecations.
16197 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16200 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
16201 * configure: Regenerate.
16203 2018-03-21 Tom de Vries <tom@codesourcery.com>
16205 PR tree-optimization/83126
16206 * tree-parloops.c (num_phis): New function.
16207 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
16209 2018-03-21 Nathan Sidwell <nathan@acm.org>
16211 * doc/extend.texi (Deprecated Features): Update deprecated flags,
16212 mention anon-struct/union members and trailing attributes.
16214 2018-03-21 Bin Cheng <bin.cheng@arm.com>
16216 PR tree-optimization/84969
16217 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
16218 builtin memset partitions if they set different rhs values.
16220 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16222 PR rtl-optimization/84989
16223 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
16224 VEC_DUPLICATE with scalar result mode.
16226 2018-03-21 Martin Liska <mliska@suse.cz>
16229 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
16230 not intended return statement.
16232 2018-03-21 Martin Liska <mliska@suse.cz>
16235 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
16236 (chkp_find_bound_slots_1): Limit number of iterations.
16238 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
16241 * Minor grammar fixes for x86 options.
16243 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16246 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
16247 holding REG_CFA_RESTORE notes, instead turn them into a USE.
16249 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
16252 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
16253 (altivec_lvx_<mode>_1op): Likewise.
16254 (altivec_stvx_<mode>_2op): Likewise.
16255 (altivec_stvx_<mode>_1op): Likewise.
16256 (altivec_lvx_<VM2:mode>): New define_expand.
16257 (altivec_stvx_<VM2:mode>): Likewise.
16258 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
16259 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
16260 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
16261 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
16262 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
16263 (rs6000_gen_lvx): Likewise.
16264 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
16265 (altivec_expand_stv_builtin): Likewise.
16266 (altivec_expand_builtin): Likewise.
16267 * config/rs6000/vector.md: Likewise.
16269 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16272 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
16275 2018-03-20 Richard Biener <rguenther@suse.de>
16278 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
16279 sign-conversions as zero, fall back to standard scalar_stmt
16282 2018-03-20 Martin Liska <mliska@suse.cz>
16285 * predict.c (rebuild_frequencies): Handle case when we have
16286 PROFILE_ABSENT, but flag_guess_branch_prob is false.
16288 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16291 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
16292 flag_section_anchors.
16293 * varasm.c (use_blocks_for_decl_p): Remove hack for
16294 dw2_force_const_mem.
16297 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
16299 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
16300 be created, use lowpart_subreg of operands[0] rather than operands[0]
16302 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
16303 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
16304 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
16305 and n constraint instead of aarch64_shift_imm_di and Usd.
16306 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
16307 (*aarch64_<optab>_reg_minus<mode>3): ... this.
16309 2018-03-20 Sudakshina Das <sudi.das@arm.com>
16312 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
16313 to favor GPR over NEON registers.
16314 (<shift>di3_neon): Likewise.
16316 2018-03-20 Tom de Vries <tom@codesourcery.com>
16319 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
16320 (nvptx_process_pars): Emit bar.sync asap and alap.
16322 2018-03-20 Tom de Vries <tom@codesourcery.com>
16325 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
16326 seen_label if seen_label is already set.
16328 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16331 * config/i386/i386.c (fold_builtin_cpu): For features above 31
16332 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
16333 Use 1U instead of 1. Formatting fixes.
16336 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
16337 instead of TREE_TYPE (s1) for the return value.
16339 2018-03-19 Jakub Jelinek <jakub@redhat.com>
16341 PR tree-optimization/84946
16342 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
16343 bitsize + bitsize in poly_uint64 rather than poly_int64.
16346 * dwarf2asm.c: Include fold-const.c.
16347 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
16348 of decl rather than decl itself.
16350 PR rtl-optimization/84643
16351 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
16353 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
16356 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
16357 calling assemble_variable.
16359 2018-03-19 Sudakshina Das <sudi.das@arm.com>
16362 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
16363 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
16365 2018-03-19 Jim Wilson <jimw@sifive.com>
16368 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
16369 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
16370 (riscv_first_stack_step): Likewise.
16371 (riscv_option_override): Use STACK_BOUNDARY instead of
16372 MIN_STACK_BOUNDARY.
16373 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
16374 MIN_STACK_BOUNDARY.
16375 (BIGGEST_ALIGNMENT): Set to 128.
16376 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
16377 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
16380 2018-03-19 Richard Biener <rguenther@suse.de>
16382 PR tree-optimization/84933
16383 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
16384 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
16386 2018-03-19 Richard Biener <rguenther@suse.de>
16388 PR tree-optimization/84859
16389 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
16390 (cond_if_else_store_replacement): Perform sinking operation on
16391 single-store BBs regardless of MAX_STORES_TO_SINK setting.
16392 Generalize what a BB with a single eligible store is.
16394 2018-03-19 Richard Biener <rguenther@suse.de>
16396 PR tree-optimization/84929
16397 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
16398 chrec_is_positive against non-chrec arg.
16400 2018-03-19 Tamar Christina <tamar.christina@arm.com>
16403 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
16405 2018-03-18 Martin Liska <mliska@suse.cz>
16407 PR rtl-optimization/84635
16408 * regrename.c (build_def_use): Use matches_mode only when
16411 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
16413 PR tree-optimization/84913
16414 * tree-vect-loop.c (vectorizable_reduction): Don't try to
16415 vectorize chains of COND_EXPRs.
16417 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
16419 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
16421 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
16423 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
16425 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
16427 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
16429 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
16430 Kito Cheng <kito.cheng@gmail.com>
16432 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
16433 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
16434 (nds32_adjust_reg_alloc_order): New function.
16435 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
16437 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
16439 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
16440 nds32_print_operand, nds32_print_operand_address): Use
16441 HOST_WIDE_INT_PRINT_DEC instead.
16443 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
16445 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
16447 2018-03-17 Jakub Jelinek <jakub@redhat.com>
16450 * config/i386/i386.c (initial_ix86_tune_features,
16451 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
16452 unsigned long long.
16453 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
16454 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
16455 rather than 1u << ix86_tune. Formatting fix.
16456 (ix86_option_override_internal): Change ix86_arch_mask from
16457 unsigned int to unsigned HOST_WIDE_INT, initialize to
16458 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
16459 (ix86_function_specific_restore): Likewise.
16461 2018-03-16 Jakub Jelinek <jakub@redhat.com>
16464 * postreload.c (reload_combine_recognize_pattern): Perform
16465 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
16466 truncate_int_for_mode the result for the destination's mode.
16469 * hsa-gen.c (mem_type_for_type): Fix comment typo.
16470 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
16472 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
16475 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
16478 * lra-assigns.c (lra_split_hard_reg_for): Don't use
16479 regno_allocno_class_array and sorted_pseudos.
16480 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
16481 insns where regno is used.
16483 2018-03-16 Martin Liska <mliska@suse.cz>
16486 * multiple_target.c (create_dispatcher_calls): Redirect
16487 reference in the symbol table.
16489 2018-03-16 Martin Liska <mliska@suse.cz>
16492 * multiple_target.c (create_dispatcher_calls): Redirect also
16495 2018-03-16 Jakub Jelinek <jakub@redhat.com>
16499 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
16500 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
16501 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
16503 2018-03-16 Julia Koval <julia.koval@intel.com>
16505 * doc/invoke.texi (Skylake Server): Add CLWB.
16506 Cannonlake): Remove CLWB.
16508 2018-03-16 Jakub Jelinek <jakub@redhat.com>
16510 PR tree-optimization/84841
16511 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
16513 (FLOAT_ONE_CONST_TYPE): Define.
16514 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
16515 (sort_by_operand_rank): Put entries with higher constant_type last
16516 rather than first to match comments.
16518 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
16520 * config/nios2/nios2.md (movsi_internal): Fix thinko in
16523 2018-03-15 Jakub Jelinek <jakub@redhat.com>
16526 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
16527 check and use address of target always.
16529 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
16532 * config/i386/i386.c (indirect_thunk_needed): Update comments.
16533 (indirect_thunk_bnd_needed): Likewise.
16534 (indirect_thunks_used): Likewise.
16535 (indirect_thunks_bnd_used): Likewise.
16536 (indirect_return_needed): New.
16537 (indirect_return_bnd_needed): Likewise.
16538 (output_indirect_thunk_function): Add a bool argument for
16540 (output_indirect_thunk_function): Don't generate alias for
16541 function return thunk.
16542 (ix86_code_end): Call output_indirect_thunk_function to generate
16543 function return thunks.
16544 (ix86_output_function_return): Set indirect_return_bnd_needed
16545 and indirect_return_needed instead of indirect_thunk_bnd_needed
16546 and indirect_thunk_needed.
16548 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
16550 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
16551 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
16552 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
16554 2018-03-15 David Malcolm <dmalcolm@redhat.com>
16555 Paul Hua <paul.hua.gm@gmail.com>
16558 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
16560 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
16562 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
16563 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
16564 resp. SFmode cases.
16566 2018-03-15 Tamar Christina <tamar.christina@arm.com>
16569 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
16570 instead of GET_MODE_SIZE when comparing Units.
16572 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
16575 * varasm.c (hash_section): Return an unchangeble hash value
16576 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
16577 Return !aarch64_can_use_per_function_literal_pools_p ().
16579 2018-03-15 Jakub Jelinek <jakub@redhat.com>
16582 * optabs.c (emit_conditional_move): Pass address of cmode's copy
16583 rather than address of cmode as last argument to prepare_cmp_insn.
16585 2018-03-15 Julia Koval <julia.koval@intel.com>
16587 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
16588 F_AVX512VNNI, F_AVX512BITALG): New.
16590 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
16593 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
16594 insn for floating-point loads and stores.
16596 2018-03-14 Carl Love <cel@us.ibm.com>
16598 * config/rs6000/rs6000-c.c: Add macro definitions for
16599 ALTIVEC_BUILTIN_VEC_PERMXOR.
16600 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
16601 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
16602 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
16603 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
16605 * config/doc/extend.texi: Add prototypes for vec_permxor.
16607 2018-03-14 David Malcolm <dmalcolm@redhat.com>
16610 * diagnostic-show-locus.c (class layout_point): Convert m_line
16611 from int to linenum_type.
16612 (line_span::comparator): Use linenum "compare" function when
16613 comparing line numbers.
16614 (test_line_span): New function.
16615 (layout_range::contains_point): Convert param "row" from int to
16617 (layout_range::intersects_line_p): Likewise.
16618 (layout::will_show_line_p): Likewise.
16619 (layout::print_source_line): Likewise.
16620 (layout::should_print_annotation_line_p): Likewise.
16621 (layout::print_annotation_line): Likewise.
16622 (layout::print_leading_fixits): Likewise.
16623 (layout::annotation_line_showed_range_p): Likewise.
16624 (struct line_corrections): Likewise for field m_row.
16625 (line_corrections::line_corrections): Likewise for param "row".
16626 (layout::print_trailing_fixits): Likewise.
16627 (layout::get_state_at_point): Likewise.
16628 (layout::get_x_bound_for_row): Likewise.
16629 (layout::print_line): Likewise.
16630 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
16631 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
16632 * input.c (selftest::test_linenum_comparisons): New function.
16633 (selftest::input_c_tests): Call it.
16634 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
16635 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
16636 * selftest.h (ASSERT_GT): New macro.
16637 (ASSERT_GT_AT): New macro.
16638 (ASSERT_LT): New macro.
16639 (ASSERT_LT_AT): New macro.
16641 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
16643 PR rtl-optimization/84780
16644 * combine.c (distribute_links): Don't make a link based on pc_rtx.
16646 2018-03-14 Martin Liska <mliska@suse.cz>
16648 * tree.c (record_node_allocation_statistics): Use
16649 get_stats_node_kind.
16650 (get_stats_node_kind): New function extracted from
16651 record_node_allocation_statistics.
16652 (free_node): Use get_stats_node_kind.
16654 2018-03-14 Richard Biener <rguenther@suse.de>
16656 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
16657 that the value-set of ANTIC_IN doesn't grow.
16660 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
16662 (BB_VISITED_WITH_VISITED_SUCCS): New define.
16663 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
16665 2018-03-14 Julia Koval <julia.koval@intel.com>
16667 * config.gcc (icelake-client, icelake-server): New.
16669 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
16670 (initial_ix86_arch_features): Ditto.
16671 (PTA_SKYLAKE): Add SGX.
16672 (PTA_ICELAKE): Remove.
16673 (PTA_ICELAKE_CLIENT): New.
16674 (PTA_ICELAKE_SERVER): New.
16675 (ix86_option_override_internal): Split up icelake on icelake client and
16677 (get_builtin_code_for_version): Ditto.
16678 (fold_builtin_cpu): Ditto.
16679 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
16680 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
16681 * config/i386/i386.h (processor_type): Ditto.
16682 * doc/invoke.texi: Ditto.
16684 2018-03-14 Jakub Jelinek <jakub@redhat.com>
16687 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
16688 INTEGER_CST offset, add it together with bitpos / 8 and
16689 sign extend based on POINTER_SIZE.
16693 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
16696 * config/i386/constraints.md (Yc): New register constraint.
16697 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
16698 Use Yc constraint for alternative 2 of operand 0. Remove
16699 preferred_for_speed attribute.
16701 2018-03-14 Richard Biener <rguenther@suse.de>
16703 PR tree-optimization/84830
16704 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
16705 with the old one to avoid oscillations.
16707 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
16710 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
16712 (assign_by_spills): Return a flag of reload assignment failure.
16713 Do not process the reload assignment failures. Do not spill other
16714 reload pseudos if they has the same reg class. Update n if
16716 (lra_assign): Add a return arg. Set up from the result of
16717 assign_by_spills call.
16718 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
16719 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
16720 usage_insns if it is not NULL.
16721 (spill_hard_reg_in_range): New function.
16722 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
16723 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
16724 function prototypes.
16725 (lra_assign): Change prototype.
16726 * lra.c (lra): Add code to deal with fails by splitting hard reg
16729 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
16731 * config/riscv/riscv.opt (mrelax): New option.
16732 * config/riscv/riscv.c (riscv_file_start): Emit ".option
16733 "norelax" when riscv_mrelax is disabled.
16734 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
16736 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16739 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
16740 reassociation for int modes.
16742 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
16744 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
16745 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
16747 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
16748 * config/aarch64/aarch64-sve.md
16749 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
16750 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
16751 (*extend<mode><Vwide>2): Rename to...
16752 (aarch64_sve_extend<mode><Vwide>2): ...this.
16753 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
16754 renaming the old pattern to...
16755 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
16757 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
16758 define_expand, renaming the old pattern to...
16759 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
16760 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
16761 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
16762 account when deciding which SVE instruction the optab should use.
16763 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
16765 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
16767 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
16768 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
16769 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
16770 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
16771 (tlsdesc_small_<mode>): Turn a define_expand and use
16772 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
16773 (tlsdesc_small_advsimd_<mode>): ...this.
16774 (tlsdesc_small_sve_<mode>): New pattern.
16776 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
16778 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
16779 (UNSPEC_UMUL_HIGHPART): New constants.
16780 (MUL_HIGHPART): New int iteraor.
16781 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
16782 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
16784 (*<su>mul<mode>3_highpart): New define_insn.
16786 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
16789 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
16792 2018-03-13 Martin Liska <mliska@suse.cz>
16795 * (sem_item_optimizer::sem_item_optimizer): Initialize new
16797 (sem_item_optimizer::~sem_item_optimizer): Release it.
16798 (sem_item_optimizer::merge_classes): Register variable aliases.
16799 (sem_item_optimizer::fixup_pt_set): New function.
16800 (sem_item_optimizer::fixup_points_to_sets): Likewise.
16801 * ipa-icf.h: Declare new variables and functions.
16803 2018-03-13 Jakub Jelinek <jakub@redhat.com>
16805 PR middle-end/84834
16806 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
16807 integer_pow2p@2 and test integer_pow2p in condition.
16808 (A < 0 ? C : 0): Similarly for @1.
16810 PR middle-end/84831
16811 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
16812 characters starting at p contain '\0' character, don't look beyond
16816 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
16817 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
16820 * reg-stack.c (change_stack): Change update_end var from int to
16821 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
16822 also call set_block_for_insn on the newly added insns and rescan.
16825 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
16826 on the last operand.
16829 * tree.c (stabilize_reference_1): Return save_expr (e) for
16830 STATEMENT_LIST even if it doesn't have side-effects.
16832 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
16834 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
16836 2018-03-12 Renlin Li <renlin.li@arm.com>
16838 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
16839 aarch64_output_scalar_simd_mov_immediate.
16841 2018-03-12 Martin Sebor <msebor@redhat.com>
16843 PR tree-optimization/83456
16844 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
16845 for perfectly overlapping calls to memcpy.
16846 (gimple_fold_builtin_memory_chk): Same.
16847 (gimple_fold_builtin_strcpy): Handle no-warning.
16848 (gimple_fold_builtin_stxcpy_chk): Same.
16849 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
16851 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
16853 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
16854 parameter. Use it for SFmode.
16855 (rs6000_function_arg_advance_1): Adjust.
16856 (rs6000_function_arg): Adjust.
16857 (rs6000_gimplify_va_arg): Pass false for that new parameter.
16859 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
16861 PR rtl-optimization/84169
16862 PR rtl-optimization/84780
16863 * combine.c (can_combine_p): Check for a 2-insn combination whether
16864 the destination register is used between the two insns, too.
16866 2018-03-12 Richard Biener <rguenther@suse.de>
16868 PR tree-optimization/84803
16869 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
16870 for refs DR analysis didn't process.
16872 2018-03-12 Richard Biener <rguenther@suse.de>
16874 PR tree-optimization/84777
16875 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
16876 force-vectorize loops ignore whether we are optimizing for size.
16878 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
16880 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
16881 (TARGET_MD_ASM_ADJUST): Define.
16883 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
16884 Kito Cheng <kito.cheng@gmail.com>
16885 Chung-Ju Wu <jasonwucj@gmail.com>
16887 * config/nds32/nds32.c (nds32_compute_stack_frame,
16888 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
16889 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
16890 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
16891 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
16892 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
16893 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
16894 * config/nds32/nds32.md (prologue, epilogue): Use macro
16895 NDS32_V3PUSH_AVAILABLE_P to do checking.
16897 2018-03-11 Jakub Jelinek <jakub@redhat.com>
16900 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
16901 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
16902 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
16903 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
16904 addition of most attributes on !orig_type_die or the attribute not
16905 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
16907 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
16908 Chung-Ju Wu <jasonwucj@gmail.com>
16910 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
16911 __NDS32_VH__ macro.
16912 * config/nds32/nds32.opt (mvh): New option.
16914 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
16915 Chung-Ju Wu <jasonwucj@gmail.com>
16917 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
16919 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
16920 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
16923 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
16924 Chung-Ju Wu <jasonwucj@gmail.com>
16926 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
16928 * config/nds32/nds32-multiple.md (strlensi): New pattern.
16929 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
16931 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
16932 Kito Cheng <kito.cheng@gmail.com>
16933 Chung-Ju Wu <jasonwucj@gmail.com>
16935 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
16936 UNSPEC_FFMISM and UNSPEC_FLMISM.
16937 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
16938 for ffb, ffmism and flmism.
16939 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
16940 (unspec_ffmism): Ditto.
16941 (unspec_flmism): Ditto.
16942 (nds32_expand_builtin_impl): Check if string extension is available.
16943 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
16944 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
16946 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
16949 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
16952 * lra-assigns.c (assign_by_spills): Return a flag of reload
16953 assignment failure. Do not process the reload assignment
16954 failures. Do not spill other reload pseudos if they has the same
16956 (lra_assign): Add a return arg. Set up from the result of
16957 assign_by_spills call.
16958 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
16959 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
16960 usage_insns if it is not NULL.
16961 (spill_hard_reg_in_range): New function.
16962 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
16963 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
16964 function prototypes.
16965 (lra_assign): Change prototype.
16966 * lra.c (lra): Add code to deal with fails by splitting hard reg
16969 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
16972 * config/i386/i386.opt: Replace Enforcment with Enforcement.
16974 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
16977 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
16978 (dw_val_node): Add val_symbolic_view.
16979 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
16980 (symview_upper_bound): New.
16981 (new_line_info_table): Initialize symviews_since_reset.
16982 (dwarf2out_source_line): Count symviews_since_reset and set
16983 symview_upper_bound.
16984 (dw_val_equal_p): Handle symview.
16985 (add_AT_symview): New.
16986 (print_dw_val): Handle symview.
16987 (attr_checksum, attr_checksum_ordered): Likewise.
16988 (same_dw_val_p, size_of_die): Likewise.
16989 (value_format, output_die): Likewise.
16990 (add_high_low_attributes): Use add_AT_symview for entry_view.
16991 (dwarf2out_finish): Reset symview_upper_bound, clear
16994 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
16997 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
16998 Add strict argument and use it.
16999 (rs6000_split_multireg_move): Update for new strict argument.
17000 (mem_operand_gpr): Disallow all non-offsettable addresses.
17001 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
17003 2018-03-09 Jakub Jelinek <jakub@redhat.com>
17006 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
17007 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
17008 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
17011 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
17012 decl, use remap_type if we want to use the type.
17014 2018-03-09 Martin Sebor <msebor@redhat.com>
17016 PR tree-optimization/84526
17017 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
17019 (builtin_access::generic_overlap): Be prepared to handle non-array
17022 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
17024 PR rtl-optimization/84682
17025 * lra-constraints.c (process_address_1): Check is_address flag
17026 for address constraints.
17027 (process_alt_operands): Likewise.
17028 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
17029 preprocess_constraints.
17030 * recog.h (preprocess_constraints): Add oploc parameter.
17032 * recog.c (preprocess_constraints): Test address_operand for
17033 CT_ADDRESS constraints.
17035 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
17038 * lra-assigns.c (assign_by_spills): Return a flag of reload
17039 assignment failure. Do not process the reload assignment
17040 failures. Do not spill other reload pseudos if they has the same
17042 (lra_assign): Add a return arg. Set up from the result of
17043 assign_by_spills call.
17044 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
17045 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
17046 usage_insns if it is not NULL.
17047 (spill_hard_reg_in_range): New function.
17048 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
17049 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
17050 function prototypes.
17051 (lra_assign): Change prototype.
17052 * lra.c (lra): Add code to deal with fails by splitting hard reg
17055 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17058 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
17059 Accept complain bool parameter. Only emit errors if it is true.
17060 (arm_parse_cpu_option_name): Likewise.
17061 (arm_target_thumb_only): Adjust callers of the above.
17062 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
17063 prototype to take a default true bool parameter.
17064 (arm_parse_arch_option_name): Likewise.
17066 2018-03-09 David Malcolm <dmalcolm@redhat.com>
17067 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
17071 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
17072 * configure: Regenerate.
17073 * configure.ac ("linker --version-script option"): New.
17074 ("linker soname option"): New.
17076 2018-03-09 Richard Biener <rguenther@suse.de>
17078 PR tree-optimization/84775
17079 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
17080 immediate uses of predicate stmts and mark them modified.
17083 PR tree-optimization/84178
17084 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
17086 (version_loop_for_if_conversion): Delay update_ssa call.
17087 (tree_if_conversion): Delay update_ssa until after predicate
17090 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
17093 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
17094 when the function accesses prior frames.
17096 2018-03-08 Jakub Jelinek <jakub@redhat.com>
17099 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
17100 gen_llsym, otherwise call maybe_gen_llsym.
17102 PR inline-asm/84742
17103 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
17104 has ',' character inside of it.
17106 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17109 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
17110 as clobbering CC_REGNUM.
17112 2018-03-08 Richard Biener <rguenther@suse.de>
17114 PR middle-end/84552
17115 * tree-scalar-evolution.c: Include tree-into-ssa.h.
17116 (follow_copies_to_constant): Do not follow SSA names registered
17119 2018-03-08 Richard Biener <rguenther@suse.de>
17121 PR tree-optimization/84178
17122 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
17124 (version_loop_for_if_conversion): Delay update_ssa call.
17125 (tree_if_conversion): Delay update_ssa until after predicate
17128 2018-03-08 David Malcolm <dmalcolm@redhat.com>
17130 PR tree-optimization/84178
17131 * tree-if-conv.c (release_bb_predicate): Remove the
17132 the assertion that the stmts have NULL use_ops.
17133 Discard the statements, asserting that they haven't
17134 yet been added to a BB.
17136 2018-03-08 Richard Biener <rguenther@suse.de>
17138 PR tree-optimization/84746
17139 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
17140 (phi_translate): Pass in destination ANTIC_OUT set.
17141 (phi_translate_1): Likewise. For a simplified result lookup
17142 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
17143 (phi_translate_set): Adjust.
17144 (do_pre_regular_insertion): Likewise.
17145 (do_pre_partial_partial_insertion): Likewise.
17147 2018-03-08 Martin Liska <mliska@suse.cz>
17149 PR gcov-profile/84735
17150 * doc/gcov.texi: Document usage of profile files.
17151 * gcov-io.h: Document changes in the format.
17153 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
17157 * dwarf2out.c (struct dw_line_info_table): Update comments for
17159 (FORCE_RESET_NEXT_VIEW): New.
17160 (FORCE_RESETTING_VIEW_P): New.
17161 (RESETTING_VIEW_P): Check for -1 too.
17162 (ZERO_VIEW_P): Likewise.
17163 (new_line_info_table): Force-reset next view.
17164 (dwarf2out_begin_function): Likewise.
17165 (dwarf2out_source_line): Simplify zero_view_p initialization.
17166 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
17167 view directly. Omit view when omitting .loc at line 0.
17169 2018-03-08 Jakub Jelinek <jakub@redhat.com>
17171 PR tree-optimization/84740
17172 * tree-switch-conversion.c (process_switch): Call build_constructors
17173 only if info.phi_count is non-zero.
17175 PR tree-optimization/84739
17176 * tree-tailcall.c (find_tail_calls): Check call arguments against
17177 DECL_ARGUMENTS (current_function_decl) rather than
17178 DECL_ARGUMENTS (func) when checking for tail recursion.
17180 2018-03-07 Jakub Jelinek <jakub@redhat.com>
17182 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
17183 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
17184 Volker Reichelt's entry and add entries for people that perform
17185 GCC fuzzy testing and report numerous bugs.
17187 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
17190 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
17191 readonly data in sdata, if that is disabled.
17192 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
17193 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
17194 -mreadonly-in-sdata option.
17196 2018-03-07 Martin Sebor <msebor@redhat.com>
17198 PR tree-optimization/84468
17199 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
17200 basic block when looking for nul assignment.
17202 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
17205 * except.h (output_function_exception_table): Adjust prototype.
17206 * except.c (output_function_exception_table): Remove FNNAME parameter
17207 and add SECTION parameter. Ouput one part of the table at a time.
17208 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
17209 the first part of the exception table and emit unwind directives.
17210 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
17211 (i386_pe_seh_cold_init): Likewise.
17212 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
17213 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
17214 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
17215 (ix86_output_call_insn): Emit a nop in one more case for SEH.
17216 * config/i386/winnt.c: Include except.h.
17217 (struct seh_frame_state): Add reg_offset, after_prologue and
17218 in_cold_section fields.
17219 (i386_pe_seh_end_prologue): Set seh->after_prologue.
17220 (i386_pe_seh_cold_init): New function.
17221 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
17222 to seh->in_cold_section.
17223 (seh_emit_push): Record the offset of the push.
17224 (seh_emit_save): Record the offet of the save.
17225 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
17226 Test seh->after_prologue to disregard the epilogue.
17227 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
17228 (i386_pe_end_cold_function): New function.
17230 2018-03-07 Jakub Jelinek <jakub@redhat.com>
17233 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
17234 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
17237 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
17239 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
17240 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
17242 PR middle-end/84723
17243 * multiple_target.c: Include tree-inline.h and intl.h.
17244 (expand_target_clones): Diagnose and fail if node->definition and
17245 !tree_versionable_function_p (node->decl).
17247 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
17249 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
17251 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
17252 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17253 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
17255 2018-03-06 Jakub Jelinek <jakub@redhat.com>
17258 * combine.c (try_combine): Use reg_or_subregno instead of handling
17259 just paradoxical SUBREGs and REGs.
17261 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
17263 * config/arc/arc.c (arc_finalize_pic): Remove function.
17264 (arc_must_save_register): We use single base PIC register, remove
17265 checks to save/restore the PIC register.
17266 (arc_expand_prologue): Likewise.
17267 * config/arc/arc-protos.h (arc_set_default_type_attributes):
17269 (arc_verify_short): Likewise.
17270 (arc_attr_type): Likewise.
17271 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
17272 (walk_stores): Likewise.
17273 (arc_address_cost): Make it static.
17274 (arc_verify_short): Likewise.
17275 (branch_dest): Likewise.
17276 (arc_attr_type): Likewise.
17277 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
17278 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
17279 (arc_final_prescan_insn): Remove inserting the nops due to
17280 hardware hazards. It is done in reorg step.
17281 (insn_length_variant_t): Remove.
17282 (insn_length_parameters_t): Likewise.
17283 (arc_insn_length_parameters): Likewise.
17284 (arc_get_insn_variants): Likewise.
17285 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
17287 2018-03-06 Jakub Jelinek <jakub@redhat.com>
17289 PR inline-asm/84683
17290 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
17293 PR tree-optimization/84687
17294 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
17296 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
17298 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17300 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
17301 Rename to ppc_speculation_barrier.
17302 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
17303 __builtin_ppc_speculation_barrier.
17305 2018-03-05 Jakub Jelinek <jakub@redhat.com>
17308 * combine.c (combine_simplify_rtx): Don't try to simplify if
17309 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
17312 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
17314 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
17315 to 32 bytes when compiling for POWER9.
17317 2018-03-05 Jakub Jelinek <jakub@redhat.com>
17320 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
17321 regparm >= 3 with no arg reg available also for calls with
17322 flag_force_indirect_call. Pass decl to ix86_function_regparm.
17325 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
17327 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
17329 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
17332 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
17334 2018-03-05 Richard Biener <rguenther@suse.de>
17336 PR tree-optimization/84486
17337 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
17338 When inserting a __builtin_assume_aligned call set the LHS
17339 SSA name alignment info accordingly.
17341 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
17343 PR tree-optimization/84114
17344 * config/aarch64/aarch64.c (aarch64_reassociation_width)
17345 Avoid reassociation of FLOAT_MODE addition.
17347 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
17349 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
17350 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
17351 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
17352 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
17353 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
17354 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
17355 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
17357 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
17358 __builtin_ia32_wbinvd): New builtins.
17359 (SPECIAL_ARGS2): New.
17360 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
17361 (SPECIAL_ARGS2): New.
17362 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
17363 (ix86_valid_target_attribute_inner_p): Ditto.
17364 (ix86_init_mmx_sse_builtins): Add special_args2.
17365 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
17366 TARGET_WBNOINVD_P): New.
17367 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
17368 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
17369 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
17370 * config/i386/immintrin.h (_wbinvd): New intrinsic.
17371 * config/i386/pconfigintrin.h: New file.
17372 * config/i386/wbnoinvdintrin.h: Ditto.
17373 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
17375 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
17377 2018-03-05 Richard Biener <rguenther@suse.de>
17379 PR tree-optimization/84670
17380 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
17382 (BB_VISITED_WITH_VISITED_SUCCS): New define.
17383 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
17384 (compute_antic_aux): Only assert the number of values in ANTIC_IN
17385 doesn't grow if all successors (recursively) were visited at least
17388 2018-03-05 Richard Biener <rguenther@suse.de>
17390 PR tree-optimization/84650
17391 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
17392 if executed in the loop pipeline.
17394 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
17396 * doc/configfiles.texi (Configuration Files): Move info about
17397 conditionalizing $target-protos.h to...
17398 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
17399 differs from $target-protos.h.
17401 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
17402 Chung-Ju Wu <jasonwucj@gmail.com>
17404 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
17405 * config/nds32/nds32-multiple.md (setmemsi): Define.
17406 * config/nds32/nds32-memory-manipulation.c
17407 (nds32_gen_dup_4_byte_to_word_value): New.
17408 (emit_setmem_word_loop): New.
17409 (emit_setmem_byte_loop): New.
17410 (nds32_expand_setmem_loop): New.
17411 (nds32_expand_setmem_loop_v3m): New.
17412 (nds32_expand_setmem_unroll): New.
17413 (nds32_expand_setmem): New.
17415 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
17416 Chung-Ju Wu <jasonwucj@gmail.com>
17418 * config/nds32/nds32-memory-manipulation.c
17419 (nds32_emit_load_store): New.
17420 (nds32_emit_post_inc_load_store): New.
17421 (nds32_emit_mem_move): New.
17422 (nds32_emit_mem_move_block): New.
17423 (nds32_expand_movmemsi_loop_unknown_size): New.
17424 (nds32_expand_movmemsi_loop_known_size): New.
17425 (nds32_expand_movmemsi_loop): New.
17426 (nds32_expand_movmemsi_unroll): New.
17427 (nds32_expand_movmemqi): Rename ...
17428 (nds32_expand_movmemsi): ... to this.
17429 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
17430 (movmemsi): ... to this.
17431 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
17432 (nds32_expand_movmemsi): ... to this.
17434 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
17435 Monk Chiang <sh.chiang04@gmail.com>
17436 Chung-Ju Wu <jasonwucj@gmail.com>
17438 * config/nds32/nds32-protos.h
17439 (nds32_expand_load_multiple): New arguments.
17440 (nds32_expand_store_multiple): Ditto.
17441 (nds32_valid_multiple_load_store): Rename ...
17442 (nds32_valid_multiple_load_store_p): ... to this.
17443 * config/nds32/nds32-memory-manipulation.c
17444 (nds32_expand_load_multiple): Refine implementation.
17445 (nds32_expand_store_multiple): Ditto.
17446 * config/nds32/nds32-multiple.md
17447 (load_multiple): Update nds32_expand_load_multiple interface.
17448 (store_multiple): Update nds32_expand_store_multiple interface.
17449 * config/nds32/nds32-predicates.c
17450 (nds32_valid_multiple_load_store): Rename ...
17451 (nds32_valid_multiple_load_store_p): ... to this and refine
17453 * config/nds32/predicates.md
17454 (nds32_load_multiple_and_update_address_operation): New predicate.
17455 (nds32_store_multiple_and_update_address_operation): New predicate.
17457 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
17458 Chung-Ju Wu <jasonwucj@gmail.com>
17460 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
17461 (combo): New attribute.
17462 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
17464 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
17466 * config/nds32/nds32.opt: Change -mcmodel= default value.
17468 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
17469 Monk Chiang <sh.chiang04@gmail.com>
17470 Chung-Ju Wu <jasonwucj@gmail.com>
17472 * config/nds32/constants.md (unspec_element): New enum.
17473 * config/nds32/constraints.md (Umw): New constraint.
17474 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
17475 * config/nds32/nds32-intrinsic.md: Likewise.
17476 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
17477 (nds32_valid_smw_lwm_base_p): New.
17478 (nds32_output_smw_single_word): New.
17479 (nds32_output_lmw_single_word): New.
17480 (nds32_expand_unaligned_load): New.
17481 (nds32_expand_unaligned_store): New.
17482 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
17483 (nds32_output_smw_single_word): Declare.
17484 (nds32_output_lmw_single_word): Declare.
17485 (nds32_expand_unaligned_load): Declare.
17486 (nds32_expand_unaligned_store): Declare.
17487 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
17488 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
17489 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
17490 NDS32_BUILTIN_UASTORE_DW.
17491 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
17494 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
17495 Kito Cheng <kito.cheng@gmail.com>
17496 Chung-Ju Wu <jasonwucj@gmail.com>
17498 * config/nds32/nds32-intrinsic.c
17499 (nds32_expand_builtin_null_ftype_reg): Delete.
17500 (nds32_expand_builtin_reg_ftype_imm): Ditto.
17501 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
17502 (nds32_read_argument): New.
17503 (nds32_legitimize_target): Ditto.
17504 (nds32_legitimize_argument): Ditto.
17505 (nds32_check_constant_argument): Ditto.
17506 (nds32_expand_unop_builtin): Ditto.
17507 (nds32_expand_unopimm_builtin): Ditto.
17508 (nds32_expand_binop_builtin): Ditto.
17509 (nds32_builtin_decl_impl): Ditto.
17510 (builtin_description): Ditto.
17511 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
17512 (nds32_init_builtins_impl): Ditto.
17513 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
17514 (nds32_builtin_decl): New.
17515 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
17516 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
17518 2018-03-02 Jeff Law <law@redhat.com>
17520 * reorg.c (stop_search_p): Handle DEBUG_INSN.
17521 (redundant_insn, fill_simple_delay_slots): Likewise.
17522 (fill_slots_from_thread): Likewise.
17523 * resource.c (mark_referenced_resources): Likewise.
17524 (mark_set_resources, find_dead_or_set_registers): Likewise.
17526 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17528 * substring-locations.h (format_warning_va): Formatting fix for
17529 ATTRIBUTE_GCC_DIAG.
17530 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
17532 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
17533 * substring-locations.c: Include intl.h.
17534 (format_warning_va): Turned into small wrapper around
17535 format_warning_n_va, renamed to ...
17536 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
17537 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
17539 (format_warning_at_substring_n): New function.
17540 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
17541 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
17542 format_warning_at_substring with just a shorter name instead of
17543 const function pointer.
17544 (fmtwarn_n): New function.
17545 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
17546 appropriate, get rid of all the fmtstr temporaries, move conditionals
17547 with G_() wrapped string literals directly into fmtwarn arguments,
17548 cast dir.len to (int), formatting fixes.
17550 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
17552 * doc/invoke.texi: Remove "Cilk Plus" references.
17554 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17555 Richard Biener <rguenther@suse.de>
17558 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
17559 for error or warning attributes if CALL_FROM_THUNK_P is set.
17562 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17565 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
17566 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
17569 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
17570 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
17572 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
17573 instead of -1U in last predictors element's probability member.
17575 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
17578 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
17579 arguments if they are comparable.
17581 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
17583 PR tree-optimization/84634
17584 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
17585 masks and masked_loop_p with a single loop_masks, making sure it's
17586 null for bb vectorization.
17588 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
17590 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
17591 (vect_analyze_data_ref_access): Use loop->safe_len rather than
17592 loop->force_vectorize to check whether there is no alias.
17594 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17597 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
17599 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
17601 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
17602 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
17603 instead of a loop around prev_real_insn.
17604 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
17607 PR inline-asm/84625
17608 * config/i386/i386.c (ix86_print_operand): Use conditional
17609 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
17612 2018-03-02 Richard Biener <rguenther@suse.de>
17614 PR tree-optimization/84427
17615 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
17616 (bitmap_set_subtract_values): Rewrite to handle multiple
17619 (prune_clobbered_mems): Likewise.
17620 (phi_translate): Take edge instead of pred/phiblock.
17621 (phi_translate_1): Likewise.
17622 (phi_translate_set): Likewise. Insert all translated
17623 exprs for a value into the set, keeping possibly multiple
17624 expressions per value.
17625 (compute_antic_aux): Adjust for phi_translate changes.
17626 When intersecting union the expressions and prune those
17627 not in the final value set, keeping possibly multiple
17628 expressions per value. Do not use value-insertion
17629 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
17630 all expressions. Add verification that the value-sets
17631 only shrink during iteration.
17632 (compute_partial_antic_aux): Adjust for the phi_translate changes.
17633 (do_pre_regular_insertion): Likewise.
17634 (do_pre_partial_partial_insertion): Likewise.
17636 2018-03-02 Richard Biener <rguenther@suse.de>
17639 * config/darwin.c (saved_debug_info_level): New static global.
17640 (darwin_asm_lto_start): Disable debug info generation for LTO out.
17641 (darwin_asm_lto_end): Restore debug info generation settings.
17643 2018-03-01 Martin Liska <mliska@suse.cz>
17646 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
17647 volatile arguments.
17649 2018-03-01 Richard Biener <rguenther@suse.de>
17652 * dwarf2out.c (gen_variable_die): Properly handle late VLA
17653 type annotation with LTO when debug was disabled at compile-time.
17655 2018-03-01 Matthew Fortune <mfortune@gmail.com>
17657 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
17659 (mips_final_postscan_insn): Likewise.
17661 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
17663 PR rtl-optimization/84528
17664 * alias.c (init_alias_target): Add commentary.
17665 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
17666 a unique base value if the frame pointer is not eliminated
17667 to the stack pointer.
17669 2018-03-01 Tom de Vries <tom@codesourcery.com>
17671 PR rtl-optimization/83327
17672 * lra-int.h (hard_regs_spilled_into): Declare.
17673 * lra.c (hard_regs_spilled_into): Define.
17674 (init_reg_info): Init hard_regs_spilled_into.
17675 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
17676 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
17677 (process_bb_lives): Handle hard_regs_spilled_into.
17678 (lra_create_live_ranges_1): Before doing liveness propagation, clear
17679 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
17681 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
17683 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
17684 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
17685 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
17686 * config/rs6000/aix72.h: New file.
17688 2018-02-28 Jakub Jelinek <jakub@redhat.com>
17690 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
17691 instead of warning_at with conditional singular and plural messages
17695 * stor-layout.c (update_alignment_for_field): For
17696 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
17697 && !DECL_PACKED (field), do the alignment update, just use
17698 only desired_align instead of MAX (type_align, desired_align)
17700 (place_field): Don't do known_align < desired_align handling
17701 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
17702 is non-NULL, instead do it after rli->prev_field handling and
17703 only if not within a bitfield word. For DECL_PACKED (field)
17704 use type_align of BITS_PER_UNIT.
17706 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
17708 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
17709 superfluous parentheses and trailing spaces.
17711 2018-02-28 Richard Biener <rguenther@suse.de>
17713 PR tree-optimization/84584
17714 * graphite-scop-detection.c (scop_detection::add_scop): Discard
17715 SCoPs with fake exit edge.
17717 2018-02-28 Martin Liska <mliska@suse.cz>
17720 * timevar.c (timer::print): Fix format to properly print 100%
17723 2018-02-28 Richard Biener <rguenther@suse.de>
17725 PR middle-end/84607
17726 * genmatch.c (capture_info::walk_match): Do not mark
17727 captured expressions without operands as expr_p given
17728 they act more like predicates and should be subject to
17729 "lost tail" side-effect preserving.
17731 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
17733 PR rtl-optimization/81611
17734 * auto-inc-dec.c (attempt_change): Move dead note from
17735 mem_insn if it's the next use of regno
17736 (find_address): Take address use of reg holding
17737 non-incremented value. Add parm to limit search to the named
17739 (merge_in_block): Attempt to use a mem insn that is the next
17740 use of the original regno.
17742 2018-02-27 Martin Sebor <msebor@redhat.com>
17745 * doc/invoke.texi (-Wmissing-attributes): New option.
17746 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
17748 2018-02-27 Martin Sebor <msebor@redhat.com>
17750 PR translation/84207
17751 * diagnostic-core.h (warning_n, error_n, inform_n): Change
17752 n argument to unsigned HOST_WIDE_INT.
17753 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
17754 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
17755 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
17756 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
17758 2018-02-27 Richard Biener <rguenther@suse.de>
17760 PR tree-optimization/84512
17761 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
17762 Do not use the estimate returned from record_stmt_cost for
17763 the scalar iteration cost but sum properly using add_stmt_cost.
17765 2018-02-27 Richard Biener <rguenther@suse.de>
17767 PR tree-optimization/84466
17768 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
17769 Adjust last change to less strictly validate use operands.
17771 2018-02-27 Martin Liska <mliska@suse.cz>
17773 PR gcov-profile/84548
17774 * gcov.c (process_file): Allow partial overlap and consider it
17775 also as group functions.
17776 (output_lines): Properly calculate range of lines for a group.
17778 2018-02-27 Martin Liska <mliska@suse.cz>
17780 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
17781 'ggc' suffixes. Change first column width.
17782 (timer::print): Fix formatting of the column.
17784 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
17786 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
17787 preserve inline entry blocks for the sake of debug inline
17788 entry point markers alone.
17789 (remove_unused_locals): Suggest in comments a better place to
17790 force the preservation of inline entry blocks that are
17791 otherwise unused, but do not preserve them.
17793 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
17795 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
17797 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
17800 * config/i386/constraints.md (Bs): Replace
17801 ix86_indirect_branch_register with
17802 TARGET_INDIRECT_BRANCH_REGISTER.
17804 * config/i386/i386.md (indirect_jump): Likewise.
17805 (tablejump): Likewise.
17806 (*sibcall_memory): Likewise.
17807 (*sibcall_value_memory): Likewise.
17808 Peepholes of indirect call and jump via memory: Likewise.
17809 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
17810 (*sibcall_value_GOT_32): Likewise.
17811 * config/i386/predicates.md (indirect_branch_operand): Likewise.
17812 (GOT_memory_operand): Likewise.
17813 (call_insn_operand): Likewise.
17814 (sibcall_insn_operand): Likewise.
17815 (GOT32_symbol_operand): Likewise.
17816 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
17818 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
17820 PR rtl-optimization/83496
17821 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
17822 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
17823 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
17824 redundant insn, if any.
17825 (relax_delay_slots): Likewise.
17826 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
17828 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
17830 PR tree-optimization/83965
17831 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
17832 that grouped statements are part of a reduction chain. Return
17833 true if the statement is not marked as a reduction itself but
17834 is part of a group.
17835 (vect_recog_dot_prod_pattern): Don't check whether the statement
17836 is part of a group here.
17837 (vect_recog_sad_pattern): Likewise.
17838 (vect_recog_widen_sum_pattern): Likewise.
17840 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
17843 * final.c (rest_of_clean_state): Also look for calls inside sequences.
17845 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
17848 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
17850 (ix86_output_indirect_function_return): New prototype.
17851 (ix86_split_simple_return_pop_internal): Likewise.
17852 * config/i386/i386.c (indirect_return_via_cx): New.
17853 (indirect_return_via_cx_bnd): Likewise.
17854 (indirect_thunk_name): Handle return va CX_REG.
17855 (output_indirect_thunk_function): Create alias for
17856 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
17857 (ix86_output_indirect_jmp): Remove the bool argument.
17858 (ix86_output_indirect_function_return): New function.
17859 (ix86_split_simple_return_pop_internal): Likewise.
17860 * config/i386/i386.md (*indirect_jump): Don't pass false
17861 to ix86_output_indirect_jmp.
17862 (*tablejump_1): Likewise.
17863 (simple_return_pop_internal): Change it to define_insn_and_split.
17864 Call ix86_split_simple_return_pop_internal to split it for
17865 -mfunction-return=.
17866 (simple_return_indirect_internal): Call
17867 ix86_output_indirect_function_return instead of
17868 ix86_output_indirect_jmp.
17870 2018-02-26 Jakub Jelinek <jakub@redhat.com>
17873 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
17874 memset and value initialization afterwards.
17876 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
17878 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
17880 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17883 * common/config/aarch64/aarch64-common.c
17884 (aarch_option_optimization_table[]): Switch
17885 off fomit-frame-pointer
17887 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
17888 Chung-Ju Wu <jasonwucj@gmail.com>
17890 * config/nds32/nds32-multiple.md (load_multiple): Disallow
17892 (store_multiple): Ditto.
17894 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
17896 * config.gcc: Add --with-cpu support for nds32 target.
17897 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
17898 * config/nds32/nds32.opt: Add -mcpu= option.
17900 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
17902 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
17903 isel=yes): Warn for these deprecated options.
17905 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
17907 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
17908 ISA_2_5_MASKS_EMBEDDED.
17910 2018-02-23 Jakub Jelinek <jakub@redhat.com>
17912 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
17913 p->max as pointers rather than using iterative_hash_expr.
17915 2018-02-23 Carl Love <cel@us.ibm.com>
17917 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
17918 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
17920 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
17921 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
17922 P8V_BUILTIN_VEC_VUNSIGNED2.
17924 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
17927 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
17928 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
17930 * lra-constraints.c (curr_insn_transform): Set up
17931 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
17933 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
17934 * lra-eliminations.c (spill_pseudos): Ditto.
17935 (process_insn_for_elimination): Ditto.
17936 * lra-lives.c (reg_early_clobber_p): Use the new macros.
17937 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
17938 LRA_NON_CLOBBERED_ALT.
17940 2018-02-22 Martin Sebor <msebor@redhat.com>
17942 PR tree-optimization/84480
17943 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
17944 to maybe_diag_stxncpy_trunc. Call it.
17945 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
17946 from gimple_fold_builtin_strcpy. Print inlining stack.
17947 (handle_builtin_stxncpy): Print inlining stack.
17948 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
17950 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
17953 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
17954 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
17955 and -fcheck-pointer-bounds are used together.
17956 (indirect_thunk_prefix): New enum.
17957 (indirect_thunk_need_prefix): New function.
17958 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
17959 "_nt" instead of "_bnd" for NOTRACK prefix.
17960 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
17961 (output_indirect_thunk_function): Likewise.
17963 (ix86_code_end): Update output_indirect_thunk_function calls.
17964 (ix86_output_indirect_branch_via_reg): Replace
17965 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
17966 (ix86_output_indirect_branch_via_push): Likewise.
17967 (ix86_output_function_return): Likewise.
17968 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
17969 incompatible with -fcf-protection=branch and
17970 -fcheck-pointer-bounds.
17972 2018-02-22 Steve Ellcey <sellcey@cavium.com>
17975 * config/aarch64/aarch64.c (aarch64_print_address_internal):
17976 Change gcc_assert call to output_operand_lossage.
17978 2018-02-22 Steve Ellcey <sellcey@cavium.com>
17980 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
17982 2018-02-22 DJ Delorie <dj@redhat.com>
17983 Sebastian Perta <sebastian.perta@renesas.com>
17984 Oleg Endo <olegendo@gcc.gnu.org>
17986 * config/rx/rx.c (rx_rtx_costs): New function.
17987 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
17989 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
17991 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
17993 2018-02-22 Martin Liska <mliska@suse.cz>
17996 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
17997 Add "native" as a possible value.
17999 2018-02-22 Martin Liska <mliska@suse.cz>
18002 * config/i386/i386.c (ix86_option_override_internal):
18003 Add "native" as a possible value for -march and -mtune.
18005 2018-02-22 Jakub Jelinek <jakub@redhat.com>
18008 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
18009 to all type variants.
18011 PR tree-optimization/84503
18012 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
18013 width as info->bitpos + info->bitsize - start.
18014 (merged_store_group::merge_overlapping): Simplify width computation.
18015 (check_no_overlap): New function.
18016 (imm_store_chain_info::try_coalesce_bswap): Compute expected
18017 start + width and last_order of the group, fail if check_no_overlap
18019 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
18020 to group if check_no_overlap fails.
18022 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
18024 * config/rs6000/altivec.md: Delete contraint arguments to
18025 define_expand, define_split, and define_peephole2, and in
18026 define_insn_and_split if always unused.
18027 * config/rs6000/darwin.md: Ditto.
18028 * config/rs6000/dfp.md: Ditto.
18029 * config/rs6000/rs6000.md: Ditto.
18030 * config/rs6000/sync.md: Ditto.
18031 * config/rs6000/vector.md: Ditto.
18032 * config/rs6000/vsx.md: Ditto.
18034 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
18036 * config/rs6000/altivec.md: Write output control strings as braced
18037 blocks instead of double-quoted strings.
18038 * config/rs6000/darwin.md: Ditto.
18039 * config/rs6000/rs6000.md: Ditto.
18040 * config/rs6000/vector.md: Ditto.
18041 * config/rs6000/vsx.md: Ditto.
18043 2018-02-21 Jason Merrill <jason@redhat.com>
18045 PR c++/84314 - ICE with templates and fastcall attribute.
18046 * attribs.c (build_type_attribute_qual_variant): Remove assert.
18048 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
18050 * ipa-cp.c (determine_versionability): Fix comment typos.
18052 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
18055 * ipa-cp.c (determine_versionability): Do not version functions caling
18058 2018-02-21 Martin Liska <mliska@suse.cz>
18061 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
18062 Add "native" as a possible value.
18063 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
18064 the macro when native cpu detection is available.
18066 2018-02-21 Martin Liska <mliska@suse.cz>
18069 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
18070 Add "native" as a possible value.
18071 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
18072 when native cpu detection is available.
18074 2018-02-21 Jakub Jelinek <jakub@redhat.com>
18075 Martin Sebor <msebor@redhat.com>
18077 PR tree-optimization/84478
18078 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
18080 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
18081 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
18082 support which is conservatively correct, for 2 only stay conservative
18083 for maxlen. Formatting and comment capitalization fixes. Add STRICT
18084 argument to the 2 argument get_range_strlen, adjust 6 arg
18085 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
18087 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
18088 (gimple_fold_builtin_strlen): Pass true as last argument to
18091 2018-02-20 Martin Sebor <msebor@redhat.com>
18093 PR middle-end/84095
18094 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
18095 (builtin_memref::set_base_and_offset): Same. Handle inner references.
18096 (builtin_memref::builtin_memref): Factor out parts into
18097 set_base_and_offset and call it.
18099 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
18101 PR middle-end/84406
18102 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
18103 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
18104 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
18105 search at the associated MODE_INT.
18107 2018-02-20 Jeff Law <law@redhat.com>
18109 PR middle-end/82123
18110 PR tree-optimization/81592
18111 PR middle-end/79257
18112 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
18113 for range data rather than using global data.
18114 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
18115 range data rather than using global data.
18116 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
18117 pass it to children as needed.
18118 (struct directive::fmtresult): Similarly.
18119 (struct directive::set_width): Similarly.
18120 (struct directive::set_precision): Similarly.
18121 (format_integer, format_directive, parse_directive): Similarly.
18122 (format_none): Accept unnamed vr_values parameter.
18123 (format_percent, format_floating, format_character): Similarly.
18124 (format_string, format_plain): Similarly.
18125 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
18126 the EVRP range analyzer for range data rather than using global data.
18127 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
18128 gimple-ssa-evrp-analyze.h
18129 (class sprintf_dom_walker): Add after_dom_children member function.
18130 Add evrp_range_analyzer member.
18131 (sprintf_dom_walker::before_dom_children): Call into the EVRP
18132 range analyzer as needed.
18133 (sprintf_dom_walker::after_dom_children): New member function.
18134 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
18136 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
18137 (evrp_range_analyzer::pop_to_marker): Likewise.
18139 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
18141 PR tree-optimization/84419
18142 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
18143 with the required type if its current type is compatible but
18146 2018-02-20 Jakub Jelinek <jakub@redhat.com>
18148 PR middle-end/82004
18149 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
18150 after vectorization.
18152 2018-02-20 Martin Liska <mliska@suse.cz>
18155 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
18156 possible values if we don't have a hint.
18158 2018-02-20 Martin Liska <mliska@suse.cz>
18162 * final.c (shorten_branches): Build align_tab array with one
18164 * opts.c (finish_options): Add alignment option limit check.
18165 (MAX_CODE_ALIGN): Likewise.
18166 (MAX_CODE_ALIGN_VALUE): Likewise.
18167 * doc/invoke.texi: Document maximum allowed option value for
18168 all -falign-* options.
18170 2018-02-19 Jakub Jelinek <jakub@redhat.com>
18173 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
18174 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
18175 * var-tracking.c (emit_note_insn_var_location): Remove all references
18176 to NOTE_INSN_CALL_ARG_LOCATION.
18177 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
18178 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
18179 Use copy_rtx_if_shared.
18180 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
18181 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
18182 (dwarf2out_var_location): Remove handling of
18183 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
18185 * final.c (final_scan_insn): Remove all references to
18186 NOTE_INSN_CALL_ARG_LOCATION.
18187 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
18188 before dumping final insns.
18189 * except.c (emit_note_eh_region_end): Remove all references to
18190 NOTE_INSN_CALL_ARG_LOCATION.
18191 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
18192 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
18193 * config/arc/arc.c (hwloop_optimize): Likewise.
18194 * config/arm/arm.c (create_fix_barrier): Likewise.
18195 * config/s390/s390.c (s390_chunkify_start): Likewise.
18196 * config/sh/sh.c (find_barrier): Likewise.
18197 * config/i386/i386.c (rest_of_insert_endbranch,
18198 ix86_seh_fixup_eh_fallthru): Likewise.
18199 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
18200 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
18201 * config/frv/frv.c (frv_function_prologue): Likewise.
18202 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
18204 (note_outside_basic_block_p): Remove all references to
18205 NOTE_INSN_CALL_ARG_LOCATION.
18206 * gengtype.c (adjust_field_rtx_def): Likewise.
18207 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
18209 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
18210 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
18213 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
18216 PR tree-optimization/84452
18217 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
18218 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
18221 2018-02-19 Martin Liska <mliska@suse.cz>
18224 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
18226 2018-02-19 Martin Liska <mliska@suse.cz>
18227 Richard Sandiford <richard.sandiford@linaro.org>
18229 PR tree-optimization/82491
18230 * gimple-fold.c (get_base_constructor): Make earlier bail out
18233 2018-02-19 Carl Love <cel@us.ibm.com>
18235 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
18236 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
18238 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
18239 P8V_BUILTIN_VEC_NEG.
18241 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
18243 * config/rl78/rl78.md (movdf): New define expand.
18245 2018-02-19 Martin Liska <mliska@suse.cz>
18248 * doc/invoke.texi: Fix typo.
18249 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
18251 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
18253 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
18254 handle rs6000_single_float and rs6000_double_float specially for
18257 2018-02-16 Jeff Law <law@redhat.com>
18259 * config/rx/rx.c (add_pop_cfi_notes): New function.;
18260 (pop_regs): Use it.
18262 2018-02-16 Jakub Jelinek <jakub@redhat.com>
18265 * ipa-inline.c (inline_small_functions): Fix a typo.
18267 2018-02-16 Nathan Sidwell <nathan@acm.org>
18269 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
18271 2018-02-16 Carl Love <cel@us.ibm.com>
18273 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
18274 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
18275 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
18276 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
18277 expansion to P8V_BUILTIN_VEC_FLOAT2.
18279 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
18281 PR rtl-optimization/70023
18282 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
18283 src_regno into account.
18285 2018-02-16 Carl Love <cel@us.ibm.com>
18287 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
18288 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
18289 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
18290 * config/rs6000/rs6000.c: Remove case statements for
18291 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
18292 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
18293 and P9V_BUILTIN_VEC_VINSERT4B.
18294 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
18295 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
18296 * config/rs6000/vsx.md:
18297 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
18300 2018-02-16 Carl Love <cel@us.ibm.com>
18302 * config/rs6000/altivec.h: Add builtin names vec_extract4b
18304 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
18306 * config/rs6000/rs6000-c.c: Add the definitions for
18307 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
18308 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
18309 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
18310 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
18311 definition for insert4b and define insn *insert3b_internal.
18312 * doc/extend.texi: Add documentation for vec_extract4b.
18314 2018-02-16 Nathan Sidwell <nathan@acm.org>
18316 * doc/extend.texi (Backwards Compatibility): Mention friend
18317 injection. Note for-scope is deprecated.
18318 * doc/invoke.texi (-ffriend-injection): Deprecate.
18320 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
18322 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
18323 that moved to I2, also allow destinations that are a paradoxical
18324 subreg (instead of a normal reg).
18326 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
18329 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
18332 2018-02-16 Richard Biener <rguenther@suse.de>
18334 PR tree-optimization/84037
18335 PR tree-optimization/84016
18337 * config/i386/i386.c (ix86_builtin_vectorization_cost):
18338 Adjust vec_construct for the fact we need additional higher latency
18339 128bit inserts for AVX256 and AVX512 vector builds.
18340 (ix86_add_stmt_cost): Scale vector construction cost for
18343 2018-02-16 Richard Biener <rguenther@suse.de>
18345 PR tree-optimization/84417
18346 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
18347 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
18348 (non_rewritable_lvalue_p): Likewise, use poly-ints.
18350 2018-02-16 Martin Liska <mliska@suse.cz>
18353 * internal-fn.def (ASAN_CHECK): Set proper flags.
18354 (ASAN_MARK): Likewise.
18356 2018-02-16 Julia Koval <julia.koval@intel.com>
18358 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
18359 from PTA_CANNONLAKE.
18361 2018-02-16 Jakub Jelinek <jakub@redhat.com>
18364 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
18365 Use ++iter rather than iter++ for std::list iterators.
18366 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
18367 defer deleting them until all nodes in the forest are processed. Do
18368 free even leaf nodes. Change to_process into auto_vec.
18371 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
18372 * vec.h (vec_default_construct): Use memset instead of placement new
18373 if BROKEN_VALUE_INITIALIZATION is defined.
18374 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
18375 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
18378 PR rtl-optimization/83723
18379 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
18380 * lra.c (lra_substitute_pseudo): Likewise. If true, use
18381 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
18383 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
18385 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
18387 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
18389 PR rtl-optimization/81443
18390 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
18391 from inner REGs to paradoxical SUBREGs.
18393 2018-02-16 Richard Biener <rguenther@suse.de>
18395 PR tree-optimization/84399
18396 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
18397 For operands we can analyze at their definition make sure we can
18398 analyze them at each use as well.
18400 2018-02-16 Richard Biener <rguenther@suse.de>
18402 PR tree-optimization/84190
18403 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
18404 volatile accesses if the decl isn't volatile.
18406 2018-02-15 Jason Merrill <jason@redhat.com>
18408 PR c++/84314 - ICE with templates and fastcall attribute.
18409 * attribs.c (build_type_attribute_qual_variant): Don't clobber
18410 TYPE_CANONICAL on an existing type.
18412 2018-02-15 Jakub Jelinek <jakub@redhat.com>
18414 PR tree-optimization/84383
18415 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
18416 dstoff nor call operand_equal_p if dstbase is NULL.
18418 PR tree-optimization/84334
18419 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
18420 also a CONSTANT_CLASS_P, punt.
18422 2018-02-14 Jim Wilson <jimw@sifive.com>
18424 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
18425 first SMALL_OPERAND check. New local min_second_step. Move assert
18426 to where locals are set. Add TARGET_RVC support.
18427 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
18429 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
18431 * doc/invoke.texi: Correct -Wformat-overflow code sample.
18433 2018-02-14 Martin Sebor <msebor@redhat.com>
18435 PR tree-optimization/83698
18436 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
18437 arrays constrain the offset range to their bounds.
18438 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
18439 (builtin_access::overlap): Avoid setting the size of overlap if it's
18441 (maybe_diag_overlap): Also consider arrays when deciding what values
18442 of offsets to include in diagnostics.
18444 2018-02-14 Martin Sebor <msebor@redhat.com>
18447 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
18448 that correspond to the kind of a declaration.
18450 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
18453 * config/pa/pa.md: Load address of PIC label using the linkage table
18454 if the label is nonlocal.
18456 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
18458 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
18459 warning message if user requests -maltivec=be.
18460 * doc/invoke.texi: Document deprecation of -maltivec=be.
18462 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
18465 * config/rs6000/rs6000-c.c: Update definitions for
18466 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
18467 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
18469 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
18472 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
18473 add _get_ssp intrinsics. Remove argument from
18474 __builtin_ia32_rdssp[d|q].
18475 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
18476 * config/i386/i386-builtin.def: Remove argument from
18477 __builtin_ia32_rdssp[d|q].
18478 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
18479 ix86_expand_special_args_builtin for _rdssp[d|q].
18480 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
18481 Clear register before usage.
18482 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
18483 Add documentation for new _get_ssp and _inc_ssp intrinsics.
18485 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
18487 PR tree-optimization/84357
18488 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
18489 operand 1 of an ARRAY_REF too.
18491 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
18494 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
18495 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
18497 (set_of_reg): New struct.
18498 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
18499 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
18500 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
18502 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
18503 Split into bitclr, bitset, bitinvert patterns if appropriate.
18504 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
18505 use rx_fuse_in_memory_bitop.
18506 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
18507 to named insn, correct maximum insn length.
18509 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
18512 * machmode.def: Define a complex mode for PARTIAL_INT.
18513 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
18515 * doc/rtl.texi: Document CSPImode.
18516 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
18518 (msp430_hard_regno_nregs_with_padding): Likewise.
18520 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
18523 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
18525 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
18527 PR rtl-optimization/84169
18528 * combine.c (try_combine): New variable split_i2i3. Set it to true if
18529 we generated a parallel as new i3 and we split that to new i2 and i3
18530 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
18531 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
18532 those to i2, not i1. Partially rewrite this scan code.
18534 2018-02-13 Jakub Jelinek <jakub@redhat.com>
18537 * stor-layout.c (place_field): For variable length fields, adjust
18538 offset_align afterwards not just based on the field's alignment,
18539 but also on the size.
18541 PR middle-end/84309
18542 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
18543 of exps and logs in the use_exp2 case.
18545 2018-02-13 Jeff Law <law@redhat.com>
18547 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
18548 entry for "vector".
18550 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
18553 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
18557 * common.opt (gas-loc-support, gas-locview-support): New.
18558 (ginline-points, ginternal-reset-location-views): New.
18559 * doc/invoke.texi: Document them. Use @itemx where intended.
18560 (gvariable-location-views): Adjust.
18561 * target.def (reset_location_view): New.
18562 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
18563 (TARGET_RESET_LOCATION_VIEW): New.
18564 * doc/tm.texi: Rebuilt.
18565 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
18566 (dwarf2out_default_as_locview_support): New.
18567 (output_asm_line_debug_info): Use option variables.
18568 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
18569 (output_loc_list): Likewise.
18570 (add_high_low_attributes): Check option variables.
18571 Don't output entry view attribute in strict mode.
18572 (gen_inlined_subroutine_die): Check option variables.
18573 (dwarf2out_inline_entry): Likewise.
18574 (init_sections_and_labels): Likewise.
18575 (dwarf2out_early_finish): Likewise.
18576 (maybe_reset_location_view): New, from...
18577 (dwarf2out_var_location): ... here. Call it.
18578 * debug.h (dwarf2out_default_as_loc_support): Declare.
18579 (dwarf2out_default_as_locview_support): Declare.
18580 * hooks.c (hook_int_rtx_insn_0): New.
18581 * hooks.h (hook_int_rtx_insn_0): Declare.
18582 * toplev.c (process_options): Take -gas-loc-support and
18583 -gas-locview-support from dwarf2out. Enable
18584 -gvariable-location-views by default only with locview
18585 assembler support. Enable -ginternal-reset-location-views by
18586 default only if the target defines the corresponding hook.
18587 Enable -ginline-points by default if location views are
18588 enabled; force it disabled if statement frontiers are
18590 * tree-inline.c (expand_call_inline): Check option variables.
18591 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
18593 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
18595 PR tree-optimization/84321
18596 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
18597 handling. Also check whether the anti-range contains any values
18598 that satisfy the mask; switch to a VR_RANGE if not.
18600 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
18603 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
18605 2018-02-13 Martin Jambor <mjambor@suse.cz>
18608 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
18609 of call statements, also set location of a load to a temporary.
18611 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
18613 * config/rl78/rl78.c (add_vector_labels): New function.
18614 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
18615 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
18616 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
18617 which checks that no arguments are passed.
18618 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
18619 * doc/extend.texi: Documentation for the new attribute.
18621 2018-02-13 Andreas Schwab <schwab@suse.de>
18623 * config/riscv/linux.h (CPP_SPEC): Define.
18625 2018-02-13 Jakub Jelinek <jakub@redhat.com>
18628 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
18629 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
18630 OPTION_MASK_ISA_AES as first argument to def_builtin_const
18631 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
18632 instead of OPTION_MASK_ISA_PCLMUL as first argument to
18633 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
18634 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
18635 temporarily for AES and PCLMUL builtins.
18637 PR tree-optimization/84339
18638 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
18639 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
18642 PR middle-end/84309
18643 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
18644 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
18645 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
18647 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
18649 * omp-simd-clone.h: New file.
18650 * omp-simd-clone.c: Include omp-simd-clone.h.
18651 (expand_simd_clones): No longer static.
18652 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
18653 cgraph.h and omp-simd-clone.h.
18654 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
18655 (vect_recog_widen_shift_pattern): Formatting fix.
18656 (vect_pattern_recog_1): Don't check optab for calls.
18659 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
18660 operands[2] into a REG before using gen_lowpart on it.
18662 2018-02-12 Jeff Law <law@redhat.com>
18665 * config/sh/sh.c (find_barrier): Consider a sibling call
18668 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
18669 successfully back substituting a reg.
18671 2018-02-12 Richard Biener <rguenther@suse.de>
18673 PR tree-optimization/84037
18674 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
18675 parameter, move visited init to caller.
18676 (vect_slp_analyze_operations): Separate cost from validity
18677 check, initialize visited once for all instances.
18678 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
18680 * tree-vect-stmts.c (vect_model_simple_cost): Make early
18682 (vect_model_promotion_demotion_cost): Likewise.
18683 (vectorizable_bswap): Guard cost modeling with !slp_node
18684 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
18686 (vectorizable_call): Likewise.
18687 (vectorizable_conversion): Likewise.
18688 (vectorizable_assignment): Likewise.
18689 (vectorizable_shift): Likewise.
18690 (vectorizable_operation): Likewise.
18691 (vectorizable_store): Likewise.
18692 (vectorizable_load): Likewise.
18693 (vectorizable_condition): Likewise.
18694 (vectorizable_comparison): Likewise.
18696 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
18699 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
18700 (ASAN_MARK): Fix fnspec to account for return value, change pointer
18701 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
18703 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
18705 PR middle-end/83665
18706 * params.def (inline-min-speedup): Increase from 8 to 15.
18707 (max-inline-insns-auto): Decrease from 40 to 30.
18708 * ipa-split.c (consider_split): Add some buffer for function to
18709 be considered inlining candidate.
18710 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
18713 2018-02-12 Richard Biener <rguenther@suse.de>
18715 PR tree-optimization/84037
18716 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
18717 matched stmts if we cannot swap the non-matched ones.
18719 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
18721 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
18722 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
18723 _mm_maskz_scalef_round_ss): New intrinsics.
18724 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
18725 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
18726 __builtin_ia32_scalefss_round): Remove.
18727 (__builtin_ia32_scalefsd_mask_round,
18728 __builtin_ia32_scalefss_mask_round): New intrinsics.
18729 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
18730 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
18731 ((match_operand:VF_128 2 "<round_nimm_predicate>"
18732 "<round_constraint>")): Changed to ...
18733 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
18734 "<round_scalar_constraint>")): ... this.
18735 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
18736 %0, %1, %2<round_op3>}"): Changed to ...
18737 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
18738 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
18739 %2<round_scalar_mask_op3>}"): ... this.
18740 * config/i386/subst.md (round_scalar_nimm_predicate): New.
18742 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
18744 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
18745 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
18746 (_mm_maskz_sqrt_round_ss): New intrinsics.
18747 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
18748 (__builtin_ia32_sqrtsd_mask_round)
18749 (__builtin_ia32_sqrtss_mask_round): New builtins.
18750 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
18751 (__builtin_ia32_sqrtss_round): Remove.
18752 (__builtin_ia32_sqrtsd_mask_round)
18753 (__builtin_ia32_sqrtss_mask_round): New builtins.
18754 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
18755 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
18756 ((match_operand:VF_128 1 "vector_operand"
18757 "xBm,<round_constraint>")): Changed to ...
18758 ((match_operand:VF_128 1 "vector_operand"
18759 "xBm,<round_scalar_constraint>")): ... this.
18760 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
18761 %0, %2, %<iptr>1<round_op3>}): Changed to ...
18762 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
18763 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
18764 %<iptr>1<round_scalar_mask_op3>}): ... this.
18765 ((set_attr "prefix" "<round_prefix>")): Changed to ...
18766 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
18768 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
18771 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
18772 Cast vec_cmpeq result to correct type.
18773 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
18774 Cast vec_cmpgt result to correct type.
18776 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
18778 * final.c (final_scan_insn_1): Renamed from...
18779 (final_scan_insn): ... this. New wrapper, to recover
18780 seen from the outermost call in recursive ones.
18781 * config/sparc/sparc.c (output_return): Drop seen from call.
18782 (output_sibcall): Likewise.
18783 * config/visium/visium.c (output_branch): Likewise.
18785 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
18787 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
18790 2018-02-10 Alan Modra <amodra@gmail.com>
18793 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
18794 Specify LR as an input.
18796 2018-02-10 Jakub Jelinek <jakub@redhat.com>
18799 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
18800 remove_member_access_dummy_vars): New functions.
18801 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
18802 lower_omp_1, execute_lower_omp): Use them.
18804 PR rtl-optimization/84308
18805 * shrink-wrap.c (spread_components): Release todo vector.
18807 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
18809 PR rtl-optimization/57193
18810 * ira-color.c (struct allocno_color_data): Add member
18811 conflict_allocno_hard_prefs.
18812 (update_conflict_allocno_hard_prefs): New.
18813 (bucket_allocno_compare_func): Add a preference based on
18814 conflict_allocno_hard_prefs.
18815 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
18816 (color_allocnos): Remove a dead code. Initiate
18817 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
18819 2018-02-09 Jakub Jelinek <jakub@redhat.com>
18822 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
18823 constraint from =wa to wa. Avoid a subreg on the output operand,
18824 instead use a pseudo and subreg it in a move.
18825 (p9_xxbrd_<mode>): Changed to ...
18826 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
18827 (p9_xxbrd_v2df): New expander.
18828 (p9_xxbrw_<mode>): Changed to ...
18829 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
18830 (p9_xxbrw_v4sf): New expander.
18832 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
18834 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
18836 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
18839 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
18840 multiply in 32-bit mode.
18841 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
18842 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
18845 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
18847 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
18848 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
18849 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
18850 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
18852 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
18855 * dwarf2out.c (is_trivial_indirect_ref): New function.
18856 (dwarf2out_late_global_decl): Do not generate a location
18857 attribute for variables that have a non-trivial DECL_VALUE_EXPR
18858 and that are not defined in the current unit.
18860 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
18862 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
18863 instead of a libcall for UNORDERED.
18865 2018-02-09 Tamar Christina <tamar.christina@arm.com>
18868 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
18869 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
18871 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18874 * config/s390/s390.c (s390_set_current_function): Invoke
18875 s390_indirect_branch_settings also if fndecl didn't change.
18877 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
18879 * config/rs6000/rs6000.md (blockage): Set length to zero.
18881 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
18883 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
18885 2018-02-09 Jakub Jelinek <jakub@redhat.com>
18888 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
18889 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
18893 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
18894 PARALLEL incoming that failed vt_get_decl_and_offset check.
18896 PR middle-end/84237
18897 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
18898 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
18900 (get_variable_section): For decls in named .bss* sections pass true as
18901 second argument to bss_initializer_p.
18903 2018-02-09 Marek Polacek <polacek@redhat.com>
18904 Jakub Jelinek <jakub@redhat.com>
18907 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
18908 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
18909 Sync some changes from cxx_fold_indirect_ref.
18911 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
18913 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
18915 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
18916 (BLOCK_INLINE_ENTRY_LABEL): New.
18917 (dwarf2out_var_location): Disregard inline entry markers.
18918 (inline_entry_data): New struct.
18919 (inline_entry_data_hasher): New hashtable type.
18920 (inline_entry_data_hasher::hash): New.
18921 (inline_entry_data_hasher::equal): New.
18922 (inline_entry_data_table): New variable.
18923 (add_high_low_attributes): Add DW_AT_entry_pc and
18924 DW_AT_GNU_entry_view attributes if a pending entry is found
18925 in inline_entry_data_table. Add old entry_pc attribute only
18926 if debug nonbinding markers are disabled.
18927 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
18928 markers are enabled.
18929 (block_within_block_p, dwarf2out_inline_entry): New.
18930 (dwarf2out_finish): Check that no entries remained in
18931 inline_entry_data_table.
18932 * final.c (reemit_insn_block_notes): Handle inline entry notes.
18933 (final_scan_insn, notice_source_line): Likewise.
18934 (rest_of_clean_state): Skip inline entry markers.
18935 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
18937 * gimple.c (gimple_build_debug_inline_entry): New.
18938 * gimple.h (enum gimple_debug_subcode): Add
18939 GIMPLE_DEBUG_INLINE_ENTRY.
18940 (gimple_build_debug_inline_entry): Declare.
18941 (gimple_debug_inline_entry_p): New.
18942 (gimple_debug_nonbind_marker_p): Adjust.
18943 * insn-notes.def (INLINE_ENTRY): New.
18944 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
18945 inline entry marker notes.
18946 (print_insn): Likewise.
18947 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
18948 (INSN_DEBUG_MARKER_KIND): Likewise.
18949 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
18950 * tree-inline.c (expand_call_inline): Build and insert
18951 debug_inline_entry stmt.
18952 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
18953 inline entry blocks early, if nonbind markers are enabled.
18954 (dump_scope_block): Dump fragment info.
18955 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
18956 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
18957 (gimple_build_debug_inline_entry): New.
18958 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
18959 Enable/disable inline entry points too.
18960 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
18961 (DEBUG_INSN): Describe inline entry markers.
18963 * common.opt (gvariable-location-views): New.
18964 (gvariable-location-views=incompat5): New.
18965 * config.in: Rebuilt.
18966 * configure: Rebuilt.
18967 * configure.ac: Test assembler for view support.
18968 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
18969 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
18970 * dwarf2out.c (var_loc_view): New typedef.
18971 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
18972 (dwarf2out_locviews_in_attribute): New.
18973 (dwarf2out_locviews_in_loclist): New.
18974 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
18975 (enum dw_line_info_opcode): Add LI_adv_address.
18976 (struct dw_line_info_table): Add view.
18977 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
18978 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
18979 (zero_view_p): New variable.
18980 (ZERO_VIEW_P): New macro.
18981 (output_asm_line_debug_info): New.
18982 (struct var_loc_node): Add view.
18983 (add_AT_view_list, AT_loc_list): New.
18984 (add_var_loc_to_decl): Add view param. Test it against last.
18985 (new_loc_list): Add view params. Record them.
18986 (AT_loc_list_ptr): Handle loc and view lists.
18987 (view_list_to_loc_list_val_node): New.
18988 (print_dw_val): Handle dw_val_class_view_list.
18989 (size_of_die): Likewise.
18990 (value_format): Likewise.
18991 (loc_list_has_views): New.
18992 (gen_llsym): Set vl_symbol too.
18993 (maybe_gen_llsym, skip_loc_list_entry): New.
18994 (dwarf2out_maybe_output_loclist_view_pair): New.
18995 (output_loc_list): Output view list or entries too.
18996 (output_view_list_offset): New.
18997 (output_die): Handle dw_val_class_view_list.
18998 (output_dwarf_version): New.
18999 (output_compilation_unit_header): Use it.
19000 (output_skeleton_debug_sections): Likewise.
19001 (output_rnglists, output_line_info): Likewise.
19002 (output_pubnames, output_aranges): Update version comments.
19003 (output_one_line_info_table): Output view numbers in asm comments.
19004 (dw_loc_list): Determine current endview, pass it to new_loc_list.
19005 Call maybe_gen_llsym.
19006 (loc_list_from_tree_1): Adjust.
19007 (add_AT_location_description): Create view list attribute if
19008 needed, check it's absent otherwise.
19009 (convert_cfa_to_fb_loc_list): Adjust.
19010 (maybe_emit_file): Call output_asm_line_debug_info for test.
19011 (dwarf2out_var_location): Reset views as needed. Precompute
19012 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
19013 attribute. Set view.
19014 (new_line_info_table): Reset next view.
19015 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
19016 (dwarf2out_source_line): Likewise. Output view resets and labels to
19017 the assembler, or select appropriate line info opcodes.
19018 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
19019 (optimize_string_length): Catch it. Adjust.
19020 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
19021 dw_val_class_view_list, and remove it if no longer needed.
19022 (hash_loc_list): Hash view numbers.
19023 (loc_list_hasher::equal): Compare them.
19024 (optimize_location_lists): Check whether a view list symbol is
19025 needed, and whether the locview attribute is present, and
19026 whether they match. Remove the locview attribute if no longer
19028 (index_location_lists): Call skip_loc_list_entry for test.
19029 (dwarf2out_finish): Call output_asm_line_debug_info for test.
19030 Use output_dwarf_version.
19031 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
19032 (struct dw_val_node): Add val_view_list.
19033 * final.c (SEEN_NEXT_VIEW): New.
19034 (set_next_view_needed): New.
19035 (clear_next_view_needed): New.
19036 (maybe_output_next_view): New.
19037 (final_start_function): Rename to...
19038 (final_start_function_1): ... this. Take pointer to FIRST,
19039 add SEEN parameter. Emit param bindings in the initial view.
19040 (final_start_function): Reintroduce SEEN-less interface.
19041 (final): Rename to...
19042 (final_1): ... this. Take SEEN parameter. Output final pending
19043 next view at the end.
19044 (final): Reintroduce seen-less interface.
19045 (final_scan_insn): Output pending next view before switching
19046 sections or ending a block. Mark the next view as needed when
19047 outputting variable locations. Notify debug backend of section
19048 changes, and of location view changes.
19049 (rest_of_handle_final): Adjust.
19050 * toplev.c (process_options): Autodetect value for debug variable
19051 location views option. Warn on incompat5 without -gdwarf-5.
19052 * doc/invoke.texi (gvariable-location-views): New.
19053 (gvariable-location-views=incompat5): New.
19054 (gno-variable-location-views): New.
19056 2018-02-08 David Malcolm <dmalcolm@redhat.com>
19058 PR tree-optimization/84136
19059 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
19060 that the result of find_edge is non-NULL.
19062 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
19065 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
19066 storing integer register in SImode. Fix cost of 256 and 512
19067 byte aligned SSE register store.
19069 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
19071 * config/i386/i386.c (ix86_multiplication_cost): Fix
19072 multiplication cost for TARGET_AVX512DQ.
19074 2018-02-08 Marek Polacek <polacek@redhat.com>
19076 PR tree-optimization/84238
19077 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
19080 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19082 PR tree-optimization/84265
19083 * tree-vect-stmts.c (vectorizable_store): Don't treat
19084 VMAT_CONTIGUOUS accesses as grouped.
19085 (vectorizable_load): Likewise.
19087 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19089 PR tree-optimization/81635
19090 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
19091 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
19092 (test_round_for_mask): New functions.
19093 (wide_int_cc_tests): Call test_round_for_mask.
19094 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
19095 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
19096 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
19097 range returned by get_range_info.
19099 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
19102 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
19103 * symtab.c: Include builtins.h
19104 (symtab_node::output_to_lto_symbol_table_p): Move here
19105 from lto-streamer-out.c:output_symbol_p.
19106 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
19107 (output_symbol_p): Move all logic to symtab.c
19108 (produce_symtab): Update.
19110 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19112 * config/s390/s390-opts.h (enum indirect_branch): Define.
19113 * config/s390/s390-protos.h (s390_return_addr_from_memory)
19114 (s390_indirect_branch_via_thunk)
19115 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
19116 (enum s390_indirect_branch_type): Define.
19117 * config/s390/s390.c (struct s390_frame_layout, struct
19118 machine_function): Remove.
19119 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
19120 (indirect_branch_table_label_no, indirect_branch_table_name):
19122 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
19123 (enum s390_indirect_branch_option): Define.
19124 (s390_return_addr_from_memory): New function.
19125 (s390_handle_string_attribute): New function.
19126 (s390_attribute_table): Add new attribute handler.
19127 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
19128 (s390_indirect_branch_via_thunk): New function.
19129 (s390_indirect_branch_via_inline_thunk): New function.
19130 (s390_function_ok_for_sibcall): When jumping via thunk disallow
19131 sibling call optimization for non z10 compiles.
19132 (s390_emit_call): Force indirect branch target to be a single
19133 register. Add r1 clobber for non-z10 compiles.
19134 (s390_emit_epilogue): Emit return jump via return_use expander.
19135 (s390_reorg): Handle JUMP_INSNs as execute targets.
19136 (s390_option_override_internal): Perform validity checks for the
19137 new command line options.
19138 (s390_indirect_branch_attrvalue): New function.
19139 (s390_indirect_branch_settings): New function.
19140 (s390_set_current_function): Invoke s390_indirect_branch_settings.
19141 (s390_output_indirect_thunk_function): New function.
19142 (s390_code_end): Implement target hook.
19143 (s390_case_values_threshold): Implement target hook.
19144 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
19146 * config/s390/s390.h (struct s390_frame_layout)
19147 (struct machine_function): Move here from s390.c.
19148 (TARGET_INDIRECT_BRANCH_NOBP_RET)
19149 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
19150 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
19151 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
19152 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
19153 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
19154 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
19155 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
19156 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
19157 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
19158 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
19159 (mnemonic attribute): Add values which aren't recognized
19161 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
19162 pattern for branch conversion. Fix mnemonic attribute.
19163 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
19164 indirect branch via thunk if requested.
19165 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
19166 ("*indirect_jump"): Disable for branch conversion using out of
19168 ("indirect_jump_via_thunk<mode>_z10")
19169 ("indirect_jump_via_thunk<mode>")
19170 ("indirect_jump_via_inlinethunk<mode>_z10")
19171 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
19172 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
19173 ("casesi_jump_via_inlinethunk<mode>_z10")
19174 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
19175 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
19176 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
19177 ("*indirect2_jump"): Disable for branch conversion.
19178 ("casesi_jump"): Turn into expander and expand patterns for branch
19180 ("return_use"): New expander.
19181 ("*return"): Emit return via thunk and rename it to ...
19182 ("*return<mode>"): ... this one.
19183 * config/s390/s390.opt: Add new options and and enum for the
19186 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19188 * lra-constraints.c (match_reload): Unconditionally use
19189 gen_lowpart_SUBREG, rather than selecting between that
19190 and equivalent gen_rtx_SUBREG code.
19192 2018-02-08 Richard Biener <rguenther@suse.de>
19194 PR tree-optimization/84233
19195 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
19196 changed flag instead of boguously re-using phi_inserted.
19198 2018-02-08 Martin Jambor <mjambor@suse.cz>
19200 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
19201 static local variables.
19203 2018-02-08 Richard Biener <rguenther@suse.de>
19205 PR tree-optimization/84278
19206 * tree-vect-stmts.c (vectorizable_store): When looking for
19207 smaller vector types to perform grouped strided loads/stores
19208 make sure the mode is supported by the target.
19209 (vectorizable_load): Likewise.
19211 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
19213 * config/aarch64/aarch64.c (aarch64_components_for_bb):
19214 Increase LDP/STP opportunities by adding adjacent callee-saves.
19216 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
19218 PR rtl-optimization/84068
19219 PR rtl-optimization/83459
19220 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
19222 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
19224 PR tree-optimization/84224
19225 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
19226 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
19227 non-zero arguments.
19229 2018-02-07 Iain Sandoe <iain@codesourcery.com>
19232 * config/rs6000/altivec.md (*restore_world): Remove LR use.
19233 * config/rs6000/predicates.md (restore_world_operation): Adjust op
19234 count, remove one USE.
19236 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19238 * doc/install.texi (Configuration): Document the
19239 --with-long-double-format={ibm,ieee} PowerPC configuration
19243 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
19244 Convert from define_expand to be define_insn_and_split. Rework
19245 float/double/_Float128 conversions to QI/HI/SImode to work with
19246 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
19247 conversions to QI/HImode types did a store and then a load to
19248 truncate the value. For conversions to VSX registers, don't split
19249 the insn, instead emit the code directly. Use the code iterator
19250 any_fix to combine signed and unsigned conversions.
19251 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
19252 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
19253 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
19254 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
19255 (fix_<mode>di2_hw): Likewise.
19256 (fixuns_<mode>di2_hw): Likewise.
19257 (fix_<mode>si2_hw): Likewise.
19258 (fixuns_<mode>si2_hw): Likewise.
19259 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
19260 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
19261 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
19262 fix<uns>_trunc<SFDF:mode>si2_p8.
19263 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
19264 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19265 (fix<uns>_<mode>_mem): Likewise.
19266 (fctiw<u>z_<mode>_mem): Likewise.
19267 (fix<uns>_<mode>_mem): Likewise.
19268 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
19269 the register allocator from doing a direct move to the GPRs to do
19270 a store, and instead use the ISA 3.0 store byte/half-word from
19271 vector register instruction. For IEEE 128-bit floating point,
19272 also optimize stores of 32-bit ints.
19273 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
19275 2018-02-07 Alan Hayward <alan.hayward@arm.com>
19277 * genextract.c (push_pathstr_operand): New function to support
19279 (walk_rtx): Call push_pathstr_operand.
19280 (print_path): Support [a-zA-Z].
19282 2018-02-07 Richard Biener <rguenther@suse.de>
19284 PR tree-optimization/84037
19285 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
19286 (cse_and_gimplify_to_preheader): Declare.
19287 (vect_get_place_in_interleaving_chain): Likewise.
19288 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
19290 (_loop_vec_info::~_loop_vec_info): Delete it.
19291 (cse_and_gimplify_to_preheader): New function.
19292 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
19293 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
19294 (vectorizable_load): Likewise. For grouped stores always base
19295 the IV on the first element.
19296 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
19297 condition before gimplifying.
19299 2018-02-07 Jakub Jelinek <jakub@redhat.com>
19301 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
19302 *DIV_EXPR and *MOD_EXPR.
19304 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
19307 * config/i386/i386.c (ix86_option_override_internal): Mask out
19308 the CF_SET bit when checking -fcf-protection.
19310 2018-02-07 Tom de Vries <tom@codesourcery.com>
19313 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
19316 2018-02-07 Richard Biener <rguenther@suse.de>
19318 PR tree-optimization/84204
19319 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
19322 PR tree-optimization/84205
19323 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
19324 special-case isl_ast_op_zdiv_r.
19326 PR tree-optimization/84223
19327 * graphite-scop-detection.c (gather_bbs::before_dom_children):
19328 Only add conditions from within the region.
19329 (gather_bbs::after_dom_children): Adjust.
19331 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
19334 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
19335 * config/avr/avr.md: Only post-reload split REG-REG moves if
19336 either register is GENERAL_REG_P.
19338 2018-02-07 Jakub Jelinek <jakub@redhat.com>
19340 PR tree-optimization/84235
19341 * tree-ssa-scopedtables.c
19342 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
19343 if the subtraction is performed in floating point type where NaNs are
19344 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
19345 build 1. Formatting fix.
19347 2018-02-06 Jakub Jelinek <jakub@redhat.com>
19350 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
19351 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
19352 and skip it regardless of bb boundaries. Use CALL_P macro,
19353 don't test INSN_P (insn) together with CALL_P or JUMP_P check
19354 unnecessarily, formatting fix.
19356 2018-02-06 Michael Collison <michael.collison@arm.com>
19358 * config/arm/thumb2.md:
19359 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
19360 (*thumb_mov_notscc): Ditto.
19362 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
19365 * config/rs6000/rs6000.md (su code attribute): Use "u" for
19366 unsigned_fix, not "s".
19368 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19370 * configure.ac (gcc_fn_eh_frame_ro): New function.
19371 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
19372 correct .eh_frame permissions.
19373 * configure: Regenerate.
19375 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
19377 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
19378 irrelevant options.
19380 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19382 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19383 Display warning message for -mno-speculate-indirect-jumps.
19385 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
19387 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
19389 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
19391 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
19393 PR tree-optimization/84225
19394 * tree-eh.c (find_trapping_overflow): Only call
19395 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
19397 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
19400 * config/i386/i386.c: Reimplement the check of possible options
19401 -mibt/-mshstk conbination. Change error messages.
19402 * doc/invoke.texi: Fix a typo: remove extra '='.
19404 2018-02-06 Marek Polacek <polacek@redhat.com>
19406 PR tree-optimization/84228
19407 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
19409 2018-02-06 Tamar Christina <tamar.christina@arm.com>
19412 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
19413 emitted arch directives.
19414 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
19415 __ARM_FEATURE_COPROC before changing architectures.
19417 2018-02-06 Richard Biener <rguenther@suse.de>
19419 * config/i386/i386.c (print_reg): Fix typo.
19420 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
19422 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
19424 * configure: Regenerate.
19426 2018-02-05 Martin Sebor <msebor@redhat.com>
19428 PR tree-optimization/83369
19429 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
19432 2018-02-05 Martin Liska <mliska@suse.cz>
19434 * doc/invoke.texi: Cherry-pick upstream r323995.
19436 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
19438 * ira.c (ira_init_register_move_cost): Adjust comment.
19440 2018-02-05 Martin Liska <mliska@suse.cz>
19442 PR gcov-profile/84137
19443 * doc/gcov.texi: Fix typo in documentation.
19445 2018-02-05 Martin Liska <mliska@suse.cz>
19447 PR gcov-profile/83879
19448 * doc/gcov.texi: Document necessity of --dynamic-list-data when
19449 using dlopen functionality.
19451 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
19453 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
19454 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
19455 _mm_maskz_range_ss, _mm_mask_range_round_ss,
19456 _mm_maskz_range_round_ss): New intrinsics.
19457 (__builtin_ia32_rangesd128_round)
19458 (__builtin_ia32_rangess128_round): Remove.
19459 (__builtin_ia32_rangesd128_mask_round,
19460 __builtin_ia32_rangess128_mask_round): New builtins.
19461 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
19462 __builtin_ia32_rangess128_round): Remove.
19463 (__builtin_ia32_rangesd128_mask_round,
19464 __builtin_ia32_rangess128_mask_round): New builtins.
19465 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
19466 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
19467 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
19468 "<round_saeonly_constraint>")): Changed to ...
19469 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
19470 "<round_saeonly_scalar_constraint>")): ... this.
19471 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
19472 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
19473 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
19474 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
19475 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
19477 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
19479 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
19481 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
19482 Remove all values except native, 8540 and 8548.
19484 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
19486 * config/i386/i386.c (ix86_output_function_return): Pass
19487 INVALID_REGNUM, instead of -1, as invalid register number to
19488 indirect_thunk_name and output_indirect_thunk.
19490 2018-02-02 Julia Koval <julia.koval@intel.com>
19492 * config.gcc: Add -march=icelake.
19493 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
19494 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
19495 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
19496 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
19497 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
19498 (processor_target_table): Add icelake.
19499 (ix86_option_override_internal): Handle new PTAs.
19500 (get_builtin_code_for_version): Handle icelake.
19501 (M_INTEL_COREI7_ICELAKE): New.
19502 (fold_builtin_cpu): Handle icelake.
19503 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
19504 * doc/invoke.texi: Add -march=icelake.
19506 2018-02-02 Julia Koval <julia.koval@intel.com>
19508 * config/i386/i386.c (ix86_option_override_internal): Change flags type
19509 to wide_int_bitmask.
19510 * wide-int-bitmask.h: New.
19512 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
19515 * config/i386/i386.md: Replace Pmode with word_mode in
19516 builtin_setjmp_setup and builtin_longjmp to support x32.
19518 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
19522 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
19524 (rs6000_supported_cpu_names): New static variable.
19525 (linux_cpu_translation_table): Likewise.
19526 (elf_platform) <cpu>: Define new static variable and use it.
19527 Translate kernel AT_PLATFORM name to canonical name if needed.
19528 Error if platform name is unknown.
19530 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
19533 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
19535 2018-02-01 Jeff Law <law@redhat.com>
19538 * config/i386/i386.c (release_scratch_register_on_entry): Add new
19539 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
19540 the scratch if RELEASE_VIA_POP is false.
19541 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
19542 If we have to save a temporary register, decrement SIZE appropriately.
19543 Pass new arguments to release_scratch_register_on_entry.
19544 (ix86_adjust_stack_and_probe): Likewise.
19545 (ix86_emit_probe_stack_range): Pass new arguments to
19546 release_scratch_register_on_entry.
19548 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
19550 PR rtl-optimization/84157
19551 * combine.c (change_zero_ext): Use REG_P predicate in
19552 front of HARD_REGISTER_P predicate.
19554 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
19556 * config/avr/avr.c (avr_option_override): Move disabling of
19557 -fdelete-null-pointer-checks to...
19558 * common/config/avr/avr-common.c (avr_option_optimization_table):
19561 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19563 PR tree-optimization/81635
19564 * tree-data-ref.c (split_constant_offset_1): For types that
19565 wrap on overflow, try to use range info to prove that wrapping
19568 2018-02-01 Renlin Li <renlin.li@arm.com>
19571 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
19572 TAILCALL_ADDR_REGS.
19573 (aarch64_register_move_cost): Likewise.
19574 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
19575 TAILCALL_ADDR_REGS.
19576 (REG_CLASS_NAMES): Likewise.
19577 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
19578 TAILCALL_ADDR_REGS. Remove IP registers.
19579 * config/aarch64/aarch64.md (Ucs): Update register constraint.
19581 2018-02-01 Richard Biener <rguenther@suse.de>
19583 * domwalk.h (dom_walker::dom_walker): Add additional constructor
19584 for specifying RPO order and allow NULL for that.
19585 * domwalk.c (dom_walker::dom_walker): Likewise.
19586 (dom_walker::walk): Handle NULL RPO order.
19587 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
19589 (rewrite_update_dom_walker): Likewise.
19590 (mark_def_dom_walker): Likewise.
19592 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19594 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
19595 (aarch64_maybe_expand_sve_subreg_move): Declare.
19596 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
19597 * config/aarch64/predicates.md (aarch64_any_register_operand): New
19599 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
19600 that are semantically a reverse operation.
19601 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
19602 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
19603 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
19605 (aarch64_can_change_mode_class): For big-endian, forbid changes
19606 between two SVE modes if they have different element sizes.
19608 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19610 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
19611 the TImode handling for big-endian targets.
19613 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19615 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
19616 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
19618 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
19619 Remove BSWAP handing for big-endian targets and use the form of
19620 LD1RQ appropariate for the mode.
19622 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19624 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
19625 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
19626 duplicated element.
19628 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19631 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
19632 check for operands that need to go through aarch64_sve_reload_be.
19634 2018-02-01 Jakub Jelinek <jakub@redhat.com>
19636 PR tree-optimization/81661
19637 PR tree-optimization/84117
19638 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
19639 * tree-eh.c: Include gimplify.h.
19640 (find_trapping_overflow, replace_trapping_overflow,
19641 rewrite_to_non_trapping_overflow): New functions.
19642 * tree-vect-loop.c: Include tree-eh.h.
19643 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
19644 * tree-data-ref.c: Include tree-eh.h.
19645 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
19647 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
19649 PR rtl-optimization/84123
19650 * combine.c (change_zero_ext): Check if hard register satisfies
19651 can_change_dest_mode before calling gen_lowpart_SUBREG.
19653 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
19656 * ira.c (ira_init_register_move_cost): Remove assert.
19658 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
19660 PR rtl-optimization/84071
19661 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
19662 * doc/tm.texi: Regenerate.
19664 2018-01-31 Richard Biener <rguenther@suse.de>
19666 PR tree-optimization/84132
19667 * tree-data-ref.c (analyze_miv_subscript): Properly
19668 check whether evolution_function_is_affine_multivariate_p
19669 before calling gcd_of_steps_may_divide_p.
19671 2018-01-31 Julia Koval <julia.koval@intel.com>
19674 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
19675 * config/i386/i386.md (rdpid_rex64) New.
19676 (rdpid): Make 32bit only.
19678 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
19681 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
19682 an IDENTIFIER_NODE for FUNCTION_TYPE's.
19684 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
19687 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
19689 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
19691 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
19693 PR rtl-optimization/84071
19694 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
19695 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
19697 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
19699 * config/arc/arc.c (arc_handle_aux_attribute): New function.
19700 (arc_attribute_table): Add 'aux' attribute.
19701 (arc_in_small_data_p): Consider aux like variables.
19702 (arc_is_aux_reg_p): New function.
19703 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
19704 (arc_get_aux_arg): New function.
19705 (prepare_move_operands): Handle aux-register access.
19706 (arc_handle_aux_attribute): New function.
19707 * doc/extend.texi (ARC Variable attributes): Add subsection.
19709 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
19711 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
19712 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
19713 (arc_attribute_table): Add 'uncached' attribute.
19714 (arc_print_operand): Print '.di' flag for uncached memory
19716 (arc_in_small_data_p): Do not consider for small data the uncached
19718 (arc_is_uncached_mem_p): New function.
19719 * config/arc/predicates.md (compact_store_memory_operand): Check
19720 for uncached memory accesses.
19721 (nonvol_nonimm_operand): Likewise.
19722 * doc/extend.texi (ARC Type Attribute): New subsection.
19724 2018-01-31 Jakub Jelinek <jakub@redhat.com>
19727 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
19728 falign-loops=): Add Optimization flag.
19730 2018-01-30 Jeff Law <law@redhat.com>
19733 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
19734 INT_REGISTERS_SAVED. Check it prior to calling
19735 get_scratch_register_on_entry.
19736 (ix86_adjust_stack_and_probe): Similarly.
19737 (ix86_emit_probe_stack_range): Similarly.
19738 (ix86_expand_prologue): Corresponding changes.
19740 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19743 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
19744 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
19746 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
19749 * lra-constraints.c (curr_insn_transform): Process AND in the
19752 2018-01-30 Jakub Jelinek <jakub@redhat.com>
19754 PR rtl-optimization/83986
19755 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
19756 dependence against last_pending_memory_flush in addition to
19757 pending_jump_insns.
19759 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
19761 PR tree-optimization/81611
19762 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
19765 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19768 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
19771 2018-01-30 Richard Biener <rguenther@suse.de>
19772 Jakub Jelinek <jakub@redhat.com>
19774 PR tree-optimization/84111
19775 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
19776 inner loops added during recursion, as they don't have up-to-date
19779 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
19782 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
19783 (can_inline_edge_by_limits_p): ... here.
19784 (can_early_inline_edge_p, check_callers,
19785 update_caller_keys, update_callee_keys, recursive_inlining,
19786 add_new_edges_to_heap, speculation_useful_p,
19787 inline_small_functions,
19788 inline_small_functions, flatten_function,
19789 inline_to_all_callers_1): Update.
19791 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
19793 * profile-count.c (profile_count::combine_with_ipa_count): Handle
19796 2018-01-30 Richard Biener <rguenther@suse.de>
19798 PR tree-optimization/83008
19799 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
19800 invariant and constant vector uses in stmts when they need
19801 more than one stmt.
19803 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19806 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
19807 * configure: Regenerate.
19809 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
19811 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
19813 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
19814 Use gen_rtx_REG rather than gen_lowpart.
19816 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
19818 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
19819 rather than 0 when creating partial subregs.
19821 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
19823 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
19826 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19829 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
19830 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
19831 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
19832 flags. This restores the settings used before the 2017-07-24.
19833 Turning off pre increment/decrement/modify allows IVOPTS to
19834 optimize DF/SF loops where the index is an int.
19836 2018-01-29 Richard Biener <rguenther@suse.de>
19837 Kelvin Nilsen <kelvin@gcc.gnu.org>
19840 * tree-vect-stmts.c (vectorizable_call): Don't call
19841 targetm.vectorize_builtin_md_vectorized_function if callee is
19844 2018-01-22 Carl Love <cel@us.ibm.com>
19846 * doc/extend.tex: Fix typo in second arg in
19847 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
19849 2018-01-29 Richard Biener <rguenther@suse.de>
19851 PR tree-optimization/84086
19852 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
19853 (flush_ssaname_freelist): When SSA names were released reset
19854 the SCEV hash table.
19856 2018-01-29 Richard Biener <rguenther@suse.de>
19858 PR tree-optimization/84057
19859 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
19860 removed paths when removing edges.
19862 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
19864 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
19865 -mfunction-return=@var{choice}.
19867 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
19869 PR diagnostic/84034
19870 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
19871 Handle CR like TAB.
19872 (layout::print_source_line): Likewise.
19873 (test_get_line_width_without_trailing_whitespace): Add test cases.
19875 2018-01-27 Jakub Jelinek <jakub@redhat.com>
19877 PR middle-end/84040
19878 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
19881 2018-01-26 Jim Wilson <jimw@sifive.com>
19883 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
19885 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
19888 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19890 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
19891 and CMP + SUB-immediate -> SUBS.
19893 2018-01-26 Martin Sebor <msebor@redhat.com>
19895 PR tree-optimization/83896
19896 * tree-ssa-strlen.c (get_string_len): Rename...
19897 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
19898 Avoid assuming length is constant.
19899 (handle_char_store): Use HOST_WIDE_INT for string length.
19901 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
19904 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
19905 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
19907 2018-01-26 Richard Biener <rguenther@suse.de>
19909 PR rtl-optimization/84003
19910 * dse.c (record_store): Only record redundant stores when
19911 the earlier store aliases at least all accesses the later one does.
19913 2018-01-26 Jakub Jelinek <jakub@redhat.com>
19915 PR rtl-optimization/83985
19916 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
19917 REG_CFA_RESTORE insns.
19918 (delete_unmarked_insns): Don't ignore separate shrink wrapping
19919 REG_CFA_RESTORE insns here.
19922 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
19923 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
19925 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
19927 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
19928 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
19929 (arc_init): Likewise.
19930 (arc_override_options): Likewise.
19931 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
19933 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
19935 * config/arc/arc.h (TARGET_DBNZ): Define.
19936 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
19937 properly set the tune attribute.
19938 (dbnz): Use TARGET_DBNZ guard.
19939 * config/arc/arc.opt (mtune): Add core3 option.
19941 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
19943 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
19944 recognize new pic like addresses.
19945 (arc_delegitimize_address): Clean up.
19947 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
19949 * config/arc/arc-arches.def: Option mrf16 valid for all
19951 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
19952 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
19953 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
19954 * config/arc/arc-tables.opt: Regenerate.
19955 * config/arc/arc.c (arc_conditional_register_usage): Handle
19956 reduced register file case.
19957 (arc_file_start): Set must have build attributes.
19958 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
19959 mrf16 option value.
19960 * config/arc/arc.opt (mrf16): Add new option.
19961 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
19962 * config/arc/genmultilib.awk: Handle new mrf16 option.
19963 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
19964 * config/arc/t-multilib: Regenerate.
19965 * doc/invoke.texi (ARC Options): Document mrf16 option.
19967 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
19969 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
19970 * config/arc/arc.c (arc_handle_secure_attribute): New function.
19971 (arc_attribute_table): Add 'secure_call' attribute.
19972 (arc_print_operand): Print secure call operand.
19973 (arc_function_ok_for_sibcall): Don't optimize tail calls when
19975 (arc_is_secure_call_p): New function. * config/arc/arc.md
19976 (call_i): Add support for sjli instruction.
19977 (call_value_i): Likewise.
19978 * config/arc/constraints.md (Csc): New constraint.
19980 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
19981 John Eric Martin <John.Martin@emmicro-us.com>
19983 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
19984 * config/arc/arc.c (_arc_jli_section): New struct.
19985 (arc_jli_section): New type.
19986 (rc_jli_sections): New static variable.
19987 (arc_handle_jli_attribute): New function.
19988 (arc_attribute_table): Add jli_always and jli_fixed attribute.
19989 (arc_file_end): New function.
19990 (TARGET_ASM_FILE_END): Define.
19991 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
19992 (arc_add_jli_section): New function.
19993 (jli_call_scan): Likewise.
19994 (arc_reorg): Call jli_call_scan.
19995 (arc_output_addsi): Remove 'S' from printing asm operand.
19996 (arc_is_jli_call_p): New function.
19997 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
19999 (movhi_insn): Likewise.
20000 (movsi_insn): Likewise.
20001 (movsi_set_cc_insn): Likewise.
20002 (loadqi_update): Likewise.
20003 (load_zeroextendqisi_update): Likewise.
20004 (load_signextendqisi_update): Likewise.
20005 (loadhi_update): Likewise.
20006 (load_zeroextendhisi_update): Likewise.
20007 (load_signextendhisi_update): Likewise.
20008 (loadsi_update): Likewise.
20009 (loadsf_update): Likewise.
20010 (movsicc_insn): Likewise.
20011 (bset_insn): Likewise.
20012 (bxor_insn): Likewise.
20013 (bclr_insn): Likewise.
20014 (bmsk_insn): Likewise.
20015 (bicsi3_insn): Likewise.
20016 (cmpsi_cc_c_insn): Likewise.
20017 (movsi_ne): Likewise.
20018 (movsi_cond_exec): Likewise.
20019 (clrsbsi2): Likewise.
20020 (norm_f): Likewise.
20030 (call_i): Remove 'S' asm letter, add jli instruction.
20031 (call_value_i): Likewise.
20032 * config/arc/arc.op (mjli-always): New option.
20033 * config/arc/constraints.md (Cji): New constraint.
20034 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
20036 (subsf3_fpx): Likewise.
20037 (mulsf3_fpx): Likewise.
20038 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
20040 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
20041 function attrbutes.
20042 * doc/invoke.texi (ARC): Document mjli-always option.
20044 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
20046 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
20047 avoid addition with 0 and use incw and decw where possible.
20049 2018-01-26 Richard Biener <rguenther@suse.de>
20051 PR tree-optimization/81082
20052 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
20053 association if it requires casting to unsigned.
20054 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
20055 from fold_plusminus_mult_expr to catch important cases late when
20056 range info is available.
20058 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20060 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
20061 * configure.ac (hidden_linkonce): New test.
20062 * configure: Regenerate.
20063 * config.in: Regenerate.
20065 2018-01-26 Julia Koval <julia.koval@intel.com>
20067 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
20068 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
20069 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
20070 _mm_mask_bitshuffle_epi64_mask): Fix type.
20071 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
20072 USI_FTYPE_V4DI_V4DI_USI): Remove.
20073 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
20074 __builtin_ia32_vpshufbitqmb256_mask,
20075 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
20076 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
20077 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
20079 2018-01-26 Alan Modra <amodra@gmail.com>
20082 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
20083 UNSPEC_VBPERMQ. Sort other unspecs.
20085 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
20087 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
20089 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
20091 PR middle-end/83055
20092 * predict.c (drop_profile): Do not push/pop cfun; update also
20094 (handle_missing_profiles): Fix logic looking for zero profiles.
20096 2018-01-25 Jakub Jelinek <jakub@redhat.com>
20098 PR middle-end/83977
20099 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
20100 on functions with #pragma omp declare simd or functions with simd
20102 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
20103 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
20104 Remove trailing \n from warning_at calls.
20106 2018-01-25 Tom de Vries <tom@codesourcery.com>
20109 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
20110 for neutered workers.
20112 2018-01-24 Joseph Myers <joseph@codesourcery.com>
20115 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
20116 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
20118 2018-01-24 Jeff Law <law@redhat.com>
20121 * i386.c (get_probe_interval): Move to earlier point.
20122 (ix86_compute_frame_layout): If -fstack-clash-protection and
20123 the frame is larger than the probe interval, then use pushes
20124 to save registers rather than reg->mem moves.
20125 (ix86_expand_prologue): Remove conditional for int_registers_saved
20128 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
20131 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
20132 min/max for never referenced object.
20134 2018-01-24 Jakub Jelinek <jakub@redhat.com>
20136 PR middle-end/83977
20137 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
20139 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
20140 attributes from DECL_ATTRIBUTES (decl) without affecting
20141 DECL_ATTRIBUTES (current_function_decl).
20142 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
20143 functions with non-NULL DECL_ABSTRACT_ORIGIN.
20145 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
20147 PR tree-optimization/83979
20148 * fold-const.c (fold_comparison): Use constant_boolean_node
20149 instead of boolean_{true,false}_node.
20151 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
20153 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
20156 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20158 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
20159 Simplify the clause that sets the length attribute.
20160 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
20161 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
20162 clause that sets the length attribute.
20163 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
20165 2018-01-24 Tom de Vries <tom@codesourcery.com>
20168 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
20169 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
20170 Add strict parameter.
20171 (prevent_branch_around_nothing): Insert dummy insn between branch to
20172 label and label with no ptx insn inbetween.
20173 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
20175 2018-01-24 Tom de Vries <tom@codesourcery.com>
20178 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
20179 for neutered threads in warp.
20180 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
20182 2018-01-24 Richard Biener <rguenther@suse.de>
20184 PR tree-optimization/83176
20185 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
20188 2018-01-24 Richard Biener <rguenther@suse.de>
20190 PR tree-optimization/82819
20191 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
20192 code generating pluses that are no-ops in the target precision.
20194 2018-01-24 Richard Biener <rguenther@suse.de>
20196 PR middle-end/84000
20197 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
20199 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
20201 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
20202 to merge probabilities.
20203 * predict.c (probably_never_executed): Also mark as cold functions
20204 with global 0 profile and guessed local profile.
20205 * profile-count.c (profile_probability::combine_with_count): New
20207 * profile-count.h (profile_probability::operator*,
20208 profile_probability::operator*=, profile_probability::operator/,
20209 profile_probability::operator/=): Reduce precision to adjusted
20210 and set value to guessed on contradictory divisions.
20211 (profile_probability::combine_with_freq): Remove.
20212 (profile_probability::combine_wiht_count): Declare.
20213 (profile_count::force_nonzero):: Set to adjusted.
20214 (profile_count::probability_in):: Set quality to adjusted.
20215 * tree-ssa-tail-merge.c (replace_block_by): Use
20216 combine_with_count.
20218 2018-01-23 Andrew Waterman <andrew@sifive.com>
20219 Jim Wilson <jimw@sifive.com>
20221 * config/riscv/riscv.c (riscv_stack_boundary): New.
20222 (riscv_option_override): Set riscv_stack_boundary. Handle
20223 riscv_preferred_stack_boundary_arg.
20224 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
20225 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
20226 (STACK_BOUNDARY): Set to riscv_stack_boundary.
20227 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
20228 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
20229 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
20231 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
20234 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
20235 of struct ix86_frame.
20236 (ix86_expand_epilogue): Likewise. Add a local variable for
20237 the reg_save_offset field in struct ix86_frame.
20239 2018-01-23 Bin Cheng <bin.cheng@arm.com>
20241 PR tree-optimization/82604
20242 * tree-loop-distribution.c (enum partition_kind): New enum item
20243 PKIND_PARTIAL_MEMSET.
20244 (partition_builtin_p): Support above new enum item.
20245 (generate_code_for_partition): Ditto.
20246 (compute_access_range): Differentiate cases that equality can be
20247 proven at all loops, the innermost loops or no loops.
20248 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
20249 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
20250 (finalize_partitions, distribute_loop): Don't fuse partition of
20251 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
20252 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
20253 parloop is enabled.
20255 2018-01-23 Martin Liska <mliska@suse.cz>
20257 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
20258 order to ignore the predictor.
20259 (PRED_POLYMORPHIC_CALL): Likewise.
20260 (PRED_RECURSIVE_CALL): Likewise.
20262 2018-01-23 Martin Liska <mliska@suse.cz>
20264 * tree-profile.c (tree_profiling): Print function header to
20265 aware reader which function we are working on.
20266 * value-prof.c (gimple_find_values_to_profile): Do not print
20267 not interesting value histograms.
20269 2018-01-23 Martin Liska <mliska@suse.cz>
20271 * profile-count.h (enum profile_quality): Add
20272 profile_uninitialized as the first value. Do not number values
20273 as they are zero based.
20274 (profile_count::verify): Update sanity check.
20275 (profile_probability::verify): Likewise.
20277 2018-01-23 Nathan Sidwell <nathan@acm.org>
20279 * doc/invoke.texi (ffor-scope): Deprecate.
20281 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20283 PR tree-optimization/83510
20284 * domwalk.c (set_all_edges_as_executable): New function.
20285 (dom_walker::dom_walker): Convert bool param
20286 "skip_unreachable_blocks" to enum reachability. Move setup of
20287 edge flags to set_all_edges_as_executable and only do it when
20288 reachability is REACHABLE_BLOCKS.
20289 * domwalk.h (enum dom_walker::reachability): New enum.
20290 (dom_walker::dom_walker): Convert bool param
20291 "skip_unreachable_blocks" to enum reachability.
20292 (set_all_edges_as_executable): New decl.
20293 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
20294 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
20296 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
20297 but converting true to REACHABLE_BLOCKS.
20298 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
20300 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
20301 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
20302 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
20304 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
20305 if check_all_array_refs will be called.
20307 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20309 * tree.c (selftest::test_location_wrappers): Add more test
20312 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20314 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
20315 (selftest::test_bit_in_range): Likewise.
20317 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
20320 * doc/sourcebuild.texi (vect_float): Say that the selector
20321 only describes the situation when -funsafe-math-optimizations is on.
20322 (vect_float_strict): Document.
20324 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
20326 PR tree-optimization/83965
20327 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
20328 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
20329 instead of checking only for a reduction.
20330 (vect_recog_widen_sum_pattern): Likewise.
20332 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
20334 * predict.c (probably_never_executed): Only use precise profile info.
20335 (compute_function_frequency): Skip after inlining hack since we now
20336 have quality checking.
20338 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
20340 * profile-count.h (profile_probability::very_unlikely,
20341 profile_probability::unlikely, profile_probability::even): Set
20342 precision to guessed.
20344 2018-01-23 Richard Biener <rguenther@suse.de>
20346 PR tree-optimization/83963
20347 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
20348 Properly terminate dominator walk when crossing the exit edge not
20349 when visiting its source block.
20351 2018-01-23 Jakub Jelinek <jakub@redhat.com>
20354 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
20355 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
20357 2018-01-22 Jakub Jelinek <jakub@redhat.com>
20359 PR tree-optimization/83957
20360 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
20361 semicolon after for body surrounded by braces.
20363 PR tree-optimization/83081
20364 * profile-count.h (profile_probability::split): New method.
20365 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
20366 Use profile_probability::split.
20367 (do_compare_rtx_and_jump): Fix adjustment of probabilities
20368 when splitting a single conditional jump into 2.
20370 2018-01-22 David Malcolm <dmalcolm@redhat.com>
20372 PR tree-optimization/69452
20373 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
20376 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20378 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
20379 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
20380 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
20382 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20384 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
20386 * config/rl78/rl78.md (movdi): New define_expand.
20387 * config/rl78/rl78.c (rl78_split_movdi): New function.
20389 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
20392 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
20394 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
20395 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
20396 128-bit to produce an UNSPEC move to get the double word with the
20397 signbit and then a shift directly to do signbit.
20398 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
20399 implementation with a new version that just does either a direct
20400 move or a regular move. Move memory interface to separate insns.
20401 Move insns so they are next to the expander.
20402 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
20403 with signbit move. Split big and little endian case.
20404 (signbit<mode>2_dm_mem_le): Likewise.
20405 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
20406 (signbit<mode>2_dm2): Likewise.
20408 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20410 * config/rl78/rl78.md (anddi3): New define_expand.
20412 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20414 * config/rl78/rl78.md (umindi3): New define_expand.
20416 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20418 * config/rl78/rl78.md (smindi3): New define_expand.
20420 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20422 * config/rl78/rl78.md (smaxdi3): New define_expand.
20424 2018-01-22 Carl Love <cel@us.ibm.com>
20426 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
20427 LVX_V1TI): Add macro expansion.
20428 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
20429 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
20430 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
20431 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
20432 Change check to determine if the instruction is a byte reversing
20433 entry. Fix typo in comment.
20434 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
20435 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
20436 Add def_builtin calls for new builtins.
20437 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
20438 Add define_insn expansion.
20440 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20442 * config/rl78/rl78.md (umaxdi3): New define_expand.
20444 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20446 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
20447 for non-QImode registers.
20449 2018-01-22 Richard Biener <rguenther@suse.de>
20451 PR tree-optimization/83963
20452 * graphite-scop-detection.c (scop_detection::get_sese): Delay
20453 including the loop exit block.
20454 (scop_detection::merge_sese): Likewise.
20455 (scop_detection::add_scop): Do it here instead.
20457 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20459 * doc/sourcebuild.texi (arm_softfloat): Document.
20461 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
20464 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
20465 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
20466 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
20468 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20469 David Edelsohn <dje.gcc@gmail.com>
20472 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
20473 Change "crset eq" to "crset 2".
20474 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
20475 (*call_indirect_aix<mode>_nospec): Likewise.
20476 (*call_value_indirect_aix<mode>_nospec): Likewise.
20477 (*call_indirect_elfv2<mode>_nospec): Likewise.
20478 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
20479 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
20480 change assembly output from . to $.
20481 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
20482 (indirect_jump<mode>_nospec): Change assembly output from . to $.
20483 (*tablejump<mode>_internal1_nospec): Likewise.
20485 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
20488 * config/sh/sh_optimize_sett_clrt.cc:
20489 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
20491 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
20493 PR tree-optimization/83940
20494 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
20495 offset_dt to vect_constant_def rather than vect_unknown_def_type.
20496 (vect_check_load_store_mask): Add a mask_dt_out parameter and
20497 use it to pass back the definition type.
20498 (vect_check_store_rhs): Likewise rhs_dt_out.
20499 (vect_build_gather_load_calls): Add a mask_dt argument and use
20500 it instead of a call to vect_is_simple_use.
20501 (vectorizable_store): Update calls to vect_check_load_store_mask
20502 and vect_check_store_rhs. Use the dt returned by the latter instead
20503 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
20504 instead of calls to vect_is_simple_use. Pass the scalar rather
20505 than the vector operand to vect_is_simple_use when handling
20506 second and subsequent copies of an rhs value.
20507 (vectorizable_load): Update calls to vect_check_load_store_mask
20508 and vect_build_gather_load_calls. Use the cached mask_dt and
20509 gs_info.offset_dt instead of calls to vect_is_simple_use.
20511 2018-01-20 Jakub Jelinek <jakub@redhat.com>
20513 PR middle-end/83945
20514 * tree-emutls.c: Include gimplify.h.
20515 (lower_emutls_2): New function.
20516 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
20517 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
20518 it before further processing.
20521 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
20522 UINTVAL (trueop1) instead of INTVAL (op1).
20524 2018-01-19 Jakub Jelinek <jakub@redhat.com>
20528 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
20529 INCOMING_FRAME_SP_OFFSET if not defined.
20530 (scan_trace): Add ENTRY argument. If true and
20531 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
20532 emit a note to adjust the CFA offset.
20533 (create_cfi_notes): Adjust scan_trace callers.
20534 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
20535 INCOMING_FRAME_SP_OFFSET in the CIE.
20536 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
20537 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
20539 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
20540 * doc/tm.texi: Regenerated.
20542 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20544 PR rtl-optimization/83147
20545 * lra-constraints.c (remove_inheritance_pseudos): Use
20546 lra_substitute_pseudo_within_insn.
20548 2018-01-19 Tom de Vries <tom@codesourcery.com>
20549 Cesar Philippidis <cesar@codesourcery.com>
20552 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
20554 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
20557 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
20558 spaces for function labels.
20560 2018-01-19 Martin Liska <mliska@suse.cz>
20562 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
20563 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
20564 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
20565 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
20566 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
20567 (PRED_CONST_RETURN): Change from 69 to 65.
20568 (PRED_NULL_RETURN): Change from 91 to 71.
20569 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
20570 (PRED_LOOP_GUARD): Change from 66 to 73.
20572 2018-01-19 Martin Liska <mliska@suse.cz>
20574 * predict.c (predict_insn_def): Add new assert.
20575 (struct branch_predictor): Change type to signed integer.
20576 (test_prediction_value_range): Amend test to cover
20577 PROB_UNINITIALIZED.
20578 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
20579 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
20580 (PRED_LOOP_ITERATIONS_MAX): Likewise.
20581 (PRED_LOOP_IV_COMPARE): Likewise.
20582 * predict.h (PROB_UNINITIALIZED): Define new constant.
20584 2018-01-19 Martin Liska <mliska@suse.cz>
20586 * predict.c (dump_prediction): Add new format for
20587 analyze_brprob.py script which is enabled with -details
20589 * profile-count.h (precise_p): New function.
20591 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
20593 PR tree-optimization/83922
20594 * tree-vect-loop.c (vect_verify_full_masking): Return false if
20595 there are no statements that need masking.
20596 (vect_active_double_reduction_p): New function.
20597 (vect_analyze_loop_operations): Use it when handling phis that
20598 are not in the loop header.
20600 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
20602 PR tree-optimization/83914
20603 * tree-vect-loop.c (vectorizable_induction): Don't convert
20604 init_expr or apply the peeling adjustment for inductions
20605 that are nested within the vectorized loop.
20607 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20609 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
20612 2018-01-18 Jakub Jelinek <jakub@redhat.com>
20616 * function.h (gimplify_parameters): Add gimple_seq * argument.
20617 * function.c: Include gimple.h and options.h.
20618 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
20619 for the added local temporaries if needed.
20620 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
20621 if there are any parameter cleanups, wrap whole body into a
20622 try/finally with the cleanups.
20624 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
20627 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20628 Use GET_MODE_CLASS for scalar floating point.
20630 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
20634 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
20635 Fix call of call_cgraph_insertion_hooks.
20637 2018-01-18 Martin Sebor <msebor@redhat.com>
20639 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
20641 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
20644 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
20647 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
20650 * common.opt: (-ffile-prefix-map): New option.
20651 * opts.c (common_handle_option): Defer it.
20652 * opts-global.c (handle_common_deferred_options): Handle it.
20653 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
20654 * file-prefix-map.h: New file.
20655 (remap_debug_filename, add_debug_prefix_map): ...here.
20656 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
20657 * final.c (debug_prefix_map, add_debug_prefix_map
20658 remap_debug_filename): Move to...
20659 * file-prefix-map.c: New file.
20660 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
20661 generalize, get rid of alloca(), use strrchr() instead of strchr().
20662 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
20663 Implement in terms of add_prefix_map().
20664 (remap_macro_filename, remap_debug_filename): Implement in term of
20666 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
20667 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
20668 * dbxout.c: Include file-prefix-map.h.
20669 * varasm.c: Likewise.
20670 * vmsdbgout.c: Likewise.
20671 * xcoffout.c: Likewise.
20672 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
20673 * doc/cppopts.texi (-fmacro-prefix-map): Document.
20674 * doc/invoke.texi (-ffile-prefix-map): Document.
20675 (-fdebug-prefix-map): Update description.
20677 2018-01-18 Martin Liska <mliska@suse.cz>
20679 * config/i386/i386.c (indirect_thunk_name): Document that also
20681 (output_indirect_thunk): Document why both instructions
20682 (pause and lfence) are generated.
20684 2018-01-18 Richard Biener <rguenther@suse.de>
20686 PR tree-optimization/83887
20687 * graphite-scop-detection.c
20688 (scop_detection::get_nearest_dom_with_single_entry): Remove.
20689 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
20690 (scop_detection::merge_sese): Re-implement with a flood-fill
20691 algorithm that properly finds a SESE region if it exists.
20693 2018-01-18 Jakub Jelinek <jakub@redhat.com>
20696 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
20697 pointer_diff optimizations use view_convert instead of convert.
20699 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20701 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
20702 Generate different code for -mno-speculate-indirect-jumps.
20703 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
20704 (*call_indirect_aix<mode>): Disable for
20705 -mno-speculate-indirect-jumps.
20706 (*call_indirect_aix<mode>_nospec): New define_insn.
20707 (*call_value_indirect_aix<mode>): Disable for
20708 -mno-speculate-indirect-jumps.
20709 (*call_value_indirect_aix<mode>_nospec): New define_insn.
20710 (*sibcall_nonlocal_sysv<mode>): Generate different code for
20711 -mno-speculate-indirect-jumps.
20712 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
20714 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
20716 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
20717 long double type, set the flags for noting the default long double
20718 type, even if we don't pass or return a long double type.
20720 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
20723 * ipa-inline.c (flatten_function): Do not overwrite final inlining
20726 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
20728 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
20729 support for merge[hl].
20730 (fold_mergehl_helper): New helper function.
20731 (tree-vector-builder.h): New #include for tree_vector_builder usage.
20732 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
20733 (altivec_vmrglw_direct): Add xxmrglw insn.
20735 2018-01-17 Andrew Waterman <andrew@sifive.com>
20737 * config/riscv/riscv.c (riscv_conditional_register_usage): If
20738 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
20740 2018-01-17 David Malcolm <dmalcolm@redhat.com>
20743 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
20744 call the lto_location_cache before reading the
20745 DECL_SOURCE_LOCATION of the types.
20747 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
20748 Richard Sandiford <richard.sandiford@linaro.org>
20750 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
20751 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
20752 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
20753 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
20754 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
20756 * config/aarch64/constraints.md (aarch64_movti_operand):
20758 * config/aarch64/predicates.md (Uti): Add new constraint.
20760 2018-01-17 Carl Love <cel@us.ibm.com>
20762 * config/rs6000/vsx.md (define_expand xl_len_r,
20763 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
20764 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
20766 (define_expand, define_insn): Move the shift left from the
20767 define_insn to the define_expand for lxvl and stxvl instructions.
20768 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
20769 and XL_LEN_R definitions to PURE.
20771 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
20773 * config/i386/i386.c (indirect_thunk_name): Declare regno
20774 as unsigned int. Compare regno with INVALID_REGNUM.
20775 (output_indirect_thunk): Ditto.
20776 (output_indirect_thunk_function): Ditto.
20777 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
20778 in the call to output_indirect_thunk_function.
20780 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
20782 PR middle-end/83884
20783 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
20784 rather than the size of inner_type to determine the stack slot size
20785 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
20787 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
20790 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
20793 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
20795 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
20796 endian Linux systems to optionally enable multilibs for selecting
20797 the long double type if the user configured an explicit type.
20798 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
20799 have no long double multilibs if not defined.
20800 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
20801 warn if the user used -mabi={ieee,ibm}longdouble and we built
20802 multilibs for long double.
20803 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
20804 appropriate multilib option.
20805 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
20807 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
20808 for building long double multilibs.
20809 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
20811 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
20813 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
20816 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
20818 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
20821 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
20824 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
20827 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
20829 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
20830 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
20832 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
20834 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
20835 different rtl trees depending on TARGET_64BIT.
20836 (rs6000_gen_lvx): Likewise.
20838 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
20840 * config/visium/visium.md (nop): Tweak comment.
20841 (hazard_nop): Likewise.
20843 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20845 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
20846 -mspeculate-indirect-jumps.
20847 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
20848 for -mno-speculate-indirect-jumps.
20849 (*call_indirect_elfv2<mode>_nospec): New define_insn.
20850 (*call_value_indirect_elfv2<mode>): Disable for
20851 -mno-speculate-indirect-jumps.
20852 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
20853 (indirect_jump): Emit different RTL for
20854 -mno-speculate-indirect-jumps.
20855 (*indirect_jump<mode>): Disable for
20856 -mno-speculate-indirect-jumps.
20857 (*indirect_jump<mode>_nospec): New define_insn.
20858 (tablejump): Emit different RTL for
20859 -mno-speculate-indirect-jumps.
20860 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
20861 (tablejumpsi_nospec): New define_expand.
20862 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
20863 (tablejumpdi_nospec): New define_expand.
20864 (*tablejump<mode>_internal1): Disable for
20865 -mno-speculate-indirect-jumps.
20866 (*tablejump<mode>_internal1_nospec): New define_insn.
20867 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
20870 2018-01-16 Artyom Skrobov tyomitch@gmail.com
20872 * caller-save.c (insert_save): Drop unnecessary parameter. All
20875 2018-01-16 Jakub Jelinek <jakub@redhat.com>
20876 Richard Biener <rguenth@suse.de>
20879 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
20880 return early, inline manually is_gimple_sizepos. Make sure if we
20881 call gimplify_expr we don't end up with a gimple constant.
20882 * tree.c (variably_modified_type_p): Don't return true for
20883 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
20884 * gimplify.h (is_gimple_sizepos): Remove.
20886 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
20888 PR tree-optimization/83857
20889 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
20890 vectorizable_live_operation for pure SLP statements.
20891 (vectorizable_live_operation): Handle PHIs.
20893 2018-01-16 Richard Biener <rguenther@suse.de>
20895 PR tree-optimization/83867
20896 * tree-vect-stmts.c (vect_transform_stmt): Precompute
20897 nested_in_vect_loop_p since the scalar stmt may get invalidated.
20899 2018-01-16 Jakub Jelinek <jakub@redhat.com>
20902 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
20903 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
20904 If off is not INTEGER_CST, issue a may not be aligned warning
20905 rather than isn't aligned. Use isn%'t rather than isn't.
20906 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
20908 <case MULT_EXPR>: Improve the case when bottom and one of the
20909 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
20910 operand, in that case check if the other operand is multiple of
20911 bottom divided by the INTEGER_CST operand.
20913 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
20916 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
20917 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
20918 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
20919 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
20920 * config/pa/pa.c (pa_function_arg_advance): Likewise.
20921 (pa_function_arg, pa_arg_partial_bytes): Likewise.
20922 (pa_function_arg_size): New function.
20924 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
20926 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
20927 in a separate statement.
20929 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
20931 PR tree-optimization/83847
20932 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
20933 group gathers and scatters.
20935 2018-01-16 Jakub Jelinek <jakub@redhat.com>
20937 PR rtl-optimization/86620
20938 * params.def (max-sched-ready-insns): Bump minimum value to 1.
20940 PR rtl-optimization/83213
20941 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
20942 to last if both are JUMP_INSNs.
20944 PR tree-optimization/83843
20945 * gimple-ssa-store-merging.c
20946 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
20947 store_immediate_info for bswap/nop orig_stores.
20949 2018-01-15 Andrew Waterman <andrew@sifive.com>
20951 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
20953 <UDIV>: Increase cost if !TARGET_DIV.
20955 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
20957 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
20958 (define_attr "cr_logical_3op"): New.
20959 (cceq_ior_compare): Adjust.
20960 (cceq_ior_compare_complement): Adjust.
20961 (*cceq_rev_compare): Adjust.
20962 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
20963 (is_cracked_insn): Adjust.
20964 (insn_must_be_first_in_group): Adjust.
20965 * config/rs6000/40x.md: Adjust.
20966 * config/rs6000/440.md: Adjust.
20967 * config/rs6000/476.md: Adjust.
20968 * config/rs6000/601.md: Adjust.
20969 * config/rs6000/603.md: Adjust.
20970 * config/rs6000/6xx.md: Adjust.
20971 * config/rs6000/7450.md: Adjust.
20972 * config/rs6000/7xx.md: Adjust.
20973 * config/rs6000/8540.md: Adjust.
20974 * config/rs6000/cell.md: Adjust.
20975 * config/rs6000/e300c2c3.md: Adjust.
20976 * config/rs6000/e500mc.md: Adjust.
20977 * config/rs6000/e500mc64.md: Adjust.
20978 * config/rs6000/e5500.md: Adjust.
20979 * config/rs6000/e6500.md: Adjust.
20980 * config/rs6000/mpc.md: Adjust.
20981 * config/rs6000/power4.md: Adjust.
20982 * config/rs6000/power5.md: Adjust.
20983 * config/rs6000/power6.md: Adjust.
20984 * config/rs6000/power7.md: Adjust.
20985 * config/rs6000/power8.md: Adjust.
20986 * config/rs6000/power9.md: Adjust.
20987 * config/rs6000/rs64.md: Adjust.
20988 * config/rs6000/titan.md: Adjust.
20990 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
20992 * config/i386/predicates.md (indirect_branch_operand): Rewrite
20993 ix86_indirect_branch_register logic.
20995 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
20997 * config/i386/constraints.md (Bs): Update
20998 ix86_indirect_branch_register check. Don't check
20999 ix86_indirect_branch_register with GOT_memory_operand.
21001 * config/i386/predicates.md (GOT_memory_operand): Don't check
21002 ix86_indirect_branch_register here.
21003 (GOT32_symbol_operand): Likewise.
21005 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21007 * config/i386/predicates.md (constant_call_address_operand):
21008 Rewrite ix86_indirect_branch_register logic.
21009 (sibcall_insn_operand): Likewise.
21011 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21013 * config/i386/constraints.md (Bs): Replace
21014 ix86_indirect_branch_thunk_register with
21015 ix86_indirect_branch_register.
21017 * config/i386/i386.md (indirect_jump): Likewise.
21018 (tablejump): Likewise.
21019 (*sibcall_memory): Likewise.
21020 (*sibcall_value_memory): Likewise.
21021 Peepholes of indirect call and jump via memory: Likewise.
21022 * config/i386/i386.opt: Likewise.
21023 * config/i386/predicates.md (indirect_branch_operand): Likewise.
21024 (GOT_memory_operand): Likewise.
21025 (call_insn_operand): Likewise.
21026 (sibcall_insn_operand): Likewise.
21027 (GOT32_symbol_operand): Likewise.
21029 2018-01-15 Jakub Jelinek <jakub@redhat.com>
21031 PR middle-end/83837
21032 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
21033 type rather than type addr's type points to.
21034 (expand_omp_atomic_mutex): Likewise.
21035 (expand_omp_atomic): Likewise.
21037 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21040 * config/i386/i386.c (output_indirect_thunk_function): Use
21041 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
21042 for __x86_return_thunk.
21044 2018-01-15 Richard Biener <rguenther@suse.de>
21046 PR middle-end/83850
21047 * expmed.c (extract_bit_field_1): Fix typo.
21049 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21052 * config/arm/iterators.md (VF): New mode iterator.
21053 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
21054 Remove integer-related logic from pattern.
21055 (neon_vabd<mode>_3): Likewise.
21057 2018-01-15 Jakub Jelinek <jakub@redhat.com>
21059 PR middle-end/82694
21060 * common.opt (fstrict-overflow): No longer an alias.
21061 (fwrapv-pointer): New option.
21062 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
21063 also for pointer types based on flag_wrapv_pointer.
21064 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
21065 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
21066 opts->x_flag_wrapv got set.
21067 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
21068 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
21069 POINTER_TYPE_OVERFLOW_UNDEFINED.
21070 * match.pd: Likewise in address comparison pattern.
21071 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
21073 2018-01-15 Richard Biener <rguenther@suse.de>
21076 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
21077 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
21078 Reset type names to their identifier if their TYPE_DECL doesn't
21079 have linkage (and thus is used for ODR and devirt).
21080 (save_debug_info_for_decl): Remove.
21081 (save_debug_info_for_type): Likewise.
21082 (add_tree_to_fld_list): Adjust.
21083 * tree-pretty-print.c (dump_generic_node): Make dumping of
21084 type names more robust.
21086 2018-01-15 Richard Biener <rguenther@suse.de>
21088 * BASE-VER: Bump to 8.0.1.
21090 2018-01-14 Martin Sebor <msebor@redhat.com>
21093 * builtins.c (check_access): Avoid warning when the no-warning bit
21096 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
21098 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
21099 * ira-color (allocno_hard_regs_compare): Likewise.
21101 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
21104 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
21105 Use .pushsection/.popsection.
21107 2018-01-14 Martin Sebor <msebor@redhat.com>
21110 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
21112 2018-01-14 Jakub Jelinek <jakub@redhat.com>
21114 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
21115 entry from extra_headers.
21116 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
21117 extra_headers, make the list bitwise identical to the i?86-*-* one.
21119 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21121 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
21122 -mcmodel=large with -mindirect-branch=thunk,
21123 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
21124 -mfunction-return=thunk-extern.
21125 * doc/invoke.texi: Document -mcmodel=large is incompatible with
21126 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
21127 -mfunction-return=thunk and -mfunction-return=thunk-extern.
21129 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21131 * config/i386/i386.c (print_reg): Print the name of the full
21132 integer register without '%'.
21133 (ix86_print_operand): Handle 'V'.
21134 * doc/extend.texi: Document 'V' modifier.
21136 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21138 * config/i386/constraints.md (Bs): Disallow memory operand for
21139 -mindirect-branch-register.
21141 * config/i386/predicates.md (indirect_branch_operand): Likewise.
21142 (GOT_memory_operand): Likewise.
21143 (call_insn_operand): Likewise.
21144 (sibcall_insn_operand): Likewise.
21145 (GOT32_symbol_operand): Likewise.
21146 * config/i386/i386.md (indirect_jump): Call convert_memory_address
21147 for -mindirect-branch-register.
21148 (tablejump): Likewise.
21149 (*sibcall_memory): Likewise.
21150 (*sibcall_value_memory): Likewise.
21151 Disallow peepholes of indirect call and jump via memory for
21152 -mindirect-branch-register.
21153 (*call_pop): Replace m with Bw.
21154 (*call_value_pop): Likewise.
21155 (*sibcall_pop_memory): Replace m with Bs.
21156 * config/i386/i386.opt (mindirect-branch-register): New option.
21157 * doc/invoke.texi: Document -mindirect-branch-register option.
21159 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21161 * config/i386/i386-protos.h (ix86_output_function_return): New.
21162 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
21163 set function_return_type.
21164 (indirect_thunk_name): Add ret_p to indicate thunk for function
21166 (output_indirect_thunk_function): Pass false to
21167 indirect_thunk_name.
21168 (ix86_output_indirect_branch_via_reg): Likewise.
21169 (ix86_output_indirect_branch_via_push): Likewise.
21170 (output_indirect_thunk_function): Create alias for function
21171 return thunk if regno < 0.
21172 (ix86_output_function_return): New function.
21173 (ix86_handle_fndecl_attribute): Handle function_return.
21174 (ix86_attribute_table): Add function_return.
21175 * config/i386/i386.h (machine_function): Add
21176 function_return_type.
21177 * config/i386/i386.md (simple_return_internal): Use
21178 ix86_output_function_return.
21179 (simple_return_internal_long): Likewise.
21180 * config/i386/i386.opt (mfunction-return=): New option.
21181 (indirect_branch): Mention -mfunction-return=.
21182 * doc/extend.texi: Document function_return function attribute.
21183 * doc/invoke.texi: Document -mfunction-return= option.
21185 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21187 * config/i386/i386-opts.h (indirect_branch): New.
21188 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
21189 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
21190 with local indirect jump when converting indirect call and jump.
21191 (ix86_set_indirect_branch_type): New.
21192 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
21193 (indirectlabelno): New.
21194 (indirect_thunk_needed): Likewise.
21195 (indirect_thunk_bnd_needed): Likewise.
21196 (indirect_thunks_used): Likewise.
21197 (indirect_thunks_bnd_used): Likewise.
21198 (INDIRECT_LABEL): Likewise.
21199 (indirect_thunk_name): Likewise.
21200 (output_indirect_thunk): Likewise.
21201 (output_indirect_thunk_function): Likewise.
21202 (ix86_output_indirect_branch_via_reg): Likewise.
21203 (ix86_output_indirect_branch_via_push): Likewise.
21204 (ix86_output_indirect_branch): Likewise.
21205 (ix86_output_indirect_jmp): Likewise.
21206 (ix86_code_end): Call output_indirect_thunk_function if needed.
21207 (ix86_output_call_insn): Call ix86_output_indirect_branch if
21209 (ix86_handle_fndecl_attribute): Handle indirect_branch.
21210 (ix86_attribute_table): Add indirect_branch.
21211 * config/i386/i386.h (machine_function): Add indirect_branch_type
21212 and has_local_indirect_jump.
21213 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
21215 (tablejump): Likewise.
21216 (*indirect_jump): Use ix86_output_indirect_jmp.
21217 (*tablejump_1): Likewise.
21218 (simple_return_indirect_internal): Likewise.
21219 * config/i386/i386.opt (mindirect-branch=): New option.
21220 (indirect_branch): New.
21223 (thunk-inline): Likewise.
21224 (thunk-extern): Likewise.
21225 * doc/extend.texi: Document indirect_branch function attribute.
21226 * doc/invoke.texi: Document -mindirect-branch= option.
21228 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
21231 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
21233 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
21235 * ipa-inline.c (want_inline_small_function_p): Return false if
21236 inlining has already failed with CIF_FINAL_ERROR.
21237 (update_caller_keys): Call want_inline_small_function_p before
21239 (update_callee_keys): Likewise.
21241 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21243 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
21245 (rs6000_quadword_masked_address_p): Likewise.
21246 (quad_aligned_load_p): Likewise.
21247 (quad_aligned_store_p): Likewise.
21248 (const_load_sequence_p): Add comment to describe the outer-most loop.
21249 (mimic_memory_attributes_and_flags): New function.
21250 (rs6000_gen_stvx): Likewise.
21251 (replace_swapped_aligned_store): Likewise.
21252 (rs6000_gen_lvx): Likewise.
21253 (replace_swapped_aligned_load): Likewise.
21254 (replace_swapped_load_constant): Capitalize argument name in
21255 comment describing this function.
21256 (rs6000_analyze_swaps): Add a third pass to search for vector loads
21257 and stores that access quad-word aligned addresses and replace
21258 with stvx or lvx instructions when appropriate.
21259 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
21260 New function prototype.
21261 (rs6000_quadword_masked_address_p): Likewise.
21262 (rs6000_gen_lvx): Likewise.
21263 (rs6000_gen_stvx): Likewise.
21264 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
21265 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
21266 when memory address is aligned.
21267 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
21268 this split to select lvx instruction when memory address is aligned.
21269 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
21270 instruction when memory address is aligned.
21271 (*vsx_le_perm_load_v16qi): Likewise.
21272 (four unnamed splitters): Modify to select the stvx instruction
21273 when memory is aligned.
21275 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
21277 * predict.c (determine_unlikely_bbs): Handle correctly BBs
21278 which appears in the queue multiple times.
21280 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21281 Alan Hayward <alan.hayward@arm.com>
21282 David Sherwood <david.sherwood@arm.com>
21284 * tree-vectorizer.h (vec_lower_bound): New structure.
21285 (_loop_vec_info): Add check_nonzero and lower_bounds.
21286 (LOOP_VINFO_CHECK_NONZERO): New macro.
21287 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
21288 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
21289 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
21290 fields. Make seg_len the distance travelled, not including the
21292 (dr_direction_indicator): Declare.
21293 (dr_zero_step_indicator): Likewise.
21294 (dr_known_forward_stride_p): Likewise.
21295 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
21297 (runtime_alias_check_p): Allow runtime alias checks with
21299 (operator ==): Compare access_size and align.
21300 (prune_runtime_alias_test_list): Rework for new distinction between
21301 the access_size and seg_len.
21302 (create_intersect_range_checks_index): Likewise. Cope with polynomial
21304 (get_segment_min_max): New function.
21305 (create_intersect_range_checks): Use it.
21306 (dr_step_indicator): New function.
21307 (dr_direction_indicator): Likewise.
21308 (dr_zero_step_indicator): Likewise.
21309 (dr_known_forward_stride_p): Likewise.
21310 * tree-loop-distribution.c (data_ref_segment_size): Return
21311 DR_STEP * (niters - 1).
21312 (compute_alias_check_pairs): Update call to the dr_with_seg_len
21314 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
21315 (vect_preserves_scalar_order_p): New function, split out from...
21316 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
21317 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
21318 (vect_vfa_access_size): New function.
21319 (vect_vfa_align): Likewise.
21320 (vect_compile_time_alias): Take access_size_a and access_b arguments.
21321 (dump_lower_bound): New function.
21322 (vect_check_lower_bound): Likewise.
21323 (vect_small_gap_p): Likewise.
21324 (vectorizable_with_step_bound_p): Likewise.
21325 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
21326 depencies if the vectorization factor is 1. Convert the checks
21327 for nonzero steps into checks on the bounds of DR_STEP. Try using
21328 a bunds check for variable steps if the minimum required step is
21329 relatively small. Update calls to the dr_with_seg_len
21330 constructor and to vect_compile_time_alias.
21331 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
21333 (vect_loop_versioning): Call it.
21334 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
21336 (vect_estimate_min_profitable_iters): Account for any bounds checks.
21338 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21339 Alan Hayward <alan.hayward@arm.com>
21340 David Sherwood <david.sherwood@arm.com>
21342 * doc/sourcebuild.texi (vect_scatter_store): Document.
21343 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
21345 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
21347 * genopinit.c (main): Add supports_vec_scatter_store and
21348 supports_vec_scatter_store_cached to target_optabs.
21349 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
21350 IFN_MASK_SCATTER_STORE.
21351 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
21353 * internal-fn.h (internal_store_fn_p): Declare.
21354 (internal_fn_stored_value_index): Likewise.
21355 * internal-fn.c (scatter_store_direct): New macro.
21356 (expand_scatter_store_optab_fn): New function.
21357 (direct_scatter_store_optab_supported_p): New macro.
21358 (internal_store_fn_p): New function.
21359 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
21360 IFN_MASK_SCATTER_STORE.
21361 (internal_fn_mask_index): Likewise.
21362 (internal_fn_stored_value_index): New function.
21363 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
21364 for scatter stores.
21365 * optabs-query.h (supports_vec_scatter_store_p): Declare.
21366 * optabs-query.c (supports_vec_scatter_store_p): New function.
21367 * tree-vectorizer.h (vect_get_store_rhs): Declare.
21368 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
21369 true for scatter stores.
21370 (vect_gather_scatter_fn_p): Handle scatter stores too.
21371 (vect_check_gather_scatter): Consider using scatter stores if
21372 supports_vec_scatter_store_p.
21373 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
21374 scatter stores too.
21375 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
21376 internal_fn_stored_value_index.
21377 (check_load_store_masking): Handle scatter stores too.
21378 (vect_get_store_rhs): Make public.
21379 (vectorizable_call): Use internal_store_fn_p.
21380 (vectorizable_store): Handle scatter store internal functions.
21381 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
21382 when deciding whether the end of the group has been reached.
21383 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
21384 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
21385 (mask_scatter_store<mode>): New insns.
21387 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21388 Alan Hayward <alan.hayward@arm.com>
21389 David Sherwood <david.sherwood@arm.com>
21391 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
21392 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
21393 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
21395 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
21396 Use vect_truncate_gather_scatter_offset if we can't treat the
21397 operation as a normal gather load or scatter store.
21398 (get_group_load_store_type): Take the gather_scatter_info
21399 as argument. Try using a gather load or scatter store for
21400 single-element groups.
21401 (get_load_store_type): Update calls to get_group_load_store_type
21402 and vect_use_strided_gather_scatters_p.
21404 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21405 Alan Hayward <alan.hayward@arm.com>
21406 David Sherwood <david.sherwood@arm.com>
21408 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
21409 optional tree argument.
21410 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
21412 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
21413 but continue to use the current value as a fallback.
21414 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
21415 to compare the updates.
21416 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
21417 (get_load_store_type): Use it when handling a strided access.
21418 (vect_get_strided_load_store_ops): New function.
21419 (vect_get_data_ptr_increment): Likewise.
21420 (vectorizable_load): Handle strided gather loads. Always pass
21421 a step to vect_create_data_ref_ptr and bump_vector_ptr.
21423 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21424 Alan Hayward <alan.hayward@arm.com>
21425 David Sherwood <david.sherwood@arm.com>
21427 * doc/md.texi (gather_load@var{m}): Document.
21428 (mask_gather_load@var{m}): Likewise.
21429 * genopinit.c (main): Add supports_vec_gather_load and
21430 supports_vec_gather_load_cached to target_optabs.
21431 * optabs-tree.c (init_tree_optimization_optabs): Use
21432 ggc_cleared_alloc to allocate target_optabs.
21433 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
21434 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
21436 * internal-fn.h (internal_load_fn_p): Declare.
21437 (internal_gather_scatter_fn_p): Likewise.
21438 (internal_fn_mask_index): Likewise.
21439 (internal_gather_scatter_fn_supported_p): Likewise.
21440 * internal-fn.c (gather_load_direct): New macro.
21441 (expand_gather_load_optab_fn): New function.
21442 (direct_gather_load_optab_supported_p): New macro.
21443 (direct_internal_fn_optab): New function.
21444 (internal_load_fn_p): Likewise.
21445 (internal_gather_scatter_fn_p): Likewise.
21446 (internal_fn_mask_index): Likewise.
21447 (internal_gather_scatter_fn_supported_p): Likewise.
21448 * optabs-query.c (supports_at_least_one_mode_p): New function.
21449 (supports_vec_gather_load_p): Likewise.
21450 * optabs-query.h (supports_vec_gather_load_p): Declare.
21451 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
21452 and memory_type field.
21453 (NUM_PATTERNS): Bump to 15.
21454 * tree-vect-data-refs.c: Include internal-fn.h.
21455 (vect_gather_scatter_fn_p): New function.
21456 (vect_describe_gather_scatter_call): Likewise.
21457 (vect_check_gather_scatter): Try using internal functions for
21458 gather loads. Recognize existing calls to a gather load function.
21459 (vect_analyze_data_refs): Consider using gather loads if
21460 supports_vec_gather_load_p.
21461 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
21462 (vect_get_gather_scatter_offset_type): Likewise.
21463 (vect_convert_mask_for_vectype): Likewise.
21464 (vect_add_conversion_to_patterm): Likewise.
21465 (vect_try_gather_scatter_pattern): Likewise.
21466 (vect_recog_gather_scatter_pattern): New pattern recognizer.
21467 (vect_vect_recog_func_ptrs): Add it.
21468 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
21469 internal_fn_mask_index and internal_gather_scatter_fn_p.
21470 (check_load_store_masking): Take the gather_scatter_info as an
21471 argument and handle gather loads.
21472 (vect_get_gather_scatter_ops): New function.
21473 (vectorizable_call): Check internal_load_fn_p.
21474 (vectorizable_load): Likewise. Handle gather load internal
21476 (vectorizable_store): Update call to check_load_store_masking.
21477 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
21478 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
21479 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
21480 (aarch64_gather_scale_operand_d): New predicates.
21481 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
21482 (mask_gather_load<mode>): New insns.
21484 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21485 Alan Hayward <alan.hayward@arm.com>
21486 David Sherwood <david.sherwood@arm.com>
21488 * optabs.def (fold_left_plus_optab): New optab.
21489 * doc/md.texi (fold_left_plus_@var{m}): Document.
21490 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
21491 * internal-fn.c (fold_left_direct): Define.
21492 (expand_fold_left_optab_fn): Likewise.
21493 (direct_fold_left_optab_supported_p): Likewise.
21494 * fold-const-call.c (fold_const_fold_left): New function.
21495 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
21496 * tree-parloops.c (valid_reduction_p): New function.
21497 (gather_scalar_reductions): Use it.
21498 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
21499 (vect_finish_replace_stmt): Declare.
21500 * tree-vect-loop.c (fold_left_reduction_fn): New function.
21501 (needs_fold_left_reduction_p): New function, split out from...
21502 (vect_is_simple_reduction): ...here. Accept reductions that
21503 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
21504 (vect_force_simple_reduction): Also store the reduction type in
21505 the assignment's STMT_VINFO_REDUC_TYPE.
21506 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
21507 (merge_with_identity): New function.
21508 (vect_expand_fold_left): Likewise.
21509 (vectorize_fold_left_reduction): Likewise.
21510 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
21511 scalar phi in place for it. Check for target support and reject
21512 cases that would reassociate the operation. Defer the transform
21513 phase to vectorize_fold_left_reduction.
21514 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
21515 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
21516 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
21518 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21520 * tree-if-conv.c (predicate_mem_writes): Remove redundant
21521 call to ifc_temp_var.
21523 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21524 Alan Hayward <alan.hayward@arm.com>
21525 David Sherwood <david.sherwood@arm.com>
21527 * target.def (legitimize_address_displacement): Take the original
21528 offset as a poly_int.
21529 * targhooks.h (default_legitimize_address_displacement): Update
21531 * targhooks.c (default_legitimize_address_displacement): Likewise.
21532 * doc/tm.texi: Regenerate.
21533 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
21534 as an argument, moving assert of ad->disp == ad->disp_term to...
21535 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
21536 Try calling targetm.legitimize_address_displacement before expanding
21537 the address rather than afterwards, and adjust for the new interface.
21538 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
21539 Match the new hook interface. Handle SVE addresses.
21540 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
21541 new hook interface.
21543 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21545 * Makefile.in (OBJS): Add early-remat.o.
21546 * target.def (select_early_remat_modes): New hook.
21547 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
21548 * doc/tm.texi: Regenerate.
21549 * targhooks.h (default_select_early_remat_modes): Declare.
21550 * targhooks.c (default_select_early_remat_modes): New function.
21551 * timevar.def (TV_EARLY_REMAT): New timevar.
21552 * passes.def (pass_early_remat): New pass.
21553 * tree-pass.h (make_pass_early_remat): Declare.
21554 * early-remat.c: New file.
21555 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
21557 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
21559 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21560 Alan Hayward <alan.hayward@arm.com>
21561 David Sherwood <david.sherwood@arm.com>
21563 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
21564 vfm1 with a bound_epilog parameter.
21565 (vect_do_peeling): Update calls accordingly, and move the prologue
21566 call earlier in the function. Treat the base bound_epilog as 0 for
21567 fully-masked loops and retain vf - 1 for other loops. Add 1 to
21568 this base when peeling for gaps.
21569 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
21570 with fully-masked loops.
21571 (vect_estimate_min_profitable_iters): Handle the single peeled
21572 iteration in that case.
21574 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21575 Alan Hayward <alan.hayward@arm.com>
21576 David Sherwood <david.sherwood@arm.com>
21578 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
21579 single-element interleaving even if the size is not a power of 2.
21580 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
21581 accesses for single-element interleaving if the group size is
21584 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21585 Alan Hayward <alan.hayward@arm.com>
21586 David Sherwood <david.sherwood@arm.com>
21588 * doc/md.texi (fold_extract_last_@var{m}): Document.
21589 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
21590 * optabs.def (fold_extract_last_optab): New optab.
21591 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
21592 * internal-fn.c (fold_extract_direct): New macro.
21593 (expand_fold_extract_optab_fn): Likewise.
21594 (direct_fold_extract_optab_supported_p): Likewise.
21595 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
21596 * tree-vect-loop.c (vect_model_reduction_cost): Handle
21597 EXTRACT_LAST_REDUCTION.
21598 (get_initial_def_for_reduction): Do not create an initial vector
21599 for EXTRACT_LAST_REDUCTION reductions.
21600 (vectorizable_reduction): Leave the scalar phi in place for
21601 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
21602 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
21603 epilogue code for EXTRACT_LAST_REDUCTION and defer the
21604 transform phase to vectorizable_condition.
21605 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
21607 (vect_finish_stmt_generation): ...here.
21608 (vect_finish_replace_stmt): New function.
21609 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
21610 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
21612 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
21614 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21615 Alan Hayward <alan.hayward@arm.com>
21616 David Sherwood <david.sherwood@arm.com>
21618 * doc/md.texi (extract_last_@var{m}): Document.
21619 * optabs.def (extract_last_optab): New optab.
21620 * internal-fn.def (EXTRACT_LAST): New internal function.
21621 * internal-fn.c (cond_unary_direct): New macro.
21622 (expand_cond_unary_optab_fn): Likewise.
21623 (direct_cond_unary_optab_supported_p): Likewise.
21624 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
21625 loops using EXTRACT_LAST.
21626 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
21627 (extract_last_<mode>): ...this optab.
21628 (vec_extract<mode><Vel>): Update accordingly.
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 * target.def (empty_mask_is_expensive): New hook.
21635 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
21636 * doc/tm.texi: Regenerate.
21637 * targhooks.h (default_empty_mask_is_expensive): Declare.
21638 * targhooks.c (default_empty_mask_is_expensive): New function.
21639 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
21640 if the target says that empty masks are expensive.
21641 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
21643 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
21645 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21646 Alan Hayward <alan.hayward@arm.com>
21647 David Sherwood <david.sherwood@arm.com>
21649 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
21650 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
21651 (vect_use_loop_mask_for_alignment_p): New function.
21652 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
21653 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
21654 niters_skip argument. Make sure that the first niters_skip elements
21655 of the first iteration are inactive.
21656 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
21657 Update call to vect_set_loop_masks_directly.
21658 (get_misalign_in_elems): New function, split out from...
21659 (vect_gen_prolog_loop_niters): ...here.
21660 (vect_update_init_of_dr): Take a code argument that specifies whether
21661 the adjustment should be added or subtracted.
21662 (vect_update_init_of_drs): Likewise.
21663 (vect_prepare_for_masked_peels): New function.
21664 (vect_do_peeling): Skip prologue peeling if we're using a mask
21665 instead. Update call to vect_update_inits_of_drs.
21666 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
21668 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
21669 alignment. Do not include the number of peeled iterations in
21670 the minimum threshold in that case.
21671 (vectorizable_induction): Adjust the start value down by
21672 LOOP_VINFO_MASK_SKIP_NITERS iterations.
21673 (vect_transform_loop): Call vect_prepare_for_masked_peels.
21674 Take the number of skipped iterations into account when calculating
21676 * tree-vect-stmts.c (vect_gen_while_not): New function.
21678 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21679 Alan Hayward <alan.hayward@arm.com>
21680 David Sherwood <david.sherwood@arm.com>
21682 * doc/sourcebuild.texi (vect_fully_masked): Document.
21683 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
21684 default value to 0.
21685 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
21687 (vect_analyze_loop_2): ...here. Don't check the vectorization
21688 factor against the number of loop iterations if the loop is
21691 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21692 Alan Hayward <alan.hayward@arm.com>
21693 David Sherwood <david.sherwood@arm.com>
21695 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
21696 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
21697 (dump_groups): Update accordingly.
21698 (iv_use::mem_type): New member variable.
21699 (address_p): New function.
21700 (record_use): Add a mem_type argument and initialize the new
21702 (record_group_use): Add a mem_type argument. Use address_p.
21703 Remove obsolete null checks of base_object. Update call to record_use.
21704 (find_interesting_uses_op): Update call to record_group_use.
21705 (find_interesting_uses_cond): Likewise.
21706 (find_interesting_uses_address): Likewise.
21707 (get_mem_type_for_internal_fn): New function.
21708 (find_address_like_use): Likewise.
21709 (find_interesting_uses_stmt): Try find_address_like_use before
21710 calling find_interesting_uses_op.
21711 (addr_offset_valid_p): Use the iv mem_type field as the type
21712 of the addressed memory.
21713 (add_autoinc_candidates): Likewise.
21714 (get_address_cost): Likewise.
21715 (split_small_address_groups_p): Use address_p.
21716 (split_address_groups): Likewise.
21717 (add_iv_candidate_for_use): Likewise.
21718 (autoinc_possible_for_pair): Likewise.
21719 (rewrite_groups): Likewise.
21720 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
21721 (determine_group_iv_cost): Update after split of USE_ADDRESS.
21722 (get_alias_ptr_type_for_ptr_address): New function.
21723 (rewrite_use_address): Rewrite address uses in calls that were
21724 identified by find_address_like_use.
21726 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21727 Alan Hayward <alan.hayward@arm.com>
21728 David Sherwood <david.sherwood@arm.com>
21730 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
21732 * gimple-expr.h (is_gimple_addressable: Likewise.
21733 * gimple-expr.c (is_gimple_address): Likewise.
21734 * internal-fn.c (expand_call_mem_ref): New function.
21735 (expand_mask_load_optab_fn): Use it.
21736 (expand_mask_store_optab_fn): Likewise.
21738 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21739 Alan Hayward <alan.hayward@arm.com>
21740 David Sherwood <david.sherwood@arm.com>
21742 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
21743 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
21744 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
21745 (cond_umax@var{mode}): Document.
21746 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
21747 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
21748 (cond_umin_optab, cond_umax_optab): New optabs.
21749 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
21750 (COND_IOR, COND_XOR): New internal functions.
21751 * internal-fn.h (get_conditional_internal_fn): Declare.
21752 * internal-fn.c (cond_binary_direct): New macro.
21753 (expand_cond_binary_optab_fn): Likewise.
21754 (direct_cond_binary_optab_supported_p): Likewise.
21755 (get_conditional_internal_fn): New function.
21756 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
21757 Cope with reduction statements that are vectorized as calls rather
21759 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
21760 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
21761 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
21762 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
21763 (UNSPEC_COND_EOR): New unspecs.
21764 (optab): Add mappings for them.
21765 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
21766 (sve_int_op, sve_fp_op): New int attributes.
21768 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21769 Alan Hayward <alan.hayward@arm.com>
21770 David Sherwood <david.sherwood@arm.com>
21772 * optabs.def (while_ult_optab): New optab.
21773 * doc/md.texi (while_ult@var{m}@var{n}): Document.
21774 * internal-fn.def (WHILE_ULT): New internal function.
21775 * internal-fn.h (direct_internal_fn_supported_p): New override
21776 that takes two types as argument.
21777 * internal-fn.c (while_direct): New macro.
21778 (expand_while_optab_fn): New function.
21779 (convert_optab_supported_p): Likewise.
21780 (direct_while_optab_supported_p): New macro.
21781 * wide-int.h (wi::udiv_ceil): New function.
21782 * tree-vectorizer.h (rgroup_masks): New structure.
21783 (vec_loop_masks): New typedef.
21784 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
21785 and fully_masked_p.
21786 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
21787 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
21788 (vect_max_vf): New function.
21789 (slpeel_make_loop_iterate_ntimes): Delete.
21790 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
21791 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
21792 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
21793 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
21794 internal-fn.h, stor-layout.h and optabs-query.h.
21795 (vect_set_loop_mask): New function.
21796 (add_preheader_seq): Likewise.
21797 (add_header_seq): Likewise.
21798 (interleave_supported_p): Likewise.
21799 (vect_maybe_permute_loop_masks): Likewise.
21800 (vect_set_loop_masks_directly): Likewise.
21801 (vect_set_loop_condition_masked): Likewise.
21802 (vect_set_loop_condition_unmasked): New function, split out from
21803 slpeel_make_loop_iterate_ntimes.
21804 (slpeel_make_loop_iterate_ntimes): Rename to..
21805 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
21806 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
21807 (vect_do_peeling): Update call accordingly.
21808 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
21810 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
21811 mask_compare_type, can_fully_mask_p and fully_masked_p.
21812 (release_vec_loop_masks): New function.
21813 (_loop_vec_info): Use it to free the loop masks.
21814 (can_produce_all_loop_masks_p): New function.
21815 (vect_get_max_nscalars_per_iter): Likewise.
21816 (vect_verify_full_masking): Likewise.
21817 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
21818 retries, and free the mask rgroups before retrying. Check loop-wide
21819 reasons for disallowing fully-masked loops. Make the final decision
21820 about whether use a fully-masked loop or not.
21821 (vect_estimate_min_profitable_iters): Do not assume that peeling
21822 for the number of iterations will be needed for fully-masked loops.
21823 (vectorizable_reduction): Disable fully-masked loops.
21824 (vectorizable_live_operation): Likewise.
21825 (vect_halve_mask_nunits): New function.
21826 (vect_double_mask_nunits): Likewise.
21827 (vect_record_loop_mask): Likewise.
21828 (vect_get_loop_mask): Likewise.
21829 (vect_transform_loop): Handle the case in which the final loop
21830 iteration might handle a partial vector. Call vect_set_loop_condition
21831 instead of slpeel_make_loop_iterate_ntimes.
21832 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
21833 (check_load_store_masking): New function.
21834 (prepare_load_store_mask): Likewise.
21835 (vectorizable_store): Handle fully-masked loops.
21836 (vectorizable_load): Likewise.
21837 (supportable_widening_operation): Use vect_halve_mask_nunits for
21839 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
21840 (vect_gen_while): New function.
21841 * config/aarch64/aarch64.md (umax<mode>3): New expander.
21842 (aarch64_uqdec<mode>): New insn.
21844 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21845 Alan Hayward <alan.hayward@arm.com>
21846 David Sherwood <david.sherwood@arm.com>
21848 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
21849 (reduc_xor_scal_optab): New optabs.
21850 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
21851 (reduc_xor_scal_@var{m}): Document.
21852 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
21853 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
21854 internal functions.
21855 * fold-const-call.c (fold_const_call): Handle them.
21856 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
21857 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
21858 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
21859 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
21860 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
21861 (UNSPEC_XORV): New unspecs.
21862 (optab): Add entries for them.
21863 (BITWISEV): New int iterator.
21864 (bit_reduc_op): New int attributes.
21866 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21867 Alan Hayward <alan.hayward@arm.com>
21868 David Sherwood <david.sherwood@arm.com>
21870 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
21871 * internal-fn.def (VEC_SHL_INSERT): New internal function.
21872 * optabs.def (vec_shl_insert_optab): New optab.
21873 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
21874 (duplicate_and_interleave): Likewise.
21875 * tree-vect-loop.c: Include internal-fn.h.
21876 (neutral_op_for_slp_reduction): New function, split out from
21877 get_initial_defs_for_reduction.
21878 (get_initial_def_for_reduction): Handle option 2 for variable-length
21879 vectors by loading the neutral value into a vector and then shifting
21880 the initial value into element 0.
21881 (get_initial_defs_for_reduction): Replace the code argument with
21882 the neutral value calculated by neutral_op_for_slp_reduction.
21883 Use gimple_build_vector for constant-length vectors.
21884 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
21885 but the first group_size elements have a neutral value.
21886 Use duplicate_and_interleave otherwise.
21887 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
21888 Update call to get_initial_defs_for_reduction. Handle SLP
21889 reductions for variable-length vectors by creating one vector
21890 result for each scalar result, with the elements associated
21891 with other scalar results stubbed out with the neutral value.
21892 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
21893 Require IFN_VEC_SHL_INSERT for double reductions on
21894 variable-length vectors, or SLP reductions that have
21895 a neutral value. Require can_duplicate_and_interleave_p
21896 support for variable-length unchained SLP reductions if there
21897 is no neutral value, such as for MIN/MAX reductions. Also require
21898 the number of vector elements to be a multiple of the number of
21899 SLP statements when doing variable-length unchained SLP reductions.
21900 Update call to vect_create_epilog_for_reduction.
21901 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
21902 and remove initial values.
21903 (duplicate_and_interleave): Make public.
21904 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
21905 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
21907 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21908 Alan Hayward <alan.hayward@arm.com>
21909 David Sherwood <david.sherwood@arm.com>
21911 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
21912 (can_duplicate_and_interleave_p): New function.
21913 (vect_get_and_check_slp_defs): Take the vector of statements
21914 rather than just the current one. Remove excess parentheses.
21915 Restriction rejectinon of vect_constant_def and vect_external_def
21916 for variable-length vectors to boolean types, or types for which
21917 can_duplicate_and_interleave_p is false.
21918 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
21919 (duplicate_and_interleave): New function.
21920 (vect_get_constant_vectors): Use gimple_build_vector for
21921 constant-length vectors and suitable variable-length constant
21922 vectors. Use duplicate_and_interleave for other variable-length
21923 vectors. Don't defer the update when inserting new statements.
21925 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21926 Alan Hayward <alan.hayward@arm.com>
21927 David Sherwood <david.sherwood@arm.com>
21929 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
21930 min_profitable_iters doesn't go negative.
21932 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21933 Alan Hayward <alan.hayward@arm.com>
21934 David Sherwood <david.sherwood@arm.com>
21936 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
21937 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
21938 * optabs.def (vec_mask_load_lanes_optab): New optab.
21939 (vec_mask_store_lanes_optab): Likewise.
21940 * internal-fn.def (MASK_LOAD_LANES): New internal function.
21941 (MASK_STORE_LANES): Likewise.
21942 * internal-fn.c (mask_load_lanes_direct): New macro.
21943 (mask_store_lanes_direct): Likewise.
21944 (expand_mask_load_optab_fn): Handle masked operations.
21945 (expand_mask_load_lanes_optab_fn): New macro.
21946 (expand_mask_store_optab_fn): Handle masked operations.
21947 (expand_mask_store_lanes_optab_fn): New macro.
21948 (direct_mask_load_lanes_optab_supported_p): Likewise.
21949 (direct_mask_store_lanes_optab_supported_p): Likewise.
21950 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
21952 (vect_load_lanes_supported): Likewise.
21953 * tree-vect-data-refs.c (strip_conversion): New function.
21954 (can_group_stmts_p): Likewise.
21955 (vect_analyze_data_ref_accesses): Use it instead of checking
21956 for a pair of assignments.
21957 (vect_store_lanes_supported): Take a masked_p parameter.
21958 (vect_load_lanes_supported): Likewise.
21959 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
21960 vect_store_lanes_supported and vect_load_lanes_supported.
21961 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
21962 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
21963 parameter. Don't allow gaps for masked accesses.
21964 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
21965 and vect_load_lanes_supported.
21966 (get_load_store_type): Take a masked_p parameter and update
21967 call to get_group_load_store_type.
21968 (vectorizable_store): Update call to get_load_store_type.
21969 Handle IFN_MASK_STORE_LANES.
21970 (vectorizable_load): Update call to get_load_store_type.
21971 Handle IFN_MASK_LOAD_LANES.
21973 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21974 Alan Hayward <alan.hayward@arm.com>
21975 David Sherwood <david.sherwood@arm.com>
21977 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
21979 * config/aarch64/aarch64-protos.h
21980 (aarch64_sve_struct_memory_operand_p): Declare.
21981 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
21982 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
21983 (VPRED, vpred): Handle SVE structure modes.
21984 * config/aarch64/constraints.md (Utx): New constraint.
21985 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
21986 (aarch64_sve_struct_nonimmediate_operand): New predicates.
21987 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
21988 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
21989 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
21990 structure modes. Split into pieces after RA.
21991 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
21992 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
21994 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
21995 SVE structure modes.
21996 (aarch64_classify_address): Likewise.
21997 (sizetochar): Move earlier in file.
21998 (aarch64_print_operand): Handle SVE register lists.
21999 (aarch64_array_mode): New function.
22000 (aarch64_sve_struct_memory_operand_p): Likewise.
22001 (TARGET_ARRAY_MODE): Redefine.
22003 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22004 Alan Hayward <alan.hayward@arm.com>
22005 David Sherwood <david.sherwood@arm.com>
22007 * target.def (array_mode): New target hook.
22008 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
22009 * doc/tm.texi: Regenerate.
22010 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
22011 * hooks.c (hook_optmode_mode_uhwi_none): New function.
22012 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
22013 targetm.array_mode.
22014 * stor-layout.c (mode_for_array): Likewise. Support polynomial
22017 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22018 Alan Hayward <alan.hayward@arm.com>
22019 David Sherwood <david.sherwood@arm.com>
22021 * fold-const.c (fold_binary_loc): Check the argument types
22022 rather than the result type when testing for a vector operation.
22024 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22026 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
22027 * doc/tm.texi: Regenerate.
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 * doc/invoke.texi (-msve-vector-bits=): Document new option.
22034 (sve): Document new AArch64 extension.
22035 * doc/md.texi (w): Extend the description of the AArch64
22036 constraint to include SVE vectors.
22037 (Upl, Upa): Document new AArch64 predicate constraints.
22038 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
22040 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
22041 (msve-vector-bits=): New option.
22042 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
22043 SVE when these are disabled.
22044 (sve): New extension.
22045 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
22046 modes. Adjust their number of units based on aarch64_sve_vg.
22047 (MAX_BITSIZE_MODE_ANY_MODE): Define.
22048 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
22049 aarch64_addr_query_type.
22050 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
22051 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
22052 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
22053 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
22054 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
22055 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
22056 (aarch64_simd_imm_zero_p): Delete.
22057 (aarch64_check_zero_based_sve_index_immediate): Declare.
22058 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
22059 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
22060 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
22061 (aarch64_sve_float_mul_immediate_p): Likewise.
22062 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
22063 rather than an rtx.
22064 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
22065 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
22066 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
22067 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
22068 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
22069 (aarch64_regmode_natural_size): Likewise.
22070 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
22071 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
22073 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
22074 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
22075 for VG and the SVE predicate registers.
22076 (V_ALIASES): Add a "z"-prefixed alias.
22077 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
22078 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
22079 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
22080 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
22081 (REG_CLASS_NAMES): Add entries for them.
22082 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
22083 and the predicate registers.
22084 (aarch64_sve_vg): Declare.
22085 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
22086 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
22087 (REGMODE_NATURAL_SIZE): Define.
22088 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
22090 * config/aarch64/aarch64.c: Include cfgrtl.h.
22091 (simd_immediate_info): Add a constructor for series vectors,
22092 and an associated step field.
22093 (aarch64_sve_vg): New variable.
22094 (aarch64_dbx_register_number): Handle VG and the predicate registers.
22095 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
22096 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
22097 (VEC_ANY_DATA, VEC_STRUCT): New constants.
22098 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
22099 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
22100 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
22101 (aarch64_get_mask_mode): New functions.
22102 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
22103 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
22104 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
22105 predicate modes and predicate registers. Explicitly restrict
22106 GPRs to modes of 16 bytes or smaller. Only allow FP registers
22107 to store a vector mode if it is recognized by
22108 aarch64_classify_vector_mode.
22109 (aarch64_regmode_natural_size): New function.
22110 (aarch64_hard_regno_caller_save_mode): Return the original mode
22112 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
22113 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
22114 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
22115 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
22117 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
22118 does not overlap dest if the function is frame-related. Handle
22120 (aarch64_split_add_offset): New function.
22121 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
22122 them aarch64_add_offset.
22123 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
22124 and update call to aarch64_sub_sp.
22125 (aarch64_add_cfa_expression): New function.
22126 (aarch64_expand_prologue): Pass extra temporary registers to the
22127 functions above. Handle the case in which we need to emit new
22128 DW_CFA_expressions for registers that were originally saved
22129 relative to the stack pointer, but now have to be expressed
22130 relative to the frame pointer.
22131 (aarch64_output_mi_thunk): Pass extra temporary registers to the
22133 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
22134 IP0 and IP1 values for SVE frames.
22135 (aarch64_expand_vec_series): New function.
22136 (aarch64_expand_sve_widened_duplicate): Likewise.
22137 (aarch64_expand_sve_const_vector): Likewise.
22138 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
22139 Handle SVE constants. Use emit_move_insn to move a force_const_mem
22140 into the register, rather than emitting a SET directly.
22141 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
22142 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
22143 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
22144 (offset_9bit_signed_scaled_p): New functions.
22145 (aarch64_replicate_bitmask_imm): New function.
22146 (aarch64_bitmask_imm): Use it.
22147 (aarch64_cannot_force_const_mem): Reject expressions involving
22148 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
22149 (aarch64_classify_index): Handle SVE indices, by requiring
22150 a plain register index with a scale that matches the element size.
22151 (aarch64_classify_address): Handle SVE addresses. Assert that
22152 the mode of the address is VOIDmode or an integer mode.
22153 Update call to aarch64_classify_symbol.
22154 (aarch64_classify_symbolic_expression): Update call to
22155 aarch64_classify_symbol.
22156 (aarch64_const_vec_all_in_range_p): New function.
22157 (aarch64_print_vector_float_operand): Likewise.
22158 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
22159 "vN" for FP registers with SVE modes. Handle (const ...) vectors
22160 and the FP immediates 1.0 and 0.5.
22161 (aarch64_print_address_internal): Handle SVE addresses.
22162 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
22163 (aarch64_regno_regclass): Handle predicate registers.
22164 (aarch64_secondary_reload): Handle big-endian reloads of SVE
22166 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
22167 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
22168 (aarch64_convert_sve_vector_bits): New function.
22169 (aarch64_override_options): Use it to handle -msve-vector-bits=.
22170 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
22171 rather than an rtx.
22172 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
22173 Handle SVE vector and predicate modes. Accept VL-based constants
22174 that need only one temporary register, and VL offsets that require
22175 no temporary registers.
22176 (aarch64_conditional_register_usage): Mark the predicate registers
22177 as fixed if SVE isn't available.
22178 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
22179 Return true for SVE vector and predicate modes.
22180 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
22181 rather than an unsigned int. Handle SVE modes.
22182 (aarch64_preferred_simd_mode): Update call accordingly. Handle
22184 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
22186 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
22187 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
22188 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
22189 (aarch64_sve_float_mul_immediate_p): New functions.
22190 (aarch64_sve_valid_immediate): New function.
22191 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
22192 Explicitly reject structure modes. Check for INDEX constants.
22193 Handle PTRUE and PFALSE constants.
22194 (aarch64_check_zero_based_sve_index_immediate): New function.
22195 (aarch64_simd_imm_zero_p): Delete.
22196 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
22197 vector modes. Accept constants in the range of CNT[BHWD].
22198 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
22199 ask for an Advanced SIMD mode.
22200 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
22201 (aarch64_simd_vector_alignment): Handle SVE predicates.
22202 (aarch64_vectorize_preferred_vector_alignment): New function.
22203 (aarch64_simd_vector_alignment_reachable): Use it instead of
22205 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
22206 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
22208 (MAX_VECT_LEN): Delete.
22209 (expand_vec_perm_d): Add a vec_flags field.
22210 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
22211 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
22212 (aarch64_evpc_ext): Don't apply a big-endian lane correction
22214 (aarch64_evpc_rev): Rename to...
22215 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
22216 (aarch64_evpc_rev_global): New function.
22217 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
22218 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
22220 (aarch64_evpc_sve_tbl): New function.
22221 (aarch64_expand_vec_perm_const_1): Update after rename of
22222 aarch64_evpc_rev. Handle SVE permutes too, trying
22223 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
22224 than aarch64_evpc_tbl.
22225 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
22226 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
22227 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
22228 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
22229 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
22230 (aarch64_expand_sve_vcond): New functions.
22231 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
22232 of aarch64_vector_mode_p.
22233 (aarch64_dwarf_poly_indeterminate_value): New function.
22234 (aarch64_compute_pressure_classes): Likewise.
22235 (aarch64_can_change_mode_class): Likewise.
22236 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
22237 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
22238 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
22239 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
22240 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
22241 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
22242 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
22243 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
22245 (Dn, Dl, Dr): Accept const as well as const_vector.
22246 (Dz): Likewise. Compare against CONST0_RTX.
22247 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
22248 of "vector" where appropriate.
22249 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
22250 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
22251 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
22252 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
22253 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
22254 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
22255 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
22256 (v_int_equiv): Extend to SVE modes.
22257 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
22259 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
22260 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
22261 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
22262 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
22263 (SVE_COND_FP_CMP): New int iterators.
22264 (perm_hilo): Handle the new unpack unspecs.
22265 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
22267 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
22268 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
22269 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
22270 (aarch64_equality_operator, aarch64_constant_vector_operand)
22271 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
22272 (aarch64_sve_nonimmediate_operand): Likewise.
22273 (aarch64_sve_general_operand): Likewise.
22274 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
22275 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
22276 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
22277 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
22278 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
22279 (aarch64_sve_float_arith_immediate): Likewise.
22280 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
22281 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
22282 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
22283 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
22284 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
22285 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
22286 (aarch64_sve_float_arith_operand): Likewise.
22287 (aarch64_sve_float_arith_with_sub_operand): Likewise.
22288 (aarch64_sve_float_mul_operand): Likewise.
22289 (aarch64_sve_vec_perm_operand): Likewise.
22290 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
22291 (aarch64_mov_operand): Accept const_poly_int and const_vector.
22292 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
22293 as well as const_vector.
22294 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
22295 in file. Use CONST0_RTX and CONSTM1_RTX.
22296 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
22297 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
22298 Use aarch64_simd_imm_zero.
22299 * config/aarch64/aarch64-sve.md: New file.
22300 * config/aarch64/aarch64.md: Include it.
22301 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
22302 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
22303 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
22304 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
22305 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
22306 (sve): New attribute.
22307 (enabled): Disable instructions with the sve attribute unless
22309 (movqi, movhi): Pass CONST_POLY_INT operaneds through
22310 aarch64_expand_mov_immediate.
22311 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
22312 CNT[BHSD] immediates.
22313 (movti): Split CONST_POLY_INT moves into two halves.
22314 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
22315 Split additions that need a temporary here if the destination
22316 is the stack pointer.
22317 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
22318 (*add<mode>3_poly_1): New instruction.
22319 (set_clobber_cc): New expander.
22321 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22323 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
22324 parameter and use it instead of GET_MODE_SIZE (innermode). Use
22325 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
22326 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
22327 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
22328 Change innermode from fixed_mode_size to machine_mode.
22329 (simplify_subreg): Update call accordingly. Handle a constant-sized
22330 subreg of a variable-length CONST_VECTOR.
22332 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22333 Alan Hayward <alan.hayward@arm.com>
22334 David Sherwood <david.sherwood@arm.com>
22336 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
22337 (add_offset_to_base): New function, split out from...
22338 (create_mem_ref): ...here. When handling a scale other than 1,
22339 check first whether the address is valid without the offset.
22340 Add it into the base if so, leaving the index and scale as-is.
22342 2018-01-12 Jakub Jelinek <jakub@redhat.com>
22345 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
22346 fold_for_warn before checking if arg2 is INTEGER_CST.
22348 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
22350 * config/rs6000/predicates.md (load_multiple_operation): Delete.
22351 (store_multiple_operation): Delete.
22352 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
22353 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
22354 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
22355 guarded by TARGET_STRING.
22356 (rs6000_output_load_multiple): Delete.
22357 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22358 OPTION_MASK_STRING / TARGET_STRING handling.
22359 (print_operand) <'N', 'O'>: Add comment that these are unused now.
22360 (const rs6000_opt_masks) <"string">: Change mask to 0.
22361 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
22362 (MASK_STRING): Delete.
22363 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
22365 (load_multiple): Delete.
22372 (store_multiple): Delete.
22379 (movmemsi_8reg): Delete.
22380 (corresponding unnamed define_insn): Delete.
22381 (movmemsi_6reg): Delete.
22382 (corresponding unnamed define_insn): Delete.
22383 (movmemsi_4reg): Delete.
22384 (corresponding unnamed define_insn): Delete.
22385 (movmemsi_2reg): Delete.
22386 (corresponding unnamed define_insn): Delete.
22387 (movmemsi_1reg): Delete.
22388 (corresponding unnamed define_insn): Delete.
22389 * config/rs6000/rs6000.opt (mno-string): New.
22390 (mstring): Replace by deprecation warning stub.
22391 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
22393 2018-01-12 Jakub Jelinek <jakub@redhat.com>
22395 * regrename.c (regrename_do_replace): If replacing the same
22396 reg multiple times, try to reuse last created gen_raw_REG.
22399 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
22400 main to workaround a bug in GDB.
22402 2018-01-12 Tom de Vries <tom@codesourcery.com>
22405 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
22407 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
22409 PR rtl-optimization/80481
22410 * ira-color.c (get_cap_member): New function.
22411 (allocnos_conflict_by_live_ranges_p): Use it.
22412 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
22413 (setup_slot_coalesced_allocno_live_ranges): Ditto.
22415 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
22418 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
22419 (*saddl_se_1): Ditto.
22420 (*ssubsi_1): Ditto.
22421 (*ssubl_se_1): Ditto.
22423 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
22425 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
22426 rather than wi::to_widest for DR_INITs.
22427 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
22428 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
22429 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
22431 (vect_analyze_group_access_1): Note that here.
22433 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
22435 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
22436 polynomial type sizes.
22438 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
22440 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
22441 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
22442 (gimple_add_tmp_var): Likewise.
22444 2018-01-12 Martin Liska <mliska@suse.cz>
22446 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
22447 (gimple_alloc_sizes): Likewise.
22448 (dump_gimple_statistics): Use PRIu64 in printf format.
22449 * gimple.h: Change uint64_t to int.
22451 2018-01-12 Martin Liska <mliska@suse.cz>
22453 * tree-core.h: Use uint64_t instead of int.
22454 * tree.c (tree_node_counts): Likewise.
22455 (tree_node_sizes): Likewise.
22456 (dump_tree_statistics): Use PRIu64 in printf format.
22458 2018-01-12 Martin Liska <mliska@suse.cz>
22460 * Makefile.in: As qsort_chk is implemented in vec.c, add
22461 vec.o to linkage of gencfn-macros.
22462 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
22463 passing the info to record_node_allocation_statistics.
22464 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
22466 * ggc-common.c (struct ggc_usage): Add operator== and use
22467 it in operator< and compare function.
22468 * mem-stats.h (struct mem_usage): Likewise.
22469 * vec.c (struct vec_usage): Remove operator< and compare
22470 function. Can be simply inherited.
22472 2018-01-12 Martin Jambor <mjambor@suse.cz>
22475 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
22476 * tree-ssa-math-opts.c: Include domwalk.h.
22477 (convert_mult_to_fma_1): New function.
22478 (fma_transformation_info): New type.
22479 (fma_deferring_state): Likewise.
22480 (cancel_fma_deferring): New function.
22481 (result_of_phi): Likewise.
22482 (last_fma_candidate_feeds_initial_phi): Likewise.
22483 (convert_mult_to_fma): Added deferring logic, split actual
22484 transformation to convert_mult_to_fma_1.
22485 (math_opts_dom_walker): New type.
22486 (math_opts_dom_walker::after_dom_children): New method, body moved
22487 here from pass_optimize_widening_mul::execute, added deferring logic
22489 (pass_optimize_widening_mul::execute): Moved most of code to
22490 math_opts_dom_walker::after_dom_children.
22491 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
22492 * config/i386/i386.c (ix86_option_override_internal): Added
22493 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
22495 2018-01-12 Richard Biener <rguenther@suse.de>
22498 * dwarf2out.c (gen_variable_die): Do not reset old_die for
22499 inline instance vars.
22501 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
22504 * config/rx/rx.c (rx_is_restricted_memory_address):
22505 Handle SUBREG case.
22507 2018-01-12 Richard Biener <rguenther@suse.de>
22509 PR tree-optimization/80846
22510 * target.def (split_reduction): New target hook.
22511 * targhooks.c (default_split_reduction): New function.
22512 * targhooks.h (default_split_reduction): Declare.
22513 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
22514 target requests first reduce vectors by combining low and high
22516 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
22517 (get_vectype_for_scalar_type_and_size): Export.
22518 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
22519 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
22520 * doc/tm.texi: Regenerate.
22521 * config/i386/i386.c (ix86_split_reduction): Implement
22522 TARGET_VECTORIZE_SPLIT_REDUCTION.
22524 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
22527 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
22528 in PIC mode except for TARGET_VXWORKS_RTP.
22529 * config/sparc/sparc.c: Include cfgrtl.h.
22530 (TARGET_INIT_PIC_REG): Define.
22531 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
22532 (sparc_pic_register_p): New predicate.
22533 (sparc_legitimate_address_p): Use it.
22534 (sparc_legitimize_pic_address): Likewise.
22535 (sparc_delegitimize_address): Likewise.
22536 (sparc_mode_dependent_address_p): Likewise.
22537 (gen_load_pcrel_sym): Remove 4th parameter.
22538 (load_got_register): Adjust call to above. Remove obsolete stuff.
22539 (sparc_expand_prologue): Do not call load_got_register here.
22540 (sparc_flat_expand_prologue): Likewise.
22541 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
22542 (sparc_use_pseudo_pic_reg): New function.
22543 (sparc_init_pic_reg): Likewise.
22544 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
22545 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
22547 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
22549 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22550 Add item for branch_cost.
22552 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
22554 PR rtl-optimization/83565
22555 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
22556 not extend the result to a larger mode for rotate operations.
22557 (num_sign_bit_copies1): Likewise.
22559 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22562 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
22564 Use values-Xc.o for -pedantic.
22565 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
22567 2018-01-12 Martin Liska <mliska@suse.cz>
22570 * ipa-devirt.c (final_warning_record::grow_type_warnings):
22572 (possible_polymorphic_call_targets): Use it.
22573 (ipa_devirt): Likewise.
22575 2018-01-12 Martin Liska <mliska@suse.cz>
22577 * profile-count.h (enum profile_quality): Use 0 as invalid
22578 enum value of profile_quality.
22580 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
22582 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
22583 -mext-string options.
22585 2018-01-12 Richard Biener <rguenther@suse.de>
22587 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
22588 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
22589 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
22591 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
22593 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
22595 * configure.ac (--with-long-double-format): Add support for the
22596 configuration option to change the default long double format on
22598 * config.gcc (powerpc*-linux*-*): Likewise.
22599 * configure: Regenerate.
22600 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
22601 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
22602 used without modification.
22604 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22606 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
22607 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
22608 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
22609 MISC_BUILTIN_SPEC_BARRIER.
22610 (rs6000_init_builtins): Likewise.
22611 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
22613 (speculation_barrier): New define_insn.
22614 * doc/extend.texi: Document __builtin_speculation_barrier.
22616 2018-01-11 Jakub Jelinek <jakub@redhat.com>
22619 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
22620 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
22621 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
22623 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
22624 integral modes instead of "ss" and "sd".
22625 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
22626 vectors with 32-bit and 64-bit elements.
22627 (vecdupssescalarmodesuffix): New mode attribute.
22628 (vec_dup<mode>): Use it.
22630 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
22633 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
22634 frame if argument is passed on stack.
22636 2018-01-11 Jakub Jelinek <jakub@redhat.com>
22639 * ree.c (combine_reaching_defs): Optimize also
22640 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
22641 reg2=any_extend(exp); reg1=reg2;, formatting fix.
22643 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
22645 PR middle-end/83189
22646 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
22648 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
22650 PR middle-end/83718
22651 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
22652 after they are computed.
22654 2018-01-11 Bin Cheng <bin.cheng@arm.com>
22656 PR tree-optimization/83695
22657 * gimple-loop-linterchange.cc
22658 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
22659 reset cached scev information after interchange.
22660 (pass_linterchange::execute): Remove call to scev_reset_htab.
22662 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22664 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
22665 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
22666 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
22667 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
22668 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
22669 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
22670 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
22671 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
22672 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
22673 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
22674 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
22675 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
22676 (V_lane_reg): Likewise.
22677 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
22679 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
22680 (vfmal_lane_low<mode>_intrinsic,
22681 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
22682 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
22683 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
22684 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
22685 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
22686 vfmsl_lane_high<mode>_intrinsic): New define_insns.
22688 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22690 * config/arm/arm-cpus.in (fp16fml): New feature.
22691 (ALL_SIMD): Add fp16fml.
22692 (armv8.2-a): Add fp16fml as an option.
22693 (armv8.3-a): Likewise.
22694 (armv8.4-a): Add fp16fml as part of fp16.
22695 * config/arm/arm.h (TARGET_FP16FML): Define.
22696 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
22698 * config/arm/arm-modes.def (V2HF): Define.
22699 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
22700 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
22701 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
22702 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
22703 vfmsl_low, vfmsl_high): New set of builtins.
22704 * config/arm/iterators.md (PLUSMINUS): New code iterator.
22705 (vfml_op): New code attribute.
22706 (VFMLHALVES): New int iterator.
22707 (VFML, VFMLSEL): New mode attributes.
22708 (V_reg): Define mapping for V2HF.
22709 (V_hi, V_lo): New mode attributes.
22710 (VF_constraint): Likewise.
22711 (vfml_half, vfml_half_selector): New int attributes.
22712 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
22714 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
22715 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
22717 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
22718 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
22719 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
22720 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
22722 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
22723 Document new effective target and option set.
22725 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22727 * config/arm/arm-cpus.in (armv8_4): New feature.
22728 (ARMv8_4a): New fgroup.
22729 (armv8.4-a): New arch.
22730 * config/arm/arm-tables.opt: Regenerate.
22731 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
22732 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
22733 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
22734 Add matching rules for -march=armv8.4-a and extensions.
22735 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
22737 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
22740 * config/rx/rx.md (BW): New mode attribute.
22741 (sync_lock_test_and_setsi): Add mode suffix to insn output.
22743 2018-01-11 Richard Biener <rguenther@suse.de>
22745 PR tree-optimization/83435
22746 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
22747 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
22748 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
22750 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22751 Alan Hayward <alan.hayward@arm.com>
22752 David Sherwood <david.sherwood@arm.com>
22754 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
22756 (aarch64_classify_address): Initialize it. Track polynomial offsets.
22757 (aarch64_print_address_internal): Use it to check for a zero offset.
22759 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22760 Alan Hayward <alan.hayward@arm.com>
22761 David Sherwood <david.sherwood@arm.com>
22763 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
22764 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
22765 Return a poly_int64 rather than a HOST_WIDE_INT.
22766 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
22767 rather than a HOST_WIDE_INT.
22768 * config/aarch64/aarch64.h (aarch64_frame): Protect with
22769 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
22770 hard_fp_offset, frame_size, initial_adjust, callee_offset and
22771 final_offset from HOST_WIDE_INT to poly_int64.
22772 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
22773 to_constant when getting the number of units in an Advanced SIMD
22775 (aarch64_builtin_vectorized_function): Check for a constant number
22777 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
22779 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
22780 attribute instead of GET_MODE_NUNITS.
22781 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
22782 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
22783 GET_MODE_SIZE for fixed-size registers.
22784 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
22785 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
22786 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
22787 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
22788 (aarch64_print_operand, aarch64_print_address_internal)
22789 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
22790 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
22791 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
22792 Handle polynomial GET_MODE_SIZE.
22793 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
22794 wider than SImode without modification.
22795 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
22796 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
22797 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
22798 passing and returning SVE modes.
22799 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
22800 rather than GEN_INT.
22801 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
22802 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
22803 (aarch64_allocate_and_probe_stack_space): Likewise.
22804 (aarch64_layout_frame): Cope with polynomial offsets.
22805 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
22806 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
22807 polynomial offsets.
22808 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
22809 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
22810 poly_int64 rather than a HOST_WIDE_INT.
22811 (aarch64_get_separate_components, aarch64_process_components)
22812 (aarch64_expand_prologue, aarch64_expand_epilogue)
22813 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
22814 (aarch64_anchor_offset): New function, split out from...
22815 (aarch64_legitimize_address): ...here.
22816 (aarch64_builtin_vectorization_cost): Handle polynomial
22817 TYPE_VECTOR_SUBPARTS.
22818 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
22820 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
22821 number of elements from the PARALLEL rather than the mode.
22822 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
22823 rather than GET_MODE_BITSIZE.
22824 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
22825 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
22826 (aarch64_expand_vec_perm_const_1): Handle polynomial
22827 d->perm.length () and d->perm elements.
22828 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
22829 Apply to_constant to d->perm elements.
22830 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
22831 polynomial CONST_VECTOR_NUNITS.
22832 (aarch64_move_pointer): Take amount as a poly_int64 rather
22834 (aarch64_progress_pointer): Avoid temporary variable.
22835 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
22836 the mode attribute instead of GET_MODE.
22838 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22839 Alan Hayward <alan.hayward@arm.com>
22840 David Sherwood <david.sherwood@arm.com>
22842 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
22843 x exists before using it.
22844 (aarch64_add_constant_internal): Rename to...
22845 (aarch64_add_offset_1): ...this. Replace regnum with separate
22846 src and dest rtxes. Handle the case in which they're different,
22847 including when the offset is zero. Replace scratchreg with an rtx.
22848 Use 2 additions if there is no spare register into which we can
22849 move a 16-bit constant.
22850 (aarch64_add_constant): Delete.
22851 (aarch64_add_offset): Replace reg with separate src and dest
22852 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
22853 Use aarch64_add_offset_1.
22854 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
22855 an rtx rather than an int. Take the delta as a poly_int64
22856 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
22857 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
22858 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
22859 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
22860 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
22861 and aarch64_add_sp.
22862 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
22863 aarch64_add_constant.
22865 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22867 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22868 Use scalar_float_mode.
22870 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22872 * config/aarch64/aarch64-simd.md
22873 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
22874 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
22875 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
22876 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
22877 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
22878 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
22879 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
22880 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
22881 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
22882 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
22884 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22887 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
22888 targ_options->x_arm_arch_string is non NULL.
22890 2018-01-11 Tamar Christina <tamar.christina@arm.com>
22892 * config/aarch64/aarch64.h
22893 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
22895 2018-01-11 Sudakshina Das <sudi.das@arm.com>
22898 * expmed.c (emit_store_flag_force): Swap if const op0
22899 and change VOIDmode to mode of op0.
22901 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22903 PR rtl-optimization/83761
22904 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
22905 than bytes to mode_for_size.
22907 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
22909 PR middle-end/83189
22910 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
22911 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
22914 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
22916 PR middle-end/83575
22917 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
22918 when in layout mode.
22919 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
22920 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
22923 2018-01-10 Michael Collison <michael.collison@arm.com>
22925 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
22926 * config/aarch64/aarch64-option-extension.def: Add
22927 AARCH64_OPT_EXTENSION of 'fp16fml'.
22928 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
22929 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
22930 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
22931 * config/aarch64/constraints.md (Ui7): New constraint.
22932 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
22933 (VFMLA_SEL_W): Ditto.
22936 (VFMLA16_LOW): New int iterator.
22937 (VFMLA16_HIGH): Ditto.
22938 (UNSPEC_FMLAL): New unspec.
22939 (UNSPEC_FMLSL): Ditto.
22940 (UNSPEC_FMLAL2): Ditto.
22941 (UNSPEC_FMLSL2): Ditto.
22942 (f16mac): New code attribute.
22943 * config/aarch64/aarch64-simd-builtins.def
22944 (aarch64_fmlal_lowv2sf): Ditto.
22945 (aarch64_fmlsl_lowv2sf): Ditto.
22946 (aarch64_fmlalq_lowv4sf): Ditto.
22947 (aarch64_fmlslq_lowv4sf): Ditto.
22948 (aarch64_fmlal_highv2sf): Ditto.
22949 (aarch64_fmlsl_highv2sf): Ditto.
22950 (aarch64_fmlalq_highv4sf): Ditto.
22951 (aarch64_fmlslq_highv4sf): Ditto.
22952 (aarch64_fmlal_lane_lowv2sf): Ditto.
22953 (aarch64_fmlsl_lane_lowv2sf): Ditto.
22954 (aarch64_fmlal_laneq_lowv2sf): Ditto.
22955 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
22956 (aarch64_fmlalq_lane_lowv4sf): Ditto.
22957 (aarch64_fmlsl_lane_lowv4sf): Ditto.
22958 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
22959 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
22960 (aarch64_fmlal_lane_highv2sf): Ditto.
22961 (aarch64_fmlsl_lane_highv2sf): Ditto.
22962 (aarch64_fmlal_laneq_highv2sf): Ditto.
22963 (aarch64_fmlsl_laneq_highv2sf): Ditto.
22964 (aarch64_fmlalq_lane_highv4sf): Ditto.
22965 (aarch64_fmlsl_lane_highv4sf): Ditto.
22966 (aarch64_fmlalq_laneq_highv4sf): Ditto.
22967 (aarch64_fmlsl_laneq_highv4sf): Ditto.
22968 * config/aarch64/aarch64-simd.md:
22969 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
22970 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
22971 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
22972 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
22973 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
22974 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
22975 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
22976 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
22977 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
22978 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
22979 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
22980 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
22981 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
22982 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
22983 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
22984 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
22985 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
22986 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
22987 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
22988 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
22989 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
22990 (vfmlsl_low_u32): Ditto.
22991 (vfmlalq_low_u32): Ditto.
22992 (vfmlslq_low_u32): Ditto.
22993 (vfmlal_high_u32): Ditto.
22994 (vfmlsl_high_u32): Ditto.
22995 (vfmlalq_high_u32): Ditto.
22996 (vfmlslq_high_u32): Ditto.
22997 (vfmlal_lane_low_u32): Ditto.
22998 (vfmlsl_lane_low_u32): Ditto.
22999 (vfmlal_laneq_low_u32): Ditto.
23000 (vfmlsl_laneq_low_u32): Ditto.
23001 (vfmlalq_lane_low_u32): Ditto.
23002 (vfmlslq_lane_low_u32): Ditto.
23003 (vfmlalq_laneq_low_u32): Ditto.
23004 (vfmlslq_laneq_low_u32): Ditto.
23005 (vfmlal_lane_high_u32): Ditto.
23006 (vfmlsl_lane_high_u32): Ditto.
23007 (vfmlal_laneq_high_u32): Ditto.
23008 (vfmlsl_laneq_high_u32): Ditto.
23009 (vfmlalq_lane_high_u32): Ditto.
23010 (vfmlslq_lane_high_u32): Ditto.
23011 (vfmlalq_laneq_high_u32): Ditto.
23012 (vfmlslq_laneq_high_u32): Ditto.
23013 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
23014 (AARCH64_FL_FOR_ARCH8_4): New.
23015 (AARCH64_ISA_F16FML): New ISA flag.
23016 (TARGET_F16FML): New feature flag for fp16fml.
23017 (doc/invoke.texi): Document new fp16fml option.
23019 2018-01-10 Michael Collison <michael.collison@arm.com>
23021 * config/aarch64/aarch64-builtins.c:
23022 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
23023 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23024 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
23025 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
23026 (AARCH64_ISA_SHA3): New ISA flag.
23027 (TARGET_SHA3): New feature flag for sha3.
23028 * config/aarch64/iterators.md (sha512_op): New int attribute.
23029 (CRYPTO_SHA512): New int iterator.
23030 (UNSPEC_SHA512H): New unspec.
23031 (UNSPEC_SHA512H2): Ditto.
23032 (UNSPEC_SHA512SU0): Ditto.
23033 (UNSPEC_SHA512SU1): Ditto.
23034 * config/aarch64/aarch64-simd-builtins.def
23035 (aarch64_crypto_sha512hqv2di): New builtin.
23036 (aarch64_crypto_sha512h2qv2di): Ditto.
23037 (aarch64_crypto_sha512su0qv2di): Ditto.
23038 (aarch64_crypto_sha512su1qv2di): Ditto.
23039 (aarch64_eor3qv8hi): Ditto.
23040 (aarch64_rax1qv2di): Ditto.
23041 (aarch64_xarqv2di): Ditto.
23042 (aarch64_bcaxqv8hi): Ditto.
23043 * config/aarch64/aarch64-simd.md:
23044 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
23045 (aarch64_crypto_sha512su0qv2di): Ditto.
23046 (aarch64_crypto_sha512su1qv2di): Ditto.
23047 (aarch64_eor3qv8hi): Ditto.
23048 (aarch64_rax1qv2di): Ditto.
23049 (aarch64_xarqv2di): Ditto.
23050 (aarch64_bcaxqv8hi): Ditto.
23051 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
23052 (vsha512h2q_u64): Ditto.
23053 (vsha512su0q_u64): Ditto.
23054 (vsha512su1q_u64): Ditto.
23055 (veor3q_u16): Ditto.
23056 (vrax1q_u64): Ditto.
23057 (vxarq_u64): Ditto.
23058 (vbcaxq_u16): Ditto.
23059 * config/arm/types.md (crypto_sha512): New type attribute.
23060 (crypto_sha3): Ditto.
23061 (doc/invoke.texi): Document new sha3 option.
23063 2018-01-10 Michael Collison <michael.collison@arm.com>
23065 * config/aarch64/aarch64-builtins.c:
23066 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
23067 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23068 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
23069 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
23070 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
23071 (AARCH64_ISA_SM4): New ISA flag.
23072 (TARGET_SM4): New feature flag for sm4.
23073 * config/aarch64/aarch64-simd-builtins.def
23074 (aarch64_sm3ss1qv4si): Ditto.
23075 (aarch64_sm3tt1aq4si): Ditto.
23076 (aarch64_sm3tt1bq4si): Ditto.
23077 (aarch64_sm3tt2aq4si): Ditto.
23078 (aarch64_sm3tt2bq4si): Ditto.
23079 (aarch64_sm3partw1qv4si): Ditto.
23080 (aarch64_sm3partw2qv4si): Ditto.
23081 (aarch64_sm4eqv4si): Ditto.
23082 (aarch64_sm4ekeyqv4si): Ditto.
23083 * config/aarch64/aarch64-simd.md:
23084 (aarch64_sm3ss1qv4si): Ditto.
23085 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
23086 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
23087 (aarch64_sm4eqv4si): Ditto.
23088 (aarch64_sm4ekeyqv4si): Ditto.
23089 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
23090 (sm3part_op): Ditto.
23091 (CRYPTO_SM3TT): Ditto.
23092 (CRYPTO_SM3PART): Ditto.
23093 (UNSPEC_SM3SS1): New unspec.
23094 (UNSPEC_SM3TT1A): Ditto.
23095 (UNSPEC_SM3TT1B): Ditto.
23096 (UNSPEC_SM3TT2A): Ditto.
23097 (UNSPEC_SM3TT2B): Ditto.
23098 (UNSPEC_SM3PARTW1): Ditto.
23099 (UNSPEC_SM3PARTW2): Ditto.
23100 (UNSPEC_SM4E): Ditto.
23101 (UNSPEC_SM4EKEY): Ditto.
23102 * config/aarch64/constraints.md (Ui2): New constraint.
23103 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
23104 * config/arm/types.md (crypto_sm3): New type attribute.
23105 (crypto_sm4): Ditto.
23106 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
23107 (vsm3tt1aq_u32): Ditto.
23108 (vsm3tt1bq_u32): Ditto.
23109 (vsm3tt2aq_u32): Ditto.
23110 (vsm3tt2bq_u32): Ditto.
23111 (vsm3partw1q_u32): Ditto.
23112 (vsm3partw2q_u32): Ditto.
23113 (vsm4eq_u32): Ditto.
23114 (vsm4ekeyq_u32): Ditto.
23115 (doc/invoke.texi): Document new sm4 option.
23117 2018-01-10 Michael Collison <michael.collison@arm.com>
23119 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
23120 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
23121 (AARCH64_FL_FOR_ARCH8_4): New.
23122 (AARCH64_FL_V8_4): New flag.
23123 (doc/invoke.texi): Document new armv8.4-a option.
23125 2018-01-10 Michael Collison <michael.collison@arm.com>
23127 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23128 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
23129 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
23130 * config/aarch64/aarch64-option-extension.def: Add
23131 AARCH64_OPT_EXTENSION of 'sha2'.
23132 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
23133 (crypto): Disable sha2 and aes if crypto disabled.
23134 (crypto): Enable aes and sha2 if enabled.
23135 (simd): Disable sha2 and aes if simd disabled.
23136 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
23138 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
23139 (TARGET_SHA2): New feature flag for sha2.
23140 (TARGET_AES): New feature flag for aes.
23141 * config/aarch64/aarch64-simd.md:
23142 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
23143 conditional on TARGET_AES.
23144 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
23145 (aarch64_crypto_sha1hsi): Make pattern conditional
23147 (aarch64_crypto_sha1hv4si): Ditto.
23148 (aarch64_be_crypto_sha1hv4si): Ditto.
23149 (aarch64_crypto_sha1su1v4si): Ditto.
23150 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
23151 (aarch64_crypto_sha1su0v4si): Ditto.
23152 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
23153 (aarch64_crypto_sha256su0v4si): Ditto.
23154 (aarch64_crypto_sha256su1v4si): Ditto.
23155 (doc/invoke.texi): Document new aes and sha2 options.
23157 2018-01-10 Martin Sebor <msebor@redhat.com>
23159 PR tree-optimization/83781
23160 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
23163 2018-01-11 Martin Sebor <msebor@gmail.com>
23164 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23166 PR tree-optimization/83501
23167 PR tree-optimization/81703
23169 * tree-ssa-strlen.c (get_string_cst): Rename...
23170 (get_string_len): ...to this. Handle global constants.
23171 (handle_char_store): Adjust.
23173 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
23174 Jim Wilson <jimw@sifive.com>
23176 * config/riscv/riscv-protos.h (riscv_output_return): New.
23177 * config/riscv/riscv.c (struct machine_function): New naked_p field.
23178 (riscv_attribute_table, riscv_output_return),
23179 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
23180 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
23181 (riscv_compute_frame_info): Only compute frame->mask if not a naked
23183 (riscv_expand_prologue): Add early return for naked function.
23184 (riscv_expand_epilogue): Likewise.
23185 (riscv_function_ok_for_sibcall): Return false for naked function.
23186 (riscv_set_current_function): New.
23187 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
23188 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
23189 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
23190 * doc/extend.texi (RISC-V Function Attributes): New.
23192 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
23194 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
23195 check for 128-bit long double before checking TCmode.
23196 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
23197 128-bit long doubles before checking TFmode or TCmode.
23198 (FLOAT128_IBM_P): Likewise.
23200 2018-01-10 Martin Sebor <msebor@redhat.com>
23202 PR tree-optimization/83671
23203 * builtins.c (c_strlen): Unconditionally return zero for the empty
23205 Use -Warray-bounds for warnings.
23206 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
23207 for non-constant array indices with COMPONENT_REF, arrays of
23208 arrays, and pointers to arrays.
23209 (gimple_fold_builtin_strlen): Determine and set length range for
23210 non-constant character arrays.
23212 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
23214 PR middle-end/81897
23215 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
23218 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
23220 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
23222 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
23225 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
23226 VECTOR_MEM_ALTIVEC_OR_VSX_P.
23227 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
23228 indexed_or_indirect_operand predicate.
23229 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
23230 (*vsx_le_perm_load_v8hi): Likewise.
23231 (*vsx_le_perm_load_v16qi): Likewise.
23232 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
23233 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
23234 (*vsx_le_perm_store_v8hi): Likewise.
23235 (*vsx_le_perm_store_v16qi): Likewise.
23236 (eight unnamed splitters): Likewise.
23238 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
23240 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
23241 * config/rs6000/emmintrin.h: Likewise.
23242 * config/rs6000/mmintrin.h: Likewise.
23243 * config/rs6000/xmmintrin.h: Likewise.
23245 2018-01-10 David Malcolm <dmalcolm@redhat.com>
23248 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
23250 * tree.c (tree_nop_conversion): Return true for location wrapper
23252 (maybe_wrap_with_location): New function.
23253 (selftest::check_strip_nops): New function.
23254 (selftest::test_location_wrappers): New function.
23255 (selftest::tree_c_tests): Call it.
23256 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
23257 (maybe_wrap_with_location): New decl.
23258 (EXPR_LOCATION_WRAPPER_P): New macro.
23259 (location_wrapper_p): New inline function.
23260 (tree_strip_any_location_wrapper): New inline function.
23262 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
23265 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
23266 stack_realign_offset for the largest alignment of stack slot
23268 (ix86_find_max_used_stack_alignment): New function.
23269 (ix86_finalize_stack_frame_flags): Use it. Set
23270 max_used_stack_alignment if we don't realign stack.
23271 * config/i386/i386.h (machine_function): Add
23272 max_used_stack_alignment.
23274 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
23276 * config/arm/arm.opt (-mbranch-cost): New option.
23277 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
23280 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
23283 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
23284 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
23286 2018-01-10 Richard Biener <rguenther@suse.de>
23289 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
23290 early out so it also covers the case where we have a non-NULL
23293 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
23295 PR tree-optimization/83753
23296 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
23297 for non-strided grouped accesses if the number of elements is 1.
23299 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
23302 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
23303 * i386.h (TARGET_USE_GATHER): Define.
23304 * x86-tune.def (X86_TUNE_USE_GATHER): New.
23306 2018-01-10 Martin Liska <mliska@suse.cz>
23309 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
23310 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
23312 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
23313 CLEANUP_NO_PARTITIONING is not set.
23315 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
23317 * doc/rtl.texi: Remove documentation of (const ...) wrappers
23318 for vectors, as a partial revert of r254296.
23319 * rtl.h (const_vec_p): Delete.
23320 (const_vec_duplicate_p): Don't test for vector CONSTs.
23321 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
23322 * expmed.c (make_tree): Likewise.
23325 * common.md (E, F): Use CONSTANT_P instead of checking for
23327 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
23328 checking for CONST_VECTOR.
23330 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
23332 PR middle-end/83575
23333 * predict.c (force_edge_cold): Handle in more sane way edges
23334 with no prediction.
23336 2018-01-09 Carl Love <cel@us.ibm.com>
23338 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
23340 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
23341 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
23342 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
23343 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
23344 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
23345 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
23346 * config/rs6000/rs6000-protos.h: Add extern defition for
23347 rs6000_generate_float2_double_code.
23348 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
23350 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
23351 (float2_v2df): Add define_expand.
23353 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
23356 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
23357 op_mode in the force_to_mode call.
23359 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
23361 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
23362 instead of checking each element individually.
23363 (aarch64_evpc_uzp): Likewise.
23364 (aarch64_evpc_zip): Likewise.
23365 (aarch64_evpc_ext): Likewise.
23366 (aarch64_evpc_rev): Likewise.
23367 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
23368 instead of checking each element individually. Return true without
23370 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
23371 whether all selected elements come from the same input, instead of
23372 checking each element individually. Remove calls to gen_rtx_REG,
23373 start_sequence and end_sequence and instead assert that no rtl is
23376 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
23378 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
23379 order of HIGH and CONST checks.
23381 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
23383 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
23384 if the destination isn't an SSA_NAME.
23386 2018-01-09 Richard Biener <rguenther@suse.de>
23388 PR tree-optimization/83668
23389 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
23391 (canonicalize_loop_form): ... here, renamed from ...
23392 (canonicalize_loop_closed_ssa_form): ... this and amended to
23393 swap successor edges for loop exit blocks to make us use
23394 the RPO order we need for initial schedule generation.
23396 2018-01-09 Joseph Myers <joseph@codesourcery.com>
23398 PR tree-optimization/64811
23399 * match.pd: When optimizing comparisons with Inf, avoid
23400 introducing or losing exceptions from comparisons with NaN.
23402 2018-01-09 Martin Liska <mliska@suse.cz>
23405 * asan.c (shadow_mem_size): Add gcc_assert.
23407 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
23409 Don't save registers in main().
23412 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
23413 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
23414 * config/avr/avr.c (avr_set_current_function): Don't error if
23415 naked, OS_task or OS_main are specified at the same time.
23416 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
23418 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
23420 * common/config/avr/avr-common.c (avr_option_optimization_table):
23421 Switch on -mmain-is-OS_task for optimizing compilations.
23423 2018-01-09 Richard Biener <rguenther@suse.de>
23425 PR tree-optimization/83572
23426 * graphite.c: Include cfganal.h.
23427 (graphite_transform_loops): Connect infinite loops to exit
23428 and remove fake edges at the end.
23430 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
23432 * ipa-inline.c (edge_badness): Revert accidental checkin.
23434 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
23437 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
23438 symbols; not inline clones.
23440 2018-01-09 Jakub Jelinek <jakub@redhat.com>
23443 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
23444 hard registers. Formatting fixes.
23446 PR preprocessor/83722
23447 * gcc.c (try_generate_repro): Pass
23448 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
23449 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
23452 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
23453 Kito Cheng <kito.cheng@gmail.com>
23455 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
23456 (riscv_leaf_function_p): Delete.
23457 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
23459 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23461 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
23463 (do_ifelse): New function.
23464 (do_isel): New function.
23465 (do_sub3): New function.
23466 (do_add3): New function.
23467 (do_load_mask_compare): New function.
23468 (do_overlap_load_compare): New function.
23469 (expand_compare_loop): New function.
23470 (expand_block_compare): Call expand_compare_loop() when appropriate.
23471 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
23472 option description.
23473 (-mblock-compare-inline-loop-limit): New option.
23475 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23478 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
23479 Reverse order of second and third operands in first alternative.
23480 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
23481 of first and second elements in UNSPEC_VPERMR vector.
23482 (altivec_expand_vec_perm_le): Likewise.
23484 2018-01-08 Jeff Law <law@redhat.com>
23486 PR rtl-optimizatin/81308
23487 * tree-switch-conversion.c (cfg_altered): New file scoped static.
23488 (process_switch): If group_case_labels makes a change, then set
23490 (pass_convert_switch::execute): If a switch is converted, then
23491 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
23493 PR rtl-optimization/81308
23494 * recog.c (split_all_insns): Conditionally cleanup the CFG after
23497 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
23499 PR target/83663 - Revert r255946
23500 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
23501 generation for cases where splatting a value is not useful.
23502 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
23503 across a vec_duplicate and a paradoxical subreg forming a vector
23504 mode to a vec_concat.
23506 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23508 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
23509 -march=armv8.3-a variants.
23510 * config/arm/t-multilib: Likewise.
23511 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
23513 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23515 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
23517 (cceq_ior_compare_complement): Give it a name so I can use it, and
23518 change boolean_or_operator predicate to boolean_operator so it can
23519 be used to generate a crand.
23520 (eqne): New code iterator.
23521 (bd/bd_neg): New code_attrs.
23522 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
23523 a single define_insn.
23524 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
23525 decrement (bdnzt/bdnzf/bdzt/bdzf).
23526 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
23527 with the new names of the branch decrement patterns, and added the
23528 names of the branch decrement conditional patterns.
23530 2018-01-08 Richard Biener <rguenther@suse.de>
23532 PR tree-optimization/83563
23533 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
23536 2018-01-08 Richard Biener <rguenther@suse.de>
23538 PR middle-end/83713
23539 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
23541 2018-01-08 Richard Biener <rguenther@suse.de>
23543 PR tree-optimization/83685
23544 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
23545 references to abnormals.
23547 2018-01-08 Richard Biener <rguenther@suse.de>
23550 * dwarf2out.c (output_indirect_strings): Handle empty
23551 skeleton_debug_str_hash.
23552 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
23554 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
23556 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
23557 (emit_store_direct): Likewise.
23558 (arc_trampoline_adjust_address): Likewise.
23559 (arc_asm_trampoline_template): New function.
23560 (arc_initialize_trampoline): Use asm_trampoline_template.
23561 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
23562 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
23563 * config/arc/arc.md (flush_icache): Delete pattern.
23565 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
23567 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
23568 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
23571 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
23574 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
23575 by not USED_FOR_TARGET.
23576 (make_pass_resolve_sw_modes): Likewise.
23578 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
23580 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
23583 2018-01-08 Richard Biener <rguenther@suse.de>
23585 PR middle-end/83580
23586 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
23588 2018-01-08 Richard Biener <rguenther@suse.de>
23590 PR middle-end/83517
23591 * match.pd ((t * 2) / 2) -> t): Add missing :c.
23593 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
23595 PR middle-end/81897
23596 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
23597 basic blocks with a small number of successors.
23598 (convert_control_dep_chain_into_preds): Improve handling of
23600 (dump_predicates): Split apart into...
23601 (dump_pred_chain): ...here...
23602 (dump_pred_info): ...and here.
23603 (can_one_predicate_be_invalidated_p): Add debugging printfs.
23604 (can_chain_union_be_invalidated_p): Improve check for invalidation
23606 (uninit_uses_cannot_happen): Avoid unnecessary if
23607 convert_control_dep_chain_into_preds yielded nothing.
23609 2018-01-06 Martin Sebor <msebor@redhat.com>
23611 PR tree-optimization/83640
23612 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
23613 subtracting negative offset from size.
23614 (builtin_access::overlap): Adjust offset bounds of the access to fall
23615 within the size of the object if possible.
23617 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
23619 PR rtl-optimization/83699
23620 * expmed.c (extract_bit_field_1): Restrict the vector usage of
23621 extract_bit_field_as_subreg to cases in which the extracted
23622 value is also a vector.
23624 * lra-constraints.c (process_alt_operands): Test for the equivalence
23625 substitutions when detecting a possible reload cycle.
23627 2018-01-06 Jakub Jelinek <jakub@redhat.com>
23630 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
23631 by default if flag_selective_schedling{,2}. Formatting fixes.
23633 PR rtl-optimization/83682
23634 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
23635 if it has non-VECTOR_MODE element mode.
23636 (vec_duplicate_p): Likewise.
23638 PR middle-end/83694
23639 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
23640 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
23642 2018-01-05 Jakub Jelinek <jakub@redhat.com>
23645 * config/i386/i386-builtin.def
23646 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
23647 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
23648 Require also OPTION_MASK_ISA_AVX512F in addition to
23649 OPTION_MASK_ISA_GFNI.
23650 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
23651 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
23652 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
23653 to OPTION_MASK_ISA_GFNI.
23654 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
23655 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
23656 OPTION_MASK_ISA_AVX512BW.
23657 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
23658 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
23659 addition to OPTION_MASK_ISA_GFNI.
23660 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
23661 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
23662 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
23663 to OPTION_MASK_ISA_GFNI.
23664 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
23665 a requirement for all ISAs rather than any of them with a few
23667 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
23669 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
23670 bitmasks to be enabled with 3 exceptions, instead of requiring any
23671 enabled ISA with lots of exceptions.
23672 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
23673 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
23674 Change avx512bw in isa attribute to avx512f.
23675 * config/i386/sgxintrin.h: Add license boilerplate.
23676 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
23677 to __AVX512F__ and __AVX512VL to __AVX512VL__.
23678 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
23679 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
23681 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
23682 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
23683 temporarily sse2 rather than sse if not enabled already.
23686 * config/i386/sse.md (VI248_VLBW): Rename to ...
23687 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
23688 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
23689 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
23690 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
23691 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
23692 mode iterator instead of VI248_VLBW.
23694 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
23696 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
23697 (record_modified): Skip clobbers; add debug output.
23698 (param_change_prob): Use sreal frequencies.
23700 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
23702 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
23703 punt for user-aligned variables.
23705 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
23707 * tree-chrec.c (chrec_contains_symbols): Return true for
23710 2018-01-05 Sudakshina Das <sudi.das@arm.com>
23713 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
23714 of (x|y) == x for BICS pattern.
23716 2018-01-05 Jakub Jelinek <jakub@redhat.com>
23718 PR tree-optimization/83605
23719 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
23720 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
23723 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
23725 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
23726 * config/epiphany/rtems.h: New file.
23728 2018-01-04 Jakub Jelinek <jakub@redhat.com>
23729 Uros Bizjak <ubizjak@gmail.com>
23732 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
23733 QIreg_operand instead of register_operand predicate.
23734 * config/i386/i386.c (ix86_rop_should_change_byte_p,
23735 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
23736 comments instead of -fmitigate[-_]rop.
23738 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23741 * cgraphunit.c (symbol_table::compile): Switch to text_section
23742 before calling assembly_start debug hook.
23743 * run-rtl-passes.c (run_rtl_passes): Likewise.
23746 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23748 * tree-vrp.c (extract_range_from_binary_expr_1): Check
23749 range_int_cst_p rather than !symbolic_range_p before calling
23750 extract_range_from_multiplicative_op_1.
23752 2018-01-04 Jeff Law <law@redhat.com>
23754 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
23755 redundant test in assertion.
23757 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23759 * doc/rtl.texi: Document machine_mode wrapper classes.
23761 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23763 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
23764 using tree_to_uhwi.
23766 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23768 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
23769 the VEC_PERM_EXPR fold to fail.
23771 2018-01-04 Jakub Jelinek <jakub@redhat.com>
23774 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
23775 to switched_sections.
23777 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23780 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
23781 test for d.testing.
23783 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
23786 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
23787 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
23789 2018-01-04 Jakub Jelinek <jakub@redhat.com>
23792 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
23793 is BLKmode and bitpos not zero or mode change is needed.
23795 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23798 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
23801 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
23804 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
23805 instead of MULT rtx. Update all corresponding splitters.
23806 (*saddl_se): Ditto.
23807 (*ssub<modesuffix>): Ditto.
23808 (*ssubl_se): Ditto.
23809 (*cmp_sadd_di): Update split patterns.
23810 (*cmp_sadd_si): Ditto.
23811 (*cmp_sadd_sidi): Ditto.
23812 (*cmp_ssub_di): Ditto.
23813 (*cmp_ssub_si): Ditto.
23814 (*cmp_ssub_sidi): Ditto.
23815 * config/alpha/predicates.md (const23_operand): New predicate.
23816 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
23817 Look for ASHIFT, not MULT inner operand.
23818 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
23820 2018-01-04 Martin Liska <mliska@suse.cz>
23822 PR gcov-profile/83669
23823 * gcov.c (output_intermediate_file): Add version to intermediate
23825 * doc/gcov.texi: Document new field 'version' in intermediate
23826 file format. Fix location of '-k' option of gcov command.
23828 2018-01-04 Martin Liska <mliska@suse.cz>
23831 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
23833 2018-01-04 Jakub Jelinek <jakub@redhat.com>
23835 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
23837 2018-01-03 Martin Sebor <msebor@redhat.com>
23839 PR tree-optimization/83655
23840 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
23841 checking calls with invalid arguments.
23843 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23845 * tree-vect-stmts.c (vect_get_store_rhs): New function.
23846 (vectorizable_mask_load_store): Delete.
23847 (vectorizable_call): Return false for masked loads and stores.
23848 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
23849 instead of gimple_assign_rhs1.
23850 (vectorizable_load): Handle IFN_MASK_LOAD.
23851 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
23853 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23855 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
23857 (vectorizable_mask_load_store): ...here.
23858 (vectorizable_load): ...and here.
23860 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23862 * tree-vect-stmts.c (vect_build_all_ones_mask)
23863 (vect_build_zero_merge_argument): New functions, split out from...
23864 (vectorizable_load): ...here.
23866 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23868 * tree-vect-stmts.c (vect_check_store_rhs): New function,
23870 (vectorizable_mask_load_store): ...here.
23871 (vectorizable_store): ...and here.
23873 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23875 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
23877 (vectorizable_mask_load_store): ...here.
23879 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23881 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
23882 (vect_model_store_cost): Take a vec_load_store_type instead of a
23884 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
23885 (vect_model_store_cost): Take a vec_load_store_type instead of a
23887 (vectorizable_mask_load_store): Update accordingly.
23888 (vectorizable_store): Likewise.
23889 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
23891 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23893 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
23894 IFN_MASK_LOAD calls here rather than...
23895 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
23897 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23898 Alan Hayward <alan.hayward@arm.com>
23899 David Sherwood <david.sherwood@arm.com>
23901 * expmed.c (extract_bit_field_1): For vector extracts,
23902 fall back to extract_bit_field_as_subreg if vec_extract
23905 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23906 Alan Hayward <alan.hayward@arm.com>
23907 David Sherwood <david.sherwood@arm.com>
23909 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
23910 they are variable or constant sized.
23911 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
23912 slots for constant-sized data.
23914 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23915 Alan Hayward <alan.hayward@arm.com>
23916 David Sherwood <david.sherwood@arm.com>
23918 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
23919 handling COND_EXPRs with boolean comparisons, try to find a better
23920 basis for the mask type than the boolean itself.
23922 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23924 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
23925 is calculated and how it can be overridden.
23926 * genmodes.c (max_bitsize_mode_any_mode): New variable.
23927 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
23929 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
23932 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23933 Alan Hayward <alan.hayward@arm.com>
23934 David Sherwood <david.sherwood@arm.com>
23936 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
23937 Remove the mode argument.
23938 (aarch64_simd_valid_immediate): Remove the mode and inverse
23940 * config/aarch64/iterators.md (bitsize): New iterator.
23941 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
23942 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
23943 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
23944 aarch64_simd_valid_immediate.
23945 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
23946 (aarch64_reg_or_bic_imm): Likewise.
23947 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
23948 with an insn_type enum and msl with a modifier_type enum.
23949 Replace element_width with a scalar_mode. Change the shift
23950 to unsigned int. Add constructors for scalar_float_mode and
23951 scalar_int_mode elements.
23952 (aarch64_vect_float_const_representable_p): Delete.
23953 (aarch64_can_const_movi_rtx_p)
23954 (aarch64_simd_scalar_immediate_valid_for_move)
23955 (aarch64_simd_make_constant): Update call to
23956 aarch64_simd_valid_immediate.
23957 (aarch64_advsimd_valid_immediate_hs): New function.
23958 (aarch64_advsimd_valid_immediate): Likewise.
23959 (aarch64_simd_valid_immediate): Remove mode and inverse
23960 arguments. Rewrite to use the above. Use const_vec_duplicate_p
23961 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
23962 and aarch64_float_const_representable_p on the result.
23963 (aarch64_output_simd_mov_immediate): Remove mode argument.
23964 Update call to aarch64_simd_valid_immediate and use of
23965 simd_immediate_info.
23966 (aarch64_output_scalar_simd_mov_immediate): Update call
23969 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23970 Alan Hayward <alan.hayward@arm.com>
23971 David Sherwood <david.sherwood@arm.com>
23973 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
23974 (mode_nunits): Likewise CONST_MODE_NUNITS.
23975 * machmode.def (ADJUST_NUNITS): Document.
23976 * genmodes.c (mode_data::need_nunits_adj): New field.
23977 (blank_mode): Update accordingly.
23978 (adj_nunits): New variable.
23979 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
23981 (emit_mode_size_inline): Set need_bytesize_adj for all modes
23982 listed in adj_nunits.
23983 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
23984 listed in adj_nunits. Don't emit case statements for such modes.
23985 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
23986 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
23987 nothing if adj_nunits is nonnull.
23988 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
23989 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
23990 (emit_mode_fbit): Update use of print_maybe_const_decl.
23991 (emit_move_size): Likewise. Treat the array as non-const
23993 (emit_mode_adjustments): Handle adj_nunits.
23995 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23997 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
23998 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
23999 (VECTOR_MODES): Use it.
24000 (make_vector_modes): Take the prefix as an argument.
24002 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24003 Alan Hayward <alan.hayward@arm.com>
24004 David Sherwood <david.sherwood@arm.com>
24006 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
24007 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
24008 for MODE_VECTOR_BOOL.
24009 * machmode.def (VECTOR_BOOL_MODE): Document.
24010 * genmodes.c (VECTOR_BOOL_MODE): New macro.
24011 (make_vector_bool_mode): New function.
24012 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
24014 * lto-streamer-in.c (lto_input_mode_table): Likewise.
24015 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
24017 * stor-layout.c (int_mode_for_mode): Likewise.
24018 * tree.c (build_vector_type_for_mode): Likewise.
24019 * varasm.c (output_constant_pool_2): Likewise.
24020 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
24021 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
24022 for MODE_VECTOR_BOOL.
24023 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
24024 of mode class checks.
24025 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
24026 instead of a list of mode class checks.
24027 (expand_vector_scalar_condition): Likewise.
24028 (type_for_widest_vector_mode): Handle BImode as an inner mode.
24030 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24031 Alan Hayward <alan.hayward@arm.com>
24032 David Sherwood <david.sherwood@arm.com>
24034 * machmode.h (mode_size): Change from unsigned short to
24036 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
24037 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
24038 or if measurement_type is not polynomial.
24039 (fixed_size_mode::includes_p): Check for constant-sized modes.
24040 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
24041 return a poly_uint16 rather than an unsigned short.
24042 (emit_mode_size): Change the type of mode_size from unsigned short
24043 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
24044 (emit_mode_adjustments): Cope with polynomial vector sizes.
24045 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24047 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24049 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
24050 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
24051 * caller-save.c (setup_save_areas): Likewise.
24052 (replace_reg_with_saved_mem): Likewise.
24053 * calls.c (emit_library_call_value_1): Likewise.
24054 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
24055 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
24056 (gen_lowpart_for_combine): Likewise.
24057 * convert.c (convert_to_integer_1): Likewise.
24058 * cse.c (equiv_constant, cse_insn): Likewise.
24059 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
24060 (cselib_subst_to_values): Likewise.
24061 * dce.c (word_dce_process_block): Likewise.
24062 * df-problems.c (df_word_lr_mark_ref): Likewise.
24063 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
24064 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
24065 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
24066 (rtl_for_decl_location): Likewise.
24067 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
24068 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
24069 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
24070 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
24071 (expand_expr_real_1): Likewise.
24072 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
24073 (pad_below): Likewise.
24074 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
24075 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
24076 * ira.c (get_subreg_tracking_sizes): Likewise.
24077 * ira-build.c (ira_create_allocno_objects): Likewise.
24078 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
24079 (ira_sort_regnos_for_alter_reg): Likewise.
24080 * ira-costs.c (record_operand_costs): Likewise.
24081 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
24082 (resolve_simple_move): Likewise.
24083 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
24084 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
24085 (lra_constraints): Likewise.
24086 (CONST_POOL_OK_P): Reject variable-sized modes.
24087 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
24088 (add_pseudo_to_slot, lra_spill): Likewise.
24089 * omp-low.c (omp_clause_aligned_alignment): Likewise.
24090 * optabs-query.c (get_best_extraction_insn): Likewise.
24091 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
24092 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
24093 (expand_mult_highpart, valid_multiword_target_p): Likewise.
24094 * recog.c (offsettable_address_addr_space_p): Likewise.
24095 * regcprop.c (maybe_mode_change): Likewise.
24096 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
24097 * regrename.c (build_def_use): Likewise.
24098 * regstat.c (dump_reg_info): Likewise.
24099 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
24100 (find_reloads, find_reloads_subreg_address): Likewise.
24101 * reload1.c (eliminate_regs_1): Likewise.
24102 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
24103 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
24104 (simplify_binary_operation_1, simplify_subreg): Likewise.
24105 * targhooks.c (default_function_arg_padding): Likewise.
24106 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
24107 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
24108 (verify_gimple_assign_ternary): Likewise.
24109 * tree-inline.c (estimate_move_cost): Likewise.
24110 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
24111 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
24112 (get_address_cost_ainc): Likewise.
24113 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
24114 (vect_supportable_dr_alignment): Likewise.
24115 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24116 (vectorizable_reduction): Likewise.
24117 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
24118 (vectorizable_operation, vectorizable_load): Likewise.
24119 * tree.c (build_same_sized_truth_vector_type): Likewise.
24120 * valtrack.c (cleanup_auto_inc_dec): Likewise.
24121 * var-tracking.c (emit_note_insn_var_location): Likewise.
24122 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
24123 (ADDR_VEC_ALIGN): Likewise.
24125 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24126 Alan Hayward <alan.hayward@arm.com>
24127 David Sherwood <david.sherwood@arm.com>
24129 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
24131 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
24132 or if measurement_type is polynomial.
24133 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
24134 * combine.c (make_extraction): Likewise.
24135 * dse.c (find_shift_sequence): Likewise.
24136 * dwarf2out.c (mem_loc_descriptor): Likewise.
24137 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
24138 (extract_bit_field, extract_low_bits): Likewise.
24139 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
24140 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
24141 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
24142 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
24143 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
24144 * reload.c (find_reloads): Likewise.
24145 * reload1.c (alter_reg): Likewise.
24146 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
24147 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
24148 * tree-if-conv.c (predicate_mem_writes): Likewise.
24149 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
24150 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
24151 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
24152 * valtrack.c (dead_debug_insert_temp): Likewise.
24153 * varasm.c (mergeable_constant_section): Likewise.
24154 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
24156 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24157 Alan Hayward <alan.hayward@arm.com>
24158 David Sherwood <david.sherwood@arm.com>
24160 * expr.c (expand_assignment): Cope with polynomial mode sizes
24161 when assigning to a CONCAT.
24163 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24164 Alan Hayward <alan.hayward@arm.com>
24165 David Sherwood <david.sherwood@arm.com>
24167 * machmode.h (mode_precision): Change from unsigned short to
24169 (mode_to_precision): Return a poly_uint16 rather than an unsigned
24171 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
24172 or if measurement_type is not polynomial.
24173 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
24174 in which the mode is already known to be a scalar_int_mode.
24175 * genmodes.c (emit_mode_precision): Change the type of mode_precision
24176 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
24178 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24179 for GET_MODE_PRECISION.
24180 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24181 for GET_MODE_PRECISION.
24182 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
24184 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
24185 (expand_field_assignment, make_extraction): Likewise.
24186 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
24187 (get_last_value): Likewise.
24188 * convert.c (convert_to_integer_1): Likewise.
24189 * cse.c (cse_insn): Likewise.
24190 * expr.c (expand_expr_real_1): Likewise.
24191 * lra-constraints.c (simplify_operand_subreg): Likewise.
24192 * optabs-query.c (can_atomic_load_p): Likewise.
24193 * optabs.c (expand_atomic_load): Likewise.
24194 (expand_atomic_store): Likewise.
24195 * ree.c (combine_reaching_defs): Likewise.
24196 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
24197 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
24198 * tree.h (type_has_mode_precision_p): Likewise.
24199 * ubsan.c (instrument_si_overflow): Likewise.
24201 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24202 Alan Hayward <alan.hayward@arm.com>
24203 David Sherwood <david.sherwood@arm.com>
24205 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
24206 polynomial numbers of units.
24207 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
24208 (valid_vector_subparts_p): New function.
24209 (build_vector_type): Remove temporary shim and take the number
24210 of units as a poly_uint64 rather than an int.
24211 (build_opaque_vector_type): Take the number of units as a
24212 poly_uint64 rather than an int.
24213 * tree.c (build_vector_from_ctor): Handle polynomial
24214 TYPE_VECTOR_SUBPARTS.
24215 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
24216 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
24217 (build_vector_from_val): If the number of units is variable,
24218 use build_vec_duplicate_cst for constant operands and
24219 VEC_DUPLICATE_EXPR otherwise.
24220 (make_vector_type): Remove temporary is_constant ().
24221 (build_vector_type, build_opaque_vector_type): Take the number of
24222 units as a poly_uint64 rather than an int.
24223 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
24225 * cfgexpand.c (expand_debug_expr): Likewise.
24226 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
24227 (store_constructor, expand_expr_real_1): Likewise.
24228 (const_scalar_mask_from_tree): Likewise.
24229 * fold-const-call.c (fold_const_reduction): Likewise.
24230 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
24231 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
24232 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
24233 (fold_relational_const): Likewise.
24234 (native_interpret_vector): Likewise. Change the size from an
24235 int to an unsigned int.
24236 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
24237 TYPE_VECTOR_SUBPARTS.
24238 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
24239 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
24240 duplicating a non-constant operand into a variable-length vector.
24241 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
24242 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
24243 * ipa-icf.c (sem_variable::equals): Likewise.
24244 * match.pd: Likewise.
24245 * omp-simd-clone.c (simd_clone_subparts): Likewise.
24246 * print-tree.c (print_node): Likewise.
24247 * stor-layout.c (layout_type): Likewise.
24248 * targhooks.c (default_builtin_vectorization_cost): Likewise.
24249 * tree-cfg.c (verify_gimple_comparison): Likewise.
24250 (verify_gimple_assign_binary): Likewise.
24251 (verify_gimple_assign_ternary): Likewise.
24252 (verify_gimple_assign_single): Likewise.
24253 * tree-pretty-print.c (dump_generic_node): Likewise.
24254 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
24255 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
24256 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
24257 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
24258 (vect_shift_permute_load_chain): Likewise.
24259 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
24260 (expand_vector_condition, optimize_vector_constructor): Likewise.
24261 (lower_vec_perm, get_compute_type): Likewise.
24262 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24263 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
24264 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
24265 (vect_recog_mask_conversion_pattern): Likewise.
24266 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
24267 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
24268 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
24269 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
24270 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
24271 (vectorizable_shift, vectorizable_operation, vectorizable_store)
24272 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
24273 (supportable_widening_operation): Likewise.
24274 (supportable_narrowing_operation): Likewise.
24275 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
24277 * varasm.c (output_constant): Likewise.
24279 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24280 Alan Hayward <alan.hayward@arm.com>
24281 David Sherwood <david.sherwood@arm.com>
24283 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
24284 so that both the length == 3 and length != 3 cases set up their
24285 own permute vectors. Add comments explaining why we know the
24286 number of elements is constant.
24287 (vect_permute_load_chain): Likewise.
24289 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24290 Alan Hayward <alan.hayward@arm.com>
24291 David Sherwood <david.sherwood@arm.com>
24293 * machmode.h (mode_nunits): Change from unsigned char to
24295 (ONLY_FIXED_SIZE_MODES): New macro.
24296 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
24297 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
24298 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
24300 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
24301 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
24302 or if measurement_type is not polynomial.
24303 * genmodes.c (ZERO_COEFFS): New macro.
24304 (emit_mode_nunits_inline): Make mode_nunits_inline return a
24306 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
24307 Use ZERO_COEFFS when emitting initializers.
24308 * data-streamer.h (bp_pack_poly_value): New function.
24309 (bp_unpack_poly_value): Likewise.
24310 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24311 for GET_MODE_NUNITS.
24312 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24313 for GET_MODE_NUNITS.
24314 * tree.c (make_vector_type): Remove temporary shim and make
24315 the real function take the number of units as a poly_uint64
24316 rather than an int.
24317 (build_vector_type_for_mode): Handle polynomial nunits.
24318 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
24319 * emit-rtl.c (const_vec_series_p_1): Likewise.
24320 (gen_rtx_CONST_VECTOR): Likewise.
24321 * fold-const.c (test_vec_duplicate_folding): Likewise.
24322 * genrecog.c (validate_pattern): Likewise.
24323 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
24324 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
24325 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
24326 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
24327 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
24328 * rtlanal.c (subreg_get_info): Likewise.
24329 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
24330 (vect_grouped_load_supported): Likewise.
24331 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
24332 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
24333 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
24334 (simplify_const_unary_operation, simplify_binary_operation_1)
24335 (simplify_const_binary_operation, simplify_ternary_operation)
24336 (test_vector_ops_duplicate, test_vector_ops): Likewise.
24337 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
24338 instead of CONST_VECTOR_NUNITS.
24339 * varasm.c (output_constant_pool_2): Likewise.
24340 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
24341 explicit-encoded elements in the XVEC for variable-length vectors.
24343 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24345 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
24347 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24348 Alan Hayward <alan.hayward@arm.com>
24349 David Sherwood <david.sherwood@arm.com>
24351 * coretypes.h (fixed_size_mode): Declare.
24352 (fixed_size_mode_pod): New typedef.
24353 * builtins.h (target_builtins::x_apply_args_mode)
24354 (target_builtins::x_apply_result_mode): Change type to
24355 fixed_size_mode_pod.
24356 * builtins.c (apply_args_size, apply_result_size, result_vector)
24357 (expand_builtin_apply_args_1, expand_builtin_apply)
24358 (expand_builtin_return): Update accordingly.
24360 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24362 * cse.c (hash_rtx_cb): Hash only the encoded elements.
24363 * cselib.c (cselib_hash_rtx): Likewise.
24364 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
24365 CONST_VECTOR encoding.
24367 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24368 Jeff Law <law@redhat.com>
24371 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
24372 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
24373 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
24374 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
24377 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
24378 explicitly probe *sp in a noreturn function if there were any callee
24379 register saves or frame pointer is needed.
24381 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24384 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
24385 BLKmode for ternary, binary or unary expressions.
24388 * var-tracking.c (delete_vta_debug_insn): New inline function.
24389 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
24390 insns from get_insns () to NULL instead of each bb separately.
24391 Use delete_vta_debug_insn. No longer static.
24392 (vt_debug_insns_local, variable_tracking_main_1): Adjust
24393 delete_vta_debug_insns callers.
24394 * rtl.h (delete_vta_debug_insns): Declare.
24395 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
24396 instead of variable_tracking_main.
24398 2018-01-03 Martin Sebor <msebor@redhat.com>
24400 PR tree-optimization/83603
24401 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
24402 arguments past the endof the argument list in functions declared
24403 without a prototype.
24404 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
24405 Avoid checking when arguments are null.
24407 2018-01-03 Martin Sebor <msebor@redhat.com>
24410 * doc/extend.texi (attribute const): Fix a typo.
24411 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
24412 issuing -Wsuggest-attribute for void functions.
24414 2018-01-03 Martin Sebor <msebor@redhat.com>
24416 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
24417 offset_int::from instead of wide_int::to_shwi.
24418 (maybe_diag_overlap): Remove assertion.
24419 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
24420 * gimple-ssa-sprintf.c (format_directive): Same.
24421 (parse_directive): Same.
24422 (sprintf_dom_walker::compute_format_length): Same.
24423 (try_substitute_return_value): Same.
24425 2018-01-03 Jeff Law <law@redhat.com>
24427 PR middle-end/83654
24428 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
24429 non-constant residual for zero at runtime and avoid probing in
24430 that case. Reorganize code for trailing problem to mirror handling
24433 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24435 PR tree-optimization/83501
24436 * tree-ssa-strlen.c (get_string_cst): New.
24437 (handle_char_store): Call get_string_cst.
24439 2018-01-03 Martin Liska <mliska@suse.cz>
24441 PR tree-optimization/83593
24442 * tree-ssa-strlen.c: Include tree-cfg.h.
24443 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
24444 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
24445 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
24447 (strlen_dom_walker::before_dom_children): Call
24448 gimple_purge_dead_eh_edges. Dump tranformation with details
24450 (strlen_dom_walker::before_dom_children): Update call by adding
24451 new argument cleanup_eh.
24452 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
24454 2018-01-03 Martin Liska <mliska@suse.cz>
24457 * cif-code.def (VARIADIC_THUNK): New enum value.
24458 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
24461 2018-01-03 Jan Beulich <jbeulich@suse.com>
24463 * sse.md (mov<mode>_internal): Tighten condition for when to use
24464 vmovdqu<ssescalarsize> for TI and OI modes.
24466 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24468 Update copyright years.
24470 2018-01-03 Martin Liska <mliska@suse.cz>
24473 * ipa-visibility.c (function_and_variable_visibility): Skip
24474 functions with noipa attribure.
24476 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24478 * gcc.c (process_command): Update copyright notice dates.
24479 * gcov-dump.c (print_version): Ditto.
24480 * gcov.c (print_version): Ditto.
24481 * gcov-tool.c (print_version): Ditto.
24482 * gengtype.c (create_file): Ditto.
24483 * doc/cpp.texi: Bump @copying's copyright year.
24484 * doc/cppinternals.texi: Ditto.
24485 * doc/gcc.texi: Ditto.
24486 * doc/gccint.texi: Ditto.
24487 * doc/gcov.texi: Ditto.
24488 * doc/install.texi: Ditto.
24489 * doc/invoke.texi: Ditto.
24491 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24493 * vector-builder.h (vector_builder::m_full_nelts): Change from
24494 unsigned int to poly_uint64.
24495 (vector_builder::full_nelts): Update prototype accordingly.
24496 (vector_builder::new_vector): Likewise.
24497 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
24498 (vector_builder::operator ==): Likewise.
24499 (vector_builder::finalize): Likewise.
24500 * int-vector-builder.h (int_vector_builder::int_vector_builder):
24501 Take the number of elements as a poly_uint64 rather than an
24503 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
24504 from unsigned int to poly_uint64.
24505 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
24506 (vec_perm_indices::new_vector): Likewise.
24507 (vec_perm_indices::length): Likewise.
24508 (vec_perm_indices::nelts_per_input): Likewise.
24509 (vec_perm_indices::input_nelts): Likewise.
24510 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
24511 number of elements per input as a poly_uint64 rather than an
24512 unsigned int. Use the original encoding for variable-length
24513 vectors, rather than clamping each individual element.
24514 For the second and subsequent elements in each pattern,
24515 clamp the step and base before clamping their sum.
24516 (vec_perm_indices::series_p): Handle polynomial element counts.
24517 (vec_perm_indices::all_in_range_p): Likewise.
24518 (vec_perm_indices_to_tree): Likewise.
24519 (vec_perm_indices_to_rtx): Likewise.
24520 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
24521 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
24522 (tree_vector_builder::new_binary_operation): Handle polynomial
24523 element counts. Return false if we need to know the number
24524 of elements at compile time.
24525 * fold-const.c (fold_vec_perm): Punt if the number of elements
24526 isn't known at compile time.
24528 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24530 * vec-perm-indices.h (vec_perm_builder): Change element type
24531 from HOST_WIDE_INT to poly_int64.
24532 (vec_perm_indices::element_type): Update accordingly.
24533 (vec_perm_indices::clamp): Handle polynomial element_types.
24534 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
24535 (vec_perm_indices::all_in_range_p): Likewise.
24536 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
24538 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
24539 polynomial vec_perm_indices element types.
24540 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
24541 * fold-const.c (fold_vec_perm): Likewise.
24542 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
24543 * tree-vect-generic.c (lower_vec_perm): Likewise.
24544 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
24545 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
24546 element type to HOST_WIDE_INT.
24548 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24549 Alan Hayward <alan.hayward@arm.com>
24550 David Sherwood <david.sherwood@arm.com>
24552 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
24553 rather than an int. Use plus_constant.
24554 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
24555 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
24557 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24558 Alan Hayward <alan.hayward@arm.com>
24559 David Sherwood <david.sherwood@arm.com>
24561 * calls.c (emit_call_1, expand_call): Change struct_value_size from
24562 a HOST_WIDE_INT to a poly_int64.
24564 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24565 Alan Hayward <alan.hayward@arm.com>
24566 David Sherwood <david.sherwood@arm.com>
24568 * calls.c (load_register_parameters): Cope with polynomial
24569 mode sizes. Require a constant size for BLKmode parameters
24570 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
24571 forces a parameter to be padded at the lsb end in order to
24572 fill a complete number of words, require the parameter size
24573 to be ordered wrt UNITS_PER_WORD.
24575 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24576 Alan Hayward <alan.hayward@arm.com>
24577 David Sherwood <david.sherwood@arm.com>
24579 * reload1.c (spill_stack_slot_width): Change element type
24580 from unsigned int to poly_uint64_pod.
24581 (alter_reg): Treat mode sizes as polynomial.
24583 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24584 Alan Hayward <alan.hayward@arm.com>
24585 David Sherwood <david.sherwood@arm.com>
24587 * reload.c (complex_word_subreg_p): New function.
24588 (reload_inner_reg_of_subreg, push_reload): Use it.
24590 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24591 Alan Hayward <alan.hayward@arm.com>
24592 David Sherwood <david.sherwood@arm.com>
24594 * lra-constraints.c (process_alt_operands): Reject matched
24595 operands whose sizes aren't ordered.
24596 (match_reload): Refer to this check here.
24598 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24599 Alan Hayward <alan.hayward@arm.com>
24600 David Sherwood <david.sherwood@arm.com>
24602 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
24603 that the mode size is in the set {1, 2, 4, 8, 16}.
24605 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24606 Alan Hayward <alan.hayward@arm.com>
24607 David Sherwood <david.sherwood@arm.com>
24609 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
24610 Use plus_constant instead of gen_rtx_PLUS.
24612 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24613 Alan Hayward <alan.hayward@arm.com>
24614 David Sherwood <david.sherwood@arm.com>
24616 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
24617 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
24618 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
24619 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
24620 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
24621 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
24622 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
24623 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
24624 * config/i386/i386.c (ix86_push_rounding): ...this new function.
24625 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
24627 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
24628 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
24629 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
24630 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
24631 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
24632 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
24633 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
24634 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
24635 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
24636 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
24638 * expr.c (emit_move_resolve_push): Treat the input and result
24639 of PUSH_ROUNDING as a poly_int64.
24640 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
24641 (emit_push_insn): Likewise.
24642 * lra-eliminations.c (mark_not_eliminable): Likewise.
24643 * recog.c (push_operand): Likewise.
24644 * reload1.c (elimination_effects): Likewise.
24645 * rtlanal.c (nonzero_bits1): Likewise.
24646 * calls.c (store_one_arg): Likewise. Require the padding to be
24647 known at compile time.
24649 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24650 Alan Hayward <alan.hayward@arm.com>
24651 David Sherwood <david.sherwood@arm.com>
24653 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
24654 Use plus_constant instead of gen_rtx_PLUS.
24656 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24657 Alan Hayward <alan.hayward@arm.com>
24658 David Sherwood <david.sherwood@arm.com>
24660 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
24661 rather than an int.
24663 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24664 Alan Hayward <alan.hayward@arm.com>
24665 David Sherwood <david.sherwood@arm.com>
24667 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
24668 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
24669 via stack temporaries. Treat the mode size as polynomial too.
24671 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24672 Alan Hayward <alan.hayward@arm.com>
24673 David Sherwood <david.sherwood@arm.com>
24675 * expr.c (expand_expr_real_2): When handling conversions involving
24676 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
24677 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
24678 as a poly_uint64 too.
24680 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24681 Alan Hayward <alan.hayward@arm.com>
24682 David Sherwood <david.sherwood@arm.com>
24684 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
24686 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24687 Alan Hayward <alan.hayward@arm.com>
24688 David Sherwood <david.sherwood@arm.com>
24690 * combine.c (can_change_dest_mode): Handle polynomial
24691 REGMODE_NATURAL_SIZE.
24692 * expmed.c (store_bit_field_1): Likewise.
24693 * expr.c (store_constructor): Likewise.
24694 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
24695 and polynomial REGMODE_NATURAL_SIZE.
24696 (gen_lowpart_common): Likewise.
24697 * reginfo.c (record_subregs_of_mode): Likewise.
24698 * rtlanal.c (read_modify_subreg_p): Likewise.
24700 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24701 Alan Hayward <alan.hayward@arm.com>
24702 David Sherwood <david.sherwood@arm.com>
24704 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
24705 numbers of elements.
24707 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24708 Alan Hayward <alan.hayward@arm.com>
24709 David Sherwood <david.sherwood@arm.com>
24711 * match.pd: Cope with polynomial numbers of vector elements.
24713 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24714 Alan Hayward <alan.hayward@arm.com>
24715 David Sherwood <david.sherwood@arm.com>
24717 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
24718 in a POINTER_PLUS_EXPR.
24720 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24721 Alan Hayward <alan.hayward@arm.com>
24722 David Sherwood <david.sherwood@arm.com>
24724 * omp-simd-clone.c (simd_clone_subparts): New function.
24725 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
24726 (ipa_simd_modify_function_body): Likewise.
24728 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24729 Alan Hayward <alan.hayward@arm.com>
24730 David Sherwood <david.sherwood@arm.com>
24732 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
24733 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
24734 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
24735 (expand_vector_condition, vector_element): Likewise.
24736 (subparts_gt): New function.
24737 (get_compute_type): Use subparts_gt.
24738 (count_type_subparts): Delete.
24739 (expand_vector_operations_1): Use subparts_gt instead of
24740 count_type_subparts.
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 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
24747 (vect_compile_time_alias): ...this new function. Do the calculation
24748 on poly_ints rather than trees.
24749 (vect_prune_runtime_alias_test_list): Update call accordingly.
24751 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24752 Alan Hayward <alan.hayward@arm.com>
24753 David Sherwood <david.sherwood@arm.com>
24755 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
24757 (vect_schedule_slp_instance): Likewise.
24759 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24760 Alan Hayward <alan.hayward@arm.com>
24761 David Sherwood <david.sherwood@arm.com>
24763 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
24764 constant and extern definitions for variable-length vectors.
24765 (vect_get_constant_vectors): Note that the number of units
24766 is known to be constant.
24768 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24769 Alan Hayward <alan.hayward@arm.com>
24770 David Sherwood <david.sherwood@arm.com>
24772 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
24773 of units as polynomial. Choose between WIDE and NARROW based
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 * tree-vect-stmts.c (simd_clone_subparts): New function.
24781 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
24783 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24784 Alan Hayward <alan.hayward@arm.com>
24785 David Sherwood <david.sherwood@arm.com>
24787 * tree-vect-stmts.c (vectorizable_call): Treat the number of
24788 vectors as polynomial. Use build_index_vector for
24789 IFN_GOMP_SIMD_LANE.
24791 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24792 Alan Hayward <alan.hayward@arm.com>
24793 David Sherwood <david.sherwood@arm.com>
24795 * tree-vect-stmts.c (get_load_store_type): Treat the number of
24796 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
24797 for variable-length vectors.
24798 (vectorizable_mask_load_store): Treat the number of units as
24799 polynomial, asserting that it is constant if the condition has
24800 already been enforced.
24801 (vectorizable_store, vectorizable_load): Likewise.
24803 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24804 Alan Hayward <alan.hayward@arm.com>
24805 David Sherwood <david.sherwood@arm.com>
24807 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
24808 of units as polynomial. Punt if we can't tell at compile time
24809 which vector contains the final result.
24811 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24812 Alan Hayward <alan.hayward@arm.com>
24813 David Sherwood <david.sherwood@arm.com>
24815 * tree-vect-loop.c (vectorizable_induction): Treat the number
24816 of units as polynomial. Punt on SLP inductions. Use an integer
24817 VEC_SERIES_EXPR for variable-length integer reductions. Use a
24818 cast of such a series for variable-length floating-point
24821 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24822 Alan Hayward <alan.hayward@arm.com>
24823 David Sherwood <david.sherwood@arm.com>
24825 * tree.h (build_index_vector): Declare.
24826 * tree.c (build_index_vector): New function.
24827 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
24828 of units as polynomial, forcibly converting it to a constant if
24829 vectorizable_reduction has already enforced the condition.
24830 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
24831 to create a {1,2,3,...} vector.
24832 (vectorizable_reduction): Treat the number of units as polynomial.
24833 Choose vectype_in based on the largest scalar element size rather
24834 than the smallest number of units. Enforce the restrictions
24837 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24838 Alan Hayward <alan.hayward@arm.com>
24839 David Sherwood <david.sherwood@arm.com>
24841 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
24842 number of units as polynomial.
24844 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24845 Alan Hayward <alan.hayward@arm.com>
24846 David Sherwood <david.sherwood@arm.com>
24848 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
24849 * target.def (autovectorize_vector_sizes): Return the vector sizes
24850 by pointer, using vector_sizes rather than a bitmask.
24851 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
24852 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
24853 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
24855 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
24856 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
24857 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
24858 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
24859 * omp-general.c (omp_max_vf): Likewise.
24860 * omp-low.c (omp_clause_aligned_alignment): Likewise.
24861 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
24862 * tree-vect-loop.c (vect_analyze_loop): Likewise.
24863 * tree-vect-slp.c (vect_slp_bb): Likewise.
24864 * doc/tm.texi: Regenerate.
24865 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
24867 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
24868 the vector size as a poly_uint64 rather than an unsigned int.
24869 (current_vector_size): Change from an unsigned int to a poly_uint64.
24870 (get_vectype_for_scalar_type): Update accordingly.
24871 * tree.h (build_truth_vector_type): Take the size and number of
24872 units as a poly_uint64 rather than an unsigned int.
24873 (build_vector_type): Add a temporary overload that takes
24874 the number of units as a poly_uint64 rather than an unsigned int.
24875 * tree.c (make_vector_type): Likewise.
24876 (build_truth_vector_type): Take the number of units as a poly_uint64
24877 rather than an unsigned int.
24879 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24880 Alan Hayward <alan.hayward@arm.com>
24881 David Sherwood <david.sherwood@arm.com>
24883 * target.def (get_mask_mode): Take the number of units and length
24884 as poly_uint64s rather than unsigned ints.
24885 * targhooks.h (default_get_mask_mode): Update accordingly.
24886 * targhooks.c (default_get_mask_mode): Likewise.
24887 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
24888 * doc/tm.texi: Regenerate.
24890 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24891 Alan Hayward <alan.hayward@arm.com>
24892 David Sherwood <david.sherwood@arm.com>
24894 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
24895 * omp-general.c (omp_max_vf): Likewise.
24896 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
24897 (expand_omp_simd): Handle polynomial safelen.
24898 * omp-low.c (omplow_simd_context): Add a default constructor.
24899 (omplow_simd_context::max_vf): Change from int to poly_uint64.
24900 (lower_rec_simd_input_clauses): Update accordingly.
24901 (lower_rec_input_clauses): Likewise.
24903 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24904 Alan Hayward <alan.hayward@arm.com>
24905 David Sherwood <david.sherwood@arm.com>
24907 * tree-vectorizer.h (vect_nunits_for_cost): New function.
24908 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
24909 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
24910 (vect_analyze_slp_cost): Likewise.
24911 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
24912 (vect_model_load_cost): Likewise.
24914 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24915 Alan Hayward <alan.hayward@arm.com>
24916 David Sherwood <david.sherwood@arm.com>
24918 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
24919 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
24920 from an unsigned int * to a poly_uint64_pod *.
24921 (calculate_unrolling_factor): New function.
24922 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
24924 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24925 Alan Hayward <alan.hayward@arm.com>
24926 David Sherwood <david.sherwood@arm.com>
24928 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
24929 from an unsigned int to a poly_uint64.
24930 (_loop_vec_info::slp_unrolling_factor): Likewise.
24931 (_loop_vec_info::vectorization_factor): Change from an int
24933 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
24934 (vect_get_num_vectors): New function.
24935 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
24936 (vect_get_num_copies): Use vect_get_num_vectors.
24937 (vect_analyze_data_ref_dependences): Change max_vf from an int *
24938 to an unsigned int *.
24939 (vect_analyze_data_refs): Change min_vf from an int * to a
24941 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
24942 than an unsigned HOST_WIDE_INT.
24943 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
24944 (vect_analyze_data_ref_dependence): Change max_vf from an int *
24945 to an unsigned int *.
24946 (vect_analyze_data_ref_dependences): Likewise.
24947 (vect_compute_data_ref_alignment): Handle polynomial vf.
24948 (vect_enhance_data_refs_alignment): Likewise.
24949 (vect_prune_runtime_alias_test_list): Likewise.
24950 (vect_shift_permute_load_chain): Likewise.
24951 (vect_supportable_dr_alignment): Likewise.
24952 (dependence_distance_ge_vf): Take the vectorization factor as a
24953 poly_uint64 rather than an unsigned HOST_WIDE_INT.
24954 (vect_analyze_data_refs): Change min_vf from an int * to a
24956 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
24957 vfm1 as a poly_uint64 rather than an int. Make the same change
24958 for the returned bound_scalar.
24959 (vect_gen_vector_loop_niters): Handle polynomial vf.
24960 (vect_do_peeling): Likewise. Update call to
24961 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
24962 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
24964 * tree-vect-loop.c (vect_determine_vectorization_factor)
24965 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
24966 (vect_get_known_peeling_cost): Likewise.
24967 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
24968 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
24969 (vect_transform_loop): Likewise. Use the lowest possible VF when
24970 updating the upper bounds of the loop.
24971 (vect_min_worthwhile_factor): Make static. Return an unsigned int
24972 rather than an int.
24973 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
24974 polynomial unroll factors.
24975 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
24976 (vect_make_slp_decision): Likewise.
24977 (vect_supported_load_permutation_p): Likewise, and polynomial
24979 (vect_analyze_slp_cost): Handle polynomial vf.
24980 (vect_slp_analyze_node_operations): Likewise.
24981 (vect_slp_analyze_bb_1): Likewise.
24982 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
24983 than an unsigned HOST_WIDE_INT.
24984 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
24985 (vectorizable_load): Handle polynomial vf.
24986 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
24988 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
24990 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24991 Alan Hayward <alan.hayward@arm.com>
24992 David Sherwood <david.sherwood@arm.com>
24994 * match.pd: Handle bit operations involving three constants
24995 and try to fold one pair.
24997 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24999 * tree-vect-loop-manip.c: Include gimple-fold.h.
25000 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
25001 niters_maybe_zero parameters. Handle other cases besides a step of 1.
25002 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
25003 Add a path that uses a step of VF instead of 1, but disable it
25005 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
25006 and niters_no_overflow parameters. Update calls to
25007 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
25008 Create a new SSA name if the latter choses to use a ste other
25009 than zero, and return it via niters_vector_mult_vf_var.
25010 * tree-vect-loop.c (vect_transform_loop): Update calls to
25011 vect_do_peeling, vect_gen_vector_loop_niters and
25012 slpeel_make_loop_iterate_ntimes.
25013 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
25014 (vect_gen_vector_loop_niters): Update declarations after above changes.
25016 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
25018 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
25019 128-bit round to integer instructions.
25020 (ceil<mode>2): Likewise.
25021 (btrunc<mode>2): Likewise.
25022 (round<mode>2): Likewise.
25024 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25026 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
25027 unaligned VSX load/store on P8/P9.
25028 (expand_block_clear): Allow the use of unaligned VSX
25029 load/store on P8/P9.
25031 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25033 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
25035 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
25036 swap associated with both a load and a store.
25038 2018-01-02 Andrew Waterman <andrew@sifive.com>
25040 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
25041 * config/riscv/riscv.md (clear_cache): Use it.
25043 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
25045 * web.c: Remove out-of-date comment.
25047 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25049 * expr.c (fixup_args_size_notes): Check that any existing
25050 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
25051 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
25052 (emit_single_push_insn): ...here.
25054 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25056 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
25057 (const_vector_encoded_nelts): New function.
25058 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
25059 (const_vector_int_elt, const_vector_elt): Declare.
25060 * emit-rtl.c (const_vector_int_elt_1): New function.
25061 (const_vector_elt): Likewise.
25062 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
25063 of CONST_VECTOR_ELT.
25065 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25067 * expr.c: Include rtx-vector-builder.h.
25068 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
25069 directly on the tree encoding.
25070 (const_vector_from_tree): Likewise.
25071 * optabs.c: Include rtx-vector-builder.h.
25072 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
25073 sequence of "u" values.
25074 * vec-perm-indices.c: Include rtx-vector-builder.h.
25075 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
25076 directly on the vec_perm_indices encoding.
25078 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25080 * doc/rtl.texi (const_vector): Describe new encoding scheme.
25081 * Makefile.in (OBJS): Add rtx-vector-builder.o.
25082 * rtx-vector-builder.h: New file.
25083 * rtx-vector-builder.c: Likewise.
25084 * rtl.h (rtx_def::u2): Add a const_vector field.
25085 (CONST_VECTOR_NPATTERNS): New macro.
25086 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
25087 (CONST_VECTOR_DUPLICATE_P): Likewise.
25088 (CONST_VECTOR_STEPPED_P): Likewise.
25089 (CONST_VECTOR_ENCODED_ELT): Likewise.
25090 (const_vec_duplicate_p): Check for a duplicated vector encoding.
25091 (unwrap_const_vec_duplicate): Likewise.
25092 (const_vec_series_p): Check for a non-duplicated vector encoding.
25093 Say that the function only returns true for integer vectors.
25094 * emit-rtl.c: Include rtx-vector-builder.h.
25095 (gen_const_vec_duplicate_1): Delete.
25096 (gen_const_vector): Call gen_const_vec_duplicate instead of
25097 gen_const_vec_duplicate_1.
25098 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
25099 (gen_const_vec_duplicate): Use rtx_vector_builder.
25100 (gen_const_vec_series): Likewise.
25101 (gen_rtx_CONST_VECTOR): Likewise.
25102 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
25103 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
25104 Build a new vector rather than modifying a CONST_VECTOR in-place.
25105 (handle_special_swappables): Update call accordingly.
25106 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
25107 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
25108 Build a new vector rather than modifying a CONST_VECTOR in-place.
25109 (handle_special_swappables): Update call accordingly.
25111 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25113 * simplify-rtx.c (simplify_const_binary_operation): Use
25114 CONST_VECTOR_ELT instead of XVECEXP.
25116 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25118 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
25119 the selector elements to be different from the data elements
25120 if the selector is a VECTOR_CST.
25121 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
25122 ssizetype for the selector.
25124 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25126 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
25127 before testing each element individually.
25128 * tree-vect-generic.c (lower_vec_perm): Likewise.
25130 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25132 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
25133 * selftest-run-tests.c (selftest::run_tests): Call it.
25134 * vector-builder.h (vector_builder::operator ==): New function.
25135 (vector_builder::operator !=): Likewise.
25136 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
25137 (vec_perm_indices::all_from_input_p): New function.
25138 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
25139 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
25140 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
25141 instead of reading the VECTOR_CST directly. Detect whether both
25142 vector inputs are the same before constructing the vec_perm_indices,
25143 and update the number of inputs argument accordingly. Use the
25144 utility functions added above. Only construct sel2 if we need to.
25146 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25148 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
25149 the broadcast of the low byte.
25150 (expand_mult_highpart): Use an explicit encoding for the permutes.
25151 * optabs-query.c (can_mult_highpart_p): Likewise.
25152 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25153 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25154 (vectorizable_bswap): Likewise.
25155 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
25156 explicit encoding for the power-of-2 permutes.
25157 (vect_permute_store_chain): Likewise.
25158 (vect_grouped_load_supported): Likewise.
25159 (vect_permute_load_chain): Likewise.
25161 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25163 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
25164 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
25165 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
25166 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
25167 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
25168 (vect_gen_perm_mask_any): Likewise.
25170 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25172 * int-vector-builder.h: New file.
25173 * vec-perm-indices.h: Include int-vector-builder.h.
25174 (vec_perm_indices): Redefine as an int_vector_builder.
25175 (auto_vec_perm_indices): Delete.
25176 (vec_perm_builder): Redefine as a stand-alone class.
25177 (vec_perm_indices::vec_perm_indices): New function.
25178 (vec_perm_indices::clamp): Likewise.
25179 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
25180 (vec_perm_indices::new_vector): New function.
25181 (vec_perm_indices::new_expanded_vector): Update for new
25182 vec_perm_indices class.
25183 (vec_perm_indices::rotate_inputs): New function.
25184 (vec_perm_indices::all_in_range_p): Operate directly on the
25185 encoded form, without computing elided elements.
25186 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
25187 encoding. Update for new vec_perm_indices class.
25188 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
25189 the given vec_perm_builder.
25190 (expand_vec_perm_var): Update vec_perm_builder constructor.
25191 (expand_mult_highpart): Use vec_perm_builder instead of
25192 auto_vec_perm_indices.
25193 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
25194 vec_perm_indices instead of auto_vec_perm_indices. Use a single
25195 or double series encoding as appropriate.
25196 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
25197 vec_perm_indices instead of auto_vec_perm_indices.
25198 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
25199 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
25200 (vect_permute_store_chain): Likewise.
25201 (vect_grouped_load_supported): Likewise.
25202 (vect_permute_load_chain): Likewise.
25203 (vect_shift_permute_load_chain): Likewise.
25204 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
25205 (vect_transform_slp_perm_load): Likewise.
25206 (vect_schedule_slp_instance): Likewise.
25207 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25208 (vectorizable_mask_load_store): Likewise.
25209 (vectorizable_bswap): Likewise.
25210 (vectorizable_store): Likewise.
25211 (vectorizable_load): Likewise.
25212 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
25213 vec_perm_indices instead of auto_vec_perm_indices. Use
25214 tree_to_vec_perm_builder to read the vector from a tree.
25215 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
25216 vec_perm_builder instead of a vec_perm_indices.
25217 (have_whole_vector_shift): Use vec_perm_builder and
25218 vec_perm_indices instead of auto_vec_perm_indices. Leave the
25219 truncation to calc_vec_perm_mask_for_shift.
25220 (vect_create_epilog_for_reduction): Likewise.
25221 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
25222 from auto_vec_perm_indices to vec_perm_indices.
25223 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
25224 instead of changing individual elements.
25225 (aarch64_vectorize_vec_perm_const): Use new_vector to install
25226 the vector in d.perm.
25227 * config/arm/arm.c (expand_vec_perm_d::perm): Change
25228 from auto_vec_perm_indices to vec_perm_indices.
25229 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
25230 instead of changing individual elements.
25231 (arm_vectorize_vec_perm_const): Use new_vector to install
25232 the vector in d.perm.
25233 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
25234 Update vec_perm_builder constructor.
25235 (rs6000_expand_interleave): Likewise.
25236 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
25237 (rs6000_expand_interleave): Likewise.
25239 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25241 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
25242 to qimode could truncate the indices.
25243 * optabs.c (expand_vec_perm_var): Likewise.
25245 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25247 * Makefile.in (OBJS): Add vec-perm-indices.o.
25248 * vec-perm-indices.h: New file.
25249 * vec-perm-indices.c: Likewise.
25250 * target.h (vec_perm_indices): Replace with a forward class
25252 (auto_vec_perm_indices): Move to vec-perm-indices.h.
25253 * optabs.h: Include vec-perm-indices.h.
25254 (expand_vec_perm): Delete.
25255 (selector_fits_mode_p, expand_vec_perm_var): Declare.
25256 (expand_vec_perm_const): Declare.
25257 * target.def (vec_perm_const_ok): Replace with...
25258 (vec_perm_const): ...this new hook.
25259 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
25260 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
25261 * doc/tm.texi: Regenerate.
25262 * optabs.def (vec_perm_const): Delete.
25263 * doc/md.texi (vec_perm_const): Likewise.
25264 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
25265 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
25266 expand_vec_perm for constant permutation vectors. Assert that
25267 the mode of variable permutation vectors is the integer equivalent
25268 of the mode that is being permuted.
25269 * optabs-query.h (selector_fits_mode_p): Declare.
25270 * optabs-query.c: Include vec-perm-indices.h.
25271 (selector_fits_mode_p): New function.
25272 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
25273 is defined, instead of checking whether the vec_perm_const_optab
25274 exists. Use targetm.vectorize.vec_perm_const instead of
25275 targetm.vectorize.vec_perm_const_ok. Check whether the indices
25276 fit in the vector mode before using a variable permute.
25277 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
25278 vec_perm_indices instead of an rtx.
25279 (expand_vec_perm): Replace with...
25280 (expand_vec_perm_const): ...this new function. Take the selector
25281 as a vec_perm_indices rather than an rtx. Also take the mode of
25282 the selector. Update call to shift_amt_for_vec_perm_mask.
25283 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
25284 Use vec_perm_indices::new_expanded_vector to expand the original
25285 selector into bytes. Check whether the indices fit in the vector
25286 mode before using a variable permute.
25287 (expand_vec_perm_var): Make global.
25288 (expand_mult_highpart): Use expand_vec_perm_const.
25289 * fold-const.c: Includes vec-perm-indices.h.
25290 * tree-ssa-forwprop.c: Likewise.
25291 * tree-vect-data-refs.c: Likewise.
25292 * tree-vect-generic.c: Likewise.
25293 * tree-vect-loop.c: Likewise.
25294 * tree-vect-slp.c: Likewise.
25295 * tree-vect-stmts.c: Likewise.
25296 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
25298 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
25299 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
25300 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
25301 (aarch64_vectorize_vec_perm_const): ...this new function.
25302 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25303 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25304 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
25305 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
25306 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25307 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25308 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
25310 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
25311 check for NEON modes.
25312 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
25313 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
25314 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
25315 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
25317 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
25318 the old VEC_PERM_CONST conditions.
25319 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
25320 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
25321 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
25322 (ia64_vectorize_vec_perm_const_ok): Merge into...
25323 (ia64_vectorize_vec_perm_const): ...this new function.
25324 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
25325 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
25326 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
25327 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
25328 * config/mips/mips.c (mips_expand_vec_perm_const)
25329 (mips_vectorize_vec_perm_const_ok): Merge into...
25330 (mips_vectorize_vec_perm_const): ...this new function.
25331 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
25332 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
25333 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
25334 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
25335 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
25336 (rs6000_expand_vec_perm_const): Delete.
25337 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
25339 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25340 (altivec_expand_vec_perm_const_le): Take each operand individually.
25341 Operate on constant selectors rather than rtxes.
25342 (altivec_expand_vec_perm_const): Likewise. Update call to
25343 altivec_expand_vec_perm_const_le.
25344 (rs6000_expand_vec_perm_const): Delete.
25345 (rs6000_vectorize_vec_perm_const_ok): Delete.
25346 (rs6000_vectorize_vec_perm_const): New function.
25347 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
25348 an element count and rtx array.
25349 (rs6000_expand_extract_even): Update call accordingly.
25350 (rs6000_expand_interleave): Likewise.
25351 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
25352 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
25353 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
25354 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
25355 (rs6000_expand_vec_perm_const): Delete.
25356 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25357 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25358 (altivec_expand_vec_perm_const_le): Take each operand individually.
25359 Operate on constant selectors rather than rtxes.
25360 (altivec_expand_vec_perm_const): Likewise. Update call to
25361 altivec_expand_vec_perm_const_le.
25362 (rs6000_expand_vec_perm_const): Delete.
25363 (rs6000_vectorize_vec_perm_const_ok): Delete.
25364 (rs6000_vectorize_vec_perm_const): New function. Remove stray
25365 reference to the SPE evmerge intructions.
25366 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
25367 an element count and rtx array.
25368 (rs6000_expand_extract_even): Update call accordingly.
25369 (rs6000_expand_interleave): Likewise.
25370 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
25371 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
25373 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25375 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25377 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
25378 vector mode and that that mode matches the mode of the data
25380 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
25381 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
25382 directly using expand_vec_perm_1 when forcing selectors into
25384 (expand_vec_perm_var): New function, split out from expand_vec_perm.
25386 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25388 * optabs-query.h (can_vec_perm_p): Delete.
25389 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
25390 * optabs-query.c (can_vec_perm_p): Split into...
25391 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
25392 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
25393 particular selector is valid.
25394 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
25395 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
25396 (vect_grouped_load_supported): Likewise.
25397 (vect_shift_permute_load_chain): Likewise.
25398 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
25399 (vect_transform_slp_perm_load): Likewise.
25400 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25401 (vectorizable_bswap): Likewise.
25402 (vect_gen_perm_mask_checked): Likewise.
25403 * fold-const.c (fold_ternary_loc): Likewise. Don't take
25404 implementations of variable permutation vectors into account
25405 when deciding which selector to use.
25406 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
25407 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
25408 with a false third argument.
25409 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
25410 to test whether the constant selector is valid and can_vec_perm_var_p
25411 to test whether a variable selector is valid.
25413 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25415 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
25416 * optabs-query.c (can_vec_perm_p): Likewise.
25417 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
25418 instead of vec_perm_indices.
25419 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
25420 (vect_gen_perm_mask_checked): Likewise,
25421 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
25422 (vect_gen_perm_mask_checked): Likewise,
25424 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25426 * optabs-query.h (qimode_for_vec_perm): Declare.
25427 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
25428 (qimode_for_vec_perm): ...this new function.
25429 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
25431 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25433 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
25434 does not have a conditional at the top.
25436 2018-01-02 Richard Biener <rguenther@suse.de>
25438 * ipa-inline.c (big_speedup_p): Fix expression.
25440 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
25443 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
25446 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
25450 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
25451 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
25452 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
25453 cond_taken_branch_cost 3->4.
25455 2018-01-01 Jakub Jelinek <jakub@redhat.com>
25457 PR tree-optimization/83581
25458 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
25459 TODO_cleanup_cfg if any changes have been made.
25461 PR middle-end/83608
25462 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
25463 convert_modes if target mode has the right side, but different mode
25466 PR middle-end/83609
25467 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
25468 last argument when extracting from CONCAT. If either from_real or
25469 from_imag is NULL, use expansion through memory. If result is not
25470 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
25471 the parts directly to inner mode, if even that fails, use expansion
25474 PR middle-end/83623
25475 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
25476 check for bswap in mode rather than HImode and use that in expand_unop
25479 Copyright (C) 2018 Free Software Foundation, Inc.
25481 Copying and distribution of this file, with or without modification,
25482 are permitted in any medium without royalty provided the copyright
25483 notice and this notice are preserved.